<%BANNER%>

Relationships between Mass and Flux in Dense Nonaqueous Phase Liquid Source Zones

Permanent Link: http://ufdc.ufl.edu/UFE0011060/00001

Material Information

Title: Relationships between Mass and Flux in Dense Nonaqueous Phase Liquid Source Zones
Physical Description: Mixed Material
Copyright Date: 2008

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0011060:00001

Permanent Link: http://ufdc.ufl.edu/UFE0011060/00001

Material Information

Title: Relationships between Mass and Flux in Dense Nonaqueous Phase Liquid Source Zones
Physical Description: Mixed Material
Copyright Date: 2008

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0011060:00001


This item has the following downloads:


Full Text
xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID E20101129_AAAABW INGEST_TIME 2010-11-30T04:05:38Z PACKAGE UFE0011060_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES
FILE SIZE 3272 DFID F20101129_AABEWG ORIGIN DEPOSITOR PATH fure_a_Page_169thm.jpg GLOBAL false PRESERVATION BIT MESSAGE_DIGEST ALGORITHM MD5
d476c849ff3781a5461ce0a2401bdb5f
SHA-1
4576b851f0521d6ce6840cab9e39957343a3ad4b
13731 F20101129_AABEVS fure_a_Page_161.QC.jpg
247c63b1acbf817ac62f08493d592848
ff2b83f9cec1a0801d883da357e3bbf5b48dfdf7
25271604 F20101129_AABDTE fure_a_Page_178.tif
9a7d66fbbf6069a0b05556be33df9ce0
5c54fc04f7f5638910686625cdd1339583c24f11
55347 F20101129_AABDSQ fure_a_Page_160.jpg
9fd5ba9432177b18382cc1a712852926
b983755215207f62b63a8fef8bc2e7d2c5ba9c1b
5659 F20101129_AABEWH fure_a_Page_170thm.jpg
5659a38b42af0efc4680f0a8026ca9c4
6438b1bd4dd22b74ad83897c7abe1f8c93365b16
4829 F20101129_AABEVT fure_a_Page_161thm.jpg
5583e9c563820324453362e8f453cb5e
83bfb5aef0ed5ac914474ec571439c560a202b65
111716 F20101129_AABDTF fure_a_Page_082.jp2
ae4d464e4c4f1e67bf99363da26d40e6
536a5e77ab5a63e10ee60bf72f58958f1b1a4ce6
72747 F20101129_AABDSR fure_a_Page_077.jpg
b4a8bf39f2366cebad9aec35889b1f57
79b3d92ad7193e340a3319c88c991cd8e243796b
17044 F20101129_AABEWI fure_a_Page_171.QC.jpg
09504642baa0de4ad6dd68294715b768
6208197c399a630156c27551805fe0440e35a37c
11524 F20101129_AABEVU fure_a_Page_162.QC.jpg
515c464c9faac99cdcbe8633ac09d43e
c5c0604a136a29eccd54689ff4d200dc5bf23abe
72494 F20101129_AABDTG fure_a_Page_092.jpg
6f3363d2bff9cb0f1dd04197190e08cc
0b239bfd311e4518a23f7af4448bff991f4b78ac
2889 F20101129_AABDSS fure_a_Page_129thm.jpg
38a8893013e56c8db02fdb92d7e63c6b
004df72c4eade6f8dc5993a0b1e7268de0ee4aa5
5157 F20101129_AABEWJ fure_a_Page_171thm.jpg
51e610314df340b61a4a43123dd4bfe5
0fedcd4716825795376357852a7d7605b2f54b4a
4089 F20101129_AABEVV fure_a_Page_162thm.jpg
71bace4ed322e60bd7267ed6c80ce944
20a2e227d4afa250edddbce8a79b007aec9e53ad
56940 F20101129_AABDTH fure_a_Page_041.jpg
6c724806cc510152ff881e04983c0f51
4ce1bd45b5c1ec45934334d6ffc602be358cb959
14782 F20101129_AABDST fure_a_Page_140.QC.jpg
b7bce585c83a092d0fcec3cdba722925
90d7fbe2aa3ec7cfa2c74198bb10d8a93664c6e7
5561 F20101129_AABEWK fure_a_Page_172thm.jpg
6bd17222c8a38de2ada677c4da012539
425022140a739a0be5aacd1bf5f37ae1e56aae71
5222 F20101129_AABEVW fure_a_Page_163thm.jpg
e81e095f0ee4ba7d02837a34a1027664
cc60f89e1789f3c7f944b04d3eca691cfabc44b4
20970 F20101129_AABDTI fure_a_Page_097.QC.jpg
b69f352a3033fa68c06dbb5c8ae05d76
73eda10a3417cdf47c3acb7a7c650b729674c6d0
21139 F20101129_AABDSU fure_a_Page_039.QC.jpg
5b3267e75724781646cedbe5c90f3942
6fcdf7902ef8e7e23f06d43722d216692f2bfa4e
16706 F20101129_AABEWL fure_a_Page_173.QC.jpg
c77caf87eaa72ce691553852bdb64a90
817c48b1177f99fa49aa43ceca4b6a815fc3a65b
14130 F20101129_AABEVX fure_a_Page_164.QC.jpg
4838033d0e033714168b9982c487b2be
f437415fd4993b1de9c3a4fce63144487434a33a
616681 F20101129_AABDTJ fure_a_Page_146.jp2
2fb2836c8f84ad1e8033ecfbf2618cf2
0ada33c1311c8f6c990ab8490dcf211cf511cff3
1053954 F20101129_AABDSV fure_a_Page_045.tif
2928273d106ee4eaaa1f217c507e5b0f
05f1f4eacea7a396a9a34854911f41e3ac257c9f
6575 F20101129_AABEXA fure_a_Page_183thm.jpg
54cef03b9b71b0fbacc98eaf08793091
4eea8f27c35f13d4b2fa4e9ad6f97db7cd318858
5033 F20101129_AABEWM fure_a_Page_173thm.jpg
688d27b896b1b11f56a95df0f132e960
f9d61071fe059f1c7777d67ddc2535906e5b7d9c
4388 F20101129_AABEVY fure_a_Page_164thm.jpg
f6ecf0d94ee7837b07e2e722cf7a129a
1c45e1faee19ac1a94f73e6830dd7e05e4e02988
24790 F20101129_AABDTK fure_a_Page_081.QC.jpg
0cde19c904755fde6e6f62e934c7cef5
a98283a0e0d2d0581c91bb0ee6aad6721e226c67
F20101129_AABDSW fure_a_Page_065.tif
2bc61d1cf683e7ecbee0fc4f68c983bb
08aca238e80aaa8e891b1be874fea5969f34cad6
19578 F20101129_AABEXB fure_a_Page_184.QC.jpg
9fcfcbc8bb727fd2e9cde6f074633eb9
206491c54523fec6aa6deef9cb458b54958316ce
13145 F20101129_AABEWN fure_a_Page_175.QC.jpg
cbae1be245af4682baed4f4935145fcb
706b6fa067ff1fe0699786d4b02fa36edcd277ca
10147 F20101129_AABEVZ fure_a_Page_165.QC.jpg
4c9603f1342b65e2b122064a867fd35d
7ddccede212d5113fd38af4b974465bfb450b6da
110560 F20101129_AABDTL fure_a_Page_085.jp2
5bca3c4af7202d7b33338065d36eb441
08d21c87e7d7d181a8b43e9f54ff5bb1fb350d70
103587 F20101129_AABDSX fure_a_Page_127.jp2
4be6a07f276fbf775dad1dce6d8752dc
148512e7235aae4db46b3111057f773ae0fe288f
6045 F20101129_AABEXC fure_a_Page_185thm.jpg
cb5f4e8cedb680e42b67254c33b9a12c
1a1c231e6dc0640c788b92c5bf8ec3b58aa067c1
23144 F20101129_AABEWO fure_a_Page_176.QC.jpg
2939320bf3cc5ef67703f8f1a52c3102
a82868618f86b7df67a41537ae7de7ec3025023e
1051915 F20101129_AABDTM fure_a_Page_004.jp2
e4b04ae295883c6a14baad84e56e6753
69589c2abc1ca8c09849358ba2184882e99822b4
64618 F20101129_AABDSY fure_a_Page_110.jpg
e12daf97ebd58ef3be3b818b93bb2917
e2190d2fa50cf30359ecbc0ee92431af4b74ae30
F20101129_AABDUA fure_a_Page_185.tif
94f1e8559f92e2d3cd93a42879679fbf
bd972023d30796b01698caa91f4a9d35cf9c281a
6738 F20101129_AABEXD fure_a_Page_186thm.jpg
367732a1cb5e763da8389b4a56d1f450
752a136204a8883e541f70aac3340e56befab688
12290 F20101129_AABEWP fure_a_Page_177.QC.jpg
a7373a6320371b379d32f5037d6359ad
9b5942c912ebad5ad547d792ac04b805c26286ee
100487 F20101129_AABDTN fure_a_Page_155.jp2
4ba6f3112b0e3e6ea75584ea575aaf3a
da8f8f18f57f12a9f05e8e7f6aaf7e6aeb5e246f
14068 F20101129_AABDSZ fure_a_Page_130.QC.jpg
7b9100c28ff29bff3599320730bddc36
457ce3374293981f918d9a29085bda286646f171
F20101129_AABDUB fure_a_Page_163.tif
289ef7e40aeb590c387e61b21d011174
98cc595aa29352c03e5b0c275416f7f145f471d8
23403 F20101129_AABEXE fure_a_Page_187.QC.jpg
7921f63d7a2406e85f4b2618bc6da893
0b465857ed598206b3547bbb7d4d39c79a018e0e
3810 F20101129_AABEWQ fure_a_Page_177thm.jpg
8af673742fa17fecd2964b63ae0a4ea9
6d56b218ee7eb87ca1626c08a4f36e70690f8775
5917 F20101129_AABDTO fure_a_Page_011thm.jpg
c730335fe2adfa8e593c46edcd4862db
2bfc96a2ee4d2dac244004d7c1b5ad025cd8db41
5665 F20101129_AABDUC fure_a_Page_167thm.jpg
5e8b3b5c3c840ce6b28eec31ee61050e
a2986bc51381c5a784d605dc926acba145490641
6476 F20101129_AABEXF fure_a_Page_187thm.jpg
7df8406d669935e04dd42a5b90ecc37f
ded100b388e5d84e4c97e183d907ed08fd141a28
19970 F20101129_AABEWR fure_a_Page_178.QC.jpg
e46bf4a683064337d423afb47e51e13e
fe83a9d628bf92baae614a0eb681f475ebad9e37
20499 F20101129_AABDTP fure_a_Page_031.QC.jpg
f10d79e6848d6bdfb7f908be7a3d1bd2
d2b24a34a1804b36eb2346f565809844039be620
209372 F20101129_AABDUD UFE0011060_00001.xml FULL
e6df8ccd43239bb40428c1412811f8b3
081576e29c4bbfa41e2cd72e23d45c8dee85b7b0
4771 F20101129_AABEXG fure_a_Page_188thm.jpg
89c3b264b87c8570ae5da0d86d54f47d
f24ea0668807d77a2f830bc7e371ecca648dc46b
5725 F20101129_AABEWS fure_a_Page_178thm.jpg
116a3e42fba9fa4a304a01f3dd4e57e9
1c3ceeb2cbbf5dff831aa46130415c704babfa07
4570 F20101129_AABDTQ fure_a_Page_056thm.jpg
bdfab091a9f20dcd935547b6ad56efea
91e3d3f4a6307b4e54b6724c54a3b575bca8ea84
5698 F20101129_AABEXH fure_a_Page_189thm.jpg
5a8e9abb37ff454258a535316e5f9f12
c6f2229b39313902dd9f10d760259ba9b323c4be
20086 F20101129_AABEWT fure_a_Page_179.QC.jpg
914382f1670cf4ee72d63603f9cd738e
117bd43822fa0217fb610396bc615e32a4a99f8d
29644 F20101129_AABDTR fure_a_Page_165.jpg
5cb1e380af75a4d55c22f47ff3fc7031
8ee6501ac0ebc022e77784917278e43b01521a88
1051937 F20101129_AABEAA fure_a_Page_008.jp2
12cc3d56dbde1647412fafc89fbc7bca
2b6d40892844791f855ef1f7963910bec5c74332
23544 F20101129_AABEXI fure_a_Page_190.QC.jpg
5845dc4a3650ecca8899c15f9709efec
b4fd42d3c26473a82b8eaa68e6fd09db600f3cae
5954 F20101129_AABEWU fure_a_Page_179thm.jpg
f541c10385eb1f24af6a0f52dc62185c
7282452483714ddfb542589c33d1d862253d1ba8
18148 F20101129_AABDTS fure_a_Page_172.QC.jpg
68a9f99b9ac97f7b7b0c856936f80d22
101aaf5a5530efd2f93bb1ae31a60a443d66c063
1051969 F20101129_AABEAB fure_a_Page_009.jp2
3acca25ac2bff8da5b7aaf2cfe8b6285
cda3d78bb4d811d6620e486d5c06f1b69aef7255
10081 F20101129_AABDUG fure_a_Page_002.jpg
37f68097fa5b0e9cbd9959ace3dee0e7
781cf967f9e5f21c8f2ad369dec7b334cda75eff
6486 F20101129_AABEXJ fure_a_Page_190thm.jpg
ade906ac25448b39f7b349ac40e44c9c
113a2cdd1028c248042f7b329b79b1aeae0aeea1
24655 F20101129_AABEWV fure_a_Page_180.QC.jpg
7ff42aa27ee644b8983ed09b99d138ce
9fea9f0513b95d688fa0f0964f2500eac1ac451d
5907 F20101129_AABDTT fure_a_Page_027thm.jpg
cb943323475b7a2b0f82f5475569eece
cf3dcebd5d5ccefaccddb4f4c6132b9c5280f490
1051967 F20101129_AABEAC fure_a_Page_010.jp2
220c739798d7668dd33f310929e8b53c
7cba7ef5517ef32cb9e933a45f07893405bfc3d1
31274 F20101129_AABDUH fure_a_Page_003.jpg
0576ed317790f4787ba9be8bb1c069ce
b5400440a8f08d53ecaa858f12f69fcb19c3d88c
6774 F20101129_AABEXK fure_a_Page_191thm.jpg
c97eb67a9fe308d4d9861016421d7d6e
420374f201b623e85d0c1a4caf61f78f32800cd0
6753 F20101129_AABEWW fure_a_Page_180thm.jpg
67a785cfab932b556a81be532846f024
2f912849b6744ee33b65aed6b3ebc10efeeb6e4c
792341 F20101129_AABDTU fure_a_Page_130.jp2
0e1e035afa279dd23170c4a71d160c59
c1a534289946e1f05a413f12812588e385a8f3b4
1051980 F20101129_AABEAD fure_a_Page_011.jp2
3f01b60a3924817d874bb1ee8c037a7c
2b4f2ba29a37b127c449e61958161c3a53ab76da
81295 F20101129_AABDUI fure_a_Page_004.jpg
5081f7616407ac54914288392012a2e3
6650e7602c0ef2d1931b9d99280bcc6116baf962
6831 F20101129_AABEXL fure_a_Page_192thm.jpg
ef1c8cae69d97bcda40133dd6e992740
b74f8cd2693ba4387b02f4ec8afaf99de56e223f
22341 F20101129_AABEWX fure_a_Page_181.QC.jpg
c9ab0e94e9af8747888f57d04aa60031
fd0caaa6308328b62b9d520b446d274d153709ce
24070 F20101129_AABDTV fure_a_Page_001.jpg
7bc3d0f43b5c3aab73ac9ed8e18a65ca
f02f55bc0b74f3bcb6f9cfcbcd3681bdd933a646
1051982 F20101129_AABEAE fure_a_Page_012.jp2
58c1e8927f1e190b17e7aa86389c4b8d
1baedc7ec42d40a2f6c3c85cb1d3bc9e5fca24e4
105457 F20101129_AABDUJ fure_a_Page_005.jpg
6f735faef82bf4b60a9b4bb308e1f55d
6fe28de929fe751334f9f2a9e5b8e43705cc930a
23268 F20101129_AABEXM fure_a_Page_193.QC.jpg
dc77279b1310dc233d0bb1175bb94ad9
61d7ccb39f5749c37c7d752f4b56d1c42b35c62a
6432 F20101129_AABEWY fure_a_Page_181thm.jpg
8c8834ef9b046a94bb63d6895fda4bb9
d6c6f4b8072fe5bf3c7ee13cf4b8f7db84c905d3
6236 F20101129_AABDTW fure_a_Page_120thm.jpg
320ac38a2d48ad2db64552a7bf8b7c29
44a8d2ed8e077771ae5e67929203c90d2c2def7f
675335 F20101129_AABEAF fure_a_Page_013.jp2
04438e5ae0e533e60f1692c22fcb680e
41ee2ce91dd40908689902879488006f77d0fcdc
73182 F20101129_AABDUK fure_a_Page_006.jpg
afc593c3850b5959c7dce5835eb5787f
6e53eda861f674b28f3857b9fc1bc66e85885d5c
6341 F20101129_AABEXN fure_a_Page_193thm.jpg
11a56829cce0af6599530645506034b0
7031ca8c90ecc1905e63be6adcb4b93dbac0a24e
5699 F20101129_AABEWZ fure_a_Page_182thm.jpg
7db898266f4adba03b9580687ad1e3fb
0063a507629bb998411cd42b532f2385f1e9cbd1
66715 F20101129_AABDTX fure_a_Page_040.jp2
db648c793e5c0fb0cbcef15b542092b9
f9d5f484a3252b5bf09debde556c53772166ad29
70209 F20101129_AABEAG fure_a_Page_014.jp2
28655a13208cec222673a889545b9a82
15df41b74a569a7860cfbd65d0b5042ddd549fae
51321 F20101129_AABDUL fure_a_Page_007.jpg
40265b4b5619430e7a24edd2f41799ca
e5ef108b0431794de068e6571acf919278b1ef7d
23061 F20101129_AABEXO fure_a_Page_194.QC.jpg
8c01cf816efc354be8188387fdf7bfbe
98fcb8eeb543a7a08cd48a59c57fe9d668b04795
71544 F20101129_AABDTY fure_a_Page_083.jpg
4c5c8870a7eb323eee290c33185549ff
364bc458f9b0dd3907e4da7b767a828402963c3f
89113 F20101129_AABEAH fure_a_Page_015.jp2
3d414d14c9b5227d26de94dad09910e3
8e74846e815fbadb3bd2e2c05796a05d6cf00851
66853 F20101129_AABDVA fure_a_Page_033.jpg
41fcd1de1bcf1c844c636639987223b2
b1a8f554e4ee8f3578a377a17ce75b0291295bb7
97679 F20101129_AABDUM fure_a_Page_009.jpg
9e0bd9e0ecdcd4e68145b7508d00a28a
96bb734128fb38a61e945b03b45aac28f305f114
6396 F20101129_AABEXP fure_a_Page_194thm.jpg
558397a951642fddc4dcb6d30a800321
3d87bbcc62a8d14612b1ca7a62d71ab6e48f1757
19225 F20101129_AABDTZ fure_a_Page_038.QC.jpg
3a5940e90535d81ac5ff6e0aaa30a257
fcfcdbacf3d318484a16311766242b945c66ec35
39494 F20101129_AABDVB fure_a_Page_034.jpg
afdd314129d4a10fdb6e1ab85c888315
8538cde30462b414ae8be21f178a5729c2cb1095
93292 F20101129_AABDUN fure_a_Page_010.jpg
cebb108b89528a26faf25550a8d11982
323f292ba5bb08f03e12a3ed33375ba021126942
25116 F20101129_AABEXQ fure_a_Page_195.QC.jpg
31dcbeab6444a51f98229a8e84762d62
84110ed3843d41e9ddc11bc3ff56d806b0251a6c
109090 F20101129_AABEAI fure_a_Page_016.jp2
bc527ccf8e2a7b5c2ea437b74456cded
9b1bfe99227a5d9c562498f3025fa42c47e71b01
67836 F20101129_AABDVC fure_a_Page_036.jpg
2273dffc8761a45790066a7d29e859f5
731dbd295ae3f6b819b7f7d813624443db8effdf
29556 F20101129_AABDUO fure_a_Page_013.jpg
749ed555735118300ea3b63ca4438f0f
690a4f29830265c21cf3dedc4832919573eb29fb
6868 F20101129_AABEXR fure_a_Page_195thm.jpg
67695c38db07c0b134fefb73911133b6
16569dcb025ec5429aeb0165cd1307b78e329965
124666 F20101129_AABEAJ fure_a_Page_017.jp2
2a1095d8c9924cb7c0661fe8ce81c7d2
abdbaa69eb56bc6aeb66a2e5ee11be94cdda7f6c
49487 F20101129_AABDVD fure_a_Page_037.jpg
fe36770dbd13eebe0357d9d9e7af7989
9904c35d7fafa64a93075dc9766d1da222095eef
50009 F20101129_AABDUP fure_a_Page_014.jpg
c876eff32b73a76f873d348ad6ce2e24
5e8b5f241485e7642f5f0ac6c01106f09a5df956
17276 F20101129_AABEXS fure_a_Page_196.QC.jpg
fa5b68eb68741e667130afbab788d599
fba587499a161e794f4c5f7b34af96819a647c2f
101530 F20101129_AABEAK fure_a_Page_018.jp2
f2e0389b2736770481754af124c03653
78110d819d914f50505879519208803587f25cda
60274 F20101129_AABDVE fure_a_Page_038.jpg
60fb3829a0d57f6f8197fb3293a25dde
01417c7541179bb16f7ffbd07563b9bfb1687d1a
70491 F20101129_AABDUQ fure_a_Page_016.jpg
6390ea7310207efa845e89b790db1c8a
1193a2254a19305226f05b3f71f0257d43b71bb4
4861 F20101129_AABEXT fure_a_Page_196thm.jpg
05ec56647e3ac3694a7181637e369436
a5d7a9641180659dcc746cb244c0b749f1380981
75439 F20101129_AABEAL fure_a_Page_020.jp2
b01fb220ac91219f5e4d444ac726419e
0d9a236323941fc0a7559b6509b13a0e7ae16a8f
64386 F20101129_AABDVF fure_a_Page_039.jpg
c36acb83663608c94ca32e10b3a88f37
4d25aaa70b198368e4c9fdc0fe967979c8a265a6
82109 F20101129_AABDUR fure_a_Page_017.jpg
55c8104de0a9e7ac83a0b9ec4110a376
c6e1420bcfe037c7c4421702c14c37458aeaba06
921066 F20101129_AABEBA fure_a_Page_038.jp2
38422e72b239578befac5b08ef3780d8
cf708014098586c696f600f830eb64fb62413495
12667 F20101129_AABEXU fure_a_Page_197.QC.jpg
b51d776c3eef7646e113cb9ec40b44aa
8ed6a10b5b19c867aca915502a6a7f5217935d0f
835290 F20101129_AABEAM fure_a_Page_021.jp2
fe50ab433fe02229e73fff9bfa9d7298
4e19c7d8c61db0951f5c73a59c0869d55c880f3b
47974 F20101129_AABDVG fure_a_Page_040.jpg
b7d6a05932dae5ff0251e2d9ca379a5c
f296fb4fa9f228a39092cefe363d0e777425f7fd
66130 F20101129_AABDUS fure_a_Page_018.jpg
90616906bd66deb66fbcedb89ca26d0a
eff3efae0ad8d719c21f50be9681041bfa9ab274
96006 F20101129_AABEBB fure_a_Page_039.jp2
c23016159adf73a2fcbbabe5c55e0f00
5f222a68e1d926eda17eec98bee4c0b6463bbb83
143243 F20101129_AABEXV UFE0011060_00001.mets
4584eb18a45baccaffd8fcfd2696b29d
706319722b59c203db646d8ffd3b4fc5ec885920
107373 F20101129_AABEAN fure_a_Page_022.jp2
b81c97d345cb7ac431e184bd9b4d2602
6868307aec0f170f03b1672c2e35ac286d751b0e
42497 F20101129_AABDVH fure_a_Page_042.jpg
aced941864280d9d4a2329676fefbb6e
0c649c932425506e8892def86927d4dfc490059b
57124 F20101129_AABDUT fure_a_Page_019.jpg
812db428c9c7ccb25f2428b0eb1be2dd
66bb7f1f1147740bfffa3d0443d4154853780a15
84305 F20101129_AABEBC fure_a_Page_041.jp2
38f39f88d0e3bba0aca7d713d3c4903a
47330e2e7f033df1f110e1eac3a851d9a6284678
100271 F20101129_AABEAO fure_a_Page_023.jp2
4c870e1e0c26e668a161d67cb2b01aa2
c0857d08f77219d813789ee1510aa5d459231de6
63688 F20101129_AABDVI fure_a_Page_043.jpg
85a8ee4bfe7b09f48bccf5de998b297a
9706d10aa06599a2bf413c37445465e00d59b791
50972 F20101129_AABDUU fure_a_Page_020.jpg
de5d0f6014194e123c25a661ba41d287
9d46925fb7dd0dc05274bc192f7f5c23b1e75f87
696209 F20101129_AABEBD fure_a_Page_042.jp2
0884c1da40b121a7755ac609a0e809eb
b855e1fca7e5ab925863c61c3973985fc570ddc8
998154 F20101129_AABEAP fure_a_Page_024.jp2
97861257596c6828801aae1c2d3fe6d4
cfd744e80ec0fe8244bc8f90626f187451aed56d
74394 F20101129_AABDVJ fure_a_Page_044.jpg
6c696eaf0465070d5b63eed98219afb6
aa82da864b27169bda52984081b7fa23bd1e462a
62147 F20101129_AABDUV fure_a_Page_021.jpg
2273a01bd1faf8d89dbc17109da16ab1
0dde804caadb88bf5e5c68973fa2cd224aa956c1
93997 F20101129_AABEBE fure_a_Page_043.jp2
26c64d49c371f51d0c761f98f3cdfde4
6998609c5d562b8af74d057c53910c74099d2b82
70866 F20101129_AABEAQ fure_a_Page_025.jp2
4b8026c3843c11f0f9b3eb1e41be7051
4164b7e1291b96f7bd07c204dbc21b7adca0ff17
53372 F20101129_AABDVK fure_a_Page_045.jpg
2ee2d25991529f021d4ddf55574276d1
d7906ad8c59888368d499b8a4799e6319c2bdc26
63214 F20101129_AABDUW fure_a_Page_027.jpg
c8c122f8e2c445c1208be2a1d98a806c
8e8f8a6c9855ed86bce8884710f233461e9174cb
112327 F20101129_AABEBF fure_a_Page_044.jp2
ddde43d28db3fb992be415548321bfc8
e5daa85ab291af65bdff0b1ca6cd3148d1a4ebbd
96283 F20101129_AABEAR fure_a_Page_027.jp2
93919e28a0cc4e7595fb31896bc0ebdb
3312ca796642e9fc567dcd8cd6e63b49073b8122
43573 F20101129_AABDVL fure_a_Page_046.jpg
f221d6bdd6de2bef19abcdf65a4bb70a
b5f934e7214abc3f3cd193dcff810ccc7322877e
71668 F20101129_AABDUX fure_a_Page_028.jpg
889b99573b9eec3b0cdf646a0bfbbff1
431e3e8cdd067e68f9cb2ae15c08b8017623d7b4
79315 F20101129_AABEBG fure_a_Page_045.jp2
766f11756f52f51ff4e1f086ad459445
467df0bac86bf659e3eab81ada78e5116573d058
66651 F20101129_AABDWA fure_a_Page_068.jpg
8da1ed2d2d3ce77f86226f346b0c74c5
d96f016a3042cb4b4b73bac6d17f66b1f78c787d
107096 F20101129_AABEAS fure_a_Page_028.jp2
9b3e6a745d84fc5736678864132ce814
ec37fd077a8a700109de0722f3bdcc772f5bd0e7
71385 F20101129_AABDVM fure_a_Page_047.jpg
56f8bc4cfb7d8853f8668f0224dc688f
0ac0d6d675bbdc8b6dfc4aca382e3eec13c840a9
58361 F20101129_AABDUY fure_a_Page_029.jpg
b9e72ddb873b6b65e7f59ae5725aa77a
0ce85c627ca1d3e101c710b7ee051ef8bc041328
659959 F20101129_AABEBH fure_a_Page_046.jp2
8b432fe3cbb319639bd4354824813a00
ba8436b08122074c13576bcdd7f8bbec20eb8bba
41138 F20101129_AABDWB fure_a_Page_069.jpg
366b232413e807591ed203b5a6194676
0cc98c4cbc341ed1af02cc28bdda15fbb53d74a4
831189 F20101129_AABEAT fure_a_Page_029.jp2
01ac7d69e0b326033eb974a585dd58a1
5e6f6ffd2f644b14f779a58e54255495e829bd21
75294 F20101129_AABDVN fure_a_Page_048.jpg
c29ba4816c125290b7434462f1b3b2c1
15e95cc8d6b63c7c954f69b898276c796a2c2b85
62424 F20101129_AABDUZ fure_a_Page_031.jpg
9880009f5c9fc26b4414644f0d809b4a
e586c076c7320b894bb98b09658565a4ffbc963d
105385 F20101129_AABEBI fure_a_Page_047.jp2
510d40f6cdf50732034885d195473bc4
f0b69521e06b9ace242423592c807a510999e120
60876 F20101129_AABDWC fure_a_Page_070.jpg
01c426edbae41e797fecea3701339461
050c6e06dd52674ab9f6a9b1a26b10d1a4f1fbf5
92523 F20101129_AABEAU fure_a_Page_031.jp2
53163026cf7e7b6a8e1e52a6dd2e939a
f6bc8d5cf939a006069b30514a943ccd08c2f1f3
64857 F20101129_AABDVO fure_a_Page_049.jpg
812bd4dc2c08bb2fb74a684989b502d3
c580015d7660fe7a8b2ad5b3856355915c86db89
47167 F20101129_AABDWD fure_a_Page_071.jpg
e240d91c74f06781e7ec04118b8576a5
3e71c61de382aaf2129d35a1d481342b93e121a3
110525 F20101129_AABEAV fure_a_Page_032.jp2
43deda33c68ba6eca9779f641b6a3107
09746fc815c039cf070fec546484e640bf3e64ce
64663 F20101129_AABDVP fure_a_Page_050.jpg
d329e41135a816904a57b8dc06d90e50
bf8cee27f9af4a77d68d06f2fc13d95f1888f6e0
112599 F20101129_AABEBJ fure_a_Page_048.jp2
841efcf2672db07c59a98173058236f2
a9e846b3da70fec9857c37cc2f2acdd31f129904
62922 F20101129_AABDWE fure_a_Page_072.jpg
881694d6917061fbe1fa4c53f979e95a
109b7258ad03a63a8f864c2a813de5510e0b2e76
99608 F20101129_AABEAW fure_a_Page_033.jp2
ac9bc9b1cd83eb3bc9002c685ced0cfe
32ed797ab774ab31d1e483b115891612a8d50d60
67976 F20101129_AABDVQ fure_a_Page_051.jpg
6d9fcb2cd6d5729bb006a86ab84f65a3
6aefe7e33a3af708d9cf4f9db35cccbf6f4ea57e
1051976 F20101129_AABEBK fure_a_Page_049.jp2
512b4a904f5824c52aadfe3f01ab0c47
0caed4e4c4869c9fb926cfb285216525c6b1ca85
74499 F20101129_AABEAX fure_a_Page_035.jp2
6b22306386b89278d6de98ce2c8eba99
480cb73ab4927a73bdc3efeff0c7652c7047f7ff
50501 F20101129_AABDVR fure_a_Page_054.jpg
dc48a4ae953e889b42b767736099bd06
02facbf16784b2999ac774021c7aec8d210e503a
857418 F20101129_AABECA fure_a_Page_071.jp2
fc3174af75826933034992d7b5f1a9b5
9590698f5598dd76fe4c2fd724603bb155bee5db
101159 F20101129_AABEBL fure_a_Page_051.jp2
e613404be5d789f866a7d4ab31a4f54c
9ce2a51fa37c30489dc750089ef1e723f241f033
73095 F20101129_AABDWF fure_a_Page_073.jpg
ac70fb7a8c5f28429eefa9494d4e1334
9d669b662c886b05d5ddb2818a8769c90d8dadc6
101311 F20101129_AABEAY fure_a_Page_036.jp2
b88e0d94bdd530f344ef5574a4b9258e
d737a75e1a866c2ea1958538a8cdfd1a45f4750a
42177 F20101129_AABDVS fure_a_Page_056.jpg
cdcbb2dad24c08f4dcfee30a0c2f4cb1
246fb621d0b3712a025d0eab7349f46614c23f5a
1036362 F20101129_AABECB fure_a_Page_072.jp2
b7b6b1ec25df1278a7a4cd936363df96
81c2cfc33e94ac8cd03dfc313a7f1d08973ef512
65524 F20101129_AABEBM fure_a_Page_052.jp2
b200a2510b45ea0fbfcaf701430ed57c
4dc31dd2151383614deecfff87b80c81e614fd0d
17576 F20101129_AABDWG fure_a_Page_074.jpg
9b7057a8f0273a2bcc6fe5b432c0f380
7b377be97c2e9d0efec8cb951b38e82e5ec37a23
715452 F20101129_AABEAZ fure_a_Page_037.jp2
ffd7013e36964c3853852dd4af002547
0db74b547871eafa14d10706c99f2f9a9891339d
49546 F20101129_AABDVT fure_a_Page_057.jpg
d242c89bf53603e4598b17b97f14bcff
d21f06a5dca9b49efb2ac02a6c39b4ef0b5f0d99
111068 F20101129_AABECC fure_a_Page_073.jp2
2a78087b6f4a93b0efccf820188f8752
842c9f7c99966931c0e657762898e8da8c7e712f
800795 F20101129_AABEBN fure_a_Page_054.jp2
cc931f892ba1095ba90c0b2ce0faece2
70113044a98331adc3e53f23547743223b3df886
74124 F20101129_AABDWH fure_a_Page_076.jpg
37dd87ad349641c80c28bb85e067e3c9
05dd7830c4f7e2b309f2a079d98e9196c4694dfb
39951 F20101129_AABDVU fure_a_Page_059.jpg
d1aff00df3cbcaea9832c8993ecd0a02
14430749977e4bdbaba00cdfe12e19e848145223
19381 F20101129_AABECD fure_a_Page_074.jp2
a6d9ce5b54231064b43e852543fa2add
ff9852a87f59ac27f690924ad38d325f54ba69e7
98934 F20101129_AABEBO fure_a_Page_055.jp2
67c191580ff0957b70aa7ee17e1cebf2
59dd2019ef918f6cfa4736dfe626a90cd96f1cd6
60402 F20101129_AABDWI fure_a_Page_078.jpg
257fd6f06981539ce37a15f97b177cf1
28063540c3eb514b9ed44ad36dd46c8a6c3eff5e
45681 F20101129_AABDVV fure_a_Page_061.jpg
331da15af9afc8cf278bc816618c01cf
2b3c146b877d2ac257f5ebdb06ff15171541d3f8
94562 F20101129_AABECE fure_a_Page_075.jp2
c8606fbc1e02779cbfee0faa359f3077
e97ecfd5a9d3b135467c030a680547bf6ad74d8c
92678 F20101129_AABEBP fure_a_Page_058.jp2
8024ead8a98fa772ca2fe9fb9eeab532
ecea249a66e132e42fb53265d1aa9eeec028b557
66432 F20101129_AABDWJ fure_a_Page_079.jpg
2075b5e69e458188c7a9f1e8d55cdc0c
89c75b8fbc4c1e001521d75f735f3b221e9b158c
48189 F20101129_AABDVW fure_a_Page_062.jpg
d076304a4dbb91b72970c12a8169cc9a
1bd765443b5d5897834cfa2327eb37b4af11891c
113488 F20101129_AABECF fure_a_Page_076.jp2
e731712eebace65ac8ea43af8c693835
de1d56593a41ffe1bba74e71752650f210a82a82
603930 F20101129_AABEBQ fure_a_Page_059.jp2
a9df08066f186c9927d032b22d47639f
b8c0a6e8f0de9fa934cd50dded986778248453a6
43156 F20101129_AABDWK fure_a_Page_080.jpg
ec9ad3166c601cd8b7baa7ac73d9818a
eef0041bd8df872ab930f86d5ea3d4351b3eb04f
73521 F20101129_AABDVX fure_a_Page_064.jpg
79456c70a896079237adfd8017b40aea
6bf3313696ab2ad6b2a425f462455e8e974679ef
111145 F20101129_AABECG fure_a_Page_077.jp2
8cee4cab4f49e7aa5258c932d62f39b0
e4bcb62440ee0b0ab712c2a02b575aacd4ff2bfa
778293 F20101129_AABEBR fure_a_Page_061.jp2
319a72b63d63e44df16615b2a4422e6e
e6e34bbecffc4d8f8ef99fbfe8af1ce14f5da149
76249 F20101129_AABDWL fure_a_Page_081.jpg
070492c1cfd8c14f2f301b6a191d604f
83068de2a79831a2bd46d4911ce9b483f88447aa
60908 F20101129_AABDVY fure_a_Page_066.jpg
fe81fe0bd6c34ad24786e1435fdc8e56
63d3cf8342ee2af476cad434f1edc5751b781f7b
89604 F20101129_AABECH fure_a_Page_078.jp2
6dbdb1ed96f20bb9436f2780ce242043
8133ce345a8a0c3a69a96837802e61470a9b8615
54423 F20101129_AABDXA fure_a_Page_102.jpg
9d780269374eeb068237162b51aa69d8
dd5e629e8fff3a9f5311f16cbc54a9b203d33bfa
777733 F20101129_AABEBS fure_a_Page_062.jp2
a97d560f68f01bada6436764d8474a66
de88632517484fe90e8ec62966356107fca98e3c
74893 F20101129_AABDWM fure_a_Page_082.jpg
ced0a506ee8b766c48cf43e05b956d27
8775b76ec49750878cfa777e3ee24bc2660e4ef5
48957 F20101129_AABDVZ fure_a_Page_067.jpg
dd759672a0ce19db4cadf4697541db8d
645f0198a7f86227a9554d50faf2b46659e11ac1
99529 F20101129_AABECI fure_a_Page_079.jp2
666afd63aecdb3470474a57fba7325d5
dee75c6b2b4e75a5951a08f84044a76a58a28df2
69991 F20101129_AABDXB fure_a_Page_104.jpg
2eec42e2368b591c8754d1e4065e31d0
2dfa976f177d06b5f232d22ef649baaf2451c9ed
935671 F20101129_AABEBT fure_a_Page_063.jp2
b768c7d631372b0bb9aeb65decec87e6
6a7e22add164f688dc4a85797e8e026a88ed061b
68599 F20101129_AABDWN fure_a_Page_084.jpg
9552d77e3d9b2ed17a456c8e6921839c
7b8158d37d8392b5f6482b27fccedb72d7d660ba
543536 F20101129_AABECJ fure_a_Page_080.jp2
8cc554a812e93fd951bc31bb6f3d9ebf
a30f0081f00be81f598fdbd56ddd094a4509b421
70674 F20101129_AABDXC fure_a_Page_105.jpg
b850b518cbce2bd32fc396559035fd38
4ead94db04ecc6b123630b92426774f8b1db9146
110366 F20101129_AABEBU fure_a_Page_064.jp2
65fb7a2d0d5d2f55e71fb5db528b09ac
ad1b249f786ee6b5fac2261e52f1228578ca1c59
73215 F20101129_AABDWO fure_a_Page_085.jpg
8b124287c1ab381da11b4443baa0f291
dd4fbeb96d98908852f9ba7b3f2d524e07c6db4d
51888 F20101129_AABDXD fure_a_Page_106.jpg
6f461c30232fd3474c78a8005eaeaff5
17ff7f6795090907c9a118c8faf575e6dc406ab5
715079 F20101129_AABEBV fure_a_Page_065.jp2
693aa8ad32c5e3da7b7c1c0d12b92ee4
78757689ee85ee4bdd42ea6b78604181cb53cb0e
74891 F20101129_AABDWP fure_a_Page_086.jpg
08e86034eaf4f5a22e5cc51fdb8d3350
016f57c6cb72ed036456d80d388422fd322d8ad2
114005 F20101129_AABECK fure_a_Page_081.jp2
91759893805780120d7ba7b067b27eae
a7ae5cd7937373669a9ba436ddad030680eb8d94
39040 F20101129_AABDXE fure_a_Page_107.jpg
a9296e8f901bc4feaa37d970d5dedbf2
9855422fe391e4e64fa2ccbea69442d9d11f3e41
939788 F20101129_AABEBW fure_a_Page_066.jp2
2bf87c6e77d95c17d31576c27560a87a
c8a4d2a9e93b4d48f2c5341e8c6bfd39d6f31b94
70675 F20101129_AABDWQ fure_a_Page_088.jpg
17c95b50996a0ca2d841784fff3f4875
264ec6000995c8e50c0d5cd60ddb9d084c67f506
106903 F20101129_AABEDA fure_a_Page_105.jp2
3e942a1906008317d4cd008ed05c8a1a
f4c19e67ab84428b6a5a0df3e76a37aa16ddffc6
107827 F20101129_AABECL fure_a_Page_083.jp2
dc5ebed08dda1d22f8cbbe04459d790a
f558d6024c3af8e679f86f075ad5c4f93d6dc74d
47018 F20101129_AABDXF fure_a_Page_111.jpg
894f5d38c5bef9d6d8d1a81536102531
7f51a2dcd9a8b7e4a295c5c034468e9020d8569f
923962 F20101129_AABEBX fure_a_Page_067.jp2
11c2b0383e4939a0538e067cef05ceec
9b29f7a8367214bd365e15d15678e74819db7e91
75265 F20101129_AABDWR fure_a_Page_090.jpg
f12db7a18eef810b3101a39a5909a011
9c67254cc82b6eaf023469c6df3a4f7e0a2a4044
587626 F20101129_AABEDB fure_a_Page_107.jp2
22bfec65396f5b49d24b58a67b33b285
dfac7e1be93a0d195875345320a00e466ecb808a
113866 F20101129_AABECM fure_a_Page_086.jp2
ac3692d63c7a410bfb17b4eba50a94d2
eef3af16e64008a0fbc168ea0bd053ad408637d8
55633 F20101129_AABDXG fure_a_Page_113.jpg
095dba23370fe118fdb040b681450989
3d8db9003ca4ae934a1263d4d212b9892983eb06
100067 F20101129_AABEBY fure_a_Page_068.jp2
c81c992c0226246e0371fabce82639c7
367ce24f54dc7d3b77f33a5d92455e91ae688e65
58192 F20101129_AABDWS fure_a_Page_091.jpg
386e1932c50442dd46fb03d579cafa94
f89f8b76ff1b07d2463c38f54f092f2f66dc457e
73086 F20101129_AABEDC fure_a_Page_108.jp2
49c9b33b8adce7e042dfc7985140dfe2
42f90307eb5f03db22bbfc42307820284c9bcf36
650775 F20101129_AABECN fure_a_Page_087.jp2
fa8ed01a84b8265a62533a114ce09029
5e7cddefca698f09fbb9535ce32dd24921cc3230
58681 F20101129_AABDXH fure_a_Page_114.jpg
944a073bc2b966f335f6c9ebd30bde86
d8d397ed8db781a71860e3104ae649f828f2034d
911486 F20101129_AABEBZ fure_a_Page_070.jp2
2a601a7bdf6d95c5f687b071620ebb57
30ef768744441e08b3e762856b66238817cca95f
45496 F20101129_AABDWT fure_a_Page_093.jpg
333bffc21c91a9840b9d3c6a899e2956
e69571a3a173b329236fe5ac3e4d3d11a7ccdd39
923037 F20101129_AABECO fure_a_Page_088.jp2
62de22cd10c87c8bf29d03e11e53b021
fc538762fc9dd517fd129b06cfa8ecbca5825827
50574 F20101129_AABDXI fure_a_Page_116.jpg
10ac354ad6dc43e9060a7ca43794f99e
1be06f51bfd81d3d6a4e85c6b05b7f39723e4596
69670 F20101129_AABDWU fure_a_Page_094.jpg
04b3226577c6564b4ff1082afe0b39ab
b3e5bc0a1f23d96b9a371bf05008573b915dad8b
51899 F20101129_AABEDD fure_a_Page_109.jp2
238bc358c4a65c5e5206abebe678ee84
5c3e981a74cefc642a27a8e440d9a3e49ce686d4
549607 F20101129_AABECP fure_a_Page_093.jp2
30e1e7bf2b571d4eee41c0494760379e
a0d8f974e85eae5f1934222aff09fbab94fef212
68741 F20101129_AABDXJ fure_a_Page_117.jpg
d6025c6ede0a61bb077081b3cad8f27b
290b4a549fecd224d5ca712b8c61dc7f7cff29d3
72083 F20101129_AABDWV fure_a_Page_095.jpg
65fb432612edfaa9ec9f6e780ab2d0ed
3d7cc5e0a3df40dc6821ab65d82c25f0b39217bf
96796 F20101129_AABEDE fure_a_Page_110.jp2
94c36c69cf890581850d3b2a808301db
5da9495a8ff44658abc456b4fc10077a5caae533
936987 F20101129_AABECQ fure_a_Page_094.jp2
cf2bd9f8499a8a3656bd269863df8c5a
cf751c4915c99470e037aa0279445541d0da087c
68181 F20101129_AABDXK fure_a_Page_118.jpg
386f0fc86210d59eb87654d4f186d1f2
b382dd6bc9d33f32341ed4197d72a2e35c932137
65904 F20101129_AABDWW fure_a_Page_097.jpg
a0b26e5918981aff24021eae50c7e125
802c5c1084a24605fad33f272aee5b442ab865af
702886 F20101129_AABEDF fure_a_Page_111.jp2
d052f283e298db8c1390e48b1c5fcf54
fd2ffa3fcd2374998b7f9ace6886cc2f53c3121a
40464 F20101129_AABDYA fure_a_Page_136.jpg
1926543c8d24e51803b71de3673897cb
c698b2965a5d5847b535dc1cd9524b1bca7fcb65
108781 F20101129_AABECR fure_a_Page_095.jp2
54d80b2a4d1f9dc6bdfed5c02e0b803c
4f7ab261b40108a7634854f2eb6c28f90cbebd44
34918 F20101129_AABDXL fure_a_Page_119.jpg
7ed380bc160df5898e5ea496eac8fe3c
fceb22939229743fdd883490965b09f2068d2da9
73999 F20101129_AABDWX fure_a_Page_098.jpg
7d223e63a17bbf03b5e8cfb2acfe27fb
bd7b0b7c3bfffc7c44b1c52c0f2c2061b69a88ad
891450 F20101129_AABEDG fure_a_Page_112.jp2
debf8f4601876484d53f21cc4a9e6e8d
88b636a356bc53cfcf7a374d1d52c66bebc0a46e
558818 F20101129_AABECS fure_a_Page_096.jp2
7235c6813bd2741d6a554387b38c18ce
9f01b8a549afe043a52d6a280c853570e1d7b063
65310 F20101129_AABDXM fure_a_Page_120.jpg
bba742ec215cbec63f7f38efacced0ce
e7df5410cf69c023cffeaba2356ac466526634c1
65932 F20101129_AABDWY fure_a_Page_100.jpg
8960bc62955f490b47b4d5a81df16968
16630e4b8ee52e898427d8a62a64cf2595404b42
85614 F20101129_AABEDH fure_a_Page_114.jp2
2fd69f114bf3fdca5da991a483ab96bc
279d2f16537a4243b70a3a4a55da2b56bb9b32a2
70046 F20101129_AABDYB fure_a_Page_137.jpg
f13786fe7e1b20d63e242da4649fe35f
6ca57da029f2af21154e3e9f785a362da2043880
870090 F20101129_AABECT fure_a_Page_097.jp2
999ed388fe6545357128390c521a6e71
d64675bd4c8d6735efccecbef0e3473b0c6c1c28
30826 F20101129_AABDXN fure_a_Page_121.jpg
ada36c82e8757ca0145b8dbc9a2cfd24
f814291fb39de122a38795ec106699d0e874084e
67597 F20101129_AABDWZ fure_a_Page_101.jpg
93971f55bfacc9401b33fd22e9abf72c
71f34755bc933a27cec68be126a329f715c2c9d8
103031 F20101129_AABEDI fure_a_Page_115.jp2
e513e7b2a8fd6191886fd1089b6f8580
8dd9c3bede708a2efb54dd5b30e28da0aab4fa49
62348 F20101129_AABDYC fure_a_Page_138.jpg
53e50a3d246ab00a87986f42ffd7dde4
cec9ee66180023f7d924c4b939a0f23c2f2590f8
112077 F20101129_AABECU fure_a_Page_098.jp2
d63749f05cd5ecdb08f4c40437a4e9ae
d29eadb6e573185e5265564b337722d1205ac787
45366 F20101129_AABDXO fure_a_Page_122.jpg
bf261fddb29407a6a7ad3e1961e141d8
489a9e9181d570dd96ad38f478bb2f287c5b30a4
75591 F20101129_AABEDJ fure_a_Page_116.jp2
c2bc7b32196cfba62e577b3bd680dc93
98a6911618a884f1e482ef554b10c2e18c0de295
74031 F20101129_AABDYD fure_a_Page_139.jpg
c9a7ce466a97fb7db055d6261432e2ca
c71409d92ed0619bffbc33caaa11ff998bd83bbe
865081 F20101129_AABECV fure_a_Page_099.jp2
6545b0941fa676d9a185e75583a6e719
3418f32c02e96c76bf3516fb5040b0f8940285d6
43568 F20101129_AABDXP fure_a_Page_123.jpg
b1382ac231eccf9c3e76a0a4054cd411
a881d465d91dafa3cf8a6c0b29867a0615ae43e9
101665 F20101129_AABEDK fure_a_Page_117.jp2
2f665aab24b3904dc696219aa8c68fc1
a484907401f6f4a79020ec18f14532fe1fa813e1
45077 F20101129_AABDYE fure_a_Page_140.jpg
f10da9aab4e7cf80f899fffd26c20e22
7dc930cf96ddb065cf796c143ab5ff03b92d9bf4
891482 F20101129_AABECW fure_a_Page_100.jp2
08fa27171a21e108eed203b6f92ffd47
a6c801c3f56363e2587f6301ee91167da38d3464
44867 F20101129_AABDXQ fure_a_Page_124.jpg
bd825908bb43bbda9a867215ea595f25
64e79b5e5a36557ff7d063855f07ce462cb89954
43275 F20101129_AABDYF fure_a_Page_141.jpg
e5b32a4d44da587485399d9043c0fa24
0edec491c9635f78f1dcdae2f35bb9a1121609ca
104021 F20101129_AABECX fure_a_Page_101.jp2
bd1ad4e6344054b4f5e1bd59ac82167e
1a26cd628b02fa481bba7eac6fc14e29d7f115d1
63656 F20101129_AABDXR fure_a_Page_125.jpg
8ba0d17079451fdfe8b794a864a355bb
0ea61f8bd3711deccc660cbe84e31f870764215f
715862 F20101129_AABEEA fure_a_Page_136.jp2
eb827b76d6dbb081ae3e7bf91c6c2cac
d99732d370b97352a89088c7114121b3ca8235ad
104554 F20101129_AABEDL fure_a_Page_118.jp2
caad78432299d85436ba9879ac607ef1
8f74bee023c7b9d24accb37b7312d496f2d8bbb2
55833 F20101129_AABDYG fure_a_Page_142.jpg
010215cca10818289ad070a85b97f43f
576b2412ed4a1c54a184eaf9211fb817bc60a558
79350 F20101129_AABECY fure_a_Page_102.jp2
443b92c4035d4bbb8f23f97ac57ca358
500e15eec9933559be881545ab7a887762fcaa45
68702 F20101129_AABDXS fure_a_Page_127.jpg
7e7ab4e714f991a6fdca7372f192c660
bf27083798d439d6abee6aeb2bcf1f8eaf715a90
104371 F20101129_AABEEB fure_a_Page_137.jp2
31698f48cbc066d6c7247e2cbb392258
204f678e55582c93d53f7c335fa2562c2b78eafa
67851 F20101129_AABEDM fure_a_Page_119.jp2
49c2f9af791f497955c7823c1ab0f251
614774a5d9317d888a11de5a8a481e371d91849a
59665 F20101129_AABDYH fure_a_Page_143.jpg
2510a94c734c45e3f8efb910d6097dcf
df513525a24cccde983a3be73b743773f5dd4b08
107988 F20101129_AABECZ fure_a_Page_104.jp2
e89c6a8ed770c202a169fc1830079291
a9f682c2978dd669857866987c097e0147560ae8
72113 F20101129_AABDXT fure_a_Page_128.jpg
d654d4148a8a0b731ad9b522cba3e163
ea982a480ce8ac11e9b9adfc90c80d1176209d66
86786 F20101129_AABEEC fure_a_Page_138.jp2
2e588502273e0bb0d786bf2aeb46f114
47a66ecd92d3a817a3d2cb73e8be183b54b45671
100864 F20101129_AABEDN fure_a_Page_120.jp2
941e00e4499bacdb506ecd4e53f177c5
f49ffd5fbc959bfc90a3dca5d53eb5dd780f5f7e
43409 F20101129_AABDYI fure_a_Page_145.jpg
77a000f594f1ba5f32a45ea237b401b4
5d2fb9324ccd21495fe303878373b760deb054e9
31391 F20101129_AABDXU fure_a_Page_129.jpg
69b40f8b04a95500cc3e32e4cbace21d
dd86bfbbf9ad088886c79b32c78ed21c042844e1
111185 F20101129_AABEED fure_a_Page_139.jp2
f07885e309871401f95db6894601b20a
fce1015d48c51918cda65176b5227b23b3034705
68080 F20101129_AABEDO fure_a_Page_121.jp2
12686a6fdc4ff96b2901c89c62f2a7e3
5918705a03041d418a6cc4887638d01699f6c4f4
72610 F20101129_AABDYJ fure_a_Page_147.jpg
25ce8120a6e464b1455d2d00831720b5
8db3c6da9042bed3b47ec998d4a0a95b1c451c24
43813 F20101129_AABDXV fure_a_Page_130.jpg
57035156c0755099711cf3deb2e0198e
c51b6db827436503a3c89b5b290815cee69636f4
825109 F20101129_AABEEE fure_a_Page_140.jp2
b344c03a574f42432fa9f302badee75f
e4eba15eb732d3da12eea52638539552c10ad275
814243 F20101129_AABEDP fure_a_Page_122.jp2
f409fdff7feef0a6156de3d3700fdff7
463cde6e4926e4ecb7d2494119050ec33306b248
71765 F20101129_AABDYK fure_a_Page_148.jpg
34f28f357cacb083dc97870a39f53f7c
16d1f02f8d4df4d79b63cc5106aa057f53d7cc79
43494 F20101129_AABDXW fure_a_Page_131.jpg
847109c678b87ea2128384dc5c8f73ba
e3b1c2e9a51e0a666eecafdafd2c3aa65e3ada04
837824 F20101129_AABEEF fure_a_Page_141.jp2
60a21b57eb6d2382fea824ca06c439be
63904676ba347733479f4193ff6ea38e9b0e0c2d
840283 F20101129_AABEDQ fure_a_Page_123.jp2
c7e825fe3c64e8fee6978f3e3ae8e521
289c56c441a3f47f68e1c87cb8a7240abcb84863
40880 F20101129_AABDYL fure_a_Page_149.jpg
19b47e5fdd0da9c2c83dadca8c0d1a7e
880aa930a7c210e1dbeed4862c668a7ed8c81fa3
66425 F20101129_AABDXX fure_a_Page_132.jpg
5b15415711e6a26699972ba6bc04e573
d8da16fc4a5ad751432c1bf4edc232161d4ce60e
79476 F20101129_AABEEG fure_a_Page_142.jp2
fa98d73ad036f5abfc06df32c8c6a666
aac2c5aa8b6dc3dd4506332b754239c5e97d9a22
58672 F20101129_AABDZA fure_a_Page_172.jpg
05f1b85b2cd1f1b2e3c5e05bb75897f6
6ba715f6124f5179d7a03a7131fb4b62bb6c610f
631188 F20101129_AABEDR fure_a_Page_124.jp2
ef9b480dd54e06ac3f06aa9404ecc937
e3ac73d27e77411910c99054bace76181786cf9d
39110 F20101129_AABDYM fure_a_Page_151.jpg
cbeee6d43796ac5da282a2f9a937affa
c47fff2adc45d32097e328089144a914c3346b5e
42209 F20101129_AABDXY fure_a_Page_134.jpg
7e0c3ff3d4269c08b770ed5180d65788
43c3f25bfd145b185f2a229c0253e041417a44f4
663453 F20101129_AABEEH fure_a_Page_144.jp2
32b1e15e1efc822dbb6e5fcdb16f7366
bbd028ea62b74ab5f984164e0172a50be8327d7c
53424 F20101129_AABDZB fure_a_Page_173.jpg
5e0c52a1f0cbaea710917f1477c1ffc2
a79e9c2722a54be85f8c6eeb2ec24a96acf462b6
599214 F20101129_AABEDS fure_a_Page_126.jp2
c3d7108a6bb5f061cea85b7e254546ab
8f94f3a93421a77be02fc016291ff2c60716a849
24626 F20101129_AABDYN fure_a_Page_152.jpg
6544a2bf26a4fd0199faa774d8916c76
4194d2a808fcecd3dcffa23e98f4568da55add13
40268 F20101129_AABDXZ fure_a_Page_135.jpg
91a4e35e24f217e2d0ce36c073e89b56
41cae43b2bcc3f77f6f3399ddec7d9da9879690b
608130 F20101129_AABEEI fure_a_Page_145.jp2
4e757352baca0b016b5847f936bde79b
c80473f6d536b2985f26df46eb3f3746796012e5
107637 F20101129_AABEDT fure_a_Page_128.jp2
0138a8d911dca8587b2f84d7482db90f
7acf75732c7e13bf36ccd5cdb895d496396d8048
53368 F20101129_AABDYO fure_a_Page_154.jpg
ed446741e362b44c86c7397349670fb7
f2ad5aa5f966e7e494694bf16d5ae609fc2a27a0
109292 F20101129_AABEEJ fure_a_Page_147.jp2
c1a634678663176670170a69c1c176b5
78b2c850cb2817d3602366e39857282c4636c27d
71879 F20101129_AABDZC fure_a_Page_174.jpg
e1b936106b02742b1098e0c67bd885e3
c4bf9094266d0289214e73738990f6931fe51d00
68127 F20101129_AABEDU fure_a_Page_129.jp2
074a318c2d3531fe0c8350bda48f067b
4eeaee3a9aa68f9fa5981bcbcfad051fcab27289
49168 F20101129_AABDYP fure_a_Page_156.jpg
92d04371d0c526339efc3070efe8355a
af05b4f0bdcb1156d5afc31ae486d6a3ad796b95
107269 F20101129_AABEEK fure_a_Page_148.jp2
d93093da70141814b8302595ddf1b25d
f9b972e698a45c589bd866b5feab3aff1fa937ed
68722 F20101129_AABDZD fure_a_Page_176.jpg
92e7cb752df673989aee9c7d4923dce5
b0509d8363effdcdbbe05199ed386294a368520d
827230 F20101129_AABEDV fure_a_Page_131.jp2
e6b9c4c0553df78147fb1af9ec68aaab
9a05d48136f54169eda58533a2848b1f9974a74c
60946 F20101129_AABDYQ fure_a_Page_157.jpg
a2f7acddf2fc3c1c410f4d4f9a36ce61
290a0e5bc795636ec2b468f56fab206af11dae77
488827 F20101129_AABEEL fure_a_Page_149.jp2
ea1e3c4476035b5a1c628ec99ec9ff95
f3ce3c6e48c47d06c6d00429b591a9fd4fe95a53
39256 F20101129_AABDZE fure_a_Page_177.jpg
8cddb5745cf1037e48ef1e5c17345096
1a0b649b9531b5d3c8e87e029c7a85b8cae4594d
97286 F20101129_AABEDW fure_a_Page_132.jp2
7bcad53a4da7533606009de6255bbebf
bb64782c1de73641c4fd6e9ad89d78e2c2bc3dcd
54461 F20101129_AABDYR fure_a_Page_158.jpg
511f50deae97a65cfe95672f7fff4fde
873c60cf704f577d7a7bcd7f9820655632f99296
89701 F20101129_AABEFA fure_a_Page_168.jp2
939a073d9a68da8f6bf84d9c2aae011f
ec2a1fa73253f8840d8ab3b90cd4e1ac452cc6d5
63064 F20101129_AABDZF fure_a_Page_178.jpg
4e039fc87b760ccd87398e504e02b2b9
fb4f74dde8afc9d8684e0230f1a2f452a0d9a07c
739246 F20101129_AABEDX fure_a_Page_133.jp2
b36ab767e0a3ad53aedd686f8dac38a0
a026a42464e45af1fdd950bc9247cc6f5d2630d2
54111 F20101129_AABDYS fure_a_Page_159.jpg
0dcd14b29a3134c68701f6714bb628af
64c10bec7566ff3700f11ba0a57bd5ee5c449bf6
618884 F20101129_AABEFB fure_a_Page_169.jp2
360a2dca7b35faf827d8b451772f27a2
ef435173c7b159bffb7f4b990726b17cdecc2212
705061 F20101129_AABEEM fure_a_Page_150.jp2
df1c2b54c2347a440f98b3ddb503ce71
43d0ef53586fd3e33c7fe29461d1cd707ba93e34
61594 F20101129_AABDZG fure_a_Page_179.jpg
c2732198d78229666b7cdfe110a62aba
4559181c40dca2b90c37d94c5e3625ca05c3877e
755749 F20101129_AABEDY fure_a_Page_134.jp2
6560830cf2ca64519e9e95d48ff0f86f
779ef1877bdbc41edafea45c9e2e968745905569
42976 F20101129_AABDYT fure_a_Page_161.jpg
fd219e06791efc39cebd0478985fa452
a85f198ab9d59f8813568667387c48cc803dca9c
90495 F20101129_AABEFC fure_a_Page_170.jp2
5aa4194a452fc43e4f9c097b89e2202a
6051bee824b8bda88389ff11e48bfdb62dc74046
534960 F20101129_AABEEN fure_a_Page_151.jp2
5adee246e699ce72bc5b9332d27b3588
f7156a5ea7a46dbfc10ee9d66e5a222cb74a4d72
75404 F20101129_AABDZH fure_a_Page_180.jpg
0896d1222eadc0b24da01d1ec2e7dc59
2a60754e802ef36527b6d131a23ee6c0d0a0d469
699391 F20101129_AABEDZ fure_a_Page_135.jp2
1ae7d259a77e404acf6239640c9f9925
6b8a860b037fc11917da6f604335f6b77f9f2e53
36519 F20101129_AABDYU fure_a_Page_162.jpg
7081fed8993d2e16d344b8f9c53c0df7
c4078be34b4b5b5396003c247795d47f1e485fe5
715668 F20101129_AABEFD fure_a_Page_171.jp2
5e40751fda29e3cb73924b2085c03dc4
2add01c2c8c949b75b42ca29474a9b4f7b61e1f9
287878 F20101129_AABEEO fure_a_Page_152.jp2
390e5886612d845206745c57cfc2efac
89d7870a076bc7f2ce1cfa3b133c34ccff183051
70076 F20101129_AABDZI fure_a_Page_181.jpg
b4eaa0be9a9f617ed9cb70f4d834834b
c6a030dbd137af71a0322ce3ab7f38f4f2a2b946
43569 F20101129_AABDYV fure_a_Page_164.jpg
1be781db2a36f5636d97fe0835f48c55
f0bc7c991abce67370ae3342feecc541da0e6cac
819920 F20101129_AABEFE fure_a_Page_172.jp2
76bd9f56169e3056debcb31c03a0f8f3
c2733475fde02ae1a42937fba25b3e321c144bbc
89548 F20101129_AABEEP fure_a_Page_153.jp2
1e454f93ac55a9a91f58fbcab6599823
3ac6af62706741e821b1aa785bdb7f08f0e247a0
54645 F20101129_AABDZJ fure_a_Page_182.jpg
8ce09da921c6c02e7a4e7d0d80ce7aff
c25c7b1615a9209d07f1f091cd886f81fb82ada4
61257 F20101129_AABDYW fure_a_Page_168.jpg
1b7e5c4fe046f197017c5a2030ba83c5
548d5fd91e00923be086d3d74a9c80767ba38356
732871 F20101129_AABEFF fure_a_Page_173.jp2
f2fc2fa53b39281fa4d6cb82620b7183
dba14e98c307f34f5d620a31dd89f1b79a84c41c
663599 F20101129_AABEEQ fure_a_Page_156.jp2
6c791b041606d9358281dbad95599556
4b87deb90c353bfdcfc36ef647f874f2d9ce225d
62454 F20101129_AABDZK fure_a_Page_184.jpg
894279feff9cbf157a790dbeb0e05225
bfc8a9d3a611b96c9896b47b01ff7e2f172285f8
31442 F20101129_AABDYX fure_a_Page_169.jpg
945caf758413f621ba9cdd70323b68d2
4514f1d15f7b2c242507f73374eea7190d6ab11d
111840 F20101129_AABEFG fure_a_Page_174.jp2
2d00686450d8e9f7defb81e5694ffd74
4d00a537d72b32f8071628925052261e79232281
88771 F20101129_AABEER fure_a_Page_157.jp2
e73d0f11d4e5b726797f71d6e6204b66
d36fcd8e9fd07af97accf4c96aa0a5986feca6b0
67891 F20101129_AABDZL fure_a_Page_185.jpg
c6f3c654a28eceef6a271bacfced3dbb
753df1c26ef628eed6e40976f7f86a35a720a0be
61795 F20101129_AABDYY fure_a_Page_170.jpg
b0cef422d0bda3819b4d7d917f0a264b
9855e9cbfe1acc99b2d58564607cfef095626261
105897 F20101129_AABEFH fure_a_Page_176.jp2
f09df61933d59d135df4536383c2e07f
fc3a2502ddaa474681e0d2bae313f26953d30abe
81480 F20101129_AABEES fure_a_Page_159.jp2
03ba7880aa9b7d4d4dec430ec0823354
c03237ec874cb1a079f1ace07ab87c095daed67a
75031 F20101129_AABDZM fure_a_Page_186.jpg
ab32772802aaf3eba5741113256995dd
e10b089e5bd14c100acb075a964d7e7f7b2cf0f3
51789 F20101129_AABDYZ fure_a_Page_171.jpg
64aef9a311e94f4b0416af9acfe85a63
e66e330fd06b39446a6ddf4dc2808c610c6b937d
607603 F20101129_AABEFI fure_a_Page_177.jp2
ec34d66b77c6ce843db0833b4e298ccb
21659c28951bf11ee4ff6918da0b3cc9a273da35
81888 F20101129_AABEET fure_a_Page_160.jp2
2335ea3fd86d7b9a2649b5de61a5ab57
ebb506b401c64b9bc4450b95a271e438b31480fd
54493 F20101129_AABDZN fure_a_Page_188.jpg
1be923d613735f8d05b3eec5db3e97be
76da9290bcf95cf92cee03362fdc43c535a14af4
911881 F20101129_AABEFJ fure_a_Page_178.jp2
c8c025d13fb4cbff1046a4b7cec953a7
e9c512a998c08f4e4c4bbb56037f90b13733e630
982746 F20101129_AABEEU fure_a_Page_161.jp2
1869f43bd80769719952e2f4e1cae21b
8ebc63bbe83c42ebc29c6235ed6ed552c0ae8194
71244 F20101129_AABDZO fure_a_Page_189.jpg
d690dbeca1ea6dfbba31a73813230329
4075dc424a30fd243027bee3755dc88e70f53c73
1004303 F20101129_AABEFK fure_a_Page_179.jp2
2e87473248aede7628030e2ee7bfb58e
150c188f84388f9e92af6d6cc27133ec42aa73b1
539619 F20101129_AABEEV fure_a_Page_162.jp2
21c5573f9017ff707dee37ba3f77c0d4
bad6d5924db490b082f36f6204ce315e911ccff4
81291 F20101129_AABDZP fure_a_Page_190.jpg
d492c0087e82acfb8aa0559ac003b8df
f79e263e6c193e5c48600cc3ad3df8c7f76996cd
899120 F20101129_AABEFL fure_a_Page_182.jp2
3433d90cbc12bd7ef8c90356c6764358
daf372fc1e024386c853536619a5e425b18c6654
787792 F20101129_AABEEW fure_a_Page_163.jp2
a598908b636b46c4e6c798a38940ebc7
57380f91684bbb2cf1c5223cb135593b2896cc81
87632 F20101129_AABDZQ fure_a_Page_192.jpg
ad5561d675f00827c649c466b3d284ed
bd5f3c21b148ae5282f84a10599484c5bac68c13
109744 F20101129_AABEFM fure_a_Page_183.jp2
fbd73e65ccf0a14e22425d073ea94e20
ca2b8fe5d462c65b6a2a38dabf1deae79aeffaa4
585097 F20101129_AABEEX fure_a_Page_164.jp2
1109ad65e7fd2bcdd92c6c93abdc0846
cc185b400d57f0f7fb64751574470de33fa45749
80576 F20101129_AABDZR fure_a_Page_193.jpg
47c335bf8626cbabc306af52b4ebc121
a7ad0df6f3b44202e6b4c588787b20458864f357
F20101129_AABEGA fure_a_Page_001.tif
d1eba348446e8d0f469887d2c567fcd4
bcb27cfa58f50ad775eb6a937bbc4ab26c34ca52
472126 F20101129_AABEEY fure_a_Page_165.jp2
7d256481f142b5b5667562310e180e35
2ff4299757c1adbaac62a446a859424a5053f4d3
83174 F20101129_AABDZS fure_a_Page_194.jpg
8c46cd425c7c15a13aba612598c98437
f441382c9aaa2644424ab049800f5d2569a108c6
F20101129_AABEGB fure_a_Page_002.tif
4537038bf926df809f278e18de697381
4d2fff3161b7961176de1638f8eedcc7c86d9ddb
892323 F20101129_AABEFN fure_a_Page_184.jp2
ab63d47b626293f069c6b8baa73c58c8
0eabb7aeb4329d0315a92423b1497791b0f4a9cf
970154 F20101129_AABEEZ fure_a_Page_167.jp2
19d626ab7ed7824d43b979e82d7eedac
73ae167e7b581e323db2b67ca80fd2e467ca8da8
59761 F20101129_AABDZT fure_a_Page_196.jpg
a00d6e683c3fcaf1c30006bccafbb92f
cc3b04dff62e98ee84a20b62c1507adf3fe7bd97
F20101129_AABEGC fure_a_Page_003.tif
867a4bb78d4424511f104463294d3ec5
37f55e8b76748ad7407ceff5292f99ce1b4f1ccb
1051986 F20101129_AABEFO fure_a_Page_185.jp2
c09ad97fb6da1bb45dcf08208254dfee
bf38a71c597525101bc562e1307530d0818c82c1
24231 F20101129_AABDZU fure_a_Page_001.jp2
40abfe4ddb24c999ef2988b34715e27c
e6cc9ccee271ec9b2f68d334b3403dba1c3668c6
F20101129_AABEGD fure_a_Page_004.tif
9dcb3d8fde32c2e525040321b7b9c7e5
cbf14d04a5c9b4374b76f4528da936a527efae3c
109963 F20101129_AABEFP fure_a_Page_187.jp2
8fc05d30cdc02b47c2888b91c429716a
b692fa6db97617703a3728159a790a2bcdb75053
5559 F20101129_AABDZV fure_a_Page_002.jp2
663f9969a5a24cf8e9be2791ed15e040
a1880a475ee4b5058674d2ae8b961465719a9529
F20101129_AABEGE fure_a_Page_005.tif
73321a05564e09393ecef5e80fc0e71c
4cbd1e840fbdc71d4988a0fd2938a5f869f47d1b
843710 F20101129_AABEFQ fure_a_Page_188.jp2
5d3097286778d56db85ad9f1a16d1192
a58cccba7f517c169bdcf1da8f376dec95484f6d
40852 F20101129_AABDZW fure_a_Page_003.jp2
6b096d8a27592adcf28b62a1bf04af77
b3b25a7f6c1904d927be7f6fd5520612d2372d10
F20101129_AABEGF fure_a_Page_007.tif
9e449c01b7122ee953967c000b9e1aa7
0811a137473b34dde2fbae42922c0c98cd48bb4f
101163 F20101129_AABEFR fure_a_Page_189.jp2
8310b316ec960ca477cf3bdd33685ec6
5e70be87111f3e989a965f4d591c9a13c1eab1f6
1051964 F20101129_AABDZX fure_a_Page_005.jp2
4bc1cedb58ccd3ba4ef8e626e68a7891
c476d33c41a6524f11622f511dc0986084be5aa0
F20101129_AABEGG fure_a_Page_008.tif
4186fa5fad4498c96b4f4531b64ce2e9
9971d423de230e8e755082ee5b979162a4b64941
120609 F20101129_AABEFS fure_a_Page_190.jp2
8fc2d855b88306e96d5d997d780b3916
6ac36b279a02f8ae640a16d13a03aa2959cd66ac
1051963 F20101129_AABDZY fure_a_Page_006.jp2
01eea0a31c7e2df0edaae9a308051ff4
8d4a0ca2f90a12437e0c43b6ba62c579a339b423
F20101129_AABEGH fure_a_Page_009.tif
07623e7cc95e6810629f1c6758de21e2
65f603cfe5d859c488f47869c26a8cc6cee687f8
124454 F20101129_AABEFT fure_a_Page_191.jp2
a45a0ff118d7fe5e55847cfa996a094d
8a39f098f38f3798e13121a97bf3785780e8ac45
1051968 F20101129_AABDZZ fure_a_Page_007.jp2
bbf7e66189f0de007d2d4472d310e5d1
4dd3d8ca29a83f61d0ff39240f13433521b19ba9
F20101129_AABEGI fure_a_Page_010.tif
c5b7863f6c0ad8bdd47ef6f8da87e873
79be62f441cdba557f0bb430f60cd8f41b947dad
130504 F20101129_AABEFU fure_a_Page_192.jp2
06955b44591b65229ef89af710285741
589ecc6427930776f93d343470de2a683f634e31
F20101129_AABEGJ fure_a_Page_012.tif
4de3394cc77617bbef7afc2b979e2350
9bdd8a31605a8440fb0bdc95b402bd6ca6b1f0ca
119810 F20101129_AABEFV fure_a_Page_193.jp2
2d9ee08e2811fb87b8b574b9f5f6ac51
319047ee00cef45a2ad5cab71eff75d71bc1476e
F20101129_AABEGK fure_a_Page_013.tif
deeb94444afeb9c4e5c136b44ea805dd
c6b2cae520d89e1f079b6eba271361846a488645
119479 F20101129_AABEFW fure_a_Page_194.jp2
3fcaf54c17ba5a1ea3ec0b274764a6f2
a4dca0ceded118b51f0bf36d68a0f09eb6615ec8
F20101129_AABEGL fure_a_Page_014.tif
de28b0e3dff2bfd5641d2f70b4a791dd
f98579a930f42c16e029b05b71c5198b258ed645
134567 F20101129_AABEFX fure_a_Page_195.jp2
99841ef72598ec0c99ea8f058575957c
0559f7a36796e2ae9ad2ad619c4acd8fcc14dd68
F20101129_AABEHA fure_a_Page_030.tif
06c5c216784f0dbe1184d7c7f32340a3
255a30ece9797077f1cb284984d6c22897a7283e
F20101129_AABEGM fure_a_Page_015.tif
6defb5eaafdeda1cf361a315965fbe29
c0121b8b1953d88ab37b7c861e91e1fc2ae8713e
82347 F20101129_AABEFY fure_a_Page_196.jp2
fd876d327921a7843e48a28f31b1f05c
978b41c025fbff93101d7400e387684392c9b160
F20101129_AABEHB fure_a_Page_031.tif
cdb70b59c67faf245c5e3a4d2ca15118
cd4f83bf849b29761e1243fd0dbb759ec57f3acf
F20101129_AABEGN fure_a_Page_016.tif
6cbe254902cad2ae5e39389302b2a6df
3fbba99850f1ada75571b1306fb302a977f0dc39
51691 F20101129_AABEFZ fure_a_Page_197.jp2
c5f940a635e367049167a21e42f7caf0
d944c9eb4ccd58ee457387e369d22b7e90d92d10
F20101129_AABEHC fure_a_Page_032.tif
3f38db783b8d9a230a62078f91d773ab
31d1380752cfca9de4e893def22390af948b1f78
F20101129_AABEHD fure_a_Page_033.tif
077d3454c9274ce5f3d26e913528b2ae
0eab7458243bab035376491c0b0eac9f35bbd976
F20101129_AABEGO fure_a_Page_017.tif
7680e5e013153c1770ab6de196a77f74
0ba9d1025bc365baff8d1aa72ea17b43e826433d
F20101129_AABEHE fure_a_Page_034.tif
ebbc4845d80bc7bdae41839559ebff81
3a5201f27533ea6b292f8df0976d7fc83ae6a9b1
F20101129_AABEGP fure_a_Page_018.tif
dc7f764f1dc5b138f4367d8270c3b98a
db39bc229f8a6bf1273683eb977481c9d814f2e3
F20101129_AABEHF fure_a_Page_035.tif
9a93db07a7eb57a7a9023117de8ad9ae
4f86cd617f5448ea20cc847014538a1f30107143
F20101129_AABEGQ fure_a_Page_019.tif
f5753b7aaf1311ce54887baf5bea06f7
c0f1822f0a3ea1d8fafa4a53ad7bdf46553ebb32
F20101129_AABEHG fure_a_Page_036.tif
1a5069486bba2f9692d3485614908949
c1999236b68488a6f8ecaaa0775cc35a2b9b604d
F20101129_AABEGR fure_a_Page_020.tif
4c166d9d5c125d2580aec83ed2e9fdaf
bb59ddd70a23faf84e3bccd1e989c9fe47598336
F20101129_AABEHH fure_a_Page_037.tif
011430885c04c690089813472cae574b
d75316fcf3936049109d66515137cbbe00e4a54a
F20101129_AABEGS fure_a_Page_022.tif
642f85d8c17d15efac2308471d6fefc4
8a7bf79d6d482d8dae4d6a3e0823f27a4e6e359d
F20101129_AABEHI fure_a_Page_038.tif
8d458440cb5c949907e473ba35608b01
c2d973293354bd9795f1c6752d9065d9b64eb5cc
F20101129_AABEGT fure_a_Page_023.tif
d2f8aac5b7a906a3b3dab5982f25f5c3
5f0d36eae2de54864e2d64b2a3499fd01f8dc427
F20101129_AABEHJ fure_a_Page_039.tif
9fb6f32caa7c677415d4c99f8ddd2859
72a6aaaccbac10aebcbc68a2791545dbb4bfa47d
F20101129_AABEGU fure_a_Page_024.tif
355bdc42c7fcc7923ad6bb7b681ad02b
fc78af23b4c14dd4c903ae0d70d518f7d811002a
F20101129_AABEHK fure_a_Page_042.tif
4d92caee18e4e5ab52b32c97cfd06a0b
cb9f9e2102409adda2d23ebaf6e9462f9f55c5b1
F20101129_AABEGV fure_a_Page_025.tif
0cfbef162e042f7a70b460b55403d7d9
d45b06ecbd8a93f5abe61a66eff0abfa139dc79c
F20101129_AABEHL fure_a_Page_043.tif
0efc18fbe04bc3b27d3f5abfae30f9a8
a45a2ad802e613a63a98e948c79dbba8534b82a1
F20101129_AABEGW fure_a_Page_026.tif
1f88081a0bd8de1a0e16608e9e5ba0dc
1fac6b6e2e27425abda564bc1164413f252f768c
F20101129_AABEHM fure_a_Page_044.tif
d9934fca50f4ad9708455c9ad895133e
7e16de6439da20e122b28851e1e2dc78f9a3d6c2
F20101129_AABEGX fure_a_Page_027.tif
9532d10af21ae6b09a20969aaca8d324
1d349ac7865233bf236c973436355ee61ac40f21
F20101129_AABEIA fure_a_Page_061.tif
4d292f126ce8d153f91c5d972526a897
65a0ff5500b42251f58a2f064332f5de95276988
F20101129_AABEHN fure_a_Page_046.tif
95b706ed1d30f1d54fb7dfdab588cdac
9c098905f3d1e3693ab09e0b4c09ed57deddacde
F20101129_AABEGY fure_a_Page_028.tif
c4b2435badc99ba672fb4117904df2ac
8e7de1268c8d59beb940e2948e2a53ceac7e01da
F20101129_AABEIB fure_a_Page_062.tif
536bfae9bacf8f70fa333989fdb864af
b0a9d9048538c4d96b344a79bf4129e1a28f5b78
F20101129_AABEHO fure_a_Page_047.tif
7323dbbdaeac53c8a2da675e576757ad
7c10b0f9355659dc776f73350ea072044188a38d
F20101129_AABEGZ fure_a_Page_029.tif
90a56ed7124d438d366eb7a9c18d121d
49a9811e4134b0ab60d73ba5fd8b59d3c0089e61
F20101129_AABEIC fure_a_Page_063.tif
55dc3f75ea70d09a665282418ff5a44b
da2d4108faf9b10243047f32a13b7528ff7d5116
F20101129_AABEID fure_a_Page_064.tif
374dc6c2fac3cada571b8cd7eee748a9
b26eb819ac73debea67437b04fa8586714e13558
F20101129_AABEHP fure_a_Page_048.tif
ae68c7822ae1b82cea31fd93106f6c37
77fc0c2981ac30e251c4ce838d804d1656413914
F20101129_AABEIE fure_a_Page_066.tif
0cbedc15b93541ec050cb1bb0d129dc3
86868339f80009792550bf31547a4fe1f58b3d4b
F20101129_AABEHQ fure_a_Page_049.tif
3a355ec08db336d5673b378798b8538a
8f5b1046b783c825e829d740fefc5dd64ea32aa3
F20101129_AABEIF fure_a_Page_067.tif
54ccb4961a1b4670d5b534f6942ede1f
dbeb0247182287b764c5b40980c30b74b9d07a30
F20101129_AABEHR fure_a_Page_050.tif
03cd6b587de4b3a2fc001641e727196b
e8844c14442df42f67e37ef93950394aeee5f273
F20101129_AABEIG fure_a_Page_068.tif
4b2d26c3e13aaee0e66732eb7153d42d
12f45794a0056c1daf231b0ac2008315976c4140
F20101129_AABEHS fure_a_Page_051.tif
076f9c7b3f9168099666730c95651ddb
5ae5e7bdd3782d2697137c36636f1c93b53eac4a
F20101129_AABEIH fure_a_Page_069.tif
c0a9d1bf98f6b6e924d00cf467382a58
c4b791b0ff5b0bd35eb80996a8efdfc5eaebc86c
F20101129_AABEHT fure_a_Page_052.tif
690b280dff9d5182038052042e102ab9
aecc01bfb3178bef44ad1a8474cee517db7d137f
F20101129_AABEII fure_a_Page_070.tif
b29c2ea07a1fa7c141ca797725f5c88f
80bdedac56b77b6545ebbb3caf7acf2ca6e2e944
F20101129_AABEHU fure_a_Page_053.tif
01e0c58866929b5f85f9b20e521dfc35
79cf9841b72f92f4ff7dbcef781ee459b2e8aec2
F20101129_AABEIJ fure_a_Page_071.tif
9d3804b398506112200a6cbb0552eea2
377cdff144eabec56e1beedff5fd8a0784cdbc60
F20101129_AABEHV fure_a_Page_054.tif
c72ae1f74d3de12bfec17b88b6091f47
996affe7cabb94210b3e6ff43d0f7327ee536ee4
F20101129_AABEIK fure_a_Page_072.tif
7d54e26655b7e94d91de6c0178e42382
6925ba86aa64fef92a64e29992cd509b1711c477
F20101129_AABEHW fure_a_Page_055.tif
6a18bae249b210e746acb3b3dfab001f
8fa0ac267c22bfd3c2d8856bd1a36f0ea784d7c9
F20101129_AABEIL fure_a_Page_074.tif
0aebcd196dfb49b62ea8be603957f93d
84d4053c4b51e4fb71bcecaebaef385173bf399f
F20101129_AABEHX fure_a_Page_056.tif
1fea916e6f4220e2523506987c64e19a
15ccb979384ae12ce867410339e25e3d428c16bf
F20101129_AABEJA fure_a_Page_093.tif
1bbff8ce0ded313cf69400a0b720f745
7bcfa1eb79ab522624c6ddf9010a91cb0d1cac58
F20101129_AABEIM fure_a_Page_076.tif
e09bb3176d56d0f9a428f97659f0c66d
2908dc61ca468e55d3c9b2a269750b18cc8f0fd3
F20101129_AABEHY fure_a_Page_057.tif
b2f44fc27f4e347024e4d066f6c1d31e
076e9105951a2fee9a1fda9860442d40c2b6e2d0
F20101129_AABEJB fure_a_Page_094.tif
573065bf7a674a19b68af1a0fe960d66
6cb298b7a196fd4e1e54343b88ced56296d8f806
F20101129_AABEIN fure_a_Page_077.tif
14ff079540887cd0f35110d1c58e7f0b
13743d195ec6f550ff050f071932e3d2ca0898aa
F20101129_AABEHZ fure_a_Page_059.tif
9d12e3fe4a70fdb649a799148116916a
9469ec679c879a9f8bfd4289b6496f1704b567bd
F20101129_AABEJC fure_a_Page_095.tif
224673188deb97fd3e93c522d97a38a9
4a0eb7aaad382f0c28427e385d15765db7c73f84
F20101129_AABEIO fure_a_Page_078.tif
0e82ac7482c5f87a27db00f0106a2422
2e05e62cef9c8abd794f7c80cef46cb5ab5df446
F20101129_AABEJD fure_a_Page_096.tif
b7800a05787796abcfb0b68a42f3bc01
e91a62db712a5427c428d6250efb8137ef8fc710
F20101129_AABEIP fure_a_Page_079.tif
2d8476242e1b8eb4523709b7228ee610
4654f5ecdf032e29ce4466c0de9fbcd6b268baca
F20101129_AABEJE fure_a_Page_099.tif
5049937487cbcdf34c04803c0d92ea47
a930abf9b7ec9bb0c9205053a05b2e89237eb6bb
F20101129_AABEJF fure_a_Page_100.tif
669c59457dfad369b31640cb7dedc1ea
4f7ee4c32354ed14175529d0d0f4288711282215
F20101129_AABEIQ fure_a_Page_081.tif
cf8c449f50fb572b8db8aeade504e0cd
2d0adeb2df2ce58ac873c870b75c308a339ce496
F20101129_AABEJG fure_a_Page_101.tif
ba6c94f9e8989cf7f45c3bbe765a590a
496c792459a2c5f02fda0705649f5e87820a3387
F20101129_AABEIR fure_a_Page_082.tif
c4f5e91f15c485f68431b5db971f3e15
4bb88ee151665fd1619619525e8e772d37f0e705
F20101129_AABEJH fure_a_Page_102.tif
55d3efd151ac19650df57048235ac38d
e2299020a98fde843571bb9da9eb9022e2c8b4be
F20101129_AABEIS fure_a_Page_083.tif
d37fc15774f35f5722460af1f0d867dd
34b37a62e64e72f483c5b59d5550ade41ba7331d
F20101129_AABEJI fure_a_Page_103.tif
6fa9bdd59b20577ccf6402aff7017400
d24272a60f7411a9adab9c9d69aea852af7a9f73
F20101129_AABEIT fure_a_Page_084.tif
20773ac7f382dabd4cf401c40a595787
59ad1e8da70244b0965fd8edcd7fc412248942e5
F20101129_AABEJJ fure_a_Page_104.tif
9ff2aa9768ab23728eda31e4e701f25b
16812db704ec9b2770a5e6a1cd30547363988d70
F20101129_AABEIU fure_a_Page_085.tif
8ed4e4e0f66507cbce1679177d5e7243
18f80ecbbb3fa8cd3a2c3924a719e65c11341d4e
F20101129_AABEJK fure_a_Page_105.tif
1ba5fb1e16445288c669db957feef352
e90bef1066f45fcd2405e729c5aba088f8cd13b6
F20101129_AABEIV fure_a_Page_087.tif
2ea89fe42e10cbf7abe67d84bdef9373
0f5204cd133c459583f3134a5c85061767795f92
F20101129_AABEJL fure_a_Page_106.tif
92cef193233f2f805fc5bd7982249eed
1211b27825554bbd295144721b1d661632b86f25
F20101129_AABEIW fure_a_Page_088.tif
56adb977e417852d3b3800e364e37b65
50f17b2661b3dfc6bbc8dad05f75c779aa29ec61
F20101129_AABEKA fure_a_Page_123.tif
02bd2f6f8fdde207288471f9cbc0d800
49480a76080917b7c68f721d9789bdcac012fc23
F20101129_AABEJM fure_a_Page_107.tif
b34773b5fabfb722706797e75ee74643
0cb839fdd1e909ac60f0507274ac3316934086d6
F20101129_AABEIX fure_a_Page_089.tif
29e8106979ba1eac93513cd9557ff570
3f177e2c172e074ebb2194ea46b3e6d2eeedab1c
F20101129_AABEKB fure_a_Page_124.tif
b45ad73b88bf8b26cf5fc3619e055e65
d5f6701228f8579c96949c7b083ddb6d7c1aa40d
F20101129_AABEJN fure_a_Page_109.tif
7c6b0b0774e8d4d3c533e749f6c9ada0
5aa67edaf61176728c3aa5882c1d5a302d65ea68
F20101129_AABEIY fure_a_Page_091.tif
d575bd866980757eedc0d715f8cf9251
ad9a41c38c4717a66ed34c15c3a09ca6475600bc
F20101129_AABEKC fure_a_Page_125.tif
a7a5e2a238adddedf4d16aa2239ea0a9
83ad76cb3a2426a97c28cd7710bd726c74d399ef
F20101129_AABEJO fure_a_Page_110.tif
9aa758d0e8c2611b9d1a3b72bb625c25
9236e9a89a256aed925be624ef8a43c0cfb3e7a1
F20101129_AABEIZ fure_a_Page_092.tif
25a6b997c26ade73469bae38a58c1403
a94c57abdc3f287f87c29bbac318a95a7f58b744
F20101129_AABEKD fure_a_Page_126.tif
88a5c8601fef737fd7481bb283d389c3
e0372f625bfa523b60aa05d0e619ddfddf0c8d88
F20101129_AABEJP fure_a_Page_111.tif
ef167f4dd5377c53ec341fba278afb60
576f526b9aa4a90eb378e5cac27f2aaa2059a03e
F20101129_AABEKE fure_a_Page_128.tif
33dc24132908bc5a3e1a9cb53c65ec79
c795c87a8586aa4185ce11bd8e7e44c42e5650ec
F20101129_AABEJQ fure_a_Page_112.tif
715dd0b83847ea0f993e36ba1b106d6b
13308238d4cf5bb121759911c8fa0efd3926d26a
1054428 F20101129_AABEKF fure_a_Page_129.tif
35a188d7a5379ea580fbc9757152d6ac
4c9efb798905d1d33b729ef99e02e5dbd5cf4b9e
F20101129_AABEKG fure_a_Page_130.tif
47896e9f50e84156915378f5e57a6efd
69694f23ffd0b3eb78e7aef6fd5911d62887671b
F20101129_AABEJR fure_a_Page_113.tif
5c44b691a71ee0bb9ab5ab14c45f269d
420266ec70a7867c26a6b23b26cd9dd009c9ad3f
F20101129_AABEKH fure_a_Page_131.tif
06a84573ea4aab8e71ba6caadf8687d8
2afd276e480bc96a65f2ead57fdeb1df0c74c0fb
F20101129_AABEJS fure_a_Page_114.tif
30ab2cbbfae475300c0906ad4ae55471
b45d62f73e73eb7b9353a588e05efd09a6bdac38
F20101129_AABEKI fure_a_Page_132.tif
8760324c73c91513689e961fc99f90c9
53aeb5beb38ff2f77ee9a6d4c3beaffe80767f56
F20101129_AABEJT fure_a_Page_115.tif
1e83875853a5a8ff758ecf66761dcede
76f414449dde8c192748851556ddf7a7de69b0d4
F20101129_AABEKJ fure_a_Page_134.tif
43dfaa50424ad9cde75a6c139d39c0f5
24018e5778c735752cb1f93fb635776eb33cdd50
F20101129_AABEJU fure_a_Page_116.tif
9b8a040ee6ac12f3d03b4d3d6360494d
f01993dbab1a9e29b92b7bbc5e40a00d0a4d1ba2
F20101129_AABEKK fure_a_Page_135.tif
b195c03bdc6d22d5f59019332e97a3c7
4fe44249adadb4fc56d4e9291060c627505dca94
F20101129_AABEJV fure_a_Page_117.tif
57d5a5b0e6da86053ebbf0314f766840
572acbc32ec7025c153486dbf9a1d25301ac5892
F20101129_AABEKL fure_a_Page_136.tif
838aac66d6a48527f89520138a2595e9
99ad5307ff7097e9b31987faa17aa5f2abe31e5a
F20101129_AABEJW fure_a_Page_119.tif
1665b1631f597104f38a3a64bc15d9e9
aff86e7c070072b1afa129ea33258772c0095f27
F20101129_AABEKM fure_a_Page_138.tif
0c6ab025ae6707512b836f84762ad707
823d6277743a4dbd25a727040ef844ce05b6d7c5
F20101129_AABEJX fure_a_Page_120.tif
b122a88fa0bfd83ff8cd39803d5ef260
0dc3c6d0419fd0481e6db89e265941cc1d7fd623
F20101129_AABELA fure_a_Page_155.tif
23f31d1dd19f80234b4808e3b01dc86d
a79637c7bf62d38d1beae3888fb45704a36725b1
F20101129_AABEKN fure_a_Page_139.tif
159969fa27145790249ff08fa535e48f
16dfa1a2c5c3e008fd4e12fcedc9215c21bec9a3
F20101129_AABEJY fure_a_Page_121.tif
01e577dfb3497216c753c691bc8310cc
5ad6e4f36292c6e428ef1114574f9c15b2658056
F20101129_AABELB fure_a_Page_157.tif
bdc311fcaceb5f0d8749172c08392d54
7247b68c9af086481154f3cfba8a876b3f625b5e
F20101129_AABEKO fure_a_Page_140.tif
e0d479f3cdbe6d99c3ec0959e42a3e46
01b58b11479614c2bd66610050a729704aa7fc76
F20101129_AABEJZ fure_a_Page_122.tif
6f8cc476a319037cfe54df3ad1551210
54d9312e0ccc2ab5fbbde4350dc36a930aa920e5
F20101129_AABELC fure_a_Page_159.tif
64a8aa7a87d1f168aadc343bcfac87a8
4771241a6a2e6a40672ae1022f1f6b4ae829c7c3
F20101129_AABEKP fure_a_Page_141.tif
c82709865dd6eac8ed52207e137a0c1b
14e97dcfaa9dbd33aa6e71b14a0ba0ba969cd031
F20101129_AABELD fure_a_Page_160.tif
06cfc3b60ead864399b0cf0a50bdd2b5
0d752324825f6b8f23bd238c060940ff00600fe5
F20101129_AABEKQ fure_a_Page_142.tif
117bc448d16f9a97f1ce52eac84f6af9
64cfe9afe70caea018d81ecd81dc152af0c85dd0
F20101129_AABELE fure_a_Page_161.tif
56cb24bebaf2b30a7ca17a8165ba74b2
e381bb8d6e1393afb10ccc09c15c511fe9334330
F20101129_AABEKR fure_a_Page_143.tif
31962861a6dd4157a75bf2d042063987
1d1ce50c48b1b7613f21b4d689a4489752e0f3f5
F20101129_AABELF fure_a_Page_164.tif
2ae554d972d6581d52c47f74335f35a2
28f995bfcf0fb50c4487b2b9072b9ed2af474bc9
F20101129_AABELG fure_a_Page_165.tif
d9ef99cb4c800046260f4e71881bf07b
d45687a5adae01dc17b7da0cf4ece27e776288a3
F20101129_AABEKS fure_a_Page_144.tif
173326e507da1a315652f58f0a0a3cb6
b813b5c256a5a350da96ccb881b4e1ab9b757bbe
F20101129_AABELH fure_a_Page_166.tif
58dd1865fe7b12b5b38ec734b3ef0816
76217d828fcd4c5b2408670b2eca8e7eab4e626a
F20101129_AABEKT fure_a_Page_146.tif
2e39ac18d72f19cff441357a183db7b7
0ae06db521710d4d5cffda309d07d8f4db04c868
F20101129_AABELI fure_a_Page_167.tif
4a501454ab8c6aecbd9861a8d35feb39
393a0a4e8790ddf8355393b0c1f436edb4daec1e
F20101129_AABEKU fure_a_Page_147.tif
865d296da0fadce08299f786590bfdff
ae474a285df54950b6aa6adadb19a9db6c8743f0
F20101129_AABELJ fure_a_Page_168.tif
d90ba40336a9f85542aab721573e92ad
5e3f6f8289ed7a1d14620a18700bcfea20054f0b
F20101129_AABEKV fure_a_Page_148.tif
95414d74828b6bbe0a5caecfee39ae2c
64e8331a2f1d52a4b624201de96e218f02dbfd0b
F20101129_AABELK fure_a_Page_170.tif
9b002b43561c944c00d48ef558974d69
cd4cc12371cd8efce7f1ecc3e7f4147e112f5a2d
F20101129_AABEKW fure_a_Page_149.tif
04449a260ba1b51619542e25eee032a5
7fdfc04512bf26beb87d5b64afcbb4a275bd3133
F20101129_AABELL fure_a_Page_171.tif
abb3f0d646b7fc8a6743aa8ffcfcfa41
5fb65c08c7393d8e7a270340d958614482830965
F20101129_AABEKX fure_a_Page_152.tif
750c150cc26229570806a44a005a033b
6ee196624cd5432020f6ed59af8d6a994fcff43c
F20101129_AABEMA fure_a_Page_191.tif
9dddf66a945a4880da1bcfeedf36174b
809d48f5cfa87b3db0be8dceeaadfef4994758a4
F20101129_AABELM fure_a_Page_172.tif
1d3f6ef4a1a864808fbc7a27aeadc530
36dfb98dd79eb885f3ec735da724867469faf563
F20101129_AABEMB fure_a_Page_193.tif
9284343723d29961b542f791a00807bd
2726a35a04948ed25db78026a80c0c325f27d8a5
F20101129_AABELN fure_a_Page_173.tif
f42cb5ac90de31c9041890beee1e8bd2
427174c041582b052556637ac7d56ea97e4d860b
F20101129_AABEKY fure_a_Page_153.tif
c3f4c62254b3bbabfad4089324e25518
e4867a7f775cea6ea5dab71c00a5fd5bcde262b1
F20101129_AABEMC fure_a_Page_194.tif
f1c7c1d75526d7b9f41db09e124372b9
18dce5983a732d9585ce5008239f3bf35fd38469
F20101129_AABELO fure_a_Page_176.tif
9c6b840d40adbf88e3efe73f05f067fd
25a4498123c4959e7a391e0d151b4e5a4d4806cb
F20101129_AABEKZ fure_a_Page_154.tif
9de62ef322d34bf8f7076357d91c19f6
d5f198f0f9a9d29fca576dd833abd808918e8e45
F20101129_AABEMD fure_a_Page_195.tif
bd957ad64f396ab98f1747598af2969d
98588cac2ec9502f295ec971e5eb985f8d768465
F20101129_AABELP fure_a_Page_177.tif
6012b3c5d1fa0fa99ab5d31d683a0c80
9a9fbb03147b8b62ed8dbd5bf6eccaa0134408d0
F20101129_AABEME fure_a_Page_196.tif
7e45abf3a874819a657ed1f35b34bda9
152f9bdf0948dceba3fffb50c63fb55f18e45f7b
F20101129_AABELQ fure_a_Page_179.tif
9a3a137b965553963c138d75cec13c30
6f91164840cc5412dc9978187c86fcb20e055816
F20101129_AABEMF fure_a_Page_197.tif
9402eb305e0d50dcb8ebd505dc524015
3eab1ac8d48324b37ae1bb53f3f12017b6206451
F20101129_AABELR fure_a_Page_180.tif
9c16813575d6e870b95a55521ce447a9
6811e38bb4379a90f71b5aabe905789feef32093
2416 F20101129_AABEMG fure_a_Page_001thm.jpg
3af6eb69acd020410e346f80f006d053
54647de5851489132b12a408d251570645f49cc1
F20101129_AABELS fure_a_Page_181.tif
af3290bb1a3eaaf58cafd5fcec9f3b87
e3cf78a73f9403ff953721d1583b1be2d9ae47c9
2906307 F20101129_AABEMH fure_a.pdf
c009b3aa8af3271d7838d0ac6d5f144f
3454951dcc8be8a7f24af668cc1e0019b2acaa30
7382 F20101129_AABEMI fure_a_Page_001.QC.jpg
661af2ad8136e9d56c232e1d1c65d97b
90b07d70f2740c9d38d5d1d29449778f136907d2
F20101129_AABELT fure_a_Page_182.tif
c02f13ed0844311f1e65d6c249339ae4
bbfdc9485939e79f43f789d4c82087b4b3cb2768
3260 F20101129_AABEMJ fure_a_Page_002.QC.jpg
59cdced95cb32f6afd1957a5f0251576
e1d779979af7be61de7f2aae47a9c9b0ba754d09
F20101129_AABELU fure_a_Page_183.tif
dc4fc1ff2f73c24cb9c2324645a8b473
d899c910d5ecf32f289be5ac18131f99b5c4b77c
1357 F20101129_AABEMK fure_a_Page_002thm.jpg
9bdb837ad6ffb77b22080669b7b0d4d5
f9f739805e838f2622d9239b00ea9bfa320d56d0
F20101129_AABELV fure_a_Page_186.tif
3065f965356c8bc56acdb7d086214aff
7fe6c247e2a95cfe59ca482d805c3657e804ad3b
10170 F20101129_AABEML fure_a_Page_003.QC.jpg
310ac1c7b3f34d18893b5a91b5a59a50
624c57fd15fd6b8c28bbff4970dd823921af77b5
F20101129_AABELW fure_a_Page_187.tif
202e0a79c2f4b624bcaf9a72c5458df5
31e007a96929d6267c8301aae93da5403edabdc5
3168 F20101129_AABEMM fure_a_Page_003thm.jpg
a97e6e7fca2d6fb3cc13cdc3245b10f1
fc421f58002414ce1b062dd3d2cf909544d4d5ad
F20101129_AABELX fure_a_Page_188.tif
5b8bd4f0998308e75d729e99a89b4ab9
bde5013ccdf3cec328b61a9a08206a54575ca8ed
2678 F20101129_AABENA fure_a_Page_013thm.jpg
ef601c8fa0e3635d220696d4057f6ef4
69b070aa912258978f53a97bc6a47652baaacef6
20624 F20101129_AABEMN fure_a_Page_004.QC.jpg
e5a8385d54a4f25052d8f13c874f9489
1dea4e4f8114c69a9db537c0eb4d4669f4fcaf9d
F20101129_AABELY fure_a_Page_189.tif
07219e0c3b13465a3c684bfb78bd26ab
632e5cb6d4df577553d3affdf4a80ab23bedd367
15608 F20101129_AABENB fure_a_Page_014.QC.jpg
ddaa840a11cb645275945932656a90c5
ade975f2c66e6851b7bd6a450e3c68a9d321ff8c
5285 F20101129_AABEMO fure_a_Page_004thm.jpg
41e1c68d9d2a50779629599ba10fd9d1
4800ee590c8865580dd92a1aed09478fdf0f4ba6
F20101129_AABELZ fure_a_Page_190.tif
08a82b45bd9d66e59820c1a555ba4450
1aa3d4d4bf7bdd967c0126267d05f66781f430e4
20248 F20101129_AABENC fure_a_Page_015.QC.jpg
eeb2507321796e1134f4b58cc2a69e91
4357d15e81508cf71d40d13710e15fde29db1b93
26327 F20101129_AABEMP fure_a_Page_005.QC.jpg
89ec837465bed864c999f8eaf1a5c1f7
78abb57e5e0cc4fc0465996c70cf8eb95610e9e1
5535 F20101129_AABEND fure_a_Page_015thm.jpg
e57b26586c6fb4c4de676d625876e2fa
d238b3de259b097346eb4639e18c24993c9f9cfa
4713 F20101129_AABEMQ fure_a_Page_006thm.jpg
c64bb811f745156d4fe86c55783c3f28
4f7fdd7151e16bf76e730354f968294488d096c1
22973 F20101129_AABENE fure_a_Page_016.QC.jpg
efe1299a53c897723a6a251a993a1133
e1d2c7555923b67457fa13a04dfd9b9039069200
14968 F20101129_AABEMR fure_a_Page_007.QC.jpg
097beed9d10e21bda38c64acbaa63352
98bdfc1d61c1ab58e59f853a7957730a7f231391
6374 F20101129_AABENF fure_a_Page_016thm.jpg
0b7cdb8a0d53e12e0980e20d2a411753
6f4b4f101dbe4a35ac44059b3ab696abd599337c
5695 F20101129_AABEMS fure_a_Page_008thm.jpg
ccad6f8cbaae96cc5e63835a2065da76
f369c66150397ae1d64baf90acf43da4db1fac2e
25260 F20101129_AABENG fure_a_Page_017.QC.jpg
c1fc4bd1ad3634645362122e01403df2
82056d44e994d0ae154afc11fd7b9b9e78e32004
26013 F20101129_AABEMT fure_a_Page_009.QC.jpg
bc019d46a429e35a3e2ebab77e7979c8
d7f3af7844be3dff14b02e563f9443f72e0dd8df
6778 F20101129_AABENH fure_a_Page_017thm.jpg
49c6524f0ce4f38b62578da643f5232f
5241dff10188f1f4bbfad9bbd4626c3c31cd1017
21455 F20101129_AABENI fure_a_Page_018.QC.jpg
959c037e3eede2d473a4eb0186736874
456016262f116de45f7c426d3c15e0fac7c1c05e
6881 F20101129_AABEMU fure_a_Page_009thm.jpg
c6e4e259d342a6cb143793d107b47b55
c6a4992f3ea8a912a856eb5828f5e7051df7ca8f
6057 F20101129_AABENJ fure_a_Page_018thm.jpg
dcac3f3a27e53c0627744f2638bdab92
6dcdc0aed99dc941059f07164024cbbefdc0b59b
26116 F20101129_AABEMV fure_a_Page_010.QC.jpg
9fe48654d8b6561cc9082a177bbdc0b9
8defe8d4720239e6a1bee9514d094525bfb947ce
18113 F20101129_AABENK fure_a_Page_019.QC.jpg
2ceb5205c86237e7bd123ace5b056437
37b5ae91f07b046816f764e68e2a0c43a0364fdd
6693 F20101129_AABEMW fure_a_Page_010thm.jpg
d9ea8d292ddc55a4d4c40b40c0ca62ed
b4420e835437bbb6c444819ad7a56ed3e2df3136
5188 F20101129_AABENL fure_a_Page_019thm.jpg
43a4850e9d0cbcfabddd1cfefd8eb713
da64cadb14334d2f30938f27a5e27a3742042826
22155 F20101129_AABEMX fure_a_Page_011.QC.jpg
34b2f4c698ddfb01e1f3a5e339ece20f
e82af497426c7e449f1849a5ea796d29fddeef4c
16895 F20101129_AABEOA fure_a_Page_029.QC.jpg
e4f2e83451052c4da3ae1aa186b007ae
a21074e1722baa0c3b9b006a4ab8a97ecc1d34c8
16916 F20101129_AABENM fure_a_Page_020.QC.jpg
29dbeaf2fb681d4cebc956f18fa1367b
e4edcec784ff29a5b2b70198bf25a05978b1257a
24916 F20101129_AABEMY fure_a_Page_012.QC.jpg
052e9b8729a755e11bddd07341f787ef
fc83ac1ff2bbd17487587b92dbc1bf5f509be55a
5477 F20101129_AABEOB fure_a_Page_029thm.jpg
87d610b190aa6e2535f89adaf734405a
a3b7d3764d12c0ccfc5dc0883feb69cf4358810e
18990 F20101129_AABENN fure_a_Page_021.QC.jpg
68926b176a9708b11b2d53d461bdb89b
2407c55e2f2f378af42424b9f8d7c1ee70d94047
8433 F20101129_AABEMZ fure_a_Page_013.QC.jpg
cfdb6ad430251f5f72eb598c3265ee36
1a0ed40fc920e098eb36636fcb294a7dc6bc11d5
12621 F20101129_AABEOC fure_a_Page_030.QC.jpg
0a6c1c83a4f61ae38c3d4921c459d857
022531a06653e7f2ea3d0a2e7610701013739818
5538 F20101129_AABENO fure_a_Page_021thm.jpg
c8a87af62c526bdbfae50e60192cbfb2
8c1053e5a398bf4d6b3384895f76c8f972e2d3a5
4444 F20101129_AABEOD fure_a_Page_030thm.jpg
1580ab22c946807309ecf2b04c702c49
7d918387d4729ba3d3c08ca25738d04a919b0934
22806 F20101129_AABENP fure_a_Page_022.QC.jpg
1e76aeefface67a3a63c7b6d21776119
46cb2528690287a1307580006a68c95d7011f2d8
5850 F20101129_AABEOE fure_a_Page_031thm.jpg
6a6f4f777af1b2b8b41ff1ec36b4b4bb
ff5564e8df567e6544bc58fa16b1cc3f2382b7af
6321 F20101129_AABENQ fure_a_Page_022thm.jpg
f950ebbf35cf290b4428e33dbb62ccbd
ee280be41667b3997ad9534ead2da32a6b1e4b7b
23115 F20101129_AABEOF fure_a_Page_032.QC.jpg
635681084177d38b994f35c778eed86e
70076484c8244449b47d1aa7360fdf59c10862db
21548 F20101129_AABENR fure_a_Page_023.QC.jpg
46788730aab3a3185ad28ceeb009cc40
e1578d7270c4a5a40784e381e0a74960ee6686ae
22220 F20101129_AABEOG fure_a_Page_033.QC.jpg
7ec3be98de38f6c6bbc5c9c96239bbf6
b451aa192562041aef5a077669f3b7c48a18a1ec
6152 F20101129_AABENS fure_a_Page_023thm.jpg
d00ddddd7de760d562c2a8c70c4f519e
607ff9850fa1b1dc388203f7853ab60fd65692cb
6244 F20101129_AABEOH fure_a_Page_033thm.jpg
97277f43514efa3c1615926bd7677977
05e05f83c007941476065a0d242a3289f655feaa
16976 F20101129_AABENT fure_a_Page_024.QC.jpg
05976e0edb26246e7611cc3e509f038c
1aa3b07ad852c18f58ff238283c8b86c9883dc04
13230 F20101129_AABEOI fure_a_Page_034.QC.jpg
81f708a9f36db43d7040bc7f455a2b97
f9aa919d78bb72d86ee94408ef122c612a075936
15926 F20101129_AABENU fure_a_Page_025.QC.jpg
c82ce3ae5f39819d5a22cc7dbb1d0fcb
66c8461360abc10e8109adf92b5ae5369aa3fee6
17332 F20101129_AABEOJ fure_a_Page_035.QC.jpg
cfc68a24b72203ccfdd85f16370de6ef
c6b8e4711f48068994b583ed870d60cdd7db81ea
21621 F20101129_AABEOK fure_a_Page_036.QC.jpg
0f7acc005c3acec6b246b32865f034c7
458f467be62640e547f7e02751d36d60fbab5acd
4591 F20101129_AABENV fure_a_Page_025thm.jpg
1b78b936d5abee66c986074752ab6dcb
1ee2a218a0fe1a6b3ef37e4a725d8f3e5360c1ab
6171 F20101129_AABEOL fure_a_Page_036thm.jpg
2008fd1dadb4b7912fa10dc0c186ac70
c6f4706bc8b9b317522d42ebb89cab80b413bb79
21092 F20101129_AABENW fure_a_Page_026.QC.jpg
eeebc4f016f6877cd55e44cb0e3f51ca
f127aa85c79fd131d6b79aa708993127b3d1393e
4269 F20101129_AABEPA fure_a_Page_046thm.jpg
20c608d0bfb22a09ed024bd0c4069118
4fc8c3e88a93eba8f25684dfad43c4919910b757
16591 F20101129_AABEOM fure_a_Page_037.QC.jpg
c915e59f0bd265a737b58ead5fae40d5
5a5a2c3ab9d49b2eda3c899c5211e9fbbf3a1b14
5842 F20101129_AABENX fure_a_Page_026thm.jpg
490636ac40871a30533b051dd16376e3
7c495d936ab2b19b9b94bfa23c308ac846df4c67
23314 F20101129_AABEPB fure_a_Page_047.QC.jpg
9574aa36692209cac7a09907c3ab50c5
77bb67d0afaf391da1ed928479a0bfd597390e14
5861 F20101129_AABEON fure_a_Page_038thm.jpg
72a4a0b54674b1697857a4f05d12c519
2b74bb57ceb5a1bfc736fb5cdcd07a974ed093c4
21206 F20101129_AABENY fure_a_Page_027.QC.jpg
5896454d1dc7775836c0f8360ed8be48
b5248f55137380371e75584e122f14710bd192de
14035 F20101129_AABDMA fure_a_Page_096.QC.jpg
e99f52f1c90afbe6941912281eaf6426
ad878db5e7036be66f4e4fd3a93250fb2e3274ad
6377 F20101129_AABEPC fure_a_Page_047thm.jpg
7a526153466e2ec236393a368097378c
a6463c08e238822ae32f1cb7c1a5631f7530e4aa
5882 F20101129_AABEOO fure_a_Page_039thm.jpg
78a1e86efffce02a79543f635bdbdfe8
4723ef3adeab455c681c22ebe2705255f741db57
6492 F20101129_AABENZ fure_a_Page_028thm.jpg
47d2541ae7587dd9ef5562f006eb8e24
6013df7cf9e3ba2448a5b036e15601ff69a4fbf8
822846 F20101129_AABDMB fure_a_Page_158.jp2
c3aa38faf13645c426d3ef2b7a2180ea
9c80592f5bfadbaf6e7b12f529664c9457a5d056
24560 F20101129_AABEPD fure_a_Page_048.QC.jpg
b421d53c91b06eb22e6bc763e6893c1c
a857b01b1256763341e08b5a05455a18900bedee
15303 F20101129_AABEOP fure_a_Page_040.QC.jpg
4da22e1489fa739810f6b00e86ef23a3
b3e4cb8c0e7fd278b289df79d70ea5118a02db8d
F20101129_AABDMC fure_a_Page_006.tif
50063b44bd95ed4332186fcc9bdaad1c
baddc2535118b0ab21550c708e5424e0e76deace
6762 F20101129_AABEPE fure_a_Page_048thm.jpg
005b67b29c6e1dde9eb1ce587d7deace
29d195692008bdfa5a86786eaa5068845772eb02
4703 F20101129_AABEOQ fure_a_Page_040thm.jpg
593b60221bcfa0b36e9b94b8ca065ed4
14e3daf02484b09b0ec543f30043adfeb2c4b658
4610 F20101129_AABDMD fure_a_Page_146thm.jpg
48f008c9036c00c1581b410a2066008d
14c255d574c78843fb16c0b30dabb3958a713ac4
6146 F20101129_AABEPF fure_a_Page_049thm.jpg
20fea466ca0b9e97f3f953ca3721a5c0
39bf1b50c4b5e70fbea41160289aadd22833c9b6
18437 F20101129_AABEOR fure_a_Page_041.QC.jpg
e8ad8945e546b723dcdaa144efc2f67b
4f501dba348f776aad8d47be2f4c66d0b5185843
5013 F20101129_AABDME fure_a_Page_116thm.jpg
ec83109db2cf48d86edaffe8f8031b13
82bbb9208e74b8b9363a60e9d9ad56c076dd37e0
6149 F20101129_AABEPG fure_a_Page_051thm.jpg
3d3b93a2d4baaf7ff1cf7b9f50d9eff7
acf5e6333303037eb40f1bee990cf075da605e47
5588 F20101129_AABEOS fure_a_Page_041thm.jpg
66cbe7e6d3477155d7a8810237ac8d87
be1e8911b554ee1139cced725a9b3d7919f08814
56638 F20101129_AABDMF fure_a_Page_024.jpg
8798697b8c7a5565875dd54ad7f9c1f5
b8d7c2ba41d38a68c2359ae20da1b7858939c68d
14967 F20101129_AABEPH fure_a_Page_052.QC.jpg
6964edd30860d0133c2806f59513a0bb
aef77f4126cb4fe28e02d64824d74e2f784ce356
4325 F20101129_AABEOT fure_a_Page_042thm.jpg
0671676ba65d6cf5d8ddabf67c0ac861
0b1ea8a9c647b51caa979a84b6288bd694088e7b
104630 F20101129_AABDMG fure_a_Page_084.jp2
09a373f812b67fbbc0c15e8107e69dfd
c3fb0f9318fc5eb9388b676afc4cd747454644bd
4567 F20101129_AABEPI fure_a_Page_052thm.jpg
e6806a67835af1e4b26611069d2b6232
d713c58e2482e223118824ec278a1140648db0fe
20501 F20101129_AABEOU fure_a_Page_043.QC.jpg
d5faec8cb432851216c6062f17af6197
018d735329312d3ffb2ea3326a9ac04e99c3a483
5546 F20101129_AABDMH fure_a_Page_114thm.jpg
9ae450e217b91e5780ac3c83cc0c1480
a471c906ab07ae4c1aa7176deb71d9dbda9067f0
18586 F20101129_AABEPJ fure_a_Page_053.QC.jpg
7c8e93dd79fbd168b9d63173fb176d14
814d43677609fe6376772d1b192586ee34955734
6010 F20101129_AABEOV fure_a_Page_043thm.jpg
f645af11b98549ace99d6ee235035a70
4f4cc795d3c40b7ace9846353c96109d9691ac42
84042 F20101129_AABDMI fure_a_Page_053.jp2
ddef68e2c4a6c98292679649b8bda03a
0f893f46e6a8564fbb09bad3d5bd59c71c3915c9
5084 F20101129_AABEPK fure_a_Page_053thm.jpg
6205cdbbd5450526674c60d2b9069e8d
b368dd813e3ef0db1eb891fbcebcb448afb4f783
36324 F20101129_AABDMJ fure_a_Page_126.jpg
53d73163dcc342926b2ef0b109be8da8
03df24468d7a1f1681df966d0b5654705037eb6f
5818 F20101129_AABEPL fure_a_Page_054thm.jpg
45d9e2f67cc6ab4a11a2f7cdbd7b38b2
a2b32c1a899ffe60d9ac0f815ceeac302731cd7d
23702 F20101129_AABDLU fure_a_Page_183.QC.jpg
678cf4ab6c6a38283e7067bf4841b447
aa95395f4f704e90f46624b162d0147954a0af4e
24159 F20101129_AABEOW fure_a_Page_044.QC.jpg
7a65f44991a0957a9b40757b2979d2f0
30a9ab52ed044eeab55030ed33d2da6cba8a27cf
4271 F20101129_AABEQA fure_a_Page_065thm.jpg
8da0c0d688f6188d6f21e2e07555dfcc
4d7a53226a37f762eabdade865da2d200c51f7c3
72120 F20101129_AABDMK fure_a_Page_032.jpg
d576dfc67859e1f3f2e515cb6d233737
b14dbf51102a4a39633be50fc9081c8371ef5926
21793 F20101129_AABEPM fure_a_Page_055.QC.jpg
ba62a5e887f560978fdff19de48c9547
d63dac358adf5fea2e4be19e8ab6a41eb690d778
51611 F20101129_AABDLV fure_a_Page_035.jpg
aec4858a02cb22fdf8023a1d8eaff341
d0240075af79e3dad9e9b647cc1430a9106eb1f0
6548 F20101129_AABEOX fure_a_Page_044thm.jpg
b4828955c898f2180977adc02fabdc99
2f45bd76bd8f6074ce991a83d3a1bba89cbc6a19
14861 F20101129_AABEQB fure_a_Page_067.QC.jpg
0820ba02b8eca8da11d039860d073dc0
0f5d6f9c7dff529ae7a37eb9f1a06eb9f5caaaf6
6114 F20101129_AABEPN fure_a_Page_055thm.jpg
945384db563a984c8add17b8bce5ec2a
0fd4f53e6e2014de7e3e8b3fa40640e4b42b3659
22571 F20101129_AABDLW fure_a_Page_051.QC.jpg
b252a8cded0a3f99189c81c72c799703
36c08858bd5db48d37bb13349a8ac52ace5523cf
5300 F20101129_AABEOY fure_a_Page_045thm.jpg
6b71c234fa896e4bf027ca1c64997b2e
fa1b316171470f1f1dda75a14de8367c502f7f61
66732 F20101129_AABDML fure_a_Page_112.jpg
254b6c6f9a5c4baf4c3fdf99b2975f7c
6467814f1fac3b16dd0d81bf1081ac407566ddb0
4707 F20101129_AABEQC fure_a_Page_067thm.jpg
d46de1d10ac0515f4a5e95390deeaff2
edd4e54d940f7ab793697402bc11ffb14fc4e865
14529 F20101129_AABEPO fure_a_Page_056.QC.jpg
fe72971f88bad8bfb2073e37a253f0ea
6cf401b06039f902d639fe2fff5f36f6c5ec8446
F20101129_AABDLX fure_a_Page_184.tif
9040bc14ac5beb0c0fda3baa90c7f7fe
7e9905fad1dbcf46b5c9db4c7a82499a57e23ed2
13649 F20101129_AABEOZ fure_a_Page_046.QC.jpg
a28c4d5488c3ff8a87aae9a5db93ef40
f5e0e9fe1b18fdb05ed5c0f750a93841abf13c2a
70113 F20101129_AABDNA fure_a_Page_115.jpg
554596c6b67d4b9a865b9718652e3047
605fa7a5a2741267d014e3635a770843dce2b865
5832 F20101129_AABDMM fure_a_Page_184thm.jpg
87fba6a8cb32589b0736ef25422f3f11
32451e8d19298765dbdec7d05e88b9fe8a435190
6142 F20101129_AABEQD fure_a_Page_068thm.jpg
37cb3f331502c6a2ed166ba608cc3273
5b1569593603f4f241a4756b81a6d8e32dadf36b
16055 F20101129_AABEPP fure_a_Page_057.QC.jpg
4edb35c3af401910483bf5916d662e3c
a8f4f980b756edc59f80ac1a15cfae37682988ef
24019 F20101129_AABDLY fure_a_Page_186.QC.jpg
06af07160d606d1b0a168c7676702691
72dafdb8236bbe1d7ecc948f54fc7d75e97e5f47
23672 F20101129_AABDNB fure_a_Page_191.QC.jpg
6b86ed6c46c8ee26e913c3186e14d0e7
ddee0047b154a8ef33352dbd1791dc8a0be5688e
106277 F20101129_AABDMN fure_a_Page_181.jp2
c77b53318b91dcb2c060da21a09d774f
0447d905a59cc42be92b2c41363d6eb75b49257d
14836 F20101129_AABEQE fure_a_Page_069.QC.jpg
f0a241984b77c0f1e24690ad182fdcc8
28809b019a8b8f7a38a09ac54166279bd826c012
20561 F20101129_AABEPQ fure_a_Page_058.QC.jpg
86e77e043ef032274c1a21bad75ac8a5
012f79d9e2aef070fcf16fcdbf3667bbbae1dc2a
F20101129_AABDLZ fure_a_Page_021.tif
70e2b38ef30531c7464c5f758a890245
6d3a912374dabc3a3dd36891945fb388790f0255
2928 F20101129_AABDNC fure_a_Page_152thm.jpg
d113607ec01aef1a15dd6521cb6aa7c9
b88bd9cc1c085b6c608beaf05f78d82e64879e12
657055 F20101129_AABDMO fure_a_Page_069.jp2
ad1f134d68b43cbfcbb58c8ae12e5f11
966a263eee634e78f28b68c755d4aa956c6112e6
5105 F20101129_AABEQF fure_a_Page_069thm.jpg
6a9e021bfb7f6182e364d08b8803a5b2
db79df6218c47e54743e0af7e72c5faba83846cd
5862 F20101129_AABEPR fure_a_Page_058thm.jpg
09eee620633ab10936f3c56048a6e01f
c28fdf3add950e19876c6da0c987cb7042c7f6f9
5558 F20101129_AABDND fure_a_Page_154thm.jpg
99aadffc818de8c0ec5807113e4a27b1
bf5d69b18933b8d9a1f1e9a3e52550910fddb42f
71934 F20101129_AABDMP fure_a_Page_008.jpg
e1ae73921fdf6a1357a082172b6d617e
8889affeae2de4c44f279edfe26d0b7a3e4ea93e
19792 F20101129_AABEQG fure_a_Page_070.QC.jpg
c9eec0d0e749556b475be29d5d382da1
bf80b6a210fbc2aaaf0bdecd69c06213cb05b60b
13542 F20101129_AABEPS fure_a_Page_059.QC.jpg
c99821a282a3d48ac25c23fa81aa124e
1437a400b9eb936772fa888de5ae1ba72847363a
63595 F20101129_AABDNE fure_a_Page_103.jpg
92f35d56f3bbe375606029f390344c01
f3b70f0e07ba8fdd6b9b495685122e79695a0e12
4746 F20101129_AABDMQ fure_a_Page_071thm.jpg
0befafd3cb1f1e848e4f06a0e6683f65
d82b5baf57408e8e69b4c0984db494c46389edcd
14854 F20101129_AABEQH fure_a_Page_071.QC.jpg
a1cf855f17956f1ceeeb3a1c72b7e5e6
4b020e876662d3f50077ad9a85b421827a6ffaf4
4819 F20101129_AABEPT fure_a_Page_059thm.jpg
6eac0867196b2cf4a0e206e356b26325
88e7c50c0569a02361fe332d87baad6c070f26d6
81069 F20101129_AABDNF fure_a_Page_011.jpg
0a6314f5aaef418df25178fb5a7c60f8
d8f2d17ce2d3e3e17c2571bdf12edee0f0a2580a
24052 F20101129_AABDMR fure_a_Page_090.QC.jpg
8757daf2847111e778db4b05a03d0b8d
ee432659f6e95134068a604bbf190590d9b90822
20186 F20101129_AABEQI fure_a_Page_072.QC.jpg
dbe09c658d62c782b9e275ce390a3bcc
6c505f8f9103e28a21dc402ec976a2ec4148f7cf
F20101129_AABEPU fure_a_Page_060thm.jpg
e4efe0c2d7940d1c032d01c027a696d9
b359c810b2b2ea2a3a12c1b492f61a35d9617ca4
706947 F20101129_AABDNG fure_a_Page_106.jp2
e933acb5fe6980429e917238d8dc57a9
bd25d46835a91d74b4b1a4b613eb249cee77829f
45007 F20101129_AABDMS fure_a_Page_063.jpg
19e1f64e67440640b087897b11d0acfb
ddd956c3df3eac92faa4c4a10413da15d44b9441
23631 F20101129_AABEQJ fure_a_Page_073.QC.jpg
05356b416c1f4f1841c47260c6559994
651c3a22563ba5f9207e8f71f7c466e2676fbb70
14272 F20101129_AABEPV fure_a_Page_061.QC.jpg
f904f44003662aa4e2c98386c3441d3e
12c8e2f7cf5d0658cd91efefccf6aca272d2e2ec
66485 F20101129_AABDNH fure_a_Page_099.jpg
5b6f9ec0ed322dd9434ca7337df060a7
6e5d48c028543197481ce848416ed7402a523ed8
49420 F20101129_AABDMT fure_a_Page_108.jpg
78c6d2d33149455ab3067e295c986ef4
089308e890e884ce272d7fe121631f4e2d619a0b
6558 F20101129_AABEQK fure_a_Page_073thm.jpg
a7468ec59e023a654495ce01066d52fc
019ea21bc29879b25efa1228f1c19d27a148c967
69808 F20101129_AABDNI fure_a_Page_022.jpg
37b0255c6e2a85d41236461be6dc6ea6
4f74dde9faa1685df91f9e983658984e22da53f3
5750 F20101129_AABDMU fure_a_Page_153thm.jpg
be5e286bad309738c9c50609c90ce242
bf9ec1428cde5803ac19d773f957eeafa3f90b01
4668 F20101129_AABEPW fure_a_Page_061thm.jpg
b599eeb28be5afafcfe7eec9f8faad9b
8023f7b6b91f0d9be9b18bc98be3362fe6bc1eab
F20101129_AABEQL fure_a_Page_074.QC.jpg
65b27893420cb4180a2a48b6f0567ed1
50a17ce7f26753b28a1b4545fbf13c75beac3152
13029 F20101129_AABDNJ fure_a_Page_107.QC.jpg
d27f33c9297140fdd8ddd50e628c4ee8
f5a4c17bf055a422c5f79db75ad223d3b3fb948d
23567 F20101129_AABERA fure_a_Page_083.QC.jpg
bb88d1dff51e6e13ce1fd7c7b8a4a04d
c918e9f6141d1e97a5256585559223bdd619ce64
2063 F20101129_AABEQM fure_a_Page_074thm.jpg
7a4f1a24db4e39cf158cdf096a89412b
2b285cf15ce9237f12b5de9771d2918d7a8cf4c5
21825 F20101129_AABDNK fure_a_Page_094.QC.jpg
98d560c0d8d572e7b51bc3683e03ce30
0a04db104c6373c172a1bea70d1a99ba967eb0fd
58317 F20101129_AABDMV fure_a_Page_030.jp2
d7610db5a47d65a6aaf7a00766b88853
2a59eb1a88fa0a0f6a711a4674466ffdb57fdae2
4688 F20101129_AABEPX fure_a_Page_062thm.jpg
aa4a8db56b5f6633c376fc9f4c6ac704
ff774b3ee2d9d51d827fa3a0a20dc0b5170aeccf
6406 F20101129_AABERB fure_a_Page_083thm.jpg
935bfb645eabe6f76e07eb71b13dbb12
e364f759c6c5809a47569d8255b86a3d63a343f5
20558 F20101129_AABEQN fure_a_Page_075.QC.jpg
e9e08d18d6244d644d1c556c7ca312bf
dafc9f898b2960a85aa6e929484521486617ac95
8423998 F20101129_AABDNL fure_a_Page_080.tif
836edef6bd53d7a114bdc4be5c6375b4
b202123f0ad43aad639343684976b07146e0c082
22293 F20101129_AABDMW fure_a_Page_127.QC.jpg
b149e938a7af8281f8bcbbe92557fff3
9db73b9e1131e7576b6a70142d8e1d14dc21f18f
5131 F20101129_AABEPY fure_a_Page_063thm.jpg
f4674599976cf1fc35a0605719fbfdb5
984c9ffb1517003a6c08f09f58aec5cf3e2c2707
23035 F20101129_AABERC fure_a_Page_084.QC.jpg
478a22dd0b063d6a7b0f05bd09641d4d
07cad380d67b72dadaa8babbcd44451a8b3897ac
5947 F20101129_AABEQO fure_a_Page_075thm.jpg
9e4ee992c4d3ae7119d50ac55926a0c9
673d578e0d5bfa3a896c2df459e30b79ed89950a
5860 F20101129_AABDOA fure_a_Page_078thm.jpg
0f679670709148e9a1e8b2e83ecea590
dc1778414b6f4ad74d8c8c20bf2eb164bd49f62d
112849 F20101129_AABDNM fure_a_Page_090.jp2
3971575bfd4ff28c168b0109f5f005a4
dbac5a83733fef0f7538c11015091a9e1245aeef
40640 F20101129_AABDMX fure_a_Page_133.jpg
8260c52ce4b88a69aa86529aece998a8
ae3f8d6295e477bf323f9f8d5cfb796a89767eaa
23960 F20101129_AABEPZ fure_a_Page_064.QC.jpg
981535b69116c4512a15c7feac335d11
952817ea2a540e55b8457ddf645e8235653903e3
6317 F20101129_AABERD fure_a_Page_084thm.jpg
517bd0130126a07ea93d43d9cf5c3e03
c4e72008baa2ec8e3e1105fbec100041b526bad7
24373 F20101129_AABEQP fure_a_Page_076.QC.jpg
a8f099f4e6973514bd0bd1821beefda9
7d50c522d03b756e154f9d1ff95f69365d957a4d
20938 F20101129_AABDOB fure_a_Page_050.QC.jpg
178f42a071a086a21c149963c05c9905
dea28b6f501ea835b8c13ab240a9d75f8e2764cd
6643 F20101129_AABDNN fure_a_Page_064thm.jpg
5479418202244277a262b88e962dfbdd
7f05ad8f71714bc008d62c0ed40904650b41a903
15399 F20101129_AABDMY fure_a_Page_063.QC.jpg
f31515cb4713479a65c0c3d19b633655
a19784ca55069761dcf8e07dfccca35762ef7c20
23951 F20101129_AABERE fure_a_Page_085.QC.jpg
ca345ada0bacf17fc01ed559737dfdc4
6fbf03386f35f92ec9bbc3e930204c5b96196d14
6586 F20101129_AABEQQ fure_a_Page_076thm.jpg
39b93e7701c297054a89be4ffc30ff0d
17e8fb85368e59c78a36c33fa0ef1a900e665e4a
92027 F20101129_AABDOC fure_a_Page_103.jp2
5e60b1956853c3dcfb8468803c36f461
73a1fdd6d32cbb09192624e955911530946f75a6
F20101129_AABDNO fure_a_Page_075.tif
1f4a79a93a1e75a616d26f583bdecdad
220384eae74f72cff08c2d856432ac822854c41d
16998 F20101129_AABDMZ fure_a_Page_163.QC.jpg
c6b94b0dfbb01343394eb2dd699debe4
d1fc5cd58003b0b6297599f88e9c22dcf50e2178
6539 F20101129_AABERF fure_a_Page_085thm.jpg
85ef2fcc75e0790d49362b32d424c1f5
1bb75d2ddb1e291d56fe5cac69def98b09377734
23865 F20101129_AABEQR fure_a_Page_077.QC.jpg
bd440fc25f04ae8dc375946da391c152
d8954e9003171d96c48eb20de1d3abac3ba1b499
40965 F20101129_AABDOD fure_a_Page_175.jpg
e3367403b83b2048368212ffba25a1fc
e67ede5bc1213804d30da014652bc2e9d1ba5ca0
16761 F20101129_AABDNP fure_a_Page_054.QC.jpg
5d9f9031fa77935379f941b53a552083
d17812a92a4d2c8c0461355cc5c7568c725fa424
24658 F20101129_AABERG fure_a_Page_086.QC.jpg
abd14070f11a7defe4d0a81b8e7eeb8c
38601dac600c7bca7e2b2d39369984d409dd1e2a
6563 F20101129_AABEQS fure_a_Page_077thm.jpg
109f71533f835b291378052f1f5b29e0
a44ab523eb27b28fa320ec3c30aa4ba4ba291331
F20101129_AABDOE fure_a_Page_058.tif
3b2e3b1b5174c8236ba99520f26c3da1
e09dc992159c8d366825e3114299c4d3df4533ad
53200 F20101129_AABDNQ fure_a_Page_163.jpg
54ffd97646befd97a18156d3903cb08f
9e1e9049e6a385415e4cbc795ac4f52aefe2c12b
6772 F20101129_AABERH fure_a_Page_086thm.jpg
38c1345ddb8b35ca87011d0a810480ff
819e719134dca59a8eef5b3baaee9978a50d8fb5
19984 F20101129_AABEQT fure_a_Page_078.QC.jpg
6769a47ab8e6b451b85d81cfd8d2fb46
8f938682f034fc1f46497839c53c8a891597f19d
21315 F20101129_AABDOF fure_a_Page_008.QC.jpg
3235820393ed02b5ccd48892657f6a81
79cebb7e356144c4284261862a51b0032663829e
13189 F20101129_AABDNR fure_a_Page_144.QC.jpg
e59c67285dff7c3bbe448ec6dafabd7d
3896199df1049f2cfd1281c86543d0818cffd019
16687 F20101129_AABERI fure_a_Page_087.QC.jpg
072cdfc59f0c1fc08fab232f1aea79ac
bf67f5044b568ff5527f649e4bf43d31e9597f3d
21448 F20101129_AABEQU fure_a_Page_079.QC.jpg
57cf0130477675e5ce28006dce6a243c
58366b78cee0f1720c07adaf4367a3ce09aa554f
4616 F20101129_AABDOG fure_a_Page_140thm.jpg
eb5ffc0f939ed1f784ae734029107df7
efe9fdfdd4611fd95ae80fc1e1efc229c4d925af
67503 F20101129_AABDNS fure_a_Page_155.jpg
2e1d9fa340561c31ca28d9e2eb781995
8e9a0a0bb8032d3ccc656e568e9529fb49b67936
5269 F20101129_AABERJ fure_a_Page_087thm.jpg
66f7edb1f332a3aea3be06d406371cad
5bd294da325c7bace8c616bc02238e59258506d1
6105 F20101129_AABEQV fure_a_Page_079thm.jpg
3a5acf3d362ad1b437746b9e59810cae
2e7b1f08a50cdb157f884031f1ddca1c751fec37
48395 F20101129_AABDOH fure_a_Page_025.jpg
4bdedf6b41020181075e8b564d5da0da
9a1716cce344295a21c33b7af8105585f09d1738
86783 F20101129_AABDNT fure_a_Page_012.jpg
6ef5a146b25faa28ed9295d94a3b2631
80de35ef2787ddead20935753067848d239d0ed1
22969 F20101129_AABERK fure_a_Page_088.QC.jpg
8efdc113f16ea632f28dcd50e47aeb78
78671f876ac662207e2c3ad6ff579ff48d3af99e
14311 F20101129_AABEQW fure_a_Page_080.QC.jpg
4186d695cdc316d996da80634c97697f
4f627f4f1addd1977272fe42af1e161eacc2742e
711578 F20101129_AABDOI fure_a_Page_060.jp2
80edb052186b5e26de26adf2036a1690
1a81d9cc7139c371704ccbe060b84487df7efcfe
97092 F20101129_AABDNU fure_a_Page_026.jp2
eadbaead0128c69718eaa7b40c1bec6f
885087f8720678acbd7a0af0054d89cbf8d1194d
6389 F20101129_AABERL fure_a_Page_088thm.jpg
90b8f071ac32c2d0213367c6f5e19316
42b5429cbf63e80742a5a6d068aa411b28b38325
6814 F20101129_AABEQX fure_a_Page_081thm.jpg
92c42cccc7cee0ec2d222f9157c6ea6f
e4dcca6f562a84c92c86752098bab669399b6f95
48912 F20101129_AABDOJ fure_a_Page_060.jpg
840a52de24970dfb6d9a8041c8e961a5
2d668eecd0325609cdc2c031f85c9e7ac385e031
17306 F20101129_AABDNV fure_a_Page_188.QC.jpg
00939509a69da27ba8dfd4a9b31747c0
a99094d8534dcf41fa2f623fb18ba4c6b40a023d
20528 F20101129_AABESA fure_a_Page_099.QC.jpg
3bbd11454b18acbee08703bca47cf7c5
bb09bea6e033fca7666527a381bf46f0d0a4e271
22559 F20101129_AABERM fure_a_Page_089.QC.jpg
1513b55dd393b5bf63cd5cca92130e30
97479c8f61fd448b389e3d07de24cae2c199230d
18222 F20101129_AABDOK fure_a_Page_006.QC.jpg
c5c04785cfda4a86177ff461aa5495ee
a44d0bb5ef0f2d5b35510216fb46aae147c4babe
5911 F20101129_AABESB fure_a_Page_099thm.jpg
008c96f75cb5cb7183690474b27ea3e4
fb191748f86719f0890255cae85dbfc4bc9c7ced
6538 F20101129_AABERN fure_a_Page_089thm.jpg
00c59efcf0babc72f875cd1e8b0c3611
853193feaaebcf4571c09489d1c05b23823198fe
24312 F20101129_AABEQY fure_a_Page_082.QC.jpg
8712ff9321687df18d9c8a665930ccc4
6334489c3ad6d61926219de7b80fa1b84637f6e2
4226 F20101129_AABDOL fure_a_Page_136thm.jpg
6cb8c6809ea0a4111c1366e7ff8564d6
f56f542de194e9d15c2d0837ea7f8fe6b4b1a296
4110 F20101129_AABDNW fure_a_Page_007thm.jpg
df3f372774e91d8f5ffc37cab8ce06a2
da114b22ee37eac782d25c88b6d09abe2e44a588
20417 F20101129_AABESC fure_a_Page_100.QC.jpg
299aa36b3484b131189f286f66ddb153
4c1f7b47958013f8a485392c61c19ad562592e5d
6731 F20101129_AABERO fure_a_Page_090thm.jpg
c1e5879c4365c0ad87e0aabbf0f4293e
6c77d5c23fd795470add8cc94c40c86d906ba3a9
6652 F20101129_AABEQZ fure_a_Page_082thm.jpg
7e6437cba53e4bdbe110b5333dc91bac
9eeacde60986740452a77f6066ba8f574ea0d381
4127 F20101129_AABDOM fure_a_Page_135thm.jpg
b7b0bf7d81ae91c3de7cf8aebc2c0494
5b5eeff17274ea8d61a1645a4212fac2e4740a28
11785 F20101129_AABDNX fure_a_Page_109.QC.jpg
23ebe879ca7e42e252959c7c1ca80ae4
7be5f493dc243826be897673d50153f32a0bf893
34781 F20101129_AABDPA fure_a_Page_109.jpg
6ed9c6b12ddd03f7993a0814a6e96fea
c37b742587b47888e1a8920925254cd01c782a88
6064 F20101129_AABESD fure_a_Page_100thm.jpg
39441e3a4154102564f8fb39df26bf7c
7545e579d91af7d9395fdeec53e5a6ef6b9df030
17485 F20101129_AABERP fure_a_Page_091.QC.jpg
0a610e01f7f1c1ad33ab3c9ad52b0b07
0b6629b747cbd3679f8378b2a81a29991a75d84b
9581 F20101129_AABDON fure_a_Page_121.QC.jpg
fe45958946bc1e9dd7082ac0fb4e10a9
c94c7e45f8aca19e88f646fec769c5eb61cbc60b
6522 F20101129_AABDNY fure_a_Page_174thm.jpg
f01b2169cdef2a093d621c79cff7847e
1b82f5e72b9a03a827c5b735252a853f6a23d0cc
6600 F20101129_AABDPB fure_a_Page_012thm.jpg
34b2cd6ddc5a8d9dbedaf8457c93e2ab
62ddb62834d3275e736d22381afeab3e0caa502b
22002 F20101129_AABESE fure_a_Page_101.QC.jpg
a09a1d712cb9a9208ace555744300335
c1f2f7a8d750d593d7b8f7ef32e66457acb82e0a
5411 F20101129_AABERQ fure_a_Page_091thm.jpg
3855c1dde15e20230a5c6ca3ba4049ce
36464abbb2ea9188e6b78c343fb2f45ce5d0beb5
23401 F20101129_AABDOO fure_a_Page_028.QC.jpg
2fd7362f01938cd6d945d95835e88d8a
cb150a24e4ec77912a1b952cc21b201955b1ac01
19838 F20101129_AABDNZ fure_a_Page_066.QC.jpg
8f8608e8de6f6f86a85855c30eff7a30
86373f8f57ac13448ee4d4556002cce816eb5f6c
888484 F20101129_AABDPC fure_a_Page_143.jp2
d241b2f59379093fc5cf9e6d4e80ae0a
5050f0bf6a8ce6390dd8e5495ec28eb12a2fca49
6376 F20101129_AABESF fure_a_Page_101thm.jpg
563a4f08ee53eccdc416ffe3c067f8a1
20ea6c10fd661cba6ad75c1241798a0388260fb4
6658 F20101129_AABERR fure_a_Page_092thm.jpg
68b576eee604a0c4c09efa6bf25d7c4c
2576a9469ab428bb7a5a485fcad84e107ffd7822
F20101129_AABDOP fure_a_Page_133.tif
d587d607e65ccd56fe885c3e79ab6f66
382e6519b953a70b840367cef58968f0935275c0
F20101129_AABDPD fure_a_Page_137.tif
58869efcf688769574e69a7b11f48f56
4d55161f07e7f7b31dee2a8f280dc362f6f43a13
17126 F20101129_AABESG fure_a_Page_102.QC.jpg
a6a1ec884b79977ae6db0c9c1c171283
a8a5fc24d4bf0d56d4551bb5c282472496208445
15133 F20101129_AABERS fure_a_Page_093.QC.jpg
fd635b9db8817b0f00c18cc257d767cb
1c18e3f1022a16b4a92e5c03aca28960eb6a9f3d
56567 F20101129_AABDOQ fure_a_Page_053.jpg
8b715e8a491ed5f118ac50b3a917852c
77ad4cbb19ecc11ef84511253b547bbbdf6b50f1
F20101129_AABDPE fure_a_Page_090.tif
bbc8c7b372fe0d22bdaedda89b12bed1
4833c309d92246c33d57ad4a5bc3810f433b01f2
20382 F20101129_AABESH fure_a_Page_103.QC.jpg
493f1c16e2fcd0c2888297349be59a19
c32675cc853abf7692d4b3c4cfb1467000b461d0
F20101129_AABERT fure_a_Page_093thm.jpg
9ec86f876f3e1dbed77059f5c6c00856
d4ffecc5626085d9af4ac498268586d46c477414
13723 F20101129_AABDOR fure_a_Page_065.QC.jpg
1003abd19522e1cf5b4710fa7584a43e
e8b51124670af9102aa4dd9ab98ee663101c2db5
6761 F20101129_AABDPF fure_a_Page_147thm.jpg
1f73a35f6f91bbbbe660b42602379d2f
740a0247c67cfb72fbe957f5718baecdef7738d3
23237 F20101129_AABESI fure_a_Page_105.QC.jpg
0ab897c6fde769b5a141d926319018c1
0fe378e1598b6d5d45e00e3e570244fe5385365a
23497 F20101129_AABERU fure_a_Page_095.QC.jpg
3e59115a44933f87d8726630deb242d7
ccd046f3d98a96603311791f1e6a576b7a0a9f9e
952483 F20101129_AABDOS fure_a_Page_125.jp2
b5032985f4d51328f019e5122c47d4f6
65e50049878afd12939730028952f8ced6435396
15433 F20101129_AABDPG fure_a_Page_060.QC.jpg
5eeaa0e89a030c81021b3b83789e93b2
f9f661fc8ebbc49ea5d88fb6a2e8794d1ac15bd8
6411 F20101129_AABESJ fure_a_Page_105thm.jpg
06f9b3492532890d8bcf8b80caa08aa5
331bb370517d6bebf54926c8c1bcfe4d8513164a
6620 F20101129_AABERV fure_a_Page_095thm.jpg
de81e26be62382027d79617a56382e0d
a3c9e91283fffcfefeedba0b4185c5f199e6650d
F20101129_AABDOT fure_a_Page_162.tif
ade58c62febf90efc22578aa77f3aa2a
fc25096f7c52cbbc46ca0e9136b930367094b75f
40471 F20101129_AABDPH fure_a_Page_144.jpg
5cb2feb960b55f93b10b996a66a74cbb
b103d2382c24503e2f92037f3ab8895dcb862258
16638 F20101129_AABESK fure_a_Page_106.QC.jpg
583e594ffca7874d4657a06b0724488a
07ea764a4e78f35a4f3c177ab4ffb3457788d599
4400 F20101129_AABERW fure_a_Page_096thm.jpg
1565c8378ba0bad38c56b0dc5c3f7e20
dc35492baa77d96bc4f199b392e164135a4d9e4a
19704 F20101129_AABDOU fure_a_Page_170.QC.jpg
74b2d0a80727813c67be68b59b363cdd
2773743eeeedb57c81e4acc54422c18b862d69cc
5015 F20101129_AABDPI fure_a_Page_037thm.jpg
39438cd0786c884fb3eb0aaa3246a15d
e4dd4af9a1ce2ec47869c983a9954552f843c969
5050 F20101129_AABESL fure_a_Page_106thm.jpg
d1eb6b812e6dd5a6b11f1285d28590df
5fcb81af6d923390d81104ce5d50d10b2d46ce9d
6002 F20101129_AABERX fure_a_Page_097thm.jpg
cabddff8b44707b7672e4563c4948c31
86171892d8acfd2f61838790a1769f0bc9566e82
F20101129_AABDOV fure_a_Page_073.tif
4a8ec74e6c916020cb9cd2e2b72a4bc1
e6ef10fe34e03329dcacc7f74c78ae247f55235e
62571 F20101129_AABDPJ fure_a_Page_167.jpg
de4b3104f9dc4c90b0a92453d7b38698
195cb0743084b4e944a4d6a8cac2d31cb434d048
6566 F20101129_AABETA fure_a_Page_115thm.jpg
ef8749543e594249cc00f743a0990308
2060931c7b19a5dd1dcb2819201b4be9409cb2a1
4599 F20101129_AABESM fure_a_Page_107thm.jpg
e2753d79edf5f68eb352912c1facd5a7
ffb9577e5893c5cdd381475599ab71a95afa2d2f
23723 F20101129_AABERY fure_a_Page_098.QC.jpg
64060af61395e60ab0b0e994c22d9c20
34c9b73117cc53490f0f3003fc85e89b1ef8b3fb
12474 F20101129_AABDOW fure_a_Page_126.QC.jpg
f53fb9d2b8fb686a9cbf09bc2c7ffda8
9c7bfb95eab5a80a6addf1cae8a4c3b06b6e17d7
5898 F20101129_AABDPK fure_a_Page_072thm.jpg
5dd643ae65bdb3bbffa1e950e188ab65
94a8860665bb04a126274ac7abcdc12fdca43d9b
16860 F20101129_AABETB fure_a_Page_116.QC.jpg
b3c53070eddb24c36c96234b68f30121
56d75e401032c42dbe31449330f03e60d8021a96
16100 F20101129_AABESN fure_a_Page_108.QC.jpg
fff5b3fcc085e008734b413370a79f5a
3822b32976cfa2df66ef1dd0f3014d99bb800b1b
F20101129_AABDPL fure_a_Page_098.tif
9718051a5916359951adad58103da4c7
9442f68bb743c8ead29bc1598f40d04312343ef5
22139 F20101129_AABETC fure_a_Page_117.QC.jpg
2cf0ef4679abdd79026fcd5189a33287
d50ff9b0d7da340a8221e47919c42e932da6d249
5220 F20101129_AABESO fure_a_Page_108thm.jpg
3ea2c6e7ad0879e08c2fe5230d229a38
de6e27f52213455dc098af0cebe23672deaa96e4
6775 F20101129_AABERZ fure_a_Page_098thm.jpg
77510de4ad8c71700da3e93e02f61986
2b56583c4f3c0f03373258e448f1c6c4ed6383aa
22429 F20101129_AABDOX fure_a_Page_104.QC.jpg
6ffbed0b211c3ada3d963ca86311878b
45c4f46f25079e1e6a13c61e3766ef524b603ef8
63171 F20101129_AABDQA fure_a_Page_058.jpg
cc85900ed66ec6951875a92c4f779892
61af09dfca4d184c3393d9d3c4cb7d3662ba366c
20969 F20101129_AABDPM fure_a_Page_185.QC.jpg
d0a09d64044dd8f9536668d06949db69
83efadcc088e965b6583df612384d04ceba73f02
22684 F20101129_AABETD fure_a_Page_118.QC.jpg
5b66c7f32b7d02241aa327c57ba20842
11432e1bfeb6ee1c90c8217c0ce3038a0be1b401
4068 F20101129_AABESP fure_a_Page_109thm.jpg
a2e155178cde76fc1f511cfe94eda77d
e0f93817e9bff02d5676cb64021bcf8d3cc233d8
66195 F20101129_AABDOY fure_a_Page_055.jpg
96774116effa346349cfa3dff5d1ba48
29ae64b1421919519d770e4760e1d80362a6f1ab
5103 F20101129_AABDQB fure_a_Page_024thm.jpg
49b67de5c5f6c3c0dde4fb760edeb7fd
dc0f1368d906ddb58e2ed36e894715213355ed50
F20101129_AABDPN fure_a_Page_097.tif
5b8bfa2b35c152bc65d062a3ab7eddf5
ab632aab7d82d0619c0a20e3ce6ee79143528975
F20101129_AABETE fure_a_Page_118thm.jpg
76bc9f5888d2242b5adf88ca80523bbe
4ffe0ff9ee41fa4f57f1f6d8cc1ad6a57ace984d
20727 F20101129_AABESQ fure_a_Page_110.QC.jpg
0bd0838189370d051f509df2a7e6418d
4154e0edb4093bfa7f95bb69f51c76806f62ac6f
6390 F20101129_AABDOZ fure_a_Page_117thm.jpg
73873badd294f853edd2475927e1932f
907191648d0a8eef11f9b404e68bf9bc8cd047df
91403 F20101129_AABDQC fure_a_Page_195.jpg
7dc90d655302b03ea107a1ac9e8f65fc
e0325af0560371c15dab5df8b85fa1d5856f37f6
37908 F20101129_AABDPO fure_a_Page_197.jpg
61cf29ab1c1a2ad1ad6684c003dd4956
98c68a0b42f02df6457db98ac49acd4c1c91036a
10599 F20101129_AABETF fure_a_Page_119.QC.jpg
e056e3074380401ceb4fb9c3b1f4625c
1b92ee5b51ea938bc6d13342a10015515a258224
5892 F20101129_AABESR fure_a_Page_110thm.jpg
7fffe0aa6f1fe2b7c831b2582079253f
3ee654ec24af0d316722797202fa6c172997b43c
20644 F20101129_AABDQD fure_a_Page_189.QC.jpg
134bb9bb3bbc9342d3628fee87da06d0
f69713a09e6894a72d3e830931de6881a31847b3
65774 F20101129_AABDPP fure_a_Page_026.jpg
7724fdd3a262688a585b7d6e9bd15bd8
ffee70b476c852609763c9feb54582b2aea8b559
3254 F20101129_AABETG fure_a_Page_119thm.jpg
67a6db7988cab7e958f61de10ff95896
566885ca5fdee4f8da4e43dc5b7a8f4f4e0da03a
15387 F20101129_AABESS fure_a_Page_111.QC.jpg
d257db67d76876f6944f091788550b75
da667829d35d4726487ee434a4ede3e597e3a274
44262 F20101129_AABDQE fure_a_Page_096.jpg
0cfb33dbbf4b2776d8da08adba35c09c
b6d1500dedfed685b1e84533f0da21e5c785c415
987557 F20101129_AABDPQ fure_a_Page_092.jp2
2983da0d2bf66103b28c0b374fae9197
a0a3aa15487579b3f8da014d9bab63f17a220154
21704 F20101129_AABETH fure_a_Page_120.QC.jpg
1ed3498b598c35e2c0453476b7539c07
e8f51798dd785c053d20a767bf48ec97a2a56043
4978 F20101129_AABEST fure_a_Page_111thm.jpg
cf1a508897ac46b2a03cd10e70b94509
804b1ce804b722a25bbc3a342ae1b2d15ed40310
10027 F20101129_AABDQF fure_a_Page_169.QC.jpg
b9d8a23346c5a5ed84f9f423aa0b99cf
4eacc75ab52c1a5bc65af54304f5e16c3b23702b
23620 F20101129_AABDPR fure_a_Page_174.QC.jpg
c98bf7a834c3964d15184ad94b40e9ee
b1b37df905bb7f29abff5b47dcf0e5f80692933f
2913 F20101129_AABETI fure_a_Page_121thm.jpg
f7cb9738e281caba4997c622c5631055
b7e6404e2112bca8e3e7faa4a33b24566e99f475
19941 F20101129_AABESU fure_a_Page_112.QC.jpg
c27bbc9cfa6f7caa573dbcd1c0545050
88416f855be319d9b9eb9efd2875522d3e2d1206
659412 F20101129_AABDQG fure_a_Page_175.jp2
00cc0d8ea35b06020f88183873a2ff0a
de11a517143cca7c60009a8a3ada50c3a225b393
61318 F20101129_AABDPS fure_a_Page_015.jpg
58d2eb03c613b62ac99c9f9a565e7fbd
8e5d28b44841cfa59c131c9ed4e6acdd008b882a
14719 F20101129_AABETJ fure_a_Page_122.QC.jpg
a8b0e84debd980f0b45c3ac639a4c151
12e8b7fa928d56b3473c6ce290097785a26e3d27
5746 F20101129_AABESV fure_a_Page_112thm.jpg
b0a7762db138d7734cdd933177ddcda8
6e402dbc1f6da75007446596b0eaea8b4a87f432
827219 F20101129_AABDQH fure_a_Page_091.jp2
39e9dd531aa6208ce0a4cc5fe5afa56a
4b8d179066304eda81f5a26628321a39116d2472
F20101129_AABDPT fure_a_Page_174.tif
f882fb2daac11d45ea08012134f01106
cb5a9a47baa0bf789987c9184cdb058b75fd37e4
4541 F20101129_AABETK fure_a_Page_122thm.jpg
438c98cd985700de7410acab6d4ac4f5
bfe3c556863e8aad823e32a5a1fa5f45bb1881b6
17971 F20101129_AABESW fure_a_Page_113.QC.jpg
c6ae4acd3c06efc04d5c96c878618d35
34c42d9d4ec65c3f66fd4bc50a9ce54890c1c284
15760 F20101129_AABDQI fure_a_Page_062.QC.jpg
ad70ffb6230b90821c3431ca6bb46aa9
c78832aefec5473df6536f434f3fdafdee9b5d9c
6025 F20101129_AABDPU fure_a_Page_132thm.jpg
7e3a2f78db3f08ce944d648b68bf326f
9318744c6a01d24f4f49f739b1ae82c664a09ba4
14343 F20101129_AABETL fure_a_Page_123.QC.jpg
c151748629075a822ed12da58225fe15
024c9915cc443e0569eaf584b97e2840ad8ad6b6
5356 F20101129_AABESX fure_a_Page_113thm.jpg
3aa50f016613c0d2da30ef965a432160
673f802d5a7e29021e90e90a29899752c77d24f7
765027 F20101129_AABDQJ fure_a_Page_019.jp2
013e5e2e803e09b10391a98904f7cd6b
69322a41cd776bdfd7adf9a596f90a07cfcbf32d
670771 F20101129_AABDPV fure_a_Page_154.jp2
0986a906224f12eb31ca8977525e3823
2c3010650e35f216106ff0257c4e6d6b92037ee2
13917 F20101129_AABEUA fure_a_Page_133.QC.jpg
66188178bb2220369013a9f8c74276e8
c944546979ff48f144e91ab0d8f50fc039ddbd68
4442 F20101129_AABETM fure_a_Page_123thm.jpg
66f71e852190dff4296d1ad432ec0c30
a47547f32364e8a2d8fd764871c4629038118cb1
19624 F20101129_AABESY fure_a_Page_114.QC.jpg
b36dd0b6f98193e33645f81f4e9d620a
8c1a0c36308084c7575006d113b8b751a9a3b903
113337 F20101129_AABDQK fure_a_Page_186.jp2
747ac1406fb31634b521ebc536e46cff
103bc769422b723286d4e43a4453ca06ce366cfc
F20101129_AABDPW fure_a_Page_040.tif
77fca037a00c9f391d596547b38f1a8a
115f3aca0738e2e0032f039b9fe1285695d21993
4261 F20101129_AABEUB fure_a_Page_133thm.jpg
65ea460e2ba93eb2c270352df9411a9f
f603d8f9c316639b51d541afa4a11abbc50095b3
13979 F20101129_AABETN fure_a_Page_124.QC.jpg
6f83657d6c3ac17004378a4f99f3cd9d
768ef222b9c82afbf8c5aabe716359c716407242
22648 F20101129_AABESZ fure_a_Page_115.QC.jpg
2f33ab0ead151d918259d3abce97c3e8
20eb58926d1013177c7b9074116348be906e0bb3
F20101129_AABDQL fure_a_Page_108.tif
1b3369557ac31b24b7dabc4a5af4a554
006d3550724c72366a8f956678e7d9e2e4dbec48
5090 F20101129_AABDPX fure_a_Page_102thm.jpg
94414f6a998a9add641d2430275beab5
ad0e8bac3940826289b16a0e640af3618b00912f
13797 F20101129_AABEUC fure_a_Page_134.QC.jpg
b18d4db10bdba8416bed68be5c951b1e
7a2e5bea4b1f4fb6094315b50ec42bc4866a615d
4487 F20101129_AABETO fure_a_Page_124thm.jpg
545e5da2ac8bb720c43be286e0f1de2d
4b3409ca87c5c3f7b4c350edabe41b56c836e973
F20101129_AABDRA fure_a_Page_158.tif
46da7a724b8550a4a26dc9aba8990965
6cac674bf9f879abcd01acf9585aebc1381842b6
679272 F20101129_AABDQM fure_a_Page_056.jp2
0ea7363db66e4a3691d8bb9de5e919d8
272d58ac5cc8353fdf49b2d65bb0b94074801699
4194 F20101129_AABEUD fure_a_Page_134thm.jpg
dc16ee22a0fcac8f1fe02e17d1ff9fb7
3d2240cde61807784a236a10a3c06387b1c451e9
20390 F20101129_AABETP fure_a_Page_125.QC.jpg
23c64f794c766178229fce5a01fcbb5e
c2861151747ef7820edff43a320f7af1190ad4f4
33418 F20101129_AABDRB fure_a_Page_150.jpg
45592b5033bd6c18b037c914173f36e6
bcf5e4d6b6e777b17999f3040ab8bd1dfdc5720f
45268 F20101129_AABDQN fure_a_Page_052.jpg
d47be773bc8597447488862977da54ad
141deb764e28b2c459a4103258df6ac8014befb9
6482 F20101129_AABDPY fure_a_Page_104thm.jpg
4b03df8c61c9ead148984173ca07dd14
351e2186c4216d2268e391a426c2977b6ce5ed60
13257 F20101129_AABEUE fure_a_Page_135.QC.jpg
22e055081bdf408c03a2c1cb71b93556
97fc5923eefdb9b5f3f4e82fa0aab3dddbc732e3
5739 F20101129_AABETQ fure_a_Page_125thm.jpg
ef6c301680739fb948fa48c1e927ec4b
4f00d443a6a1a3014d6fd88072538fac58ddf6a6
52796 F20101129_AABDRC fure_a_Page_087.jpg
7d0298ee617ae98abd17d79a632b04f1
4bf2e2ab0b99b1752d584500b85a53012df63610
66788 F20101129_AABDQO fure_a_Page_023.jpg
1bce041ca226755629aebc75712a700d
b25023a53050cbedba53a0d0722550022b195ddc
61060 F20101129_AABDPZ fure_a_Page_153.jpg
7daecad9d50e1790dacd4c28ee1df957
53b056fd2bedac6d4ca0e7c820600f7d82bd7574
13376 F20101129_AABEUF fure_a_Page_136.QC.jpg
dd0f008c5c15d7f40858c73ea6c4e229
b4dadf0e5cc5ab3946d000621f290c326b486b25
3925 F20101129_AABETR fure_a_Page_126thm.jpg
1d6130b3b4a16425c677fd3a7b71610a
c0c9d549e3125ede8a956dac32d831c71c415fc4
6420 F20101129_AABDRD fure_a_Page_094thm.jpg
1b240b5d9a59f77d197db69b57e944eb
d6481e61c2e234d9e9ecc7143776b550ca07d241
4822 F20101129_AABDQP fure_a_Page_020thm.jpg
5c6934c872c593cc841cf18cd5690e7e
036c3326568ddc0b4e43cba2571eebc57df07050
22192 F20101129_AABEUG fure_a_Page_137.QC.jpg
3d9f8fa154549684c2098485743e800c
9d90c175878e3c8fc19eba26c75ac03100f9131b
6380 F20101129_AABETS fure_a_Page_127thm.jpg
298378f614a1613e445ef869c8ce4837
012179340c5d87a26c2d517d8a941bf6bf9e8d24
24544 F20101129_AABDRE fure_a_Page_192.QC.jpg
f436b5ace48791a907ee0e2af50eae45
cdefc2a43e11208e0e1639100fb93f15b46ea3da
F20101129_AABDQQ fure_a_Page_127.tif
abcac8b0948222c649d46c7a5a3f0af9
764ef962a5390efa7b64b2944e8bab23716303fc
6565 F20101129_AABEUH fure_a_Page_137thm.jpg
931ed3683cbafb913ceccd8c4057bbe8
54011644acb5680443efe45c86b8837e1f03edbb
23206 F20101129_AABETT fure_a_Page_128.QC.jpg
a4eb5a2d8d7d5eaaeec88335d32b418a
f4109cf9ddb33a2931c4421335134c29f5f60400
4117 F20101129_AABDRF fure_a_Page_080thm.jpg
499e1c605c0ebaede2c76489bd4a6e95
af542876c68e5f573ede19ded9484af36e51e908
912343 F20101129_AABDQR fure_a_Page_089.jp2
4a519c1ab6eb4bd2ea9599fb805c9275
af0a8360712ecc5847b743f32be5dcb5b7330a19
20762 F20101129_AABEUI fure_a_Page_138.QC.jpg
5e8d17acf98ebf59d82e547cd0440a2c
ce491df86b76644b7eb094488550a1d732eb8694
6625 F20101129_AABETU fure_a_Page_128thm.jpg
2e52723c19a9dddb30cc81a25bd8b01e
afd168363c730864713774308fe2268524f08d80
F20101129_AABDRG fure_a_Page_151.tif
7522b0bf37d895e52413bc24928536d3
5e297d812d37e80fea2f9f20346026a494a8a4f7
41056 F20101129_AABDQS fure_a_Page_030.jpg
4d5d7b666ff435865b2c7f67a3bb378e
f7dc8b9c127149dc7c470f9bbec35baf39154dcf
5600 F20101129_AABEUJ fure_a_Page_138thm.jpg
6f5656c7ee1da4c5bfd46b27c4d16f59
386f05b7ec309e9d5190081398024ad927b58d20
9589 F20101129_AABETV fure_a_Page_129.QC.jpg
96c322cd4f79b250f5c9bcf29e030263
f2c635fdb06c97a1b76e3885b9b918f03de8c9e7
F20101129_AABDRH fure_a_Page_041.tif
7bf390f4b71a0fe2b1b47991f0598739
b5ea00061fc05b9db5bbf2fe4ff70e0fb2561705
554135 F20101129_AABDQT fure_a_Page_034.jp2
3d3aac4c5a5a72537cc7f1b75f3828e8
b58f93d6573c9f4d0b15a78612bc1dccdd559b97
6536 F20101129_AABEUK fure_a_Page_139thm.jpg
84279614ebdede5409e375a225f2b684
269ebae0167a7c509131221db4a15476ff44eeb7
4445 F20101129_AABETW fure_a_Page_130thm.jpg
e7f6dfc415c17c6351b143b480c6e73c
4e5346dfc623453eeae31a8dd42aaf0321b005ae
4545 F20101129_AABDQU fure_a_Page_014thm.jpg
7d7e819d84a914b692633b9a1a32178a
254960e44d6407444b16644994f44b117e19ac72
46269 F20101129_AABDRI fure_a_Page_146.jpg
ad02cc51fbefdc5403680b034f736492
d5cfa049189531a6cdf7664b0862676e8d84295b
14164 F20101129_AABEUL fure_a_Page_141.QC.jpg
7697cd8656a965cf0cd3e6fec169b68c
f8045aa8894792ef77d9bb27d114e2efe9eb8a70
14358 F20101129_AABETX fure_a_Page_131.QC.jpg
602d219fc76f36d7b374cd551cf52264
7a59a27d88cb6fd7d03a52c579158de875f33689
4207 F20101129_AABDQV fure_a_Page_175thm.jpg
9ad5a50422f7bcfe28278d4f80f7fcca
b3d8d81c848bc3d6443dc7d2c7116f1793a29967
14157 F20101129_AABDRJ fure_a_Page_042.QC.jpg
0bf619b82ea1893c39cc8a3cf5ab7722
593ea5401606c3bdc175a97112856aff983cdb8d
3928 F20101129_AABEVA fure_a_Page_150thm.jpg
cf1b17df34db8b81229ffbacfbf2af61
6769688f5f602805d9f86aaca24215568757dfbb
17726 F20101129_AABEUM fure_a_Page_142.QC.jpg
de95a51fbb39f9e2cd3b8ec4f5d5f4f7
95eac0c072a1ef713f488fd5b12dfd94ff9ec5c5
4500 F20101129_AABETY fure_a_Page_131thm.jpg
7e9604314ddde46e24851b9c8f8228dc
84ca9c33f5dc06424cafde08f54280c244325791
F20101129_AABDQW fure_a_Page_011.tif
9c7d6d81b774485f9262cc20b80d72f7
a16b8ffa388a874b6eb0d2cbab76f9fcbafaac58
F20101129_AABDRK fure_a_Page_086.tif
2dd8d911273bf253e48f62d87c064dca
7d986665443c56e076ad7ca101ab84eb087c4414
12160 F20101129_AABEVB fure_a_Page_151.QC.jpg
939a1ea5a30758de0ea70d822bc53f10
45b81dbf86e3856da06e27af1798a93d5cb0df4d
5081 F20101129_AABEUN fure_a_Page_142thm.jpg
559750b5cd911b90b945ae6088dd3d85
2de1432eeaa06636e5b810aa32c85cf1a3f83ee9
21347 F20101129_AABETZ fure_a_Page_132.QC.jpg
5055cf16c744db1863b08baa0df70795
1e40ce6ccb0d8df3a8fcfb73dc6a06d0cb5e81dc
789759 F20101129_AABDQX fure_a_Page_113.jp2
7721330a995119949198ee743884540f
54b9665690d3b0b3c7a9860e01b2ae4ba061cc1b
5019 F20101129_AABDRL fure_a_Page_035thm.jpg
d6d3d726120d41301ad1427c51e9fc04
5e6c09483dedaabf980df9870c3f11658a260bca
4034 F20101129_AABEVC fure_a_Page_151thm.jpg
8fabad5fd389976df76c31259235107e
b8d93cd6b83a34b9546fb5469187decf5d7c7fef
19478 F20101129_AABEUO fure_a_Page_143.QC.jpg
144d3f88099565ac0622d037958a9618
4456a8c87dc68cdbae7bd7bc45181d4e8137dac3
41330 F20101129_AABDQY fure_a_Page_065.jpg
557e0d319588d111ac8b5888accac769
673028341dd39403fa06ab1e9a3ff01aa595e6eb
F20101129_AABDSA fure_a_Page_150.tif
01531039dc21669296274b17fedf3733
b1ae6eae3c53aec77e7c9327bf92da4ac3a2a323
5728 F20101129_AABDRM fure_a_Page_103thm.jpg
a162381a5b12eb5a16823046fb394636
9531ec7fb6e5772f177f4de431ccf4f10ebc58d7
8347 F20101129_AABEVD fure_a_Page_152.QC.jpg
df6cdc7fc2b132d9a27a60a54fe3cdbd
c8e20c72c933859df0b73f6675031e0482948bd6
5626 F20101129_AABEUP fure_a_Page_143thm.jpg
ca9a0d85a22b8297b939768542051c3c
f1c5d1b05d55f49729ba8d1c647db838840a4824
F20101129_AABDSB fure_a_Page_192.tif
e5fc9ba3e0a08c90fcf3e3a0125ac71a
2c3ed60318bba8d589363c918ac2ab078249d541
73620 F20101129_AABDRN fure_a_Page_166.jpg
79a5c25905c17e3d1f6da36e921d41c5
0670138dbf638b606e464c64cb541bf02db3c6a2
20184 F20101129_AABEVE fure_a_Page_153.QC.jpg
03d1219d3f4e836cf7bc3d04bd9081b4
00b4c5c3fbef6dc9b7eefdc1681a26be141cb941
4141 F20101129_AABEUQ fure_a_Page_144thm.jpg
80420e7a5c96b972228cf4d3116ce6bf
b24f39fb043b80d9903245723ee3eca397307e8c
72075 F20101129_AABDQZ fure_a_Page_183.jpg
786eaf1ac73827883ef3ec68442e15c6
884df86e67e754fe32dbf3bbcd41f3b60785be2e
16913 F20101129_AABDSC fure_a_Page_045.QC.jpg
283a974dbf97445d0cc8ee4062ec37a6
f5c19cb50c17d84b270a6ed634c98f1b198464b1
70903 F20101129_AABDRO fure_a_Page_187.jpg
202961c08517ef1705e4fc7d75acfd08
f181a5e0a3a5af2b8d375a7171d9579765433be0
16750 F20101129_AABEVF fure_a_Page_154.QC.jpg
d1526b8ba176e8d86fb04e8f4c838a73
d210244773f17d3dea0457de1a94d3cea6ff2e4f
13858 F20101129_AABEUR fure_a_Page_145.QC.jpg
e2857343d5cff73ce2cbd26fc492e765
f3e1b518d9432be0f0f44f9e034268de6ef2600b
81880 F20101129_AABDSD fure_a_Page_191.jpg
f7d15af48e50e21239d7f8cfda853647
cd7047562ea7e4264893a8c8efb518c62fa4c71b
F20101129_AABDRP fure_a_Page_145.tif
9de6ce9027912d202a95a000e6771688
611de63890ed5ebf8db3b0ed016524c02b624b5a
22510 F20101129_AABEVG fure_a_Page_155.QC.jpg
58817a4a837f8cecc1916660ffe98836
25aa1108df98f7dc492c4d890466592382438c1f
4419 F20101129_AABEUS fure_a_Page_145thm.jpg
5316ae61ca20bf4768150277ec6ea284
14f568046fa51de986fd5292618cfda04029fbb9
F20101129_AABDSE fure_a_Page_182.QC.jpg
5b3112e6935515c1575ab11cd2c29248
3f5f6aef04bb5b802bddf3470dd86089339941a1
4795 F20101129_AABDRQ fure_a_Page_057thm.jpg
5939416c902f6ef91c7b514e2c45302e
8ddf9b328c78886a3781b95b8562034f5efea6da
6351 F20101129_AABEVH fure_a_Page_155thm.jpg
fd01921978fc8ad43f007158039b3e28
9c4ee86e5639ee84aec1715b575ced3f64bb831b
14576 F20101129_AABEUT fure_a_Page_146.QC.jpg
49c4a1e650e03f385dc4dfb8d6480fd9
29dd089b287ed67150b59b47bf2d16776f46e756
F20101129_AABDSF fure_a_Page_118.tif
327492eedaa2cfdbe8f873652f8a732b
983971bb3cb776f4cf487f11ede6633920d6876b
6557 F20101129_AABDRR fure_a_Page_005thm.jpg
069735736b597dff3373af123fa3cac1
4ab99b9fc6cf6bf68deb8109e122901138c8f761
16824 F20101129_AABEVI fure_a_Page_156.QC.jpg
4647cd9d0e236daa4db19d9cb5ec0c16
27ad87ec5a1aa97836ec5acafd0ea98b6a2f6131
22987 F20101129_AABEUU fure_a_Page_147.QC.jpg
0a3a7eb24407cd987954ad8e49a26277
7c82a958ee47e1eb528085d1a7203093fa1789e5
F20101129_AABDSG fure_a_Page_175.tif
a3679a512ed9ed80de734088ee7345d5
871bd31a584407534d9d9c9aaf7fc38c2aae5504
111685 F20101129_AABDRS fure_a_Page_166.jp2
5826a24602e5acf83495b6d2bf708b73
3c3daa96f9d30693904e1e8a07a8343b79565599
4924 F20101129_AABEVJ fure_a_Page_156thm.jpg
94ec07dbe4128b7f05f95f6ca614c559
186da417ba886feb1a03458435bfa9b77e0bbc9d
23278 F20101129_AABEUV fure_a_Page_148.QC.jpg
9065dffb7c463e8be106a8f5a0f56a3f
b4ac89cf5c80b3f1608f37cd33f9e61f691691e9
63641 F20101129_AABDSH fure_a_Page_075.jpg
02fdde1dcae0542fd8d3308aae1cfc04
64a4c7cfc17d9b91cd9f8de80471cbf737e1aea1
4420 F20101129_AABDRT fure_a_Page_141thm.jpg
9896b8324cecfedee1eb7c72545b4099
38964226255c31fba7300f4c71ed9184520f388c
19696 F20101129_AABEVK fure_a_Page_157.QC.jpg
3107c012a22b343c8a5455960f33bc48
27e50848ea868da72281d914fa8e5b306ab671dc
6852 F20101129_AABEUW fure_a_Page_148thm.jpg
072cc995fe07a042d07242bfb1dfadce
3a575d550206e7784885db548e429718d66d6eb5
F20101129_AABDSI fure_a_Page_060.tif
eb4338e7ff30b1571db92cb5dcc9d949
ad69e959783d9792f8e19c62bff4bcfc786bde14
24300 F20101129_AABDRU fure_a_Page_139.QC.jpg
58df01af1254c8e15aea374b237e0759
4f3700380cead99c305215468a5088ff6dc8513e
5737 F20101129_AABEVL fure_a_Page_157thm.jpg
10121105c4e5d961e5811c090784c8d0
49c5c414e11e705f68ff7ba58a2efa3311503e22
13224 F20101129_AABEUX fure_a_Page_149.QC.jpg
d358ecd4bd23abcdd8b647f5283500a7
3e0536adeddc49a78322b1894a75a2d0d380c1aa
917842 F20101129_AABDSJ fure_a_Page_057.jp2
47075befccbe97267528bd00c0f8e964
aeb7d6ab11bc8ab2320adb50b81ff9b190c623dc
3736 F20101129_AABDRV fure_a_Page_197thm.jpg
f48106e06878fcf4987a371bc662b30f
43513ee6efef76510fab1023ec036e63e97e807c
3564 F20101129_AABEWA fure_a_Page_165thm.jpg
c54ccd7226222b23c9ad20c3d83bf362
f32083d93b0e30a8e8ac9138aac35ab42eece6ed
18264 F20101129_AABEVM fure_a_Page_158.QC.jpg
dfaa5beb308e8e4c197ded16e8787e6e
47ffa385e36050472e1b3c7f27cfcf0815030e38
4030 F20101129_AABEUY fure_a_Page_149thm.jpg
64ae198052ab675f776e039e760d4800
81fb88dd1bae33851675b798659f72ff642a85be
4533 F20101129_AABDSK fure_a_Page_034thm.jpg
6c26469a51e62cf26017c1fe1c28e440
ea4d97b11558934ae8fe13914ae35c4b8b01eab7
F20101129_AABDRW fure_a_Page_032thm.jpg
74b466e4e7cc5086868e0de071ada327
0b825fdc63fca6c3dcf13b9c3dacbd16e56debcb
23921 F20101129_AABEWB fure_a_Page_166.QC.jpg
85121885c5908cf3796428ef6a5aafee
5cdd9d565d6bde2db2bf6b86e10176d82d29584f
5497 F20101129_AABEVN fure_a_Page_158thm.jpg
fb2ffd82151f0c2a5b5bb916a8052ec8
c44226e332c421db882256c30bfe03ed280a1f2b
11335 F20101129_AABEUZ fure_a_Page_150.QC.jpg
87e61af5a5800e75a7983aee178a3e87
64d77347d846bfe268b5ccc5a7df1d8a1e5a3f15
22056 F20101129_AABDSL fure_a_Page_068.QC.jpg
6ad798eea0446e57b33f5862379f129c
28caeb5f50a4676d55529196e9a1b38df83d9221
97067 F20101129_AABDRX fure_a_Page_050.jp2
f82091cd50a5c0b8c090ce176989c2d4
6e48a9dbf5bd4184c55d7a9228d5990ce7f6d350
6458 F20101129_AABEWC fure_a_Page_166thm.jpg
b4e104168c280ff08f37f954a2125c96
7d7a9f94a908bf5d760064b352c7e31b26d5f960
17887 F20101129_AABEVO fure_a_Page_159.QC.jpg
6218eedc8a46d1c30db0fed50c6bd37c
80ab409c1a2b0b373068afa62e05f825315ebc79
F20101129_AABDTA fure_a_Page_169.tif
f91eea53bc176bcf96b3098fb24f9b72
f1df688ab8c9aab06ef790dfce8b3a0159186f66
18829 F20101129_AABDSM fure_a_Page_049.QC.jpg
0a9887c6ee61def2ca2760d9fd945ffe
2407fd2db2fcc625dc07f0ae54fcd5e6abb7f04a
6081 F20101129_AABDRY fure_a_Page_066thm.jpg
355b6fe1d68e08cde85e91dc971fbd11
a22785cf4ead21996acd6a3ef62bc78054693c7d
19514 F20101129_AABEWD fure_a_Page_167.QC.jpg
91df53694dbe703f1160681b3796e6aa
d50ebdc588980d1711d6d4cd5e04d83f9209134b
5475 F20101129_AABEVP fure_a_Page_159thm.jpg
f07999697ec5f3a466651e9758b41f6c
d0fafcd9d147d9ef9731d2d9563f05bf75133009
23198 F20101129_AABDTB fure_a_Page_092.QC.jpg
9ade72bff578a95f2829329b9e7290b0
4bafe3b6b5057e7e0b2fd75a10438438288bb2fb
5974 F20101129_AABDSN fure_a_Page_050thm.jpg
60cb36740434a33a29302819eafd165a
1f18532d215f030ccddc9165e298efe3a1ea4b47
115537 F20101129_AABDRZ fure_a_Page_180.jp2
15f68d3c1951968142996125b095f05f
b2faf8f9e80201d5015d5c2d7040729711893298
20202 F20101129_AABEWE fure_a_Page_168.QC.jpg
ad107484d10c9e6ec4db30404eaa6838
c0114588788759c88184727951bb81435448e259
18887 F20101129_AABEVQ fure_a_Page_160.QC.jpg
06eb6d36400b3602f4e5452dc131548a
900d4ff7d080ad8d4bbbd186f5fa54f4a6e5c171
70633 F20101129_AABDTC fure_a_Page_089.jpg
5c68ed1f333acaa27ccf42372d86f9bd
8561b13c2aec58bc1068b297c53b24deae5d3d81
F20101129_AABDSO fure_a_Page_176thm.jpg
1443bb7164c7dfd7de4e56a1e4a0cc47
c3eeaf1f49a5d7f4f99eec76ceaf63b04ec443eb
F20101129_AABEWF fure_a_Page_168thm.jpg
bae80f577d5db781b730fc828e24cd4b
f05c69c8042daa8b6266db96d5141389cc587ad4
5240 F20101129_AABEVR fure_a_Page_160thm.jpg
147d5d6a748927f1eaac80ea9699d3fa
9d1491fb010be97cdad8605bab1b194582456662
5643 F20101129_AABDTD fure_a_Page_070thm.jpg
ff6a54b35220e61c5759397110cd01a9
29e4278e8ed43dd4b4fa4d9d315a1333d1eabfed
F20101129_AABDSP fure_a_Page_156.tif
e9341faf562267673073fff7d9b9e58e
ac85c92269eddecb6753349c115fa7fafa301de9



PAGE 1

RELATIONSHIPS BETWEEN MASS AND FLUX IN DENSE NONAQUEOUS PHASE LIQUID SOURCE ZONES By ADRIAN D. FURE A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2005

PAGE 2

Copyright 2005 by Adrian D. Fure

PAGE 3

ACKNOWLEDGMENTS Primary thanks goes to my committee members, Drs. Mike Annable, Jim Jawitz, Kirk Hatfield, and Suresh Rao. I would especially like to thank my cochairmen Drs. Mike Annable and Jim Jawitz for their support and encouragement and in general just being a lot of fun to work with. I want to thank Dr. Annable for being supportive when I decided to switch research topics after my first year and always encouraging me to pursue avenues of inquiry which I found most interesting. I want to thank Dr. Jawitz for getting me fired up about streamtube models and turning me on to some alternative ways to approach contaminant transport problems. iii

PAGE 4

TABLE OF CONTENTS page ACKNOWLEDGMENTS .................................................................................................iii LIST OF TABLES ............................................................................................................vii LIST OF FIGURES .........................................................................................................viii ABSTRACT .....................................................................................................................xiv 1 PARADIGM SHIFTS: THE CHALLENGE OF MANAGING DENSE NONAQUEOUS PHASE LIQUID CONTAMINATED SITES.................................1 Thomas Kuhn and Paradigm shifts ...............................................................................1 A Crisis of Technological Limitations .........................................................................3 Definition of Risk .........................................................................................................4 Alternative Performance Metrics for Source Zone Remediation .................................8 Technical Considerations ............................................................................................10 2 SIMPLIFIED SOURCE DEPLETION MODELS.....................................................12 Introduction .................................................................................................................12 General Mathematical Framework .............................................................................13 Streamtube Models .....................................................................................................14 Lagrangian Definition of NAPL Architecture.....................................................14 The Lognormal Distribution ................................................................................15 Equilibrium Streamtube Model ...........................................................................17 Concept of contribution time ........................................................................19 Derivation of contribution time ....................................................................19 Expected value for streamtube ensemble .....................................................20 Breakthrough curves and mass reduction/flux reduction relationships .......21 Rate Limited Streamtube Model .........................................................................25 Advection Dispersion Model ...............................................................................26 BTCs and Mass Reduction/Flux Reduction ........................................................27 Power Function Model ........................................................................................27 BTCs and Mass Reduction/Flux Reduction ........................................................31 Comparison of Source Depletion Models with Numerical Simulations ....................33 Constitutive Relationships ...................................................................................33 Simulation Setup .................................................................................................37 Simulation Set 1 ..................................................................................................38 iv

PAGE 5

Simulation domain and random field generation .........................................38 NAPL architectures ......................................................................................40 Dissolution profiles, model fits, and mass reduction/flux reduction relationships ..............................................................................................40 Simulation Domain and Grid Setup for Simulation Sets 2 through 4 .................44 Simulation Set 2 ..................................................................................................45 NAPL architectures ......................................................................................45 Dissolution profiles, model fits, and mass reduction/flux reduction relationships ..............................................................................................46 Simulation Set 3 ..................................................................................................48 NAPL architectures ......................................................................................48 Dissolution profiles, model fits, and mass reduction/flux reduction relationships ..............................................................................................49 Simulation Set 4 ..................................................................................................54 NAPL architectures ......................................................................................54 Dissolution profiles, model fits, and mass reduction/flux reduction relationships ..............................................................................................54 Conclusions and Recommendations for Further Research .........................................58 3 LABORATORY STUDIES OF RELATIONSHIPS BETWEEN ARCHITECTURE AND FLUX IN MODEL SOURCE ZONES..............................61 Introduction .................................................................................................................61 Modeling Approach ....................................................................................................64 Experimental Methods ................................................................................................64 Flow Cell Design .................................................................................................65 Porous Media Packing and NAPL Injection .......................................................67 Image Analysis ....................................................................................................69 Experimental Results, Modeling, and Discussion ......................................................70 NAPL Migration and General Dissolution Behavior ..........................................70 Flux Response to Changes in NAPL Architecture ..............................................72 Source Depletion Models ....................................................................................77 Additional NAPL Architectures ..........................................................................83 Conclusions .................................................................................................................87 4 AN EVALUATION OF PARTITIONING TRACERS FOR PARAMETERIZING SOURCE DEPLETION MODELS............................................................................89 Overview of Partitioning Tracers................................................................................89 Moment Based Parameter Estimation.........................................................................91 Measured moments..............................................................................................93 First and second moments of nonreactive tracer..................................................94 Domain average trajectory integrated NAPL content..........................................94 Excess spreading..................................................................................................96 Moment equations for quantifying Lagrangian NAPL architecture....................97 Inverse Modeling Based Parameterization (Lognormal Distribution)........................99 Additional Contribution Time Parameterization Methods.......................................101 v

PAGE 6

ADE based paramterization...............................................................................101 Inverse modeling with no travel time variability...............................................102 Simulation Setup.......................................................................................................103 Results.......................................................................................................................103 Predicted domain averaged NAPL saturation....................................................103 Method of moments dissolution predictions......................................................104 Inverse lognormal dissolution predictions.........................................................113 ADE dissolution predictions...............................................................................123 Comparison of techniques...................................................................................125 5 EXTENDING SOURCE DEPLETION MODELS TO FLUSHING BASED REMEDIATION.......................................................................................................139 Streamtube Models .................................................................................................139 Advection Dispersion Model ..................................................................................142 Extensions for Dispersive Mixing ..........................................................................143 Experimental Setup .................................................................................................146 Initial Architectures ................................................................................................148 Modeling of Effluent BTCs ....................................................................................149 Remedial Endpoints ................................................................................................152 6 SYNTHESIS: A FLUX BASED FRAMEWORK FOR MANAGING DNAPL CONTAMINATED SITES......................................................................................156 Three Definitions of Efficiency ..............................................................................156 Combining Three Efficiency Definitions ................................................................164 Source Identification (Pre-Reactor Based Characterization) ..................................165 Transitional Reactor-Based Characterization: Mass Flux as a Remedial Screening Tool and Qualitative Estimator of Aging ......................................166 Reactor-Based Characterization ..............................................................................168 LIST OF REFERENCES .................................................................................................175 BIOGRAPHICAL SKETCH ...........................................................................................183 vi

PAGE 7

LIST OF TABLES Table page 2-1. Parameters used in UTCHEM simulations. ................................................................38 2-2. Grid and permeability field parameters. .....................................................................39 2-3. Parameters assigned to each indicator class. ..............................................................40 2-4. Grid/random field parameters for simulation sets 2-4. ...............................................44 3-1. Summary of experimental conditions for conducted experiments. ............................65 4-1. Domain averaged NAPL saturations estimated with the different techniques. ........105 4-2. Tracer derived parameters using method of moments. .............................................107 4-3. Moment derived contribution time mean and variance. ...........................................110 4-4. Inverse lognormal derived mean and variance of contribution time. .......................118 4-5. Tracer derived parameters for ADE method. ...........................................................124 4-6. Tracer determined source depletion parameters for ADE model. ............................128 4-7. Moment derived parameters for dissolution experiment with tracers. .....................138 5-1. Fitted model parameters from UTCHEM surfactant simulations. ...........................154 vii

PAGE 8

LIST OF FIGURES Figure page 1-1. Components to consider in remedial design. ................................................................5 1-2. General approaches for managing risk at contaminated sites. ......................................7 1-3. Source zone TMDL approach of Einarson and Mackay (2001) ................................10 2-1. Eulerian and Lagrangian definitions of NAPL architecture. ......................................15 2-2. Illustration of the tau concept. ....................................................................................20 2-3. BTCs for increasing variability in contribution time (m2). ......................................23 2-4. Mass reduction/flux reduction relationships for BTCs in figure 2-3. .........................24 2-5. BTCs for increasing Peclet number. ...........................................................................28 2-6. Mass reduction/flux reduction relationships for BTCs in figure 2-5 ..........................28 2-7. BTCs for the Da model. ..............................................................................................32 2-8. Mass reduction/flux reduction relationships for Da model ........................................32 2-9. Comparison of NAPL spill using Brooks-Corey and Van Genuchten models. .........35 2-10. Raining effect observed when running UTCHEM in restart mode. .........................35 2-11. NAPL spills for set 1 (contour range: 0 to 25% saturation). ....................................40 2-12. Streamtube model fit of UTCHEM dissolution data (set 1). ....................................42 2-13. ADE model fit of UTCHEM dissolution data (set 1). ..............................................42 2-14. Mass reduction/flux reduction fit for streamtube model (set 1). ..............................43 2-15. Mass reduction/flux reduction fit for ADE model (set 1). .......................................43 2-16. NAPL spills for set 2 (contour range 0 to 25% saturation). .....................................45 2-17. Streamtube model fit of UTCHEM dissolution data (set 2). ....................................46 viii

PAGE 9

2-18. ADE model fit of UTCHEM dissolution data (set 2). ..............................................47 2-19. Mass reduction/flux reduction fit for streamtube model (set 2). ..............................47 2-20. Mass reduction/flux reduction fit for ADE model (set 2). .......................................48 2-21. NAPL spills for set 3 (contour range: 0-25% saturation). ........................................49 2-22. Streamtube model fit of UTCHEM dissolution data (Set 3). ...................................51 2-23. ADE model fit of UTCHEM dissolution data (Set 3). .............................................51 2-24. Discontinuous fit of simulation 4 (set 3). .................................................................52 2-25. Streamtube model fit of mass reduction/flux reduction relationships. .....................53 2-26. ADE model fit of mass reduction/flux reduction relationships. ...............................53 2-27. NAPL spills for set 4 (contour range: 0 to 25% saturation) .....................................55 2-28. Streamtube model fit of UTCHEM dissolution data (set 4) .....................................56 2-29. ADE model fit of UTCHEM dissolution data. .........................................................57 2-30. Streamtube model fit mass reduction/flux reduction relationships (set 4). ..............57 2-31. ADE model fit of mass reduction/flux reduction relationships (set 4). ....................58 3-1. Schematic representation of segmented box. .............................................................66 3-2. Flux plane response to changes in NAPL architecture for experiment DCA-1. Note that MF is the mass fraction ((mass at time t)/(initial mass at t = 0)) and FF is the flux fraction ((flux at time t) /(initial flux at t = 0)). .......................................73 3-3. Flux plane response to changes in NAPL architecture for experiment TCE-1. Note that MF is the mass fraction ((mass at time t)/(initial mass at t = 0)) and FF is the flux fraction ((flux at time t) /(initial flux at t = 0)). ................................................74 3-4. Flux plane response to changes in NAPL architecture for experiment DCA-2. Note that MF is the mass fraction ((mass at time t)/(initial mass at t = 0)) and FF is the flux fraction ((flux at time t) /(initial flux at t = 0)). .......................................75 3-5. Cumulative reductions in mass fraction and flux fraction with increasing dissolution time for selected ports from experiments DCA-1, TCE-1, and DCA-2. Note that the cumulative reductions are expressed in relation to the total mass reduction and the total flux reduction (e.g. 27% of the total reduction in flux for experiment DCA-2 comes from port 5). ..................................................................77 ix

PAGE 10

3-6. Mass reduction/flux reduction relationships for the entire system (upper three graphs) and for selected well segments (bottom three graphs). ...............................78 3-7. Streamtube mass fractions estimated from the image analysis technique. .................79 3-8. Comparison of the rate-limited streamtube, equilibrium streamtube, and effective Damkohler approaches for modeling source depletion from experiments DCA-1, TCE-1, and DCA-2. .................................................................................................80 3-9. ADE model fit for source depletion experiments DCA-1, TCE-1, and DCA-2. ........82 3-10. Initial NAPL distribution and effluent BTC for experiment DCA-3. ......................83 3-11. Mass reduction flux reduction relationship for experiment DCA-3 and hypothetical system developed by the superposition of selected segmented BTCs. ........................................................................................................................85 3-12. Comparison of the rate-limited streamtube, equilibrium streamtube, effective Damkohler, and ADE approaches for modeling effluent BTCs from experiments DCA-3. .....................................................................................................................86 4-1. UTCHEM generated reactive tracer fit with product of two lognormals. ..................92 4-2. RMS matrix for tracer data in figure 4-1. ...................................................................93 4-3. BTCs representing transport of a nonreactive tracer, a reactive tracer with homogenously distributed NAPL, and a reactive tracer with heterogeneously distributed NAPL. ....................................................................................................97 4-4. Summary of process used to predict dissolution using partitioning tracers and moment analysis. ......................................................................................................98 4-5. Cubic spline interpolation of arbitrary UTCHEM generated tracer data. ..................99 4-6. Set 1 tracer fit (method of moments). .......................................................................108 4-7. Set 2 tracer fit (method of moments). .......................................................................108 4-8. Set 3 tracer fits (method of moments). .....................................................................109 4-9. Set 4 tracer fits (method of moments). .....................................................................109 4-10. Set 1 predicted source depletion (method of moments ...........................................110 4-11. Set 2 predicted source depletion (method of moments). ........................................111 4-12. Set 3 predicted source depletion (method of moments). ........................................112 4-13. Set 4 predicted source depletion (method of moments). ........................................112 x

PAGE 11

4-14. Set 1 tracer fit (inverse lognormal). ........................................................................116 4-15. Set 2 tracer fits (inverse lognormal). ......................................................................116 4-16. Set 3 tracer fit (inverse lognormal). ........................................................................117 4-17. Set 4 tracer fit (inverse lognormal). ........................................................................117 4-18. Set 1 predicted source depletion (inverse lognormal). ...........................................119 4-19. Set 1 predicted source depletion (inverse lognormal-no travel time variability). ..119 4-20. Set 2 predicted source depletion (inverse lognormal). ...........................................120 4-21. Set 2 predicted source depletion (inverse lognormal w/ no travel time variability). .............................................................................................................120 4-22. Set 3 predicted source depletion (inverse lognormal). ...........................................121 4-23. Set 3 predicted source depletion (inverse lognormal w/ no travel time variability). .............................................................................................................121 4-24. Set 4 predicted source depletion (inverse lognormal). ...........................................122 4-25. Set 4 predicted source depletion (inverse lognormal w/ no travel time variability). .............................................................................................................122 4-26. Set 1 tracer fit (ADE). .............................................................................................126 4-27. Set 2 tracer fit (ADE). .............................................................................................126 4-28. Set 3 tracer fit (ADE). .............................................................................................127 4-29. Set 4 tracer fits. (method of moments). ..................................................................127 4-30. Set 1 predicted source depletion (ADE). ................................................................129 4-31. Set 2 predicted source depletion (ADE). ................................................................130 4-32. Set 3 predicted source depletion (ADE). ................................................................130 4-33. Set 4 predicted source depletion (ADE). ................................................................131 4-34. Root mean square errors for parameterization techniques. Points indicate average RMS for simulation set. Lines indicate average RMS for all simulations. 131 4-35. Impact of increased longitudinal dispersivity on tracer response and dissolution profile. ....................................................................................................................132 xi

PAGE 12

4-36. Relationship between ln(K) and contribution time parameters. .............................135 4-37. Relationship between travel time, S and contribution time variability. ...............136 4-38. Initial TCE distribution for dissolution experiment with tracers. ...........................136 4-39. Tracer BTCs (top graph) and tracer BTCs selected for moment analysis (bottom graph). ....................................................................................................................137 4-40. Experimental and predicted dissolution data. .........................................................138 5-1. Dissolution profiles for aqueous dissolution and flushing based remediation. ........140 5-2. Flushing BTCs for increasing second moment of .................................................142 5-3. Flushing BTCs for increasing second moment of .................................................144 5-4. Initial PCE distribution for experiments 1 and 2. .....................................................147 5-5. Model fit of ethanol BTCs for experiments 1 (top graph) and 2. .............................148 5-6. Model fit of experiment 1 data. ................................................................................149 5-7. Equivalent aqueous dissolution source depletion for experiment 1 .........................150 5-8. Model fit of experiment 2 data. ................................................................................151 5-9. Equivalent aqueous dissolution source depletion for box 2. ....................................151 5-10. Pre and post flux-averaged concentrations in experiments 1 and 2. ......................153 5-11. Model fits of UTCHEM surfactant floods. .............................................................155 6-1. Comparison of source zones A, B, & C using mass reduction efficiency. ...............157 6-2. Comparison of source zones A, B, & C using reduction in source discharge per reduction in mass. ...................................................................................................158 6-3. Comparison of source zones A, B, & C using reduction in source discharge efficiency. ...............................................................................................................159 6-4. Aging effects on relationships between reduction in mass and reductions in source discharge for source zone A, B, & C. .....................................................................161 6-5. Aging effects on source discharge efficiency for source zone A, B, & C. ...............163 6-6. Source discharge profiles for a fixed initial mass and alternative initial source discharge. ................................................................................................................164 xii

PAGE 13

6-7. Figure 6-2 modified to include net work input and site specific flux-based performance metric. ...............................................................................................165 6-8. Source discharge profile and down-gradient flux planes at selected time steps. .....168 6-9. Key assumptions required when using spatially resolved flux information for flux mapping. .................................................................................................................170 6-10. Use of unique nonreactive tracers for flux mapping. .............................................171 6-11. Conceptual example of using figure 6-5 in the remedial design process. ..............174 xiii

PAGE 14

Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy RELATIONSHIPS BETWEEN MASS AND FLUX IN DENSE NONAQUEOUS PHASE LIQUID SOURCE ZONES By Adrian D. Fure August, 2005 Chair: Michael D. Annable Cochair: James W. Jawitz Major Department: Environmental Engineering Sciences The relationship between reductions in dense nonaqueous phase liquid (DNAPL) mass and the corresponding reduction in contaminant source discharge is evaluated in experimental and numerical studies. Analytical solutions for modeling the relationships between mass and source discharge under natural gradient conditions are developed. The models are extended to also consider flushing based source zone remediation. The models are compared with numerical and experimental studies of natural gradient dissolution and flushing based remediation. The use of reactive and nonreactive tracers for parameterizing the analytical models is also evaluated. The utilization of the models in a flux based framework for addressing DNAPL contaminated sites is discussed. xiv

PAGE 15

CHAPTER 1 PARADIGM SHIFTS: THE CHALLENGE OF MANAGING DENSE NONAQUEOUS PHASE LIQUID CONTAMINATED SITES In this chapter I will provide a brief historical sketch of the approaches utilized to manage sites contaminated with dense nonaqueous phase liquids (DNAPLs). I have attempted to frame this discussion in terms of Thomas Kuhns The Structure of Scientific Revolutions as I believe the transition from mass-based to flux-based performance metrics for DNAPL source zones is a paradigm shift, marked by many of the characteristics outlined by Kuhn. Another reason I have elected to incorporate Kuhn is his notion of normal science or paradigm articulation. I would consider most of my own work as fitting into this classification, i.e. I feel that much of my work has been directed towards advancing and refining an already introduced flux-based paradigm (Rao et al., 2001). Thomas Kuhn and Paradigm shifts The notion of paradigmsthe idea that the direction of scientific inquiry and the interpretation of results is significantly influenced by background contextscan be to traced to Thomas Kuhns influential book The Structure of Scientific Revolutions. Kuhn was a graduate student in theoretical physics at Harvard within sight of the end of [his] dissertation when he became involved with an experimental college course treating the physical sciences for the non-scientist. This course would drastically alter his career path, taking him from physics to the history of science and eventually back to the 1

PAGE 16

2 philosophical underpinnings of science. He would publish The Structure of Scientific Revolutions 15 years later in 1962. Kuhn suggests that paradigm shifts share two essential characteristics, they are sufficiently unprecedented to attract a group of adherents away from competing modes of scientific inquiry and they are open-ended enough to leave all sorts of problems for the redefined group of practitioners to resolve (Kuhn refers to the resolution of these problems as normal science or paradigm articulation). Although Kuhn dealt only with truly revolutionary paradigm shifts, such as the shift from the Ptolemaic system of astronomy (which placed the earth at the center of the solar system) to Copernican astronomy (which placed the sun at the center of the solar system), the same patterns of paradigm shifts have been applied extensively and perhaps often inappropriately to other less revolutionary, non-scientific modes of inquiry that have undergone fundamental change in the background context within which they operate. In my opinion, a shift from mass-based to flux-based performance metrics for dense nonaqueous phase liquid (DNAPL) source zones constitutes a paradigm shift which, to borrow again from Kuhn, requires the reconstruction of prior theory and the re-evaluation of prior fact. This is especially true with regards to surfactant-based flushing technologies, a technology adapted from the oil industry which, by definition, has been previously designed and optimized to most effectively remove mass from the system, not to yield the largest reduction flux. The mass reduction/flux reduction relationships discussed throughout serve as a sort of paradigm transfer function seeking to take at least some of the large body of work that has operated within the mass reduction paradigm and cast it into the new flux based paradigm.

PAGE 17

3 A Crisis of Technological Limitations Kuhn suggests that paradigm shifts occur as the result of crisis or the inability of an existing paradigm to describe certain observed behavior. In the case of Ptolemaic astronomy, the crisis arose from complications describing astronomical observations outside of planetary positions, such as equinoxes. Kuhn suggests that out of crises arise any number of competing paradigms, which vie for preeminence within the scientific community. The initial stages of a paradigm shift are invariably characterized by elevated tension between members belonging to the previous paradigm and those members aligning themselves with any one of the competing paradigms. Eventually, a dominant paradigm arises (e.g. that of Copernican astronomy) leading to the gradual disappearance of both the original paradigm and any competing paradigms. When, in the development of a natural science, an individual or group first produces a synthesis able to attract most of the next generations practitioners, the older schools gradually disappear. In part their disappearance is caused by the members conversion to the new paradigm. But there are always some men who cling to one or another of the old views, and they are simply read out of the profession, which thereafter ignores their work. The new paradigm implies a new and more rigid definition of the field. Those unwilling to or unable to accommodate their work to it must proceed in isolation or attach themselves to some other group. The crisis that precipitated the proposed shift from a mass based to a flux based paradigm is not one in which the preexisting paradigm was unable to describe a given set of physical observations, but rather a crisis of technological limitations. In this case, the technological limitations are those associated with the suite of potential remediation technologies available to those charged with managing sites contaminated with DNAPLs, none of which perform as a silver bullet technology capable of efficiently removing or destroying all of the contaminant mass from the subsurface and thereby, prima facie, appearing as failures. A paradigm shift occurs by considering these same technologies

PAGE 18

4 not for their ability to remove or destroy mass, but for their ability to reduce risk, which does not require the complete removal of mass. Such a shift, as Kuhn suggests, has been characterized by initial resistance, both from those within the regulatory and scientific communities. I hope the flux paradigm will soon be characterized, also as Kuhn suggests, by a large influx of research that advances and refines the new paradigm (paradigm articulation) as it gains acceptance as the more rational paradigm. Definition of Risk Of critical importance to the flux-based paradigm is the contention that there is no direct linkage between risk and the amount of DNAPL mass present at a given site as a receptor will virtually never come into direct contact with a contaminant in the nonaqueous phase. The risk associated with a given DNAPL contaminated site is thereby derived from that particular sites ability to transfer contaminant mass from the nonaqueous phase to aqueous or gaseous phases which can migrate beyond the boundaries of the source zone and potentially interface with contaminant receptors. An evaluation of risk would thereby require an evaluation of the rate at which contaminant mass is being transferred from the nonaqueous to the aqueous and/or gaseous phases (source discharge) coupled with predictive contaminant fate and transport analysis to determine potential contaminant receptors, exposure pathways, and associated toxicological impacts. While acknowledging gaseous phase transport, the following analysis is primarily directed towards aqueous phase transport. Therefore, unless specifically stated the term risk will be used hereafter to specifically refer to receptor exposure to a dissolved contaminant plume.

PAGE 19

5 Historical Sketch of Approaches for Managing DNAPL Contaminated Sites Based upon the discussion outlined in the previous section there are three components to consider when managing the risk associated with a DNAPL contaminated site: the source, the receptor, and the pathway by which the contaminant is transferred from the source to the receptor (figure 1). Management strategies can conceptually be delineated into three groups based upon which of the three components (the source zone, the exposure pathway, or the receptor) are targeted (figure 2). Source ZoneTransport Vehicle (e.g. flowing groundwater) ReceptorExposure Pathway Nonaqueous PhaseMass Transfer Figure 1-1. Components to consider in remedial design. Institutional based approaches provide an immediate short-term reduction in risk by removing the receptor from the exposure pathway. An example of such an institutional-based approach would be taking a drinking water well offline due to the interaction of the well with a dissolved contaminant plume. Institutional approaches of course make no

PAGE 20

6 attempt to remediate or restore sites to a level where the site can again be used for industrial, residential, or recreational purposes. Plume management based approaches target the exposure pathway by either removing/stripping the contaminant from the transport vehicle (e.g. reactive barriers), removing the transport vehicle from the exposure pathway (e.g. pump and treat), or by enhancing transformation of the contaminant to a more benign derivative as it migrates along the exposure pathway (e.g. enhanced bioremediation). Pump and treat, which extracts contaminated groundwater from the subsurface and treats it above ground, historically has been the approach most employed at DNAPL contaminated sites (NRC, 1994). The major limitation associated with pump and treat technologies when applied to DNAPL sites stems from the longevity of most DNAPL source zones, which tend to function as sources of groundwater contamination for extended periods of time. Wide recognition in the early to mid 90s that pump and treat was by and large ineffective at most DNAPL sites due to the longevity of DNAPL source zones (NRC, 1994; Travis and Doty, 1990) sparked interest in approaches directed towards the removal or destruction of the source of contamination.

PAGE 21

7 Source Zone Remediation Plume Management Institutional Eliminate source discharge by complete removal/destruction (e.g. surfactant flushing; chemical oxidation)Reduce source discharge via partial removal/destructionReduce source discharge by limiting access of transport vehicle to source (e.g. slurry walls)Remove transport vehicle from exposure pathway (e.g. pump and treat)Remove contaminant from transport vehicle in-situ (e.g. permeable reactive barriers)Enhance in-situ transformation of contaminant as it travels along the exposure pathway (e.g. enhanced bioremediaton)Source ZoneExposure PathwayReceptorRemove receptor from exposure pathway (e.g. taking drinking water well offline) Figure 1-2. General approaches for managing risk at contaminated sites. It was of course initially hoped for that the research and funding directed towards source zone remediation would yield technologies capable of complete contaminant removal/destruction. Source zone remediation is inherently complex however, from the initial characterization of the site to the eventual implementation of the given technology, and has, as a result, prohibited the development of a silver bullet type technology which can efficiently remove or destroy all of the contaminant mass in the source zone (Soga et al., 2004). Several source zone remediation technologies have however shown the ability to remove/destroy a large portion of the contaminant mass which introduces questions as to the benefits of partial source removal and whether those benefits are

PAGE 22

8 commensurate to the often considerable cost required to achieve partial source removal (E.P.A., 2003). Alternative Performance Metrics for Source Zone Remediation A recent E.P.A. report (EPA, 2003) groups source zone remediation performance metrics into three categories. Type I metrics are measurements that can be reliably acquired and are commonly used such as the total mass of DNAPL removed from the subsurface and changes in resident groundwater concentrations. Type II metrics are metrics that can sometimes be measured, but are not in wide use such as remaining DNAPL mass and DNAPL architecture. Type III metrics are metrics that are theoretically possible and under development such as mass flux and mass discharge. To consider partial source removal as a viable remedial alternative requires an alternative performance metric that is not based upon reducing local resident concentrations in the source zone below regulatory maximum contaminant levels (MCLs) (type I metrics), but rather on reducing risk to down-gradient receptors below a given threshold. A risk-based analysis requires consideration of both the source strength (type III metrics) and fate and transport modeling of the dissolved contaminant plume. Rao et al. (2001) discuss attenuation based performance metrics for aggressive source zone remediation technologies based on whether the contaminant flux is reduced to a level where the following four criteria are met: (a) the spatial extent of the existing dissolved plume is stable or decreases; (b) the total contaminant mass within the plume is constant or diminishing; (c) both the average concentration and the range in concentrations is diminishing; (d) contaminant fluxes decrease at succeeding control planes along the dissolved plume. If these conditions are met, then the dissolved plume is stable (source

PAGE 23

9 strength equals attenuation capacity) or decreasing (source strength is less than the attenuation capacity) (Rao et al., 2001). When considering the ecological and human impacts of a migrating dissolved contaminant plume, one is typically concerned with communication between the plume and down-gradient surface water bodies and/or drinking water supply wells. In relation to surface water bodies, where there are potentially multiple contaminant sources contributing to the overall contaminant load, a mass discharge approach for evaluating and prioritizing source zone remediation is clearly more appropriate than an MCL approach as a mass discharge approach would fit into the preexisting Total Maximum Daily Load (TMDL) framework. Einarson and Mackay (2001) suggest a similar TMDL type approach for the protection of water supply wells with the pollutant mass balance being performed on the capture zone as opposed to the surface water body. The approach used in the source zone TMDL approach to estimate the contaminant concentration at a drinking water supply well is to divide the total pollutant discharge to the capture zone by the total extraction flow rate from the capture wells (figure 3). As noted by Einerson and Mackay (2001), the TMDL approach, if implemented, would establish a more rational framework for addressing multiple sources and coping with the reality that the resources available for addressing groundwater contamination are substantially smaller than the scope of the problem. In the case of multiple sources, a TMDL approach would also provide an opportunity for emissions trading in manner directly analogous to the current approach utilized for air and water based emissions.

PAGE 24

10 MdMd Capture ZoneDrinking Water Supply Wells wellsjjsourcesidwellQMCi11Dissolved Contaminant Plumes Q Figure 1-3. Source zone TMDL approach of Einarson and Mackay (2001) Technical Considerations The purpose of this opening chapter is to provide a socio-political context for the more technical portion that follows. I want to emphasize that a flux-based paradigm/source zone TMDL approach is not widely accepted and is somewhat controversial. Aside from mentioning this controversy in passing and simply stating that the source zone TMDL approach is, in my opinion, the most rational and appropriate way to approach DNAPL contaminated sites, I will not evaluate the socio-political components of this controversy in any sort of systematic fashion. My main focus here is to look at some of the technical ramifications of adopting a flux based paradigm with my main thesis being that many of the technical challenges associated with the management of DNAPL contaminated sites are more tractable in a flux-based paradigm.

PAGE 25

11 In the following chapters, analytical models are introduced which are useful for addressing some of the difficult questions surrounding the benefits of partial source removal. In chapter 2, the models are introduced and compared with numerical simulations conducted with the multiphase flow and transport simulator UTCHEM to determine if the analytical models are capable of reproducing the range of dissolution profiles produced by UTCHEM. Chapter 3 consists of experiments designed to qualitatively investigate linkages between source discharge and NAPL architecture. The experiments of chapter 3 were also designed to validate some of the underlying assumptions of the analytical models. In chapter 4, the use of tracers to parameterize the analytical models was investigated. The overall performance of the tracers served as an additional validation of the model assumptions. In chapter 5, the models are extended to handle flushing-based (surfactant/cosolvent) remediation. In chapter 6, after the models have been validated, a framework for utilizing the models is suggested. The models are also used to answer some important questions related to the benefits of partial source depletion.

PAGE 26

CHAPTER 2 SIMPLIFIED SOURCE DEPLETION MODELS Introduction The decision process for risk-based corrective action at sites contaminated with DNAPLs introduces several difficult questions, many of which do not have answers that are widely accepted throughout the technical and regulatory communities. Of critical importance when considering plume management vs. aggressive source zone remediation is the prediction of the long term natural-gradient dissolution of the source zone to determine the required duration of plume management. The use of numerical simulators for the prediction of long term source depletion is perhaps impractical due to the sensitivity of a migrating DNAPL to small scale changes in permeability, which leads to complex entrapment configurations (e.g., Keuper et al., 1993) which are next to impossible to characterize using currently available technologies (Parker and Park, 2004). The practical limitations of conventional numerical simulators has led researchers to propose simplified screening models which serve as analogs to more sophisticated numerical simulators (Parker and Park, 2004; Zhu and Sykes, 2004). These screening models require relatively few parameters and can be efficiently coupled to contaminant fate and transport models as source terms. Screening models are also useful for evaluating uncertainty in key input parameters (e.g., NAPL mass) and the associated effect of that uncertainty on predicted source depletion. In this chapter three simplified NAPL source depletion models are considered, a streamtube model, an advection dispersion model, and the power function model of Parker and Park (2004). 12

PAGE 27

13 There are two key questions to consider in relation to utilizing simplified source depletion models in the decision process for DNAPL contaminated sites. The first question is whether the model is capable of reproducing the range of source depletion behavior observed in field, laboratory and numerical studies of source depletion. The second question is how would the parameters for these models be determined in order to be of use to site managers? There are two approaches that can be employed. The first approach is to evaluate an expected range of model parameters based on typical values encountered in field, laboratory and numerical studies and incorporate that range/uncertainty into the site specific decision process. The second approach is to try and physically measure the parameters in the field. In order to do this, the parameters must take on some physical meaning (as opposed to being just fitting parameters) and there must be an available technology to measure the parameter. In this chapter I will investigate part of the first question: whether these simplified source depletion models can serve as mathematical analogs to more sophisticated numerical simulators. In chapter 3, I will evaluate these models with respect to experimental studies of source depletion. In chapter 4, I will investigate the second question, whether these models can be parameterized using existing technologies. General Mathematical Framework The three models introduced below take on a similar formulation where the flux-averaged concentration exiting a source zone is related to a source depletion term (S D ) as in equation 3-1 below. )(1)(TSCfTCDscf (2-1)

PAGE 28

14 Where C f is the flux averaged concentration exiting the source zone [M/L 3 ], C s is the solubility of the NAPL in groundwater (or flushing solution) [M/L 3 ], and T is the flushing duration. The contaminated fraction, f c is the fraction of streamtubes initially containing NAPL. It is expected that the flux averaged concentration exiting a source zone will be well below solubility (MacKay et al., 1985) resulting from both a physical nonequilibrium brought about by streamtubes that do not intersect with the contaminated fraction of the source zone (manifested in f c ) and are subsequently blended in the extraction well with contaminated streamtubes, and to a lesser extent by chemical nonequilibrium brought about by rate limited mass transfer from the NAPL to the flowing aqueous phase. The three model types introduced below share the same general form as equation 3-1, differing only in the way in which the source depletion term (S D ) is formulated. Streamtube Models Lagrangian Definition of NAPL Architecture Sale and McWhorter (2001) use the term source zone architecture to refer to the geometry, spatial distribution, and saturation of a collection of subzones comprising a source zone. Quantitatively there are two ways to define NAPL architecture, an Eulerian approach and a Lagrangian approach. In the Eulerian definition of source zone architecture, point values of NAPL saturation are expressed in terms of their magnitude and location in an arbitrary coordinate system. The concept behind the Lagrangian-based definition of NAPL architecture is to resolve the source zone into a collection of streamtubes with variability in NAPL saturation (and mass transfer rate coefficients) along each streamtube integrated and transformed into an effective value for each streamtube (Cvetkovic et al., 1998; Jawitz et al., 2003a). Following Jawitz et al. (2003a),

PAGE 29

15 this integrated parameter is referred to as the trajectory integrated NAPL content, wNSS (where is the trajectory integrated NAPL saturation, is the porosity [L NS 3 L -3 ], and w is the water content [L 3 L -3 ]). The paramteris utilized in place of because it exhibits a range of (0,) as opposed to the (0,1) range exhibited by which is consistent with lognormal and gamma probability distributions. The Lagrangian definition of source zone architecture would then be defined through some form of a probability distribution which would describe the distribution of trajectory integrated NAPL contents among the streamtube network. S NS NS NAPL saturation/velocity defined in terms of x,y,zlocation Eulerian Definition of NAPL Architecture Source Zone discretized into a network of streamtubes Mass FluxLagrangian Definition of Source Zone Architecture Figure 2-1. Eulerian and Lagrangian definitions of NAPL architecture. The Lognormal Distribution In the streamtube models discussed below, the travel time, t, and are considered random variables, described by a lognormal probability density function (pdf). S

PAGE 30

16 222))(ln(exp21)(xxxf (2-2) Where is the mean of the log transformed variable x and is the standard deviation. When considering process models with two random variables, a joint pdf must be utilized if the random variables are correlated. )1(22exp121),(2222yyxxyxYYYYxyyxf (2-3) xxxxY)ln( (2-4) yyyyY )ln( (2-5) Where is the correlation coefficient between x and y. If x and y are uncorrelated then (2-3) can be expressed as the product of f(y) and f(x). The expected value of x is given by the following: (2-6) dxxfxxEnn0)( The analytical solution of (2-6) is the well known moment generating function for a lognormal distribution (e.g. Jury and Roth 1990): 2exp22NNmxEXNN (2-7) Given the first two moments, any higher moments can be determined by using (2-7) and solving for and 2)ln()ln(221XxXmm (2-8)

PAGE 31

17 (2-9) )ln(2)ln(12XXXmm A final property of the lognormal distribution that has relevance to the streamtube models below is that the product of two lognormals, Z = XY, can be substituted for with another lognormal with mean and variance: YXZ (2-10) 222YYXXZ (2-11) Equilibrium Streamtube Model The first model considered is a version of a model presented in Jawitz et al. (2005) that has been modified for natural gradient source depletion. The model is built upon the contention that the Lagrangian-based NAPL architecture and its relation to the velocity field exert primary influence over dissolution dynamics, where additional mechanisms not explicitly accounted for in the streamtube model, such as rate-limited mass transfer, transverse dispersion, and relative permeability exert only secondary influence, causing perturbation about the mean behavior. The assumption that a parcel of water that traverses the length of a contaminated source zone, from injection plane to extraction plane, is at solubility by the time it reaches the extraction plane is well supported in the literature as column studies of aqueous dissolution have shown the spatio-temporal scales required to reach equilibrium concentrations to be on the cm scale for typical groundwater velocities (Miller et al., 1990; Powers et al., 1994). The fundamental assumption of the streamtube models is that dissolution dynamics are governed primarily by the NAPL architecture and its relation to the velocity field such that additional mechanisms not accounted for in the mathematical formulation such as relative

PAGE 32

18 permeability and transverse dispersion introduces negligible error. The major advantage of simplifying dissolution dynamics by neglecting mechanisms such as transverse dispersion and relative permeability is that it allows for the development of relatively compact analytical expressions that can be parameterized with tracer tests (Jawitz et al., 2000a) and are potentially of more practical value to site managers (Parker and Park, 2004). The assumption that NAPL architecture is a principal controlling mechanism of dissolution dynamics is well supported by previous numerical modeling studies of NAPL infiltration and dissolution/remediation (Mayer and Miller, 1996; Dekker and Abriola, 2000; Lemke et al., 2004b; Parker and Park, 2004). The effects of neglecting transverse mixing between streamtubes has been investigated previously by Fiori (1996) and Berglund and Fiori (1997). Berglund and Fiori (1997) determined that the advection-only approach is appropriate for sorbing solutes under typical field conditions. Other researchers have incorporated mixing effects into streamtube formulations (Cirpka and Kitanidis, 2000; Ginn 2001) in order to model reaction processes that are highly dependent on transverse mixing (Kapoor et al., 1997; Kapoor et al., 1998; Cirpka et al., 1999). Laboratory studies have documented the importance of transverse mixing in the dissolution of NAPL pools (Chrysikopoulos et al., 2000; Eberhardt and Grathwohl, 2002), yet the effect of transverse mixing when considering dissolution dynamics at the field scale are somewhat less certain. For example, in the numerical simulations of surfactant-based DNAPL source zone remediation of Lemke et al. (2004b), the mass flux and PCE recovery were found to be insensitive to transverse dispersivity values ranging from .0075m to 0.06m, values extending well beyond the upper limit reported in field and

PAGE 33

19 laboratory experiments (Klenk and Grathwohl 2002). Further research is needed to determine the effect of neglecting transverse dispersion on field scale dissolution dynamics. The effect of neglecting relative permeability is also uncertain and an area for further research. If relative permeability effects are large the flow paths (i.e. the travel time and NAPL associated with that travel time) may be temporally dynamic, changing throughout the dissolution process (Geller and Hunt, 1993; Powers et al., 1998). Concept of contribution time The model formulation below is based on the concept of contribution time, denoted throughout as The parameter can be thought of as the time at which a given streamtube is depleted of NAPL under a set of specific dissolution/flushing conditions. For an arbitrary flushing/dissolution duration (T), all streamtubes with values less than T are clean, while all streamtubes with values greater than T are still contaminated. Under the equilibrium assumption, the concentration discharging from the contaminated streamtubes is at solubility while the concentration discharging from the clean streamtubes is zero. This concept is illustrated in figure 2-2. Derivation of contribution time Utilizing the assumptions listed above it is possible to consider a simple mass balance on an individual streamtube: StKvACLASQCMfwsNnsn (2-12) where M n [M] is the NAPL mass in the individual streamtube, Q is the flow rate [L 3 /T], n is the NAPL density, v is the velocity in the streamtube [L/T], L is the length of the streamtube, A is the cross sectional area of the streamtube [L 2 ], t is the residence time or the travel time and K f is ratio of the NAPL density to the solubility ( n /C s ). Utilizing

PAGE 34

20 Figure 2-2. Illustration of the tau concept. equations 2-10 and 2-11 and substituting a lognormal random variable t S for the product of two random lognormals t and, equation 2-12 reduces to S SftK with mean and variance given by: tStS 222ttSStS (2-13) Expected value for streamtube ensemble Using the concept of contribution time, the ensemble average for a collection of streamtubes is given by equation 2-1 with S D equal to the cumulative distribution function of :

PAGE 35

21 )(1)(TSCfTCDscf with: )()( pTSD (2-14) where p() is the cumulative distribution function of the contribution time (). The mean and variance for the contribution time can be determined by employing equations 2-7 through 2-9: 2)ln()ln(221mm (2-15) (2-16) )ln(2)ln(12mm (2-17) StfmKm11 (2-18) StfmKm222 Breakthrough curves and mass reduction/flux reduction relationships The magnitude of the average contribution time (represented by m 1 ) increases with a decrease in average velocity or an increase in the domain averaged NAPL saturation. Contribution time heterogeneity (manifested in m 2 ) increases by introducing more variability to the flow field and/or the NAPL distribution. Contribution time heterogeneity also increases through an increase in positive correlation between travel time and NAPL saturation. The physical meaning of a positive correlation between NAPL saturation and travel time is that more of the NAPL mass would be present in the lower conductivity, lower velocity media. Qualitative studies of NAPL spills (e.g., Keuper et al., 1993; Schwille, 1998) have demonstrated that a migrating NAPL is extremely sensitive to small scale changes in permeability, moving preferentially through

PAGE 36

22 courser grained media, suggesting a negative correlation between travel time and NAPL saturation. A suite of breakthrough curves (BTCs) for a fixed average contribution time (fixed m 1 ; i.e., fixed NAPL mass) and increasing variability in contribution time (expressed through increasing m 2 ) is displayed below in figure 2-3. Within the streamtube framework outlined above, the shape of the BTC exiting the source zone is controlled by the distribution of NAPL saturations and velocities among the streamtubes rather than the equation that describes transport at the scale of the individual streamtube. For the case of a homogenous distribution of NAPL saturations in a uniform flow field (m 2 = 1.01 in figure 2-3), the dissolution profile resembles a step function with each streamtube depleted of mass at an identical flushing duration. This type of dissolution behavior would most resemble column-based experiments of uniformly distributed residual NAPL saturations (Imhoff et al., 1994; Powers et al., 1994), the PCE component of the relatively homogenous Borden emplaced source (Frind et al., 1999), and the theoretical analysis of Sale and McWhorter (2001). In this scenario there is limited reduction in contaminant discharge until most of the NAPL mass has been depleted (figure 2-4). With increased variability in contribution time ( distribution) the dissolution profile moves from a step function, to more of an exponentially decaying dissolution profile, and eventually to a dissolution profile at high variability that is characterized by a rapid drop in flux followed by extensive tailing. It is possible to consider the BTCs of figure 2-3 as having two distinct components: an accessible component associated with an initial flux decrease and a recalcitrant

PAGE 37

23 component associated with the late-term tailing portion. As variability in contribution time increases, more mass is shifted to the recalcitrant late-term tailing portion of the Figure 2-3. BTCs for increasing variability in contribution time (m2). BTC, leading to a larger initial reduction in flux for a given reduction in mass. A similar concept used recently by Lemke et al. (2004b) and Christ et al. (2005) is the ganglia to pool ratio (GTP), which describes the ratio of NAPL existing in a residual phase to that existing as higher saturation pools. This is somewhat analogous to describing the NAPL architecture using the Lagrangian approach outlined previously. A high variance lognormal distribution of trajectory integrated NAPL contents is one in

PAGE 38

24 0.3 PV0.6 PV1 PV2 PVcumulative reduction in masscumulative reduction in source discharge Figure 2-4. Mass reduction/flux reduction relationships for BTCs in figure 2-3. which the majority of the NAPL mass is associated with a small number of high streamtubes. In numerical simulations of DNAPL spills conducted by Christ et al. (2005), results indicated that as the variance of the permeability field increased, the GTP ratio decreased, meaning that the NAPL distribution became more skewed towards the presence of NAPL in pools as opposed to residual. The results of Christ et al. (2005) are in agreement with the previous work of Mayer et al. (1996) and Dekker et al. (2001). Numerical simulations of aqueous dissolution have indicated that that as the NAPL distribution becomes increasingly skewed towards more recalcitrant pools, there is increased tailing in the BTC (Parker and Park, 2004; Mayer et al., 1996), which is in agreement with the streamtube models outlined above. In the high resolution simulation S

PAGE 39

25 of Parker and Park (2004), the NAPL architecture was characterized by residual zones of varying lengths (responsible for the large initial decrease in flux) overlying a laterally extensive pool perched on top of a clay aquitard (responsible for long term tailing). It is instructive to notice that m 1 is equal to the contribution time for a step function dissolution profile, that is m 1 is equal to the total mass divided by the initial mass discharge. An interesting property of the streamtube model (lognormal distribution) outlined above is that for a given value of m 1 the range of possible BTCs for different values of m 2 always intersect at a flushing duration equal to m 1 and at a flux-averaged concentration equal to 0.309 (figure 2-3), meaning that the tie line for a flushing duration equal to m 1 is a parallel to the mass reduction axis and equal to a flux reduction of .691 (see figure 2-4 for a flushing duration of 1 PV). An additional property of interest is that as m 2 approaches infinity, the total mass left of the intersection point approaches 0.309. The implication of this is that the lognormal distribution cannot produce mass reduction/flux reduction relationships that extend beyond the .309 mass reduction/(1-.309) flux reduction point in the upper left portion of the mass reduction/flux reduction plot. This is entirely a property of the lognormal distribution and has no physical meaning. Cases of more extreme tailing could be accounted for with alternative probability distributions. Rate Limited Streamtube Model Berglund (1997) presented a stochastic-advective model for rate-limited nonaqueous phase liquid dissolution. tStC T CN with:

PAGE 40

26 CCSktSsN (2-19) Where k is the mass transfer rate coefficient that is linearly related toSwith units of [T -1 ]. The analytical solution of (2-19) is given below in (2-20). 1)(expexp)(exp1NsNsscftTkCtSktTkCCfC (2-20) Advection Dispersion Model Jury (1990), among others, have demonstrated the mathematical similarity of the advection dispersion equation and the lognormal travel time distribution. An alternative to utilizing the cumulative distribution function of for S D (T) would then be the dimensionless analytical solution for the advection dispersion equation (step injection; injection and detection in flux). )(1)(TSCfTCDscf with: )(4)exp(21)(421)(TRRTPerfcPTRRTPerfcTSD (2-21) Where P is the Peclet number and is used here in a manner analogous to the second moment of the contribution time, functioning as variability index to describe the combined effects of hydrodynamics and NAPL architecture. The retardation factor, R, is used to describe the total mass of NAPL in the source zone in a manner similar to the first

PAGE 41

27 moment of contribution time. A useful property of equation 2-21 is that the resulting equation integrates to R. This property is widely known throughout the chemical engineering literature as column holdup time. (2-22) 0)(1RTSD Using equation 3-22, R can be used to represent the total NAPL mass in the source zone. NfSKR (2-23) Where S N is the domain average NAPL saturation. BTCs and Mass Reduction/Flux Reduction A suite of BTCs and mass reduction/flux reduction relationships for a fixed NAPL mass (fixed R) and increased contribution time heterogeneity (increased P) is displayed below in figures 2-5 and 2-6. As shown in figure 2-5 and 2-6, the advection dispersion model produces dissolution profiles and mass reduction/flux reduction relationships that are similar to the streamtube model. One difference of note is that the ADE equation is capable of producing more severe tailing, leading to mass reduction/flux reduction relationships that extend farther into the upper-left corner of the mass reduction/flux reduction plot. Power Function Model Parker and Park (2004) presented a simplified model for estimating DNAPL source zone depletion based on the concept of an effective Damkohler number ( qLkDaeff/( ) where eff [T -1 ] is the effective mass transfer coefficient, L is the source zone length in the mean flow direction, and q [LT -1 ] is the average Darcy flux for the source zone). The

PAGE 42

28 Figure 2-5. BTCs for increasing Peclet number. 0.3 PV0.6 PV1 PV2 PVcumulative reduction in masscumulative reduction in source discharge Figure 2-6. Mass reduction/flux reduction relationships for BTCs in figure 2-5

PAGE 43

29 concept of a effective Damkohler number was also utilized previously by Mayer and Miller (1996). For brevity this approach will be referred to as the Da approach or Da model throughout. The concept of the Da model is to relate changes in the field-scale mass transfer rate coefficient to changes in the global values of NAPL mass and the average groundwater velocity as in equations 2-24 and 2-252 below (Parker and Park, 2004). qLCtCeffsoutexp1)( (2-24) 21)(osoeffMtMKq (2-25) Where C out [ML -3 ] is the flux averaged concentration exiting the source zone, L is the length of the source zone, sK [LT -1 ] is the average saturated hydraulic conductivity of the source zone, M(t) is the NAPL mass in the source zone at time t, M o is the initial NAPL mass in the source zone, and o [T -1 ] 1 and 2 are fitting parameters. Conceptually, there are similarities between the Da approach and the streamtube approaches outlined above. The first similarity noted is that of the 1 parameter. For cases where 1 = 1, as in the high resolution simulation conducted by Parker and Park (2004), the flux-averaged concentration exiting the source zone is independent of the mean groundwater velocity, meaning that changes in contact time between advecting groundwater and the NAPL phase brought about by temporal changes in groundwater velocity do not result in rate-limiting effects. In this sense, the 1 =1 case for the Da model is similar to the equilibrium streamtube model where the flux-averaged concentration exiting the source zone is insensitive to changes in the residence time of the

PAGE 44

30 advecting groundwater. The second conceptual similarity is between o and f c As noted previously, in the streamtube formulation f c is a measure of the fraction of total streamtubes intersecting the contaminated portion of the source zone. Mathematically, both o of f c act to scale the solubility of the NAPL to the initial flux-averaged concentration exiting the source zone. Although o is treated in Parker and Park (2004) as a fitting parameter, it is suggested that the physical meaning of o is similar to f c and is primarily a measure of the fraction of streamtubes intersecting the contaminated portion of the source zone. The third and most important similarity between the Da model and the streamtube models is the relationship between 2 and the way in which the velocity and NAPL distributions are described in the streamtube models. In the Da model, 2 is referred to by Parker and Park (2004) as a mass depletion exponent, which they describe as being related to the groundwater velocity distribution, the DNAPL geometry, and the correlation between the two. The main difference then, between the Da model and the streamtube models, is the specificity in which the velocity and NAPL distributions are described. In the streamtube model, variability in velocity and NAPL saturation along a single streamtube is integrated and transformed into an effective value (contribution time ) for the individual streamtube, with the variability of this distribution controlling dissolution behavior. In the Da model, like the ADE model, variability in velocity and NAPL saturation is integrated not just along the individual streamtube, but over the entire source zone domain and transformed into a single parameter, 2, which relates changes in the effluent concentration to reductions in global NAPL mass. Mathematically, 2 is therefore related to the second moment of the distribution and the Peclet number in the ADE model. Two special cases of the Da model are noted. When 2

PAGE 45

31 is equal to 1, the equation describing the dissolution profile reduces to exponential decay, C(T) = C o e -kT with C o equal to the initial flux-averaged concentration exiting the source zone and k equal to a rate constant that can be solved for in terms of the initial mass and flux. The second case of note is when 2 = 0, in which case the dissolution profile reduces to a step function where the effluent concentration remains constant until all the mass within the source zone has been depleted. For steady flow conditions the Da model can be expressed in a similar fashion to the streamtube and ADE model using equation 3-1 with the source depletion term given below in 2-26. )(1)(TSCfTCDscf with: 2)(exp)(ooDMTMTS (2-26) When o is large, the combination of 2-26 and 2-1 reduces to 2-27 below (Parker and Park, 2004). 2)()(ocsMTMfCTC (2-27) Note that 2-27 is the same formulation as that of Zhu and Sykes (2004). BTCs and Mass Reduction/Flux Reduction BTCs and mass reduction/flux reduction plots for the Da model are shown below in figures 2-7 and 2-8 for a range of 2 Like the Peclet number in the ADE model and the second moment of in the streamtube model, a larger value of 2 connotes a more heterogeneous velocity field/NAPL architecture. The Da model produces similar shaped

PAGE 46

32 Figure 2-7. BTCs for the Da model. 0.3 PV0.6 PV1 PV2 PVcumulative reduction in masscumulative reduction in source discharge Figure 2-8. Mass reduction/flux reduction relationships for Da model

PAGE 47

33 BTCs for 2 > 0.7. For 2 < 0.7, the BTCs are characterized by a single inflection point and are therefore unable to simulate both an initial stage of constant concentration and a later term tailing portion. Comparison of Source Depletion Models with Numerical Simulations In this section the streamtube model and the ADE model are evaluated as mathematical analogues to the multiphase flow and transport simulator UTCHEM (Deslshad et al., 1996). If the source depletion models are robust in terms of being able to represent the range of dissolution profiles observed in field, laboratory, and numerical studies, then an alternative to trying to physically measure relevant parameters is to evaluate a typical range of observed behavior and then incorporate that range/uncertainty into the site-specific decision process. To produce a range of initial architectures in order to evaluate the robustness of the models outlined above, four simulation sets were conducted, each with a different set of parameters describing a spatially correlated random permeability field. Each simulation set contained four realizations. Random fields for the first set were generated using indicator Markov Chain techniques. Random fields for sets 2 through 4 were generated using the turning bands code of Tompson et al. (1989). In sets 2 through 4 the variance of ln(K), where ln(K) is the log transformed hydraulic conductivity in units of cm/s, was sequentially increased from 0.1 to 0.37 to 0.7 to yield progressively more heterogeneous flow fields and NAPL architectures. Constitutive Relationships UTCHEM offers several options for the various constitutive relationships required to model multiphase flow and transport. The decision as to which relationship to utilize is nontrivial as it has a significant impact on simulation results. UTCHEM offers four

PAGE 48

34 different capillary pressure/saturation (Pc/S) models, the Brooks-Corey (BC) model, the Van Genuchten (VG) model, and two versions of a hysteretic VG model, the Parker and Lenhard model (Parker et al., 1987) and the Kalurachchi and Parker model (Kalurachchi and Parker, 1990). The approach used in UTCHEM for incorporating hysteretic effects when utilizing the BC or VG model is to model the NAPL spill event in the main drainage direction for the entire spill. After the primary spill event, the model can be run in restart mode, for any subsequent water or surfactant flooding. For water or surfactant flooding the direction of the capillary pressure/saturation curve is taken to be in the imbibition direction. Hysteretic effects are incorporated by assigning different parameters to the Pc/S model (e.g. lambda parameter for BC) in the imbibition direction than were assigned in the main drainage direction (Delshad et al., 1996). A series of recent papers by Gerhard and Keuper (Gerhard and Kueper, 2003 a,b,c) investigated spill behavior for both the BC and VG capillary pressure and relative permeability models. They concluded that for accurate simulations of NAPL spills, a hysteretic BC model should be utilized. The major limitation of the VG model, which has significant implications to the simulations here, is that the continuous nature of the VG Pc/S curve (meaning it does not have a fixed entry pressure like the BC model) allows for unrealistic penetration of NAPL into finer grained NAPL (Gerhard and Keuper, 2003a). This is clearly inconsistent with virtually all studies of NAPL spills, which have shown a migrating NAPL to be extremely sensitive to small scale changes in permeability (Illangasekare et al., 1995; Kueper et al., 1993; Oostrom et al., 1999; Schwille, 1988; Taylor et al., 2001; Walker et al., 1998). A comparison of the BC and VG models for identical permeability fields and similar shaped Pc/S curves is shown below in figure 2-9.

PAGE 49

35 As shown in figure 2-9, which model is selected has a significant impact on the resulting NAPL architecture and in turn on dissolution dynamics. Van GenuchtenBrooks Corey Figure 2-9. Comparison of NAPL spill using Brooks-Corey and Van Genuchten models. As mentioned previously, one option in UTCHEM is to model the primary spill in the main drainage direction and then to conduct water or surfactant flooding in restart mode with the Pc/S curve now in the direction of imbibition. Several preliminary Figure 2-10. Raining effect observed when running UTCHEM in restart mode.

PAGE 50

36 simulations were conducted to evaluate this option. When running the simulator in restart mode a strange raining effect occurs which was determined to be unrealistic. This raining effect is shown below in figure 2-10. Note that the BC simulation shown in the main drainage direction in figure 2-9 was run in restart mode to generate the two plots spills shown in figure 2-10. In order to generate realistic NAPL spills, the code in UTCHEM should be modified to include a hysteretic BC model (Christ et al., 2005). A linear interpolation approach similar to the Parker and Lenhard model (Parker et al., 1987) and the Kalurachchi and Parker model (Kalurachchi and Parker, 1990) for the BC model was used here. The NAPL spill was conducted in the main drainage direction. The maximum NAPL saturation achieved during main drainage (denoted by the subscript max_sim below) was then used to calculate the corresponding organic residual saturation (S r ). ltheoreticasimrrSSSSmax_max_max_ (2-28) Where the subscript r_max is used to denote the maximum residual saturation and the subscript max_theoretical is the theoretical maximum saturation achievable during main drainage. A theoretical maximum organic saturation (S max_theoretical ) of 0.8 and a maximum organic residual of 0.25 (S r_max ) were utilized in all simulations. A maximum organic residual of 0.25 is towards the upper end of the typical values used in numerical simulations of NAPL spills which incorporate hysteretic effects (Christ et al., 2005; Dekker and Abriola, 2000a; Gerhard and Kueper, 2003a; Lemke et al., 2004a; Mayer and Miller, 1996; Parker and Park, 2004; Rathfelder and Abriola, 1998).

PAGE 51

37 The entry pressure in UCTCHEM is calculated based on Leverett scaling: kChd (2-29) where C is a user defined coefficient which scales the entry pressure (h d ) to the relative permeability (k), with k in units of millidarcies (md), and is the porosity. Simulation Setup Parameters and simulation conditions coincident to all simulations are outlined in this section. The NAPL was introduced into the simulation domain through an injection well at a rate of 15 L/day over a period of 100 days for simulations sets 1 and 2 and 150 days for simulation set 3 and 4. The maximum saturations after the 100 day period were then scaled external to the simulator to the appropriate residual saturation using equation 2-28 above. For certain simulations in set 1 and 2, the permeability field was such that NAPL had reached the lateral no flow boundary at the base of the simulation domain leading to extensive pooling. In those instances the NAPL distribution from a time step less than 100 days was utilized in order to generate dissolution profiles that were representative of the permeability field and not an artificial byproduct of the lateral no flow boundaries. There are certainly cases in the field in which extensive lateral pooling occurs, such as on top of clay aquitard. While recognizing this, the focus here is on NAPL architectures governed by the statistics of the permeability field. A list of parameters utilized is listed below in Table 2-1. The values selected were designed to represent typical values, consistent with recent numerical simulations of multiphase flow and transport (Christ et al., 2005; Dekker and Abriola, 2000a; Gerhard and Kueper, 2003a; Lemke et al., 2004a; Mayer and Miller, 1996; Parker and Park, 2004; Rathfelder and Abriola, 1998).

PAGE 52

38 After the NAPL spill, the dissolution process was initiated through a pair of injection and extraction wells comprising the left and right boundaries respectively. The injection and extraction rate was 10 m 3 /day. Table 2-1. Parameters used in UTCHEM simulations. Parameter Value Notes General Media Properties Porosity 0.3 NAPL fluid properties Density 1.625 g/cm 3 Viscosity 0.89 cp Solubility 1000 mg/L increased to cut down on simulation time for aqueous dissolution Water Properties Density 1 gm/cm 3 Viscosity 1 cp Capillary Pressure C 2 (1/md) 0.5 scales entry pressure to permeability 1.7 BC pore size distribution index S max_theoretical 0.8 S r_max 0.25 Relative Permeability 2.85 exponent relating NAPL saturation to relative permeability NAPL injection well Flow rate 15 L/day Duration of Injection 100 days Injection and Extraction Wells Flow rate 10 m 3 /day Dispersivity Longitudinal 0.15 m Transverse .0005 m Simulation Set 1 Simulation domain and random field generation For simulation set 1, a model domain of 20m x 10m x 1m was utilized. The grid discretization selected was 10cm in the z direction and 20cm in the x direction. Correlation lengths were assumed to be 4m is the x direction and 0.25 m in the z direction. Grid parameters are summarized in table 2-2.

PAGE 53

39 Table 2-2. Grid and permeability field parameters. Value x dimension 20 m z dimension 10 m y dimension 1 m discretization x 20 cm discretization z 10 cm discretization y 1 m correlation length x 4 m correlation length z 0.25 m Random fields for the four simulations of set 1 were generated using Transition Probability Geostatistical Software (T-Progs). T-Progs is a Markov chain based geostatistical package that is designed to simplify the process of interpreting geostatistical data. For actual sites, the user would be required to delineate the porous media into five or fewer classes, inputting the sampling location and class into the simulator. T-Progs then calculates the volume percentage of each class, transition probabilities, correlation lengths and other relevant parameters and generates a user specified number of realizations that are conditioned to the measured data. For theoretical simulations, such as those presented here, the only input data required is the volume percentage of each class and the correlations lengths. One advantage of the indicator approach is that specific values for each class can be specified (e.g. entry pressure, pore size distribution, residual saturation), as opposed to having to come up with scaling relationships. This approach was employed by Lemke and collaborators (Lemke and Abriola, 2003; Lemke et al., 2004a) at a DNAPL contaminated site in Michigan. Parameters for each class are listed below in table 2-3.

PAGE 54

40 Table 2-3. Parameters assigned to each indicator class. Class 1 Class 2 Class 3 Class 4 Class 5 Volume Fraction 0.1 0.5 0.25 0.1 0.05 Permeability (md) 8000 5000 3000 1000 100 Hydraulic Conductivity (m/day) 7.7 4.8 2.9 1 .01 BC Lambda Parameter 2.1 1.9 1.4 1.1 1 Max Residual Oil Saturation .21 .22 .23 .24 .25 NAPL architectures NAPL architectures for the four set 1 realizations are displayed below in figure 2-11. The plots display NAPL saturations contoured over a range of 0 to 25% saturation. Simulation 3Simulation 4 Simulation 1 Simulation 2 Figure 2-11. NAPL spills for set 1 (contour range: 0 to 25% saturation). Dissolution profiles, model fits, and mass reduction/flux reduction relationships The UTCHEM generated dissolution profiles for the four simulations along with model fits for the streamtube and ADE model are shown below in figure 2-12 and 2-13. The R 2 value and the fitted parameters (P and 2 ) are also shown. The objective function

PAGE 55

41 selected for curve fitting was a simple root mean square function given by equation 2-30 below. 21UTCHEMSDUTCHEMCCCRMS (2-30) Where the subscript UTCHEM denotes the UTCHEM model output and SD denotes the fit value from the source depletion models (ADE and streamtube). A value of 1 was added to the denominator to avoid dividing by zero. As evidenced in figure 2-12 and 2-13, both models provide fits of the UTCHEM generated dissolution data with R 2 values > 99%, suggesting that these simplified source depletion models can serve as mathematical analogs for more sophisticated numerical simulators such as UTCHEM. Despite differences in NAPL architectures (figure 2-11), the dissolution profiles are quite similar, suggesting the variance of the distribution and the Peclet number may be highly correlated to the statistics of the porous media. There is a strong conceptual foundation for this as the permeability field is the primary governing mechanism of both the NAPL architecture (e.g., Dekker and Abriola, 2001a; Mayer and Miller, 1996) and velocity field (e.g. Dagan, 1989; Gelhar, 1993). The mass reduction/flux reduction plots for the simulations of set 1 along with the model fits from the streamtube and ADE model are shown below in figures 2-14 and 2-15 (Note that the dissolution profiles and not the mass reduction/flux reduction plots are being fit). The perturbation about the mean dissolution behavior, which is somewhat evident in figures 2-12 and 2-13 is more apparent in the mass reduction/flux reduction plots below as the perturbation occurs over a larger mass reduction interval than a flushing duration interval. This is especially true for simulation 4.

PAGE 56

42 2= 0.58R2 = 0.999 2= 0.49R2 = 0.999Simulation 3Simulation 4Simulation 1Simulation 22= 0.54R2 = 0.9982= 0.72R2 = 0.990Cumulative Pore VolumesC/Co Source Depletion ModelUTCHEM Simulation Figure 2-12. Streamtube model fit of UTCHEM dissolution data (set 1). P = 5.3R2 = 0.999 P = 7.7R2 = 0.999 P = 7.3R2 = 0.998Simulation 3Simulation 4Simulation 1Simulation 2P = 4.9R2 = 0.990Cumulative Pore VolumesC/Co Source Depletion ModelUTCHEM Simulation Figure 2-13. ADE model fit of UTCHEM dissolution data (set 1).

PAGE 57

43 Simulation 3Simulation 4Simulation 1Simulation 2 Source Depletion ModelUTCHEM SimulationCumulative Reduction in MassCumulative Reduction in Source Discharge Figure 2-14. Mass reduction/flux reduction fit for streamtube model (set 1). Simulation 3Simulation 4Simulation 1Simulation 2 Source Depletion ModelUTCHEM SimulationCumulative Reduction in MassCumulative Reduction in Source Discharge Figure 2-15. Mass reduction/flux reduction fit for ADE model (set 1).

PAGE 58

44 Simulation Domain and Grid Setup for Simulation Sets 2 through 4 For simulation sets 2 through 4, a model domain of 30 m x 10 m x 1 m was utilized. For simulation set 3 (which was the first simulation set run), the grid discretization selected was 5cm in the z direction and 50 cm in the x direction. Grid discretization was determined based on the recommendation of Russo (1991) (cited in: Mayer and Miller, 1996) who suggest at least 4 nodes per correlation length and 15 correlation lengths over the entire domain to maintain ergodicity. The same sized domain used for simulation set 3 was also used for simulation set 2 and simulation set 4. The only difference between simulation set 3 and sets 2 and 4 was that the grid discretization was increased from 5cm in the z direction to 8 cm. This resulted in a factor of 5 decrease in computation time. For example, the dissolution simulations, which took approximately 30 hrs of run time for simulation set 3, were decreased to around 6 hrs for simulation set 4. Random fields were generated using the turning bands code of Tompson (1989). Table 2-4. Grid/random field parameters for simulation sets 2-4. Set 2 Set 3 Set 4 Units X dimension 30 30 30 m Y dimension 1 1 1 m Z dimension 10 10 10 m grid cells X 60 60 60 grid cells y 1 1 1 grid cells Z 125 200 200 dx 50 50 50 cm dy 100 100 100 cm dz 8 5 5 cm variance ln(K) 0.1 0.36 0.7 cm/s mean ln(K) -5.62 -5.62 -5.62 cm/s correlation length x 2 2 2 m correlation length z 0.2 0.2 0.2 m Geostatistical parameters for set 3, which functioned as the base case, were adapted from Jensen et al. (1993). For simulation set 2 the variance of ln(K) was decreased from the

PAGE 59

45 base case value of 0.36 to 0.1. For simulation set 4, the variance of ln(K) was increased from the base case value of 0.36 to 0.7. A summary of the grid/random field parameters used for simulation sets 2 through 4 is displayed below in table 2-4. Simulation Set 2 NAPL architectures NAPL architectures for the four set 2 realizations are shown below in figure 2-16. As shown in figure 2-16, the low variability in the permeability field leads to a relatively uniform NAPL architecture. Simulation 3Simulation 4Simulation 1Simulation 2 Figure 2-16. NAPL spills for set 2 (contour range 0 to 25% saturation).

PAGE 60

46 Dissolution profiles, model fits, and mass reduction/flux reduction relationships The UTCHEM generated dissolution profiles for set 2 simulations along with model fits of the dissolution profile and mass reduction/flux reduction relationship are displayed below in figures 2-17 through 2-20. As in simulation set 1, the models fits of the UTCHEM generated dissolution data were > 99%. Similar to simulation set 1, the dissolution profiles for all simulations in set 2 are quite similar, suggesting the Peclet number and variance of contribution time may be highly correlated to the permeability field statistics. 2= 0.43R2 = 0.9992= 0.4R2 = 0.998Simulation 3Simulation 4Simulation 1Simulation 22= 0.5R2 = 0.9992= 0.46R2 = 0.999Cumulative Pore VolumesC/Co Source Depletion ModelUTCHEM Simulation Figure 2-17. Streamtube model fit of UTCHEM dissolution data (set 2).

PAGE 61

47 P = 9.9R2 = 0.999P = 11.9R2 = 0.998P = 7.2R2 = 0.999Simulation 3Simulation 4Simulation 1Simulation 2P = 8.6R2 = 0.999Cumulative Pore VolumesC/Co Source Depletion ModelUTCHEM Simulation Figure 2-18. ADE model fit of UTCHEM dissolution data (set 2). Simulation 3Simulation 4Simulation 1Simulation 2 Source Depletion ModelUTCHEM SimulationCumulative Reduction in MassCumulative Reduction in Source Discharge Figure 2-19. Mass reduction/flux reduction fit for streamtube model (set 2).

PAGE 62

48 Simulation 3Simulation 4Simulation 1Simulation 2 Source Depletion ModelUTCHEM SimulationCumulative Reduction in Source DischargeCumulative Reduction in Mass Figure 2-20. Mass reduction/flux reduction fit for ADE model (set 2). Simulation Set 3 NAPL architectures NAPL architectures for the four set 3 realizations are displayed below in figure 2-21. The plots display NAPL saturations contoured over a range of 0 to 25%. Consistent with previous numerical simulations of NAPL migration, figure 2-21 illustrates that an increase in the variability of the permeability field leads to a decrease in the total penetration depth and a distribution more skewed towards the presence of NAPL in pools (Christ et al., 2005; Dekker and Abriola, 2000a; Mayer and Miller, 1996).

PAGE 63

49 Simulation 3Simulation 4Simulation 1Simulation 2 Figure 2-21. NAPL spills for set 3 (contour range: 0-25% saturation). Dissolution profiles, model fits, and mass reduction/flux reduction relationships The UTCHEM generated dissolution profiles for the set 3 simulations along with model fits for the streamtube and ADE model are shown below in figure 2-22 and 2-23. The models perform well, with fits characterized by R 2 values >= 99%. The trend of decreased penetration depth and an initial architecture more skewed towards NAPL

PAGE 64

50 existing in pools effects the flux averaged BTC in two important ways, both of which lead to increased longevity of the source zone. First, with a decrease in penetration depth, the fraction of streamtubes contaminated with NAPL, f c decreases. For a given amount of NAPL mass, as f c decreases, the amount of mass in each streamtube increases which of course leads to an increase in average contribution time for each streamtube. Second, a NAPL architecture more skewed towards pools is equivalent to an increase in the variability in the trajectory averaged NAPL content,. As noted previously this highly variable architecture causes substantial tailing in the BTC as the lowstreamtubes are depleted of mass relatively early with the laterally extensive, high streamtubes continuing to produce a small portion of the initial flux for several additional pore volumes. When comparing the dissolution profiles of simulation set 3 to simulation set 2, the general trend predicted by the streamtube models of increased spreading of the BTC with increased heterogeneity in the combined effects of the Lagrangian NAPL architecture and velocity field (lower Peclet number/higher variance) is evident (see figures 2-3 and 2-5). S S S The dissolution profile of simulation 4 proved somewhat more difficult to fit with the simplified source depletion models. The NAPL architecture for simulation 4 was characterized by an extensive pool in the vicinity of the injection point (figure 2-24 below), leading to a discontinuous distribution with the majority of the streamtubes associated with mixed length residual components (comparatively smaller) and a small fraction of streamtubes associated with the laterally extensive pool component (much higher ). The discontinuity arises from a large gap in values from the residual components and the pooled components, i.e. the distribution is relatively smooth and S S S S

PAGE 65

51 2= 0.98R2 = 0.9982= 1.03R2 = 0.992Simulation 3Simulation 4Simulation 1Simulation 22= 0.86R2 = 0.9942= 0.924R2 = 0.989Cumulative Pore VolumesC/Co Source Depletion ModelUTCHEM Simulation Figure 2-22. Streamtube model fit of UTCHEM dissolution data (Set 3). P = 1.5R2 = 0.998P = 1.3R2 = 0.994P = 2.12R2 = 0.994Simulation 3Simulation 4Simulation 1Simulation 2P = 1.8R2 = 0.989Cumulative Pore VolumesC/Co Source Depletion ModelUTCHEM Simulation Figure 2-23. ADE model fit of UTCHEM dissolution data (Set 3).

PAGE 66

52 continuous until all the residual mass has been depleted upon which there is a large jump to the next highest streamtube. As noted previously this highly variable architecture causes substantial tailing in the BTC as the lowstreamtubes are depleted of mass relatively early with the laterally extensive, high streamtubes continuing to produce a small portion of the initial flux for several additional pore volumes. To illustrate the discontinuous nature of the BTC, the fit in figure 2-24 was obtained by fitting the first portion of the BTC associated with the residual component and the portion associated with the pooled component each with a separate Peclet number. The respective architectures for each component of the BTC are also inset into figure 2-24. This type of dissolution behavior, with two distinct components, one associated with the mixed length residual component and the other with a laterally extensive pool component, was also evident in simulation 1 of set 4. S S S Pore VolumesC/Cs UTCHEM dissolution data distribution fit to mixed length residual componentdistribution fit to pool component Figure 2-24. Discontinuous fit of simulation 4 (set 3).

PAGE 67

53 Simulation 3Simulation 4Simulation 1Simulation 2 Source Depletion ModelUTCHEM SimulationCumulative Reduction in Source DischargeCumulative Reduction in Mass Figure 2-25. Streamtube model fit of mass reduction/flux reduction relationships. Simulation 3Simulation 4Simulation 1Simulation 2 Source Depletion ModelUTCHEM SimulationCumulative Reduction in Source DischargeCumulative Reduction in Mass Figure 2-26. ADE model fit of mass reduction/flux reduction relationships.

PAGE 68

54 The difficulty in fitting the extensive late term tailing behavior is further magnified in the mass reduction flux reduction plots (figure 2-24; 2-25). The discontinuous nature of the simulation 4 NAPL architecture is more enhanced in the mass reduction/flux reduction plots where there is a sharp jump to the late term tailing behavior once the residual component has been depleted. Simulation Set 4 NAPL architectures NAPL architectures for the set 4 realizations are displayed below in figure 2-27. The plots display NAPL saturations contoured over a range of 0 to 25%. When compared with the simulation sets 2 and 3, simulation set 4 architectures are characterized by a decrease in the total penetration depth and a distribution more skewed towards the presence of NAPL in pools. Dissolution profiles, model fits, and mass reduction/flux reduction relationships The UTCHEM generated dissolution profiles for the set 4 simulations, along with model fits for the streamtube and ADE model, are shown below in figure 2-28 and 2-29. The models provide reasonably good fits of the UTCHEM simulations, yet not as accurate as the more homogenous systems (sets 1 through 3). Comparing the dissolution profiles of simulation set 4 to the previous simulation sets, the trend of decreased penetration depth and enhanced pooling increasing source longevity is again evident. The general trend predicted by the streamtube models of increased spreading of the BTC with increased heterogeneity in the combined effects of the Lagrangian NAPL architecture and velocity field (lower Peclet number/higher variance) is also evident. Although there is the expected trend of increased spreading of the BTC, the variability between individual

PAGE 69

55 realizations also increases (e.g. compared with set 2 where all of the BTCs were quite similar). This is consistent with general probability theory that suggests that as the Simulation 3Simulation 4Simulation 1Simulation 2 Figure 2-27. NAPL spills for set 4 (contour range: 0 to 25% saturation) variance of the permeability field increases, more simulations would be required to converge to the expected value, i.e. mean BTC for the particular permeability field

PAGE 70

56 statistics. An exception to the increased variability trend was simulation 3, where unlike previous simulations where the late term tailing was dominated by a single large pool in the vicinity of the injection point, the permeability field was such that three pools of comparable size formed. The effect of these three pools of comparable size was to dampen variability in the dissolution profile, such that the BTC more resembled the simulations of set 1 and 2. The initial NAPL architecture and dissolution profile of simulation 1 was quite similar to simulation 4 of set 3, with two distinct BTC components, one associated with a residual component, and another associated with a late term tailing portion as the large pool was gradually depleted of mass. 2= 0.97R2 = 0.9822= 1.16R2 = 0.994Simulation 3Simulation 4Simulation 1Simulation 22= 0.31R2 = 0.9942= 1.3R2 = 0.937Cumulative Pore VolumesC/Cs Source Depletion ModelUTCHEM Simulation Figure 2-28. Streamtube model fit of UTCHEM dissolution data (set 4) Mass reduction flux reduction plots for simulation set 4 are shown below in figures 2-30 and 2-31. Difficulty in fitting the late term tailing portion of the BTC is evident in

PAGE 71

57 P = 1.5R2 = 0.981P = 1.0R2 = 0.996P = 20.67R2 = 0.995Simulation 3Simulation 4Simulation 1Simulation 2P = .754R2 = 0.937Cumulative Pore VolumesC/Co Source Depletion ModelUTCHEM Simulation Figure 2-29. ADE model fit of UTCHEM dissolution data. Simulation 3Simulation 4Simulation 1Simulation 2 Source Depletion ModelUTCHEM SimulationCumulative Reduction in Source DischargeCumulative Reduction in Mass Figure 2-30. Streamtube model fit mass reduction/flux reduction relationships (set 4).

PAGE 72

58 simulation 1 and simulation 3 where there is a large jump to the late term tailing behavior as in simulation 4 of set 3. Simulation 3Simulation 4Simulation 1Simulation 2 Source Depletion ModelUTCHEM Simulation Cumulative Reduction in Source DischargeCumulative Reduction in Mass Figure 2-31. ADE model fit of mass reduction/flux reduction relationships (set 4). Conclusions and Recommendations for Further Research In this chapter simplified source depletion models which may serve as mathematical analogs to more sophisticated multiphase flow and transport simulators where developed. The models are conceptually appealing as a single parameter defines the total NAPL mass and a single parameter defines the combined variability of the Lagrangian NAPL architecture and the velocity field. The fundamental assumption is that dissolution dynamics are primarily governed by the Lagrangian NAPL architecture and the velocity field, whereby processes such as relatively permeability, transverse dispersion and mass transfer rate coefficients, when neglected, introduce negligible error. The general trend predicted by the models is with increased variability in the NAPL

PAGE 73

59 architecture and/or the velocity field, the dissolution profile exhibits more spreading and more severe tailing, leading to increased source longevity and a more favorable relationship between reductions in mass and reductions in source discharge. The source depletion models were compared with UTCHEM simulations, comparing favorably, especially when considering a sophisticated numerical simulator with run times in the vicinity of 30 hrs has been replaced with a relatively compact analytical solution. Consistent with previous numerical investigations, increased variability in the permeability field led to increased pooling, which in turn led to enhanced tailing of the dissolution profile, suggesting that the Peclet number/variance of contribution time (the parameters that control spreading in the simplified source depletion models) are correlated with the variance of ln(K). The higher variance ln(K) simulations would benefit from further simulations to come closer to arriving at a mean dissolution profile for a given set of permeability field statistics. More simulations would perhaps yield an empirical estimator of the Peclet number or variance of contribution time based on the statistics of the permeability field. A more in depth investigation of the impact of neglecting relative permeability, transverse mixing between streamtubes, and mass transfer rate coefficients is warranted. The impact of transverse mixing in relation to bioreactive transport has been investigated by Cirpka and Kitanidis (2000) and Ginn (2001). Berglund and Fiori (1997) studied the impact of transverse mixing on a sorbing solute. In general, the limited amount of work directed towards streamtube modeling of NAPL dissolution processes (Berglund, 1997; Jawitz et al., 2005) has been of the theoretical modeling type and has not investigated the error associated with the underlying assumptions. I have attempted to address the impact

PAGE 74

60 of the underlying assumptions somewhat here but this is a potentially large area for further research. Investigating bimodal distributions and other types of probability distributions, especially for the higher variance ln(K) simulations is also an area for further research.

PAGE 75

CHAPTER 3 LABORATORY STUDIES OF RELATIONSHIPS BETWEEN ARCHITECTURE AND FLUX IN MODEL SOURCE ZONES Introduction Of critical importance when considering plume management vs. source zone remediation is the expected performance of various source zone remediation technologies under consideration, both in terms of the expected reduction in mass (and the effect on long term plume management costs) and the expected reduction in flux (and the effect on near term risk reduction). Remedial forecasting in terms of flux reduction performance is difficult in that virtually all field scale demonstrations and laboratory studies of aggressive source zone remediation to date have focused on a given technologys ability to remove mass from the source zone and have not quantified the associated reduction in contaminant mass flux. A flux-based evaluation of previous source zone remediation technologies is further complicated by the fact that relationships between reductions in mass and flux are potentially highly variable depending on both the site characteristics and the technology selected. In the case of technologies designed to remove contaminant mass from the subsurface (e.g. surfactant or cosolvent flushing), continuous measurements of mass reduction are available via sampling of the remedial fluid for the desired constituent. In contrast to mass reduction, there are typically only two points where measurements of flux are available (again note that these measurements are typically not taken); an initial measurement before the given technology is implemented and another measurement after 61

PAGE 76

62 the completion of remediation when the system has returned to a condition thought to be representative of natural flowing groundwater. Additional flux reduction data points would require the iterative process of removing a portion of the source zone mass via the given technology and then allowing the system to return to natural flowing groundwater conditions where a flux measurement could be taken. Aqueous dissolution experiments on the other hand, because they do not require switching back and forth between remediation conditions and natural flowing groundwater, have the capability of yielding nearly continuous mass reduction /flux reduction data. The experiments presented here can serve as a pump and treat baseline for comparison with bench scale experiments of aggressive remediation technologies. More challenging is directly interpreting the near continuous mass reduction/flux reduction relationships of aqueous dissolution in order to lend insight to the mass reduction/flux reduction relationships of systems in which aggressive source zone remediation has resulted in partial mass removal. The limited body of work that has addressed the issue of reductions in flux brought about by partial reductions in contaminant mass from a NAPL source zone has been of the theoretical modeling type and directed towards systems undergoing aqueous dissolution (Sale and McWhorter, 2001; Falta, 2003; Rao and Jawitz, 2003; Parker and Park, 2004), sometimes with the implicit assumption that the conclusions reached regarding mass reduction/flux reduction relationships are transferable to other types of aggressive source zone remediation (note that Rao and Jawitz (2003) consider in situ flushing with a mathematical framework that would be identical for aqueous dissolution; i.e. the flushing solution travels along identical flow paths to that of natural flowing groundwater). Based on a steady-state mathematical formulation, Sale and McWhorter

PAGE 77

63 (2001) conclude that the vast majority of the contaminant mass must be removed to achieve significant reductions in flux. Rao and Jawitz (2003) suggest that a more favorable relationship between reductions in mass and flux may be realized with increased heterogeneity of the velocity field. Falta (2003) conducted a high resolution transient simulation of the NAPL architecture considered by Sale and McWhorter (2001) and found reductions in contaminant flux were realized for partial reductions in DNAPL mass. Parker and Park (2004) suggest a continuum of relationships between mass and flux are possible depending on the site specific NAPL distribution, groundwater velocity field, and correlation between the two. It is noted here that depending on the given technology and the way in which the technology is implemented (e.g. vertical circulation vs. line drive pumping), the order and way in which mass is removed from the source zone and the associated reduction in flux could be significantly different when compared to a system which is undergoing aqueous dissolution. It is therefore inappropriate to extrapolate mass reduction/flux reduction relationships directly from aqueous dissolution work to all types of aggressive source zone technologies. Aqueous dissolution work can however be used to gain further understanding as to the way in which the NAPL architecture must be altered in order to bring about reductions in flux. That is, aqueous dissolution experiments can be constructed in a way in order to investigate which components of the initial source zone mass, if efficiently targeted and removed via a given technology, would yield the largest (or any) reduction in flux. Accordingly, the experiments presented here are designed for three major purposes: first, for comparison to the theoretical work briefly discussed above with a specific focus on the role that NAPL architecture plays in controlling

PAGE 78

64 dissolution behavior; second, to serve as a baseline of source depletion under natural flowing groundwater (or pump and treat) from which to compare flushing-based remediation experiments currently being conducted in similar systems; and third, to evaluate simplified modeling approaches for simulating long-term source depletion. Modeling Approach In this chapter, the streamtube models discussed in the previous chapter are parameterized in a deterministic manner by dividing the source zone into n streamtubes and assigning an estimated value of to each streamtube based on image analysis of the NAPL distribution. The superposition of n solutions of the model for an individual streamtube is then used to model the dissolution profile for the entire system. The dissolution profile for the equilibrium streamtube model was estimated using: S (3-1) )()(1iniscTHCfTC where H is the heaviside step function. The dissolution profile for the rate-limited streamtube model was estimated in a similar fashion. niNisiiNissctTkCtSktTkCCfTC11)(expexp)(exp1)( (3-2) For comparison purposes, the experimental results were also modeled using the Da model and the ADE model using traditional curve fitting techniques. Experimental Methods A total of four aqueous dissolution experiments were conducted in two-dimensional flow cells. Three experiments were conducted with 1,2-Dichloroethane (DCA) and one experiment with trichloroethene (TCE). A summary of the experimental

PAGE 79

65 conditions can be found in table 3-1. For brevity the four experiments will hereafter be referred to as DCA-1, DCA-2, DCA-3, and TCE-1. Tabl e 3-1. Summary of experimental conditions for conducted experiments. Experiment 1 Experiment 2 Experiment 3 Experiment 4 Name DCA-1 TCE-1 DCA-2 DCA-3 Volume injected (ml) 10 10 12 10 Injections single single double single Box design segmented segmented segmented in-line Packing discrete layers discrete layers continuous layering fine gradation Pore water velocity (cm/hr) 22.9 18.7 18.2 17.3 Flow Cell Design Two flow cell designs were utilized for the following experiments. The first design featured a segmented extraction well that was designed to provide spatially resolved flux information about temporally changing up-gradient NAPL distributions. The segmented flux measurements, coupled with visualization of the up-gradient NAPL distribution provided a means of investigating the relationship between temporal changes in NAPL architecture and reductions in flux. The design for the segmented box was similar to that of Jawitz et al. (1998a) where a frame of 1.5cm square aluminum tubing was enclosed by two sheets of glass. The 32 cm high by 40 cm long frame, which also served as an injection and extraction well, was slotted at a width of 0.03 cm and a frequency of 4 slots per cm. The integrated injection well was connected via Teflon tubing and a switch valve to either a constant head reservoir or a syringe pump, which was used for tracer injections. The extraction well was segmented at intervals of 4 cm. Teflon tubing originating from each individual well segment was terminated at a constant elevation sampling apparatus. A tray of gas chromatography vials was inserted into the apparatus to

PAGE 80

66 collect simultaneous samples from each segment. When not sampling, effluent was collected in a fraction collector and routed to a series of waste bottles, one for each segment. The individual waste bottles were weighed periodically for an additional flow measurement and for mass balance purposes. The well segments were numbered sequentially from the bottom of the box upwards (i.e. the bottom segment is referred to throughout as port 1 where the top port is referred to as port 7). Segmented WellMedia heterogeneities Constant head reservoir Syringe pump for tracer injectionsValve x Syringe pump injection Integrated well Constant elevation sampling apparatus for segmented water and solute flux measurements Waste routed to collection bottles. One bottle per segment Figure 3-1. Schematic representation of segmented box.

PAGE 81

67 An additional experiment was conducted in a 61 cm long x 41 cm high flow cell with an integrated extraction well and inline sampling system for gas chromatography (GC) analysis (Jawitz et al., 2002). A piston pump was used to continuously deliver effluent to a modified autosampler tray. Sub-samples from the flowing effluent were injected into the GC at a user defined frequency, eliminating the need for manual sampling of the effluent. The temporal discretization of the sampling could be as small as the time required for the GC to analyze an individual sample. The inline system enabled automation at the expense of spatially resolved effluent information and was therefore used only for experiment DCA-3, where due to the porous media packing, the NAPL was expected to be highly concentrated at a particular elevation. Porous Media Packing and NAPL Injection The well-characterized Accusand (Unimin Corp.) was used in all experiments (Schroth et al., 1996). Sand was introduced into the box through a funnel. A small depth of standing water and continuous vibration were maintained throughout the packing process to ensure a saturated system. Three different types of packs were used in the four experiments. In experiments DCA-1 and TCE-1 discrete low permeability lenses of 40/50 and 50/70 sands were inserted into otherwise homogenous 20/30 sand. In experiment DCA-2, larger, more continuous layers of low permeability (40/50 sand) were developed. In experiment DCA-3, four different sand types (20/30, 30/40, 40/50, and 50/70) were mixed together to create a homogenous mixture. The mixture was introduced into the box under standing water while the funnel was slowly moved back and forth across the box. The different settling velocities of the particles produced fine gradations of alternating grain sizes that spanned the width of the box. In all four experiments a syringe pump was used to inject 10 mL of NAPL into the upper portion of the domain at a rate of

PAGE 82

68 0.5mL/min. In addition to the initial 10mL NAPL injection, 2mL of NAPL was injected into one of the low permeability layers in experiment DCA-2. It was expected based on experiments conducted in similar systems that NAPL migration and eventual distribution would be exclusive to the higher permeability 20/30 sand (Oostrom et al., 1999; Taylor et al., 2001) meaning that despite the existence of heterogeneity in the porous media, the NAPL would be present in a relatively uniform velocity field. The presence of all of the NAPL in the higher permeability background media introduces two questions. The first question is whether there are mechanisms for NAPL entry into lower permeability media that may be prevalent at a given field site that are not represented in the experimental system used here and secondly if there are mechanisms that are not represented in the experimental system, should and how could those mechanisms be best represented in the experimental system. One possible explanation for DNAPL entry into lower permeability media is a system that transitions from unsaturated to saturated conditions with the initial NAPL migration occurring under unsaturated conditions and thus promoting retention of the NAPL (which is assumed to be the wetting fluid with respect to air) in lower permeability media (Illangasekare et al., 1995). This was the justification used by Brusseau et al. (2000) who utilized an emplaced source technique where the NAPL and the lower permeability media were mixed external to the experimental system prior to emplacement of the low permeability layers in the flow cell. Another possible explanation for DNAPL entry into a lower permeability layer is a system in which portions of the domain exhibit wettability properties such that NAPL is wetting with respect to water instead of vice versa (Bradford et al., 2003). In such cases the experimental system may be modified by altering the wettability properties of

PAGE 83

69 selected media (Bradford et al., 1999). A third possible explanation for DNAPL entry into lower permeability layers is perhaps tenable due to scaling issues resulting from a possibly large discrepancy in spill rates between a field site and laboratory prototype. Numerical simulations conducted by Kueper and Gerhard (1995) and Dekker and Abriola (2000) suggest that increases in spill rates lead to larger capillary pressures achieved during migration of the NAPL, which in turn allow for entry into increasingly less permeable media. It is therefore reasonable to speculate that for a field site with a similar grain size distribution and a several order of magnitude larger NAPL spill rate that NAPL migration and eventual distribution would not be exclusive to the highest permeability sand. To circumvent the exclusive migration of the NAPL in the highest permeability sand, an additional 2ml of NAPL were injected into one of the lower permeability layers in experiment DCA-2 in order to create flow variability in the NAPL contaminated portion of the flow cell. Mass balances for all experiments were in the range of 80-90%. Possible sources of mass balance errors include volatilization during sampling, inaccuracies in the measurement of flow, and residual NAPL remaining in the flow cell upon completion of the experiment. Mass removal as a function of time and total initial mass were determined by integrating the breakthrough curve without attempting to correct or add back in the various experimental mass losses. Image Analysis A simple image analysis technique was utilized to provide semi-quantitative information of NAPL architecture. Digital images of the system were taken at each sampling acquisition. Note that images were taken from only one side of the flow cell but compared favorably (in terms of similar NAPL distributions) with the side of the flow

PAGE 84

70 cell that was not imaged. The images were imported into an image processing program and overlain with a finely discretized grid. A binary map of the NAPL distribution was developed based on the presence (in which case the grid cell was filled black) or absence (in which case the grid cell was filled white) of NAPL in each individual grid cell. The black and white image was then imported into Mathcad using an image processing function which returns a matrix containing one of two possible numeric values, one value corresponding to a white grid cell (no NAPL) and another value corresponding to a black grid cell (NAPL). Trajectory integrated NAPL contents were estimated by assuming that flow was completely horizontal, such that a trajectory could be represented by one row in the binary matrix. A simple algorithm was used to count the number of elements in each row of the matrix that had a numeric value corresponding to the positive identification of NAPL. This process returns an array containing the total number of NAPL containing grid cells in each trajectory. By assuming a uniform saturation, the mass fraction in each trajectory was then estimated by dividing the number of NAPL-containing grid cells in each row by the total number of NAPL-containing grid cells. The mass fractions were converted to values of and inserted into equation 3-1 and 3-2 to model source depletion. The contaminated fraction, f S c was also estimated from this procedure. Experimental Results, Modeling, and Discussion NAPL Migration and General Dissolution Behavior The migration of the nonaqueous phase and the eventual distribution for experiments DCA-1, DCA-2, and TCE-1 (see figures 3-2a, 3-3a, and 3-4a) was as expected: the NAPL migrated vertically until a low permeability layer was encountered

PAGE 85

71 upon which the NAPL pooled and moved horizontally, eventually spilling over the side where it again began to migrate vertically downward until encountering another low permeability unit. This migration behavior led to an eventual distribution where the NAPL resided entirely in the 20/30 sand. Based on the observed dissolution behavior, the saturation of the pooled NAPL perched on top of the low permeability layers and at the base of the flow cell was concluded to not be high enough to significantly restrict flow through the interior portion of the pool due to reductions in relative permeability. An exception was the pool that formed up-gradient of well segments 5 and 6 in the vicinity of the injection point for experiment TCE-1. Again based on the dissolution behavior, advecting water was primarily restricted to the perimeter of this pool resulting in the thinning of the pool in the vertical direction and shortening of the pool in the mean direction of flow with increased dissolution time. While flow of the aqueous phase was not restricted to the exterior of the NAPL perched on top of the low permeability layers (with the exception of the one pool in the TCE experiment), this is not to suggest that relative permeability effects were insignificant. In fact, several of the segments experienced temporary flux increases as the dissolution process progressed in time, behavior likely attributable to increases in relative permeability and the exposure of additional NAPL interfaces to the flowing aqueous phase (Geller and Hunt, 1993; Powers et al., 1998; Nambi and Powers, 2000). These flux increases were somewhat masked in the integrated BTCs as other portions of the source zone were simultaneously experiencing flux decreases in response to depleting mass. In experiment DCA-2, the lower permeability layers into which the NAPL was injected were intentionally aligned with selected effluent well segments. The flow rates

PAGE 86

72 out of these segments were approximately 0.7 ml/min where the flow rates out of the well segments aligned with the higher permeability background media (20/30 sand) were approximately 0.9 ml/min. The discrepancy in flow rates between the respective ports was smaller than expected based on hydraulic conductivities of 15.0 cm/min for the 20/30 sand and 4.3 cm/min for the 40/50 sand reported by Schroth et al. (1996). It was expected that there would be detectable flow bypassing around the low permeability zone injected with NAPL, causing the NAPL to persist within the low permeability unit for extended periods of time. The small discrepancy in flow rates did not indicate a large degree of bypass flow around this zone possibly due to flow convergence at the extraction well. The contaminant fluxes out of the ports associated with the low permeability zone were characterized by a delayed flux increase, presumably due to an increase in the relative permeability as the up-gradient NAPL mass was depleted, enhancing dissolution of the down-gradient NAPL in the low permeability zone and leading to a gradual increase in the relative permeability. Flux Response to Changes in NAPL Architecture The flux plane response to temporal changes in the NAPL architecture is shown in figures 3-2 to 3-4. The bar graphs plot the fraction of the total initial flux discharged from each extraction well segment. Results are shown for six selected time steps. The fraction of the total mass remaining (MF) and the fraction of the total initial flux (FF) are also displayed for each time step. The segmented flux data is intended to be coupled with the up-gradient NAPL distribution to investigate the linkages between NAPL architecture and flux. For discussion purposes it is useful to consider each well segment (or port) as corresponding with a bundle of streamtubes with a given distribution of trajectory integrated NAPL contents. It is also useful to recall that based on the image analysis

PAGE 87

73 procedure and the definition of, the more elongated the NAPL morphology in the mean flow direction, the larger the estimated value of the trajectory integrated NAPL content,(e.g., a NAPL pool would have a much larger estimated value of than a vertical finger). S S S Figure 3-2. Flux plane response to changes in NAPL architecture for experiment DCA-1. Note that MF is the mass fraction ((mass at time t)/(initial mass at t = 0)) and FF is the flux fraction ((flux at time t) /(initial flux at t = 0)).

PAGE 88

74 Figure 3-3. Flux plane response to changes in NAPL architecture for experiment TCE-1. Note that MF is the mass fraction ((mass at time t)/(initial mass at t = 0)) and FF is the flux fraction ((flux at time t) /(initial flux at t = 0)). Figures 3-2 to 3-4 indicate that reductions in flux were realized for partial reductions in mass as certain well segments showed flux decreases at earlier dissolution times than other well segments. The well segments that continued to produce flux at later dissolution times were predominantly associated with streamtube bundles characterized by larger Svalues (NAPL morphologies that were elongated in the mean flow direction). The well segments that were depleted of mass at earlier dissolution times were predominantly associated with streamtube bundles characterized by smaller values (NAPL morphologies that were less elongated in the mean flow direction). Figures 3-2 to 3-4 indicate that variability inyields reductions in flux for partial reductions in mass as the smaller S streamtubes are depleted of mass at earlier dissolution times. S S

PAGE 89

75 Figure 3-4. Flux plane response to changes in NAPL architecture for experiment DCA-2. Note that MF is the mass fraction ((mass at time t)/(initial mass at t = 0)) and FF is the flux fraction ((flux at time t) /(initial flux at t = 0)). Variability in the distribution associated with each well segment resulted in ports where the reduction in the mass fraction (expressed as the mass removed from the individual well segment divided by the total initial mass in the model source zone) was larger than the reduction in the flux fraction (expressed as the flux discharging from the individual well segment divided by the total initial flux discharging from the entire model source zone), ports where the reduction in the mass fraction and the flux fraction were approximately equal, and ports where the reduction in the mass fraction was less than the reduction in the flux fraction. These effects are illustrated in figure 3-5, which displays cumulative reductions in the mass fraction and the flux fraction for selected ports. For all S

PAGE 90

76 three experiments, port 1 was selected as the well segment in which the reduction in flux was smaller than the reduction in mass (figure 3-5a, 3-5d, 3-5g). The streamtube bundles associated with these well segments were characterized by larger trajectory integrated NAPL contents resulting from lateral migration of the NAPL along the base of the flow cell. The higher streamtubes required the depletion of more up-gradient NAPL mass to achieve reductions in flux when compared to the smaller streamtubes. In accordance, when considered in relation to the other well segments, the reduction in the flux fraction from port 1 is smaller than the reduction in the mass fraction (figure 3-5a, 3-5d, 3-5g) where the opposite is true (the reduction in the flux fraction is larger than the reduction in the mass fraction) for ports associated with streamtube bundles characterized by smaller values which require the removal of less mass to achieve reductions in flux (figure 3-5c, 3-5f, 3-5i). The ports where the reductions in mass and flux were approximately equal (figure 3b, 3e, 3h) were not necessarily associated with a medium length NAPL morphology or a bundle of medium S streamtubes, but rather more of a triangular distribution overlaying an elongated portion on top of a low permeability layer. Put in terms of the streamtube bundle analogy, the distribution of initial in the streamtubes was such that the cumulative reduction in flux and mass were approximately equal. S S S S This variability among different portions of the source zone is further illustrated in figure 3-6 by plotting the nine ports displayed in figure 3-5 in a mass reduction/flux reduction framework along with the integrated mass reduction/flux reduction relationship for the entire system. For all three experiments, the integrated mass reduction/flux reduction behavior for the entire system was approximately linear, with a 1:1 relationship between reductions in mass and reductions in flux. This 1:1 behavior however was the

PAGE 91

77 Figure 3-5. Cumulative reductions in mass fraction and flux fraction with increasing dissolution time for selected ports from experiments DCA-1, TCE-1, and DCA-2. Note that the cumulative reductions are expressed in relation to the total mass reduction and the total flux reduction (e.g. 27% of the total reduction in flux for experiment DCA-2 comes from port 5). result of a combination of segmented mass reduction/flux reduction relationships that differed significantly from the 1:1 behavior of the entire system, with segments that were both less than (elongated or highstreamtubes) and greater than (shorter or small streamtubes) the 1:1 mass reduction/flux reduction relationship of the entire system. S S Source Depletion Models The estimated streamtube mass fractions (see image analysis section) for experiments DCA-1, TCE-1, and DCA-2 are displayed in figure 3-7. In order to simulate

PAGE 92

78 dissolution behavior, the mass fractions displayed in figure 3-7 were converted to values of and inserted into equation 3-1 and 3-2. The data was modeled under the assumption S Figure 3-6. Mass reduction/flux reduction relationships for the entire system (upper three graphs) and for selected well segments (bottom three graphs). that the dissolution dynamics were primarily controlled by the distribution. A nonreactive travel time, t, equal to the mean nonreactive travel time for the entire system was assumed for each streamtube (measured mean travel times for the three experiments were 1.2 hr for DCA-1, 1.7 hr for TCE-1, and 1.3 hr for DCA-2). The superposition of n solutions, with n being the number of trajectories from the image analysis procedure, was then used to develop the plots in figure 3-8. S The nonequilibrium model required fitting one parameter and the Da model required fitting 2 parameters. Two approaches were available for estimating f c : if equilibrium conditions were assumed, f c could be estimated from the initial flux-averaged concentration, otherwise f c could be estimated using figure 3-7. For the equilibrium streamtube model, estimating f c from figure 3-7, based on the number of trajectories

PAGE 93

79 Figure 3-7. Streamtube mass fractions estimated from the image analysis technique. containing NAPL, resulted in an overestimation of C o Explanations for the overestimation of C o when using an f c obtained from the image analysis include velocities high enough to induce rate-limiting effects in the smaller trajectories and incomplete span of the NAPL across the entire thickness of the box in certain locations, which would S

PAGE 94

80 result in bypass flow that was not captured in the two-dimensional image analysis technique. After estimating f c from C o and using NAPL distribution information obtained from the image analysis, the equilibrium streamtube models matched the observed dissolution behavior closely (figure 3-8), which supports the argument that the NAPL architecture was the primary factor controlling dissolution behavior. Additional processes not explicitly accounted for in the model, such as relative permeability, transverse dispersivity, and velocity variability are thought to exert a secondary influence on dissolution behavior. Figure 3-8. Comparison of the rate-limited streamtube, equilibrium streamtube, and effective Damkohler approaches for modeling source depletion from experiments DCA-1, TCE-1, and DCA-2. The rate-limited streamtube model required the estimation of the mass transfer rate coefficient, k, to obtain the fits shown in figure 3-8. With the incorporation of rate

PAGE 95

81 limiting effects into the model, the value for f c measured from the image analysis procedure could be utilized directly. As a comparison of the values of k used here, TCE dissolution data from a one-dimensional column study of uniform residual saturation presented in Imhoff et al. (1994) (experiment 6 in their paper) was modeled using equation 3-2 above. Using an estimated travel time from their experimental conditions, equation 3-2 provides an excellent fit of their data with a value of k equal to 65/hr, which is in the range of the values used here. It is noted that based on equation 3-2, increased accuracy from incorporating rate-limiting effects at the laboratory scale would likely be insignificant at the field-scale for typical groundwater velocities and source zone dimensions. For example, for a travel time of two days and a value of k = 42/hr, equation 3-2 does not predict rate-limiting effects until the trajectory average NAPL content,, is reduced to a value 0.002 upon which the concentration drops rapidly, closely approximating a step function. S The Da model also provided a good fit to the data with 2 = 1 and o scaled to reflect C o Again note that for the case of 2 = 1, the Da model is reduced to an exponentially-decaying function. The case of 2 = 1 is attractive because there is possibly a wide range of long-term dissolution behavior that can be reasonably approximated with an exponentially-decaying type model. The parameterization of the exponential model is also attractive as k can be expressed in terms of the initial mass, meaning that measurements of the mean groundwater velocity, the initial flux averaged concentration, and the initial mass would be sufficient to parameterize the model. For cases where 12 there is little guidance as to how to go about estimating 2 a priori. However, based on the effectiveness of the streamtube models, the parameter 2 could be

PAGE 96

82 estimated empirically using reactive travel time statistics in a manner similar to the techniques discussed in the following chapter. ADE model fits for DCA-1, DCA-2, and TCE-1 are shown below in figure 3-9. The parameter optimized in the curve fitting procedure was the Peclet number, which again note controls the shape of the BTC in a similar fashion to the second moment of the contribution time distribution. Like the Da model, the effectiveness of the streamtube models suggests that the Peclet number can perhaps be estimated using tracer techniques which are discussed in the following chapter. Figure 3-9. ADE model fit for source depletion experiments DCA-1, TCE-1, and DCA-2.

PAGE 97

83 Additional NAPL Architectures As a demonstration of NAPL architectures that would differ significantly from the 1:1 mass reduction/flux reduction behavior of experiments DCA-1, TCE-1, and DCA-2, a fourth experiment, DCA-3 is introduced. The packing procedure in this experiment was designed to create small alternating sand layers spanning the entire width of the box. During injection, the NAPL immediately migrated laterally along two 20/30 (the highest permeability sand used) layers in the vicinity of the injection point resulting in the highly elongated distribution shown in figure 3-10. The dissolution process resulted in the continual shortening of these two elongated portions with effluent concentrations remaining constant until a large fraction of the mass had been removed (figure 3-10). The mass reduction/flux reduction relationship for this system is displayed in figure 3-11. Note the similarities between the NAPL distribution of DCA-3 and the NAPL distribution up-gradient of segmented port 1 in the previous three experiments. As expected, the dissolution behavior and resulting mass reduction/flux reduction relationship of DCA-3 (figure 3-10; 3-11) was similar to port 1 in DCA-1, DCA-2, and TCE-1 (figure 3-6). Figure 3-10. Initial NAPL distribution and effluent BTC for experiment DCA-3.

PAGE 98

84 In contrast to DCA-3, it is possible to conceptualize a system where the relationship between mass and flux is quite favorable. Consider a source zone similar to the one simulated by Parker and Park (2004) where residual zones of various length overlie a large, laterally extensive pool. In this type of system a large portion of the initial flux is associated with the overlying residual zones while a large portion of the initial mass is associated with the laterally extensive pool at the base of the simulation domain. As discussed previously, the dissolution profile for such a system is similar to the high-variance lognormal distribution of case in the streamtube formulation where there is a large initial decrease in flux as the overlying residual zones are depleted of mass followed by extensive tailing resulting from the laterally extensive pool. The resulting mass reduction/flux reduction for such a system is quite favorable where there is a large decrease in flux for a relatively small decrease in mass followed by limited reduction in flux as the laterally extensive pool is slowly depleted of mass. Experimentally, this case is difficult to construct as a large laterally extensive pool invariably migrates into the injection and extraction wells. As a substitute, consider a hypothetical experiment where 75% of the well segments are characterized by BTCs identical to the BTC of port 5 in DCA-2 and 25% of the well segments are characterized by a BTC identical to that of port 1 in experiment DCA-2 (see figure 3-5). The resulting BTC is similar to the high-variance lognormal streamtube case with a large initial drop in flux followed by extensive tailing. The mass reduction/flux reduction relationship from the superposition of these data sets is displayed alongside that of DCA-3 in figure 3-11. S Experiment DCA-3 did not conform to the model assumptions as well as the other experiments because the flow paths were likely not parallel with the bottom of the flow

PAGE 99

85 cell but rather followed the curvilinear direction of the alternating sand layers. In order to estimate the streamtube mass fractions for modeling purposes, the algorithm used to process the binary image was abandoned in favor of manual counting of the NAPL containing grid cells along the curvilinear layers of alternating grain size. In contrast to the previous three experiments, f c did not need to be measured using C o for the equilibrium streamtube model, suggesting that rate-limiting effects were less important Figure 3-11. Mass reduction flux reduction relationship for experiment DCA-3 and hypothetical system developed by the superposition of selected segmented BTCs. due to the elongated distribution of the NAPL. The mass transfer rate coefficient, k, in the rate-limited streamtube model therefore acted not to scale C o to the appropriate initial value, but rather to provide a smoother fit of the data. As in the previous DCA experiments, a value of k = 42/hr was selected for the mass transfer rate coefficient. Both the equilibrium and rate-limited streamtube models provide reasonably good fits of the data, supporting the argument that the NAPL architecture is the primary mechanism controlling dissolution behavior in the systems studied. The data/modeling of DCA-3

PAGE 100

86 supports the general trend that increased variability in the distribution leads to more favorable relationships between reductions in mass and flux. In DCA-3, there is limited variability in the distribution due to most of the NAPL mass concentrated in a small number of streamtubes, which results in a less favorable relationship between mass and flux when compared to experiments DCA-1, DCA-2, and TCE-1, which exhibit more variability in terms of their distributions. S S S Figure 3-12. Comparison of the rate-limited streamtube, equilibrium streamtube, effective Damkohler, and ADE approaches for modeling effluent BTCs from experiments DCA-3. The Da model again required scaling o to the initial concentration. Unlike the previous experiments, 2 1, meaning that 2 had to be estimated by fitting the dissolution data, as opposed to solving for the rate constant of an exponential decay model in terms of the initial flux and the initial mass. A value of 2 < 1 for laterally extensive portions of the source zone was consistent with Parker and Park (2004) who fit their model to both the upper portion of their simulation domain, which consisted of

PAGE 101

87 residual zones of variable lengths, with a 2 = 1.1, and to the lower zone, which consisted of a laterally extensive pool, with a 2 = 0.4. A value of 0.1 for 2 was used in this work to obtain the fit displayed in figure 3-12. As shown in figure 3-12, the model is not capable of simulating both the initial stage of constant concentration and the late-term tailing behavior, which is a limitation of all power function-type models. Aside from the inability to simulate the late-term tailing behavior, the Da model provides a reasonable fit of the data and outlines a trend of more favorable relationships between mass and flux for larger values of 2 The general relationship between 2 and is that S 2 increases with increased variability in the distribution. S The ADE model, unlike the Da model is capable of producing the type of dissolution profile characteristic of DCA-3 and therefore provides an excellent fit of the observed data. As discussed in the previous chapter, the streamtube model (with a lognormal distribution) and the ADE model produce similarly shaped BTCs. The streamtube model therefore would also provide an excellent fit of the data by fitting the second moment of the tau distribution. Conclusions Four aqueous dissolution experiments in two-dimensional heterogeneous systems were presented with an emphasis on the dissolved contaminant flux response to reductions in DNAPL mass. The relationship between mass and flux was found to be primarily controlled by the NAPL architecture, with mechanisms such as relative permeability, velocity variability, transverse dispersion, and mass transfer rate limitations exerting only secondary influences. A Lagrangian-type definition of NAPL architecture was utilized which incorporates the concept of a trajectory integrated NAPL content,. S

PAGE 102

88 The distribution of S amongst streamtubes governed the relationship between mass and flux. For experiments DCA-1, DCA-2, and TCE-1 the relationship between reductions in mass and flux was found to be approximately 1:1. In addition to experiments DCA-1, DCA-2, and TCE-1, a fourth experiment, DCA-3, and a theoretical case, developed from the superposition of BTCs from selected well segments, were used to illustrate cases where mass reduction/flux reduction would deviate significantly from the 1:1 behavior of DCA-1, DCA-2, and TCE-1. The dissolution profiles from the four experiments were modeled using three simplified source depletion models; a streamtube model, an advection dispersion model and an effective Damkohler number model previously presented in Parker and Park (2004). The distribution of amongst the streamtubes was estimated using a simplified image analysis procedure. Based on the effectiveness of the streamtube models, dissolution dynamics for the systems studied were determined to be primarily controlled by the distribution with mechanisms such as relative permeability and velocity variability causing perturbation about the mean behavior. In the next chapter a combination tracer techniques are evaluated for there ability to parameterize these models for prediction of long-term source depletion. S S

PAGE 103

CHAPTER 4 AN EVALUATION OF PARTITIONING TRACERS FOR PARAMETERIZING SOURCE DEPLETION MODELS In this chapter partitioning tracers are investigated as a potential parameterization technique for the models introduced in chapter 2. Two common parameter estimation techniques are evaluated: the method of moments where measured moments from an extraction well BTC are equated with derived moment equations, and inverse modeling where BTCs are fit to a transport model using specified objective functions. In addition, the ADE source depletion model of chapter 2 is parameterized using a hybrid approach where the total mass is determined using moment analysis and the Peclet number is determined using traditional curve fitting techniques. For brevity these approaches will be referred to as the moment, inverse lognormal, and ADE approach respectively. Theory Overview of Partitioning Tracers Partitioning tracers are a characterization tool for estimating NAPL saturation between pairs of injection and extraction wells (Annable et al., 1998; Brooks et al., 2002; Cain et al., 2000; Falta et al., 1999; Jawitz et al., 1998b; Jawitz et al., 2000; Meinardus et al., 2002; Nelson and Brusseau, 1996). Reactive tracers that partition into the NAPL phase are injected with one or more nonreactive tracers. The retardation of the reactive/partitioning tracer in relation to the nonreactive tracer is then used to estimate the integral NAPL saturation in the swept volume between the injection well and the extraction well. An integrated measurement of NAPL saturation is conceptually 89

PAGE 104

90 appealing as it circumvents the inherent difficulty of interpolating between highly spatially variable point measurements of NAPL saturation, which is required when employing conventional point measurement tools such as soil coring. Although partitioning tracers are conceptually appealing because they provide an integrated measurement of an often highly heterogeneous saturation field, there are some questions as to the reliability of the measurements obtained from partitioning tracers. Rao et al. (2000) and Brooks et al. (2002) outline most of the technical issues associated with the use of the partitioning tracers. Some of the issues, such as partitioning nonlinearity, can be overcome through modifications to the tracer test itself (e.g., injection of multiple tracers) while other issues may be minimized through data analysis techniques (e.g., extrapolation of BTC tails). Other issues, such as hydrodynamic accessibility of the tracer to low permeability zones and interference from organic matter and mineral components are inherent limitations of the technique (Rao et al., 2000). Practical issues of concern associated with partitioning tracers are its high cost and possible regulatory issues (EPA, 2003). A comparison of the reliability of NAPL saturation estimates obtained from tracers with more conventional point measurement techniques should not simply consider the accuracy of the point measurement vs. the accuracy of the integrated measurement, but should also include the error associated with interpolating between point measurements. The centimeter scale sensitivity of a migrating NAPL to very slight changes in permeability creates highly irregular entrapment configurations (Keuper et al., 1993), which poses questions as to the reliability of conventional techniques, regardless of how accurate the individual point measurements are.

PAGE 105

91 Moment Based Parameter Estimation The moment equations developed below are based on equations presented in Jawitz et al. (2003a). The equations of Jawitz et al. (2003a) can be simplified considerably by assuming no correlation between the travel time and NAPL saturation and the contaminated fraction parameter, (f c ), is equal to the flux-averaged contaminant concentration measured in the extraction well during the tracer test. For example, if a partitioning tracer test was conducted in a PCE contaminated source zone and the measured flux-averaged PCE concentration in the extraction well during the partitioning tracer test was 60% of solubility, it is assumed that 60% of the streamtubes are contaminated. The assumption that the travel time and trajectory integrated NAPL content, are uncorrelated is somewhat supported by the recent work of Lemke et al. (2004b) who found that the extent of correlation between organic liquid saturation and permeability that has been postulated by others (e.g. Anderson et al., 1992; Berglund, 1997; and Rao et al., 2000) was not supported by their simulation results. It should be noted however that the analysis of Lemke et al. (2004b) was based on an Eulerian type comparison, plotting point values of NAPL saturations against the corresponding hydraulic conductivity for that particular point. While investigation of a correlation structure is useful for theoretical simulations there is a practical concern when dealing with mathematical formulations that include the product of two correlated lognormals. From inspection of equation 2-11 (also 4-18 below) it is clear that there are an infinite number of and combinations of the individual lognormals that lead to the same aggregate for the product of two lognormals. To illustrate this consider fitting the reactive tracer data shown below in figure 4-1. S

PAGE 106

92 Cumulative Pore VolumesFlux Averaged Concentration UTCHEM generated reactive tracer data Model fit using correlated lognormals Figure 4-1. UTCHEM generated reactive tracer fit with product of two lognormals. The 3-D plot in figure 4-2 below shows the root mean square for a matrix of variances and correlation coefficients. Evident in figure 4-2 is that for a function which is a product of two correlated lognormals, there are an infinite number of individual combinations of correlation and variance that produce the same aggregate variance for the product of two lognormals. Therefore, from a practical standpoint, assuming no correlation does not limit in any way the range of BTCs the model can produce and dramatically simplifies the equations. An exception to this statement is that if the nonreactive tracer exhibits more spreading than the partitioning tracer, a negative correlation would be needed in order to obtain a reasonable fit of the partitioning tracer (this will be discussed further below).

PAGE 107

93 correlationvariance RRoot Mean Square combinations of correlation and variance yielding same solution Figure 4-2. RMS matrix for tracer data in figure 4-1. Measured moments The absolute moments of a measured BTC can be determined by substituting the measured concentration at the extraction well, C(T), for the lognormal probability distribution in equation 2-6. (4-1) 0)(dTTCTMNN The N th normalized moment (m N ) is determined by dividing the N th absolute moment by the 0 th absolute moment. 0MMmNN (4-2)

PAGE 108

94 For step injections the first and second normalized moments can be expressed with equation 4-3 and 4-4 (Sardin et al., 1991). An advantage of the moment equations for step injections and step decreases is that the second moment lacks the higher order term (T 2 ) and leads to decreased truncation error (Young and Ball, 2000). (4-3) dTTCm01)(1 (4-4) dTTCtm02)(12 First and second moments of nonreactive tracer The mean and variance of the travel time distribution can be determined from the first and second measured moments of the nonreactive/nonpartitioning tracer (denoted by the superscript np): 211onptTmm (4-5) 42122onponptTmTmm (4-6) where 4-5 and 4-6 are corrected for the pulse duration T o The first and second moments of the travel time distribution can then be inserted into equations 2-8 and 2-9 to determine the mean and variance of the travel time distribution. Domain average trajectory integrated NAPL content For an individual streamtube, the arrival time of a partitioning tracer is given by: Rttnpp (4-7) where the subscript p denotes partitioning tracer and the subscript np denotes nonpartitioning tracer. After Jin (1995), R is given by the following equation.

PAGE 109

95 NNnSSKR11 (4-8) with: wnnCCK (4-9) Where C n is the concentration of the tracer in the NAPL phase and C w is the concentration of the tracer in water. Note that equation 4-8 differs from Jin (1995) in that the trajectory integrated NAPL saturation along a streamtube,, has been substituted for the domain average saturation S NS N Equation 4-9 can be expressed in terms of the trajectory integrated NAPL content,, in order to be consistent with the streamtube models presented in chapter 2 by making the following substitution. S SKRn1 with: wNS1 (4-10) Assuming no correlation between R and the nonreactive tracer the expected value of 4-7 can be determined by: 22111onpopRTmTmm (4-11) The expected value ofcan then be expressed through the measured moments as: S nRSKmm111 (4-12)

PAGE 110

96 Excess spreading The traditional evaluation of partitioning tracers has been limited to analysis of the first moment (domain average saturation). Information related to the spatial distribution of the NAPL has historically been limited to soil coring (Jawitz et al., 2000; Meinardus et al., 2002; Rao et al., 1997) or partitioning tracer tests conducted at multiple sampling locations (Jawitz et al., 2000; Sillan et al., 1998). Jawitz et al. (2003a) presented equations utilizing partitioning tracer higher moments to obtain information regarding NAPL spatial distribution between pairs of injection and extraction wells. The higher moment technique is based on the concept of excess spreading. This concept is perhaps best illustrated through use of the advection dispersion equation. If we were to fit a nonreactive tracer BTC with the analytical solution to the ADE (equation 2-21injection and detection in flux) by adjusting the Peclet number, a partitioning tracer response for a homogenously distributed NAPL should then theoretically be described by the same Peclet number determined from the nonreactive tracer test and an R value equal to equation 2-5 (e.g., Valochi, 1985). Any additional spreading beyond that predicted with the Peclet number from the nonreactive tracer test is considered excess spreading potentially resulting from heterogeneity in the NAPL distribution (or any other reactive parameter). Theoretically, this excess spreading should manifest itself in the higher moments of the partitioning tracer, allowing for quantification of variability in the Lagrangian-based NAPL architecture. The excess spreading concept is illustrated below in figure 4-3.

PAGE 111

97 Pore VolumesConcentrationtheoretical value assuming uniform NAPL distribution excess spreading caused by heterogeneity in the NAPL distribution nonreactive tracer Figure 4-3. BTCs representing transport of a nonreactive tracer, a reactive tracer with homogenously distributed NAPL, and a reactive tracer with heterogeneously distributed NAPL. Moment equations for quantifying Lagrangian NAPL architecture The second moment of the retardation factor R can be expressed as: 442122122oonpnpooppRTTmmTTmmm (4-13) The second moment of the trajectory integrated NAPL content can then be expressed as: 212221nSnRSKmKmm (4-14) The first and second moments of the trajectory integrated NAPL content can then be inserted into equations 2-8 and 2-9 to determine the mean and variance of the distribution. As outlined previously in chapter 2, the mean and variance of the S S

PAGE 112

98 distribution parameters can then be combined with the mean and variance of the travel time distribution parameters to determine the mean and variance of the contribution time () distribution, which in turn is used to predict source depletion under natural gradient conditions. A summary of the steps used to determine the mean and variance of the t and distributions is shown below in figure 4-4. S Nonreactive Tracer Data Determine first and second moments of twith equations 5-5 and 5-6 Reactive Tracer Data Determine first and second moments of Rwith equations 5-11 and 5-13 Determine first and second moments of the trajectory integrated NAPL content with equations 5-12 and 5-14 Determine mean and variance of contribution time using equation 3-13 and equations 3 -15 through 3-18. Determine mean and variance of the trajectory integrated NAPL content with equations 3-8 and 3-9 Determine mean and variance of twith equations 3-8 and 3-9 Forward predict dissolution profile with equation 3-14 and 3-1 Figure 4-4. Summary of process used to predict dissolution using partitioning tracers and moment analysis. Moments for the partitioning and nonpartitioning BTCs were fit to a cubic spline function in Mathcad (figure 4-5). Cubic spline interpolation passes a curve through a set of data points such that the first and second derivatives are continuous across each point. There are two primary advantages of replacing the discrete tracer data with a continuous cubic spline function. The first advantage is that a continuous function allows for the use

PAGE 113

99 of Mathcads more accurate built in numerical integration schemes such as adaptive and Romberg integration as opposed to using a trapezoidal rule calculation in a spreadsheet type application. The second advantage is the spline function extrapolates the tracer data beyond the last sampling point in cases of incomplete data collection of the BTC tail due to substantial tailing. (Note that the level of extrapolation accomplished with the spline function may be insufficient for typical field applications where the truncation error is significantly larger than the idealized numerical simulations presented here.) Data Cubic splineinterpolation Figure 4-5. Cubic spline interpolation of arbitrary UTCHEM generated tracer data. Inverse Modeling Based Parameterization (Lognormal Distribution) An alternative parameter estimation technique to the method of moments is inverse modeling where the extraction well BTC is fit to a process model. Considering the arrival time of the nonreactive tracer as a lognormally distributed random variable allows for expression of the extraction well BTC (C f (T)) using a lognormal probability distribution

PAGE 114

100 function (pdf) for Dirac inputs (equation 4-15) and a cumulative distribution function (cdf) for step inputs (equation 4-16). )()(TfTCf (4-15) )()(TpTCf (4-16) Where f and p are used to denote a pdf and a cdf respectively. For pulse inputs (or any other type of input), 4-15 can be utilized in conjunction with a convolution integral to obtain the mean and variance of t. (4-17) TdttftTCTC0)()()( The travel of a partitioning tracer (denoted by the subscript p in equation 4-18 and 4-19) is equal to the product of t and R, meaning that equations 4-15 through 4-17 can still be utilized by making use of equation 2-10 and 2-11 to define the mean and variance of the partitioning tracer arrival time (Jury and Roth; 1990). tRp (4-18) 22ttRRp (4-19) Again note that the assumption is made that R is uncorrelated with travel time. The correlation coefficient was preserved in equation 4-19 to again emphasize that there are and infinite number of combinations of variance and correlation of individual lognormals that yield identical variances for the product of two lognormals. The exception to the previous statement in noted again as well. From inspection of 4-19, if the product of a negative correlation coefficient and the standard deviations of R and t is larger than the variance of R, the reactive tracer would exhibit less spreading than the nonreactive tracer, meaning that the best fit could not be obtained without use of the correlation coefficient.

PAGE 115

101 With R determined, the mean and variance of for uncorrelated cases can be determined using the moment generating function 2-8 and equation 4-11 and 4-14. S The inverse modeling process utilized here was to first fit the nonreactive tracer using equation 4-16 in order to determine t and t 2 Values for R and R 2 were then determined by fitting the partitioning tracer with equation 4-16 using values of t and t 2 determined from the nonreactive tracer and a and equal to equation 4-18 and 4-19. The values for R and R 2 were then inserted into the moment generating function 2-7 to determine the first and second moments of R. The first and second moments of R were then used in conjunction with equations 4-12 and 4-13 to obtain the first and second moments of. The first and second moment of, together with the first and second moments of the travel time distribution were then used to determine the first and second moments of the contribution time distribution using the equations outlined in chapter 2. S S Additional Contribution Time Parameterization Methods In simulation sets one and two the above techniques tended to over predict spreading of the dissolution profile. Because of this, two additional techniques were investigated that deemphasized velocity variability (this will be discussed in further detail in the results section). ADE based paramterization The ADE approach involved using traditional moment analysis to determine the pore volume (such that R for the nonreactive tracer was 1) and the retardation factor for the reactive tracer. The retardation factor was then used to calculate the domain average NAPL saturation, which was used in equation 2-23 in conjunction with f c to determine R for the ADE source depletion model. After retardation factors were calculated, traditional

PAGE 116

102 curve fitting techniques were used to determine Peclet numbers for the nonreactive (P nr ) and reactive (P r ) tracers. The simple estimator displayed below in equation 4-20 was used to estimate the Peclet number for the source depletion model (P SD ) based on the nonreactive and reactive tracers. Equation 4-20 is semi-empirical measure of excess spreading as defined by the nonreactive and reactive tracers: rnrSDrrnrSDPPPRPPP for P nr P r > 2 (4-20) for P nr P r < 2 where R r is the retardation factor of the reactive/partitioning tracer. Inverse modeling with no travel time variability For simulation sets 1 and 2 the inverse modeling approach over predicted dissolution profile spreading. Because of this, an approach identical to the inverse lognormal procedure outlined above, except without the second moment of travel time (travel time variability) incorporated into the second moment of the contribution time, was investigated. Using this approach, equation 2-18 can be expressed as: StfmmKm22122 (4-21) UTCHEM Simulations Prior to the NAPL dissolution simulations detailed in chapter 2, a partitioning tracer test was simulated to evaluate the ability of the techniques presented above to parameterize the source depletion models of chapter 2.

PAGE 117

103 Simulation Setup A nonpartitioning and a partitioning tracer were introduced into the domain via the injection well under steady-state flow at the same flow rate as the subsequent water flood/aqueous dissolution simulation at a concentration of 500 mg/L. A partitioning coefficient of 50 was selected after preliminary simulations indicated negligible differences in predicted dissolution behavior when utilizing alternative partitioning coefficients. Although UTCHEM has the capability to simulate tracer sorption and degradation, these effects were not evaluated here. Results Predicted domain averaged NAPL saturation Predicted domain averaged NAPL saturations using the moment, inverse lognormal, and ADE approach are displayed below in table 4-1. Recall that the ADE approach uses the traditional moment technique (Jin, 1995) to estimate the domain averaged saturation (S N ). The inverse lognormal and moment approaches outlined above yield an estimate of the trajectory integrated NAPL content (), which was converted to S S N using equation 4-10. The moment and ADE approaches yielded significantly more accurate results than the inverse lognormal approach. The moment and ADE approaches exactly predicted the NAPL saturations for each simulation in both sets and 1 and 2. The inverse lognormal approach over predicted the NAPL saturations by an average of approximately 5% for both simulation sets 1 and 2. For simulations sets 3 and 4 the moment and ADE approaches under predicted the NAPL saturations by an average of approximately 1% for simulation set 3 and 3% for simulation set 4. The under prediction of NAPL saturations as heterogeneity increases is likely due to a combination of two factors. The first factor

PAGE 118

104 that could lead to an underestimation of NAPL saturation is a decrease in hydrodynamic accessibility due to relative permeability effects as the architecture becomes more skewed towards pools. The second factor is data truncat ion due to enhanced tailing of the tracer BTCs as heterogeneity increases. This effect is likely small however as the simulations were run well into the tail portion of the BTCs, thereby minimizing truncation error. Despite the enhanced tailing of tracer BTCs and possible issues associated with hydrodynamic accessibility, the moment and ADE predictions for simulation set 3 and 4 are extremely accurate. These findings are of course influenced by the fact that the simulator produces tracer data th at is not subject to analytic al/detection limit issues and exactly conforms to the underlying assump tions (e.g., perfect linear partitioning). The inverse lognormal approach performs poorly compared to the moment based approaches, with errors of around 7% for simulations sets 3 and 4. As will be shown in subsequent sections, despite the inaccuracy of the saturation estimates, the parameters determined from the inverse lognormal approach yield excellent fits of the tracer BTCs. This suggests that the RMS object ive function yields parameter se ts that accurately fit the spreading of the tracer data yet do no t preserve the mean arrival time. Method of moments dissolution predictions Moment derived parameters describing tr acer transport for all simulations are displayed below in table 4-2. The moment determined parameters for the mean and variance of the t and R distributions were inserted into equation 4-18 and 4-19 and plotted alongside the UTCHEM simulated tracer data (figure 4-6 through 4-9) The RMS values for the moment predicted tracer BT Cs are displayed in table 4-2.

PAGE 119

105 Table 4-1. Domain averaged NAPL saturations estimated with the different techniques. Simulation Actual ADE Inverse Lognormal Moments S S Error % S Error % S Error % Set 1 1 0.0043 0.0043 0.0 0.0045 0.0045 +4.4 0.0043 0.0043 0.0 2 0.0033 0.0033 0.0 0.0036 0.0036 +8.6 0.0033 0.0033 0.0 3 0.0043 0.0043 0.0 0.0046 0.0046 +6.3 0.0043 0.0043 0.0 4 0.0030 0.0030 0.0 0.0030 0.0030 0.0 0.0030 0.0030 0.0 0.0 5.1 0.0 average Set 2 1 0.0042 0.0042 0.0 0.0043 0.0043 +2.8 0.0042 0.0042 0.0 2 0.0033 0.0033 0.0 0.0035 0.0035 +6.4 0.0033 0.0033 0.0 3 0.0033 0.0033 0.0 0.0035 0.0035 +6.4 0.0033 0.0033 0.0 4 0.0045 0.0045 0.0 0.0047 0.0047 +4.9 0.0045 0.0045 0.0 0.0 5.2 0.0 average Set 3 1 0.0044 0.0043 -2.3 0.0041 0.0041 -6.6 0.0043 0.0044 0.0 2 0.0045 0.0045 0.0 0.0047 0.0047 +4.9 0.0045 0.0045 0.0 3 0.0045 0.0044 -2.2 0.0042 0.0042 -6.3 0.0045 0.0045 0.0 4 0.0045 0.0045 0.0 0.0034 0.0034 -24.2 0.0045 0.0045 0.0 1.1 10.5 0.0 average Set 4 1 0.0065 0.0063 -3.1 0.0056 0.0056 -13.4 0.0063 0.0064 -2.2 2 0.0067 0.0067 0.0 0.0071 0.0072 +6.7 0.0068 0.0069 +1.7 3 0.0065 0.0061 -6.2 0.0064 0.0064 -1.5 0.0061 0.0061 -5.6 4 0.0062 0.0060 -3.2 0.0057 0.0057 -7.4 0.0060 0.0061 -2.3 3.1 7.2 2.9 average S S

PAGE 120

106 SS One of the problems of assuming no correlation in the moment equations is that for high travel time variances, the calculated second moment of R is either very small or negative which leads to a calculated negative second moment of (equation 4-14). This of course has no physical meaning. A negative second moment of was calculated in simulations 2 and 3 of set 4 (table 4-2). For the more homogenous simulation sets 1 and 2, the moment derived parameters significantly under predicted tracer spreading (figure 4-6 and 4-7) when inserted into the lognormal cdf. The method of moments performed better, in terms of yielding parameters that more accurately reproduced tracer spreading, as the heterogeneity of the porous media increased (figure 4-8 and 4-9). The moment derived parameters describing tracer transport were used to determine the contribution time mean and variance using the process outlined in figure 4-4. The contribution time mean and variance for all simulations are displayed below in table 4-3. The predicted dissolution profiles using the parameters in table 4-3 are displayed below in figures 4-10 through 4-13. As shown in figure 4-10 and 4-11, the moment equations provide a fairly good prediction of dissolution dynamics for simulation sets 1 and 2, predicting slightly less spreading/variability of the dissolution profile than that of UTCHEM. An under-prediction of dissolution profile spreading was expected based on the performance of the moment equations in relation to fitting the partitioning tracer BTCs, yet the under prediction of spreading for the dissolution profiles is less than that of the tracer (also see RMS values in tables 4-2 and 4-3). This suggests that there are mechanisms that cause

PAGE 121

107 Table 4-2. Tracer derived parameters using method of moments. nonreactive reactive Simulation t t 2 RMS R R 2 RMS S S 2 1-1 -0.03 0.06 0.144 0.47 0.06 0.232 -5.63 0.35 1-2 -0.03 0.06 0.236 0.21 0.01 0.380 -5.86 0.29 1-3 -0.04 0.07 0.101 0.19 0.01 0.147 -5.63 0.36 1-4 -0.06 0.12 0.039 0.19 0.02 0.072 -6.10 0.53 average -0.04 0.08 0.130 0.26 0.03 0.208 -5.81 0.38 S.D. 0.02 0.03 0.083 0.14 0.02 0.132 0.22 0.10 2-1 -0.01 0.03 0.128 0.19 0.01 0.156 -5.59 0.24 2-2 -0.01 0.02 0.131 0.15 0.01 0.115 -5.91 0.39 2-3 -0.01 0.03 0.136 1.50 0.01 0.144 -5.94 0.46 2-4 -0.01 0.02 0.132 0.20 0.01 0.147 -5.57 0.32 average -0.01 0.02 0.132 0.51 0.01 0.141 -5.75 0.35 S.D. 0.00 0.00 0.003 0.66 0.00 0.018 0.20 0.09 3-1 -0.15 0.29 0.043 0.15 0.10 0.037 -6.17 1.47 3-2 -0.16 0.32 0.032 0.19 0.02 0.063 -5.60 0.38 3-3 -0.21 0.42 0.015 0.16 0.08 0.016 -6.13 1.27 3-4 -0.23 0.46 0.007 0.13 0.14 0.023 -6.25 1.69 average -0.19 0.37 0.024 0.16 0.08 0.035 -6.04 1.20 S.D. 0.04 0.08 0.016 0.03 0.05 0.021 0.30 0.57 4-1 -0.34 0.67 0.022 0.32 0.14 0.036 -5.51 0.89 4-2 -0.42 0.86 0.047 0.33 0.01 0.022 x x 4-3 -0.43 0.86 0.021 0.27 0.05 0.014 x x 4-4 -0.37 0.74 0.015 0.23 0.07 0.008 -5.55 0.87 average -0.39 0.78 0.026 0.29 0.07 0.020 -5.53 0.88 S.D. 0.04 0.09 0.014 0.05 0.05 0.012 0.03 0.02 x = negative second moment of S

PAGE 122

108 Simulation 3Simulation 4Simulation 1Simulation 2Cumulative Pore VolumesC/Co Model Fit (nonreactive tracer)UTCHEM SimulationModel Fit (reactive tracer)UTCHEM simulation Figure 4-6. Set 1 tracer fit (method of moments). Simulation 3Simulation 4Simulation 1Simulation 2Cumulative Pore VolumesC/Co Model Fit (nonreactive tracer)UTCHEM SimulationModel Fit (reactive tracer)UTCHEM simulation Figure 4-7. Set 2 tracer fit (method of moments).

PAGE 123

109 Simulation 3Simulation 4Simulation 1Simulation 2Cumulative Pore VolumesC/Co Model Fit (nonreactive tracer)UTCHEM SimulationModel Fit (reactive tracer)UTCHEM simulation Figure 4-8. Set 3 tracer fits (method of moments). Simulation 3Simulation 4Simulation 1Simulation 2Cumulative Pore VolumesC/Co Model Fit (nonreactive tracer)UTCHEM SimulationModel Fit (reactive tracer)UTCHEM simulation Figure 4-9. Set 4 tracer fits (method of moments).

PAGE 124

110 Table 4-3. Moment derived contribution time mean and variance. SimulationRMS1-11.740.410.1231-21.50.340.1031-31.730.430.0761-41.240.660.059average1.550.460.090S.D.0.240.140.0282-11.390.260.1812-21.070.410.0362-31.050.490.0552-41.420.340.079average1.230.380.088S.D.0.200.100.0653-11.211.760.0753-21.310.70.0633-31.491.690.0633-41.122.140.174average1.281.570.094S.D.0.160.610.0544-11.651.570.1054-2xxx4-3xxx4-42.241.610.069average1.951.590.09S.D.0.420.030.025 2 Simulation 3Simulation 4Simulation 1Simulation 2Cumulative Pore VolumesC/Cs Source Depletion ModelUTCHEM Simulation Figure 4-10. Set 1 predicted source depletion (method of moments

PAGE 125

111 Simulation 3Simulation 4Simulation 1Simulation 2Cumulative Pore VolumesC/Cs Source Depletion ModelUTCHEM Simulation Figure 4-11. Set 2 predicted source depletion (method of moments). spreading of the tracers in the simulator which do not, in turn cause spreading of the dissolution profile. This is most likely due to the dispersivity parameter. In a finite difference numerical model of solute transport there is always some level of spatial averaging that must occur in order to define an average parameter value for each grid cell. The dispersivity parameter defines spatially averaged spreading that occurs as a solute move through an individual grid cell. According to the assumptions of the streamtube model, spreading of a nonreactive solute is due to differences in streamtube velocities, which in turn leads to spreading of the dissolution profile according to the equations outlined in chapter 2. In UTCHEM however, dispersivity at the scale of the individual grid cell does not lead to spreading of the dissolution profile at the grid cell scale. An alternative explanation is that dispersive mixing between streamtubes dampens spreading of the dissolution profiles This was evaluated recently by Cirpka and Kitanidis

PAGE 126

112 Simulation 3Simulation 4Simulation 1Simulation 2Cumulative Pore VolumesC/Cs Source Depletion ModelUTCHEM Simulation Figure 4-12. Set 3 predicted source depletion (method of moments). Simulation 3Simulation 4Simulation 1Simulation 2Cumulative Pore VolumesC/Cs Source Depletion ModelUTCHEM Simulationnegative second moment of Snegative second moment of S Figure 4-13. Set 4 predicted source depletion (method of moments).

PAGE 127

113 (2000) in relation to bio-reactive transport. Their approach was to incorporate longitudinal mixing in each streamtube as a surrogate to transverse mixing processes. Incorporation of longitudinal dispersion into the streamtubes dampens the spreading of the BTC when compared to the advection only case. For simulations sets 3 and 4 the method of moments now over predicts spreading of the tracer BTC. This is in agreement with the dispersivity hypothesis discussed above. If the spreading evidenced in the nonreactive tracer is not entirely due to velocity variability, a more accurate fit of nonreactive tracer spreading (see figures 4-6 to 4-9) suggests an over prediction of dissolution profile spreading. As noted previously, one of the problems of assuming no correlation in the moment equations is that for cases where the reactive tracer exhibits little to no excess spreading, the calculated second moment of R is either very small or negative which leads to a calculated negative second moment of (equation 4-14). This of course has no physical meaning. A negative second moment of was calculated in simulations 2 and 3, precluding dissolution profile prediction. S S Inverse lognormal dissolution predictions The mean and variance of the travel time, retardation coefficient, and distribution determined from the inverse lognormal approach for simulation set 1 are displayed below in table 4-4. The fit tracer data is shown below in figures 4-14 through 4-17. From the RMS values in table 4-4 and figures 4-14 through 4-17, the inverse lognormal approach provides a substantially better fit of the tracer data than the method of moments. S In the more heterogeneous systems (simulation sets 3 and 4), it became increasingly difficult for the lognormal distribution to capture the later term tailing

PAGE 128

114 S For simulations sets 1 and 2, despite a highly accurate fit of the tracer data, the model predicts more spreading of the dissolution profile predicted by than the UTCHEM output. This again suggests either there is dampening of dissolution profile spreading by dispersive mixing processes not accounted for in the advection only streamtube models, or, alternatively, that dispersive mixing leads to enhanced spreading of the tracers which is erroneously interpreted as resulting entirely from velocity variability. For simulations The predicted source depletion parameters for all simulation sets along with the RMS values are displayed below in table 4-5. As discussed previously in relation to the method of moments, an accurate fit of the partitioning tracer data leads to an over prediction of dissolution profile spreading. Because of this, the second moment of the contribution time (), was calculated without the incorporation of travel time variability using equation 4-21. The source depletion parameters for no travel time variability and associated RMS values are also displayed in table 4-5. The predicted dissolution profiles are shown in figures 4-17 through 4-24. portion of the tracer BTCs. This difficulty was most evident in simulations 3-4, 4-1, and 4-2 for the partitioning tracer. Enhanced tailing of the partitioning tracer is a result of both increased velocity variability and increased variability in the trajectory integrated NAPL content. Recall that as the heterogeneity of the permeability field increases the NAPL architecture becomes more skewed to the existence of NAPL in elongated pools, enhancing tailing of the partitioning tracer. The poor fit of the tail portion of the partitioning tracer in simulation set 3-4 led to a case where there was no excess spreading of the partitioning tracer in relation to the nonpartitioning tracer, such that the variance of was zero.

PAGE 129

115 Table 4-3. Tracer derived parameters using inverse lognormal. nonreactive reactive Simulation t t 2 RMS R R 2 RMS S S 2 1-1 -0.04 0.20 0.001 0.46 0.10 0.002 -5.660 0.50 1-2 -0.04 0.19 0.002 0.21 0.05 0.003 -6.030 0.75 1-3 -0.04 0.20 0.001 0.19 0.04 0.007 -5.770 0.77 1-4 -0.06 0.21 0.003 0.18 0.05 0.009 -6.250 0.91 average -0.04 0.20 0.002 0.26 0.06 0.005 -5.928 0.73 S.D. 0.01 0.01 0.001 0.13 0.03 0.003 0.265 0.17 2-1 -0.02 0.15 0.000 0.19 0.03 0.000 -5.720 0.58 2-2 -0.02 0.13 0.000 0.15 0.03 0.000 -6.070 0.85 2-3 -0.01 0.17 0.000 0.15 0.03 0.000 -6.100 0.89 2-4 -0.02 0.15 0.000 0.20 0.03 0.000 -5.660 0.60 average -0.02 0.15 0.000 0.17 0.03 0.000 -5.888 0.73 S.D. 0.00 0.02 0.000 0.03 0.00 0.000 0.230 0.16 3-1 -0.15 0.45 0.004 0.14 0.09 0.007 -6.200 1.40 3-2 -0.15 0.45 0.004 0.18 0.06 0.004 -5.850 1.02 3-3 -0.18 0.51 0.001 0.15 0.09 0.002 -6.170 1.39 3-4 -0.19 0.50 0.003 0.16 0.00 0.008 -5.700 0.00 average -0.17 0.48 0.003 0.16 0.06 0.005 -5.980 0.95 S.D. 0.02 0.03 0.001 0.02 0.04 0.003 0.245 0.66 4-1 -0.27 0.53 0.009 0.30 0.09 0.009 -5.570 0.74 4-2 -0.31 0.56 0.002 0.27 0.07 0.002 -5.330 0.75 4-3 -0.39 0.66 0.005 0.25 0.06 0.005 -5.410 0.70 4-4 -0.31 0.60 0.004 0.18 0.14 0.004 -5.880 1.42 average -0.32 0.59 0.005 0.25 0.09 0.005 -5.548 0.90 S.D. 0.05 0.06 0.003 0.05 0.04 0.003 0.243 0.35

PAGE 130

116 Simulation 3Simulation 4Simulation 1Simulation 2Cumulative Pore VolumesC/Co Model Fit (nonreactive tracer)UTCHEM SimulationModel Fit (reactive tracer)UTCHEM simulation Figure 4-14. Set 1 tracer fit (inverse lognormal). Simulation 3Simulation 4Simulation 1Simulation 2Cumulative Pore VolumesC/Co Model Fit (nonreactive tracer)UTCHEM SimulationModel Fit (reactive tracer)UTCHEM simulation Figure 4-15. Set 2 tracer fits (inverse lognormal).

PAGE 131

117 Simulation 3Simulation 4Simulation 1Simulation 2Cumulative Pore VolumesC/Co Model Fit (nonreactive tracer)UTCHEM SimulationModel Fit (reactive tracer)UTCHEM simulation Figure 4-16. Set 3 tracer fit (inverse lognormal). Simulation 3Simulation 4Simulation 1Simulation 2Cumulative Pore VolumesC/Co Model Fit (nonreactive tracer)UTCHEM SimulationModel Fit (reactive tracer)UTCHEM simulation Figure 4-17. Set 4 tracer fit (inverse lognormal).

PAGE 132

118 sets 1 and 2 there is therefore a substantial increase in predicted dissolution accuracy when not incorporating travel time variability into the model. For simulations 1-2, and 1-3 and all of the simulation in set 2, even when not incorporating travel time variability into the second moment of the model still predicts more spreading of the dissolution profile than the UTCHEM output, suggesting that dispersive mixing acts to dampen the dissolution process. Table 4-4. Inverse lognormal derived mean and variance of contribution time. travel time variability no travel time variability Simulation RMS RMS 1-1 1.69 0.71 0.283 1.82 0.44 0.118 1-2 1.35 0.94 0.288 1.47 0.7 0.132 1-3 1.58 0.97 0.063 1.7 0.71 0.055 1-4 1.08 1.11 0.161 1.18 0.91 0.086 average 1.43 0.93 0.199 1.54 0.69 0.098 S.D. 0.27 0.17 0.108 0.28 0.19 0.034 2-1 1.25 0.73 0.322 1.33 0.58 0.178 2-2 0.9 0.98 0.538 0.97 0.85 0.396 2-3 0.89 1.05 0.454 0.97 0.89 0.302 2-4 1.46 0.76 0.219 1.53 0.6 0.070 average 1.13 0.88 0.383 1.20 0.73 0.237 S.D. 0.28 0.16 0.141 0.28 0.16 0.142 3-1 1.19 1.86 0.091 1.4 1.41 0.030 3-2 1.07 1.47 0.065 1.3 1.01 0.007 3-3 1.43 1.89 0.089 1.68 1.39 0.034 3-4 1.7 0.5 0.049 1.95 0 0.336 average 1.35 1.43 0.074 1.58 0.95 0.102 S.D. 0.28 0.65 0.020 0.29 0.66 0.157 4-1 1.66 1.27 0.050 1.92 0.74 0.048 4-2 1.48 1.3 0.044 1.77 0.75 0.190 4-3 1.84 1.37 0.532 2.17 0.7 0.175 4-4 1.97 2.02 0.099 2.27 1.42 0.070 average 1.82 1.65 0.075 2.10 1.08 0.059 S.D. 0.22 0.53 0.235 0.25 0.48 0.072 The increased accuracy of not including travel time variability for simulation sets 3 and 4 is less than simulation sets 1 and 2. This suggests that as the heterogeneity of the hydraulic conductivity field increases, the spreading of the nonreactive tracer becomes increasingly a result of macrodispersiondispersive mixing brought about by variability

PAGE 133

119 Simulation 3Simulation 4Simulation 1Simulation 2Cumulative Pore VolumesC/Cs Source Depletion ModelUTCHEM Simulation Figure 4-18. Set 1 predicted source depletion (inverse lognormal). Simulation 3Simulation 4Simulation 1Simulation 2Cumulative Pore VolumesC/Cs Source Depletion ModelUTCHEM Simulation Figure 4-19. Set 1 predicted source depletion (inverse lognormal-no travel time variability).

PAGE 134

120 Simulation 3Simulation 4Simulation 1Simulation 2Cumulative Pore VolumesC/Cs Source Depletion ModelUTCHEM Simulation Figure 4-20. Set 2 predicted source depletion (inverse lognormal). Simulation 3Simulation 4Simulation 1Simulation 2Cumulative Pore VolumesC/Cs Source Depletion ModelUTCHEM Simulation Figure 4-21. Set 2 predicted source depletion (inverse lognormal w/ no travel time variability).

PAGE 135

121 Simulation 3Simulation 4Simulation 1Simulation 2Cumulative Pore VolumesC/Cs Source Depletion ModelUTCHEM Simulation Figure 4-22. Set 3 predicted source depletion (inverse lognormal). Simulation 3Simulation 4Simulation 1Simulation 2Cumulative Pore VolumesC/Cs Source Depletion ModelUTCHEM Simulation Figure 4-23. Set 3 predicted source depletion (inverse lognormal w/ no travel time variability).

PAGE 136

122 Simulation 3Simulation 4Simulation 1Simulation 2Cumulative Pore VolumesC/Cs Source Depletion ModelUTCHEM Simulation Figure 4-24. Set 4 predicted source depletion (inverse lognormal). Simulation 3Simulation 4Simulation 1Simulation 2Cumulative Pore VolumesC/Cs Source Depletion ModelUTCHEM Simulation Figure 4-25. Set 4 predicted source depletion (inverse lognormal w/ no travel time variability).

PAGE 137

123 in the hydraulic conductivity fieldand less a result of spatially averaged mixing occurring at the individual grid cell. For simulation 3-4, the inverse lognormal approach predicted a variance of zero for the retardation coefficient, meaning that the predicted variance was zero. This prediction of no variability can be traced to the difficulty a lognormal distribution has at fitting data characterized by extensive tailing. For the case of a variance of zero, not incorporating travel time variability of course leads to a step function dissolution profile when travel time variability is subtracted (figure 4-22). S S S ADE dissolution predictions The ADE parameter estimation method was developed based on the performance of the method of moments and the inverse lognormal approach. Because the method of moments produced better estimates of the domain averaged NAPL saturation when compared to the inverse lognormal technique, the method of moments was used to determine parameters related to the first momentpore volume; domain average NAPL saturation. Because the inverse lognormal approach yielded more accurate fits of tracer spreading when compared to the method of moments, traditional curve fitting techniques were used to determine parameters related to tracer spreadingPeclet numbers. A conditional statement was also employed (see equation 4-20) to deemphasize travel time variability at low permeability field heterogeneity. The ADE tracer parameters are shown below in table 4-6 for all simulations. Tracer fits are shown in figures 4-25 to 4-28. Like the inverse lognormal approach that also employs curve fitting to obtain parameters related to tracer spreading, the ADE approach provides a better fit than the method of moments, especially at lower heterogeneity (sets

PAGE 138

124 1 and 2). Similar to the inverse lognormal approach, simulations 3-4 and 4-1 proved somewhat more difficult to fit due to substantial tailing in the BTCs. Unlike the inverse lognormal approach however, the ADE method fits the first part and tail portion of the BTC well while not fitting the middle portion well. The quality of the tracer fit is manifested in the predicted source depletion profiles where the model predicts the early time and tailing portions of the dissolution profile well, but over predicts the middle portion. Table 4-5. Tracer derived parameters for ADE method. nonreactive reactive Simulation P nr RMS P r R r RMS 1-1 43.5 0.003 18.9 1.7 0.010 1-2 50.7 0.006 32.7 1.3 0.005 1-3 45.1 0.004 31.6 1.2 0.005 1-4 36.2 0.010 23.0 1.2 0.019 average 43.9 0.006 26.6 1.3 0.010 S.D. 6.0 0.003 6.7 0.2 0.007 2-1 85.1 0.000 62.2 1.2 0.000 2-2 108.4 0.000 72.2 1.2 0.001 2-3 73.2 0.000 51.2 1.2 0.000 2-4 85.9 0.000 59.7 1.2 0.000 average 88.2 0.000 61.3 1.2 0.000 S.D. 14.7 0.000 8.6 0.0 0.001 3-1 8.1 0.006 5.4 1.2 0.010 3-2 8.0 0.006 6.3 1.2 0.005 3-3 6.3 0.004 4.5 1.2 0.004 3-4 6.2 0.006 5.3 1.2 0.018 average 7.2 0.006 5.4 1.2 0.009 S.D. 1.0 0.001 0.7 0.0 0.006 4-1 4.8 0.010 3.1 1.5 0.018 4-2 4.2 0.008 3.4 1.3 0.008 4-3 2.9 0.008 2.5 1.3 0.009 4-4 3.8 0.006 2.5 1.3 0.006 average 3.9 0.008 2.9 1.4 0.010 S.D. 0.8 0.002 0.5 0.1 0.005 The tracer determined source depletion parameters for the ADE model are displayed below in table 4-6 for all simulations along with the RMS value for the model fits. Model fits are shown in figures 4-31 to 4-34. The low RMS values for simulation sets 1 and 2

PAGE 139

125 suggests that the employed conditional statement (equation 4-20) does an effective job of deemphasizing pore scale dispersion at lower heterogeneities. The ADE approach also provides a reasonably good prediction of dissolution behavior for set 3 (see RMS values in table 4-6 and figure 4-33). As mentioned previously, the tracer fits for simulation 3-4 and 4-1, where the first and tailing portion of the tracer BTCs were fit well, manifested. itself in the predicted dissolution, where the first and tailing portions of the dissolution profile were well predicted. Simulation 4-3 proved to be somewhat of an anomaly, such that the parameterization methods entirely broke down. Despite a good fit of the tracer data, the predicted dissolution profile was significantly different than the actual dissolution profile. Similar to the other simulations in set 4, the tracers exhibited a large degree of spreading, suggesting spreading of the dissolution profile. The actual dissolution profile however was more consistent with the less heterogeneous simulations of sets 1 and 2. The initial architecture for simulation 4-3 was characterized by three large pools of similar lengths. It is possible that the close vertical proximity of these three pools, coupled with transverse vertical dispersion acted to dampen dissolution profile spreading. Aside from simulation 4-3, the ADE approach performed well, especially when considering a sophisticated numerical model has been replaced with an analytical model that can be parameterized by a combination of nonreactive and reactive tracers. Comparison of techniques The average predicted dissolution RMS value for each simulation set is shown below in figure 4-35 along with the average RMS for all simulations. Overall, each of the techniques performed well when considering that a sophisticated multiphase flow and transport simulator has been replaced with an analytical model that can be parameterized

PAGE 140

126 Simulation 3Simulation 4Simulation 1Simulation 2Cumulative Pore VolumesC/Co Model Fit (nonreactive tracer)UTCHEM SimulationModel Fit (reactive tracer)UTCHEM simulation Figure 4-26. Set 1 tracer fit (ADE). Simulation 3Simulation 4Simulation 1Simulation 2Cumulative Pore VolumesC/Co Model Fit (nonreactive tracer)UTCHEM SimulationModel Fit (reactive tracer)UTCHEM simulation Figure 4-27. Set 2 tracer fit (ADE).

PAGE 141

127 Simulation 3Simulation 4Simulation 1Simulation 2Cumulative Pore VolumesC/Co Model Fit (nonreactive tracer)UTCHEM SimulationModel Fit (reactive tracer)UTCHEM simulation Figure 4-28. Set 3 tracer fit (ADE). Simulation 3Simulation 4Simulation 1Simulation 2Cumulative Pore VolumesC/Co Model Fit (nonreactive tracer)UTCHEM SimulationModel Fit (reactive tracer)UTCHEM simulation Figure 4-29. Set 4 tracer fits. (method of moments).

PAGE 142

128 Table 4-6. Tracer determined source depletion parameters for ADE model. Simulation P R RMS 1-1 12.5 6.9 0.129 1-2 9.9 5.3 0.045 1-3 6.8 7 0.025 1-4 8 4.8 0.135 average 9.3 6.0 0.084 S.D. 2.5 1.1 0.057 2-1 9.8 4.5 0.016 2-2 24.3 3.58 0.070 2-3 13.6 4.2 0.076 2-4 12.9 4.88 0.056 average 15.2 4.3 0.055 S.D. 6.3 0.5 0.027 3-1 1.5 8.1 0.002 3-2 1.7 5.5 0.003 3-3 1.9 10.3 0.014 3-4 0.9 8.9 0.038 average 1.5 8.2 0.014 S.D. 0.4 2.0 0.017 4-1 1.7 11.3 0.052 4-2 0.77 8.3 0.785 4-3 0.33 12.6 0.031 4-4 1.3 20.9 0.079 average 1.0 13.3 0.237 S.D. 0.6 5.4 0.366 with partitioning tracers. The ADE approach, because it combined traditional moment analysis (which more accurately predicted the total NAPL mass) with curve fitting techniques (which more accurately predicted tracer spreading) and also included a conditional statement to deemphasize grid scale dispersion, had the lowest average RMS error. The inverse lognormal with no travel time variability and the method of moments had similar RMS errors for all simulations sets. The least accurate approach was the inverse lognormal approach. This is predominantly because of the first two simulation sets. In simulation sets 1 and 2 the inverse lognormal approach accurately predicted spreading of the tracers. Much of the tracer spreading however was attributable to grid scale dispersion, which does not impact the dissolution process, as opposed to

PAGE 143

129 macrodispersionspreading resulting from variability in the permeability fieldwhich does impact the dissolution process. To illustrate this, the longitudinal dispersivity was increased from the base case (0.1 m) to 1 m and to 10 m. As shown in figure 4-36, increasing longitudinal dispersivity leads to enhanced spreading of the tracer. Increasing longitudinal dispersivity does virtually nothing to the dissolution profile however, illustrating the problem that arises when interpreting tracer spreading entirely as velocity variability, as opposed to the combination of grid scale dispersion and macrodispersion. Simulation 3Simulation 4Simulation 1Simulation 2Cumulative Pore VolumesC/Cs Source Depletion ModelUTCHEM Simulation Figure 4-30. Set 1 predicted source depletion (ADE). Conclusions and Suggestions for Further Research All of the techniques outlined above for predicting natural gradient source depletion perform well considering that a sophisticated multiphase flow and transport simulator with runtimes upwards to 30hrs has been substituted for with simple, compact

PAGE 144

130 Simulation 3Simulation 4Simulation 1Simulation 2Cumulative Pore VolumesC/Co Source Depletion ModelUTCHEM Simulation Figure 4-31. Set 2 predicted source depletion (ADE). Simulation 3Simulation 4Simulation 1Simulation 2Cumulative Pore VolumesC/Cs Source Depletion ModelUTCHEM Simulation Figure 4-32. Set 3 predicted source depletion (ADE).

PAGE 145

131 Simulation 3Simulation 4Simulation 1Simulation 2Cumulative Pore VolumesC/Co Source Depletion ModelUTCHEM Simulation Figure 4-33. Set 4 predicted source depletion (ADE). 00.050.10.150.20.25Set 1Set 2Set 3Set 4 ADE Inv Log Inv Log MOM ADE Inv Log Inv Log (no t) MOM Figure 4-34. Root mean square errors for parameterization techniques. Points indicate average RMS for simulation set. Lines indicate average RMS for all simulations.

PAGE 146

132 00.00010.00020.00030.00040.00050.00060.00070.00080.00090.0010510152025 DL = 0.1 DL = 1 DL= 10 0100200300400500600012345 DL = 0.1 DL = 1 DL = 10 Pore VolumesCf(mg/L)Volume Fraction Dissolution Tracer Figure 4-35. Impact of increased longitudinal dispersivity on tracer response and dissolution profile. analytical models which can be parameterized with a combination of reactive and nonreactive tracers. The inverse lognormal technique does not perform as well as the other techniques at medium to low heterogeneity of the conductivity field, where the magnitude of pore scale dispersion (which does not impact the dissolution process in the simulator) is comparable to macrodispersion (which does impact the dissolution process). It is

PAGE 147

133 important to note however that the implicit assumption has been made that UTCHEM, because it has been validated with field and experimental data, represents some form of an underlying reality which we can compare with the streamtube approaches. In reality we are really just comparing two mathematical formulations, each with embedded assumptions. The issue of dissolution mechanics at the scale of the individual grid cell, that is whether or not the physical mechanisms that lead to dispersion at the grid scale impact the dissolution process, is an interesting area for further research. Soerens et al. (1998) illustrated that column scale experiments of NAPL dissolution interpreted as rate-limited (e.g., Imhoff et al., 1994) can also be interpreted as resulting from velocity variability. Based on the success of the tracers at parameterizing the simplified source depletion models, it appears that the fundamental assumption of the streamtube models, that dissolution dynamics are largely governed by the combined effects of the NAPL architecture and the velocity field are fundamentally correct, although a more systematic evaluation of the impact of neglecting mechanisms such as transverse dispersion and relative permeability is warranted. Likewise, the predicted trend of increased variability in either velocity and/or Lagrangian NAPL architecture leading to enhanced spreading of the dissolution profile appears to be fundamentally correct as well. Figure 4-37 below plots the variance of ln(K) vs. the variance of the travel time, variance of trajectory integrated NAPL content, and the variance of the contribution time. Based on the plots in figure 4-37, both and t increase initially with increased variability in ln(K). The travel time variance is highly correlated with the variance of the permeability field where the relationship between and ln(K) variance is more complex, increasing S S

PAGE 148

134 initially before appearing to asymptotically approach a maximum variance with increased heterogeneity. Previous numerical simulations have indicated a decrease in a penetration depth and enhanced spreading, quantified by spatial moments in the horizontal direction, with increased variance of the permeability field (Christ et al., 2005; Dekker and Abriola, 2000a; Lemke et al., 2004b; Mayer and Miller, 1996). There is likely some opposing interplay between penetration depth and lateral spreading of NAPL in terms of their effect on the trajectory integrated NAPL content. More spreading of the NAPL leads to an increase in variability provided there are vertical fingers present. A decrease in penetration depth compresses the architecture into relatively few streamtubes, minimizing the contrasts between vertical fingers and lateral pools which lead to a high variance. Figure 4-38 plots the variance of travel time and vs. the contribution time variance. As evident in figure 4-38, the contribution time is more highly correlated with the distribution. S S S S Additional dissolution experiments with tracers would be useful for systematically evaluating some of the assumptions of the streamtube models, especially the issue of grid scale dispersion outlined above. One such experiment was conducted here and is outlined briefly below. The experiment was conducted with the in-line flow cell system discussed in chapter 3. 10 ml of TCE were introduced into the domain, yielding the initial distribution shown in figure 4-39 below. After the NAPL was spilled, two nonreactive tracers (methanol and TBA) were introduced into the domain along with two partitioning tracers (DMB and hexanol) at a flow rate of 5 ml/min for pulse duration of 27 min. The tracer BTCs are shown in below in figure 4-40. Based on the quality of the BTC data, TBA and hexanol were chosen for

PAGE 149

135 moment analysis (figure 4-40). Due to the small separation between the nonreactive and partitioning tracers, the other techniques (ADE, inverse lognormal) were not effective. It is suggested that these techniques are applicable only for step injections where the partitioning coefficient is large enough where there is separation in the rising limbs of the BTCs. The moment derived parameters for this experiment are listed in table 4-7. 00.10.20.30.40.50.60.700.10.20.30.40.50.60.70.8 00.20.40.60.811.21.41.600.10.20.30.40.50.60.70.8 00.511.522.500.10.20.30.40.50.60.70.8 Variance ln(K)Variance contribution timeVariance NAPL traj. int. NAPL contentVariance travel time Figure 4-36. Relationship between ln(K) and contribution time parameters.

PAGE 150

136 00.511.522.500.10.20.30.40.50.60.7 00.511.522.500.20.40.60.811.21.41.6 variance traj. int. NAPL contentvariance travel timevariance contribution time Figure 4-37. Relationship between travel time,, and contribution time variability. S Figure 4-38. Initial TCE distribution for dissolution experiment with tracers.

PAGE 151

137 The experimental dissolution data along with the moment predicted dissolution profile are shown below in figure 4-41. The method of moments provides a good fit of the data. There is a slight over of total mass, likely due to experimental mass losses. Future experiments would be beneficial in order to further evaluate the streamtube models and the parameterization techniques. 0.111010010000.1110 TBA Hex 0.11101001000100000.1110 Meth TBA 23DMB Hex 0.111010010000.1110 TBA Hex 0.11101001000100000.1110 Meth TBA 23DMB Hex Figure 4-39. Tracer BTCs (top graph) and tracer BTCs selected for moment analysis (bottom graph).

PAGE 152

138 Table 4-7. Moment derived parameters for dissolution experiment with tracers. paramtervaluet-0.03t20.06S-4.67S20.783.490.84 2 Pore VolumesC/Cs Figure 4-40. Experimental and predicted dissolution data.

PAGE 153

CHAPTER 5 EXTENDING SOURCE DEPLETION MODELS TO FLUSHING BASED REMEDIATION In this chapter, the source depletion models of chapter 2 are extended to flushing based remediation. Both surfactant and cosolvent based remediation are considered. Like the simplified source depletion models, these models are perhaps of more practical utility to contaminated site managers based on the difficulty of parameterizing more sophisticated numerical multiphase flow models. The equations below are modified versions of the equations presented in Jawitz et al. (2005). The purpose of this chapter is similar to chapter 2 where the models are evaluated based on their ability to reproduce typical behavior from experimental and numerical studies. Theory Streamtube Models The streamtube source depletion models of chapter 2 can be extended to flushing based remediation by incorporating the hydrodynamics of the flushing solution which involves lagging the initialization of the flushing-based source depletion process until the flushing solution has traveled from injection well to extraction well, such that the dissolution profile for an individual streamtube is as shown in figure 5-1. Incorporation of flushing solution hydrodynamics involves two relatively straightforward steps. The first step is adding the travel time of the flushing solution (t fl ) to the contribution time calculated using equation 2-12. Recall that equation 2-12 is a simple mass balance equation on an individual streamtube, dividing the total mass in each 139

PAGE 154

140 CflCaqarrival of flushing solutionAqueous Dissolution Flushing Cflstreamtube depleted of massCumulative Pore Volumes' Cf Figure 5-1. Dissolution profiles for aqueous dissolution and flushing based remediation. streamtube by the mass discharge to arrive at the time an individual streamtube will contribute contaminant flux under a set of specific flushing solutions. The only difference between flushing and aqueous dissolution, as it relates to equation 2-12, would be a large decrease in K f due to the large increase in contaminant solubility. After adding the arrival time of the flushing solution (second term in equation 5-1), the second step is to subtract the mass that is discharged from the streamtube at aqueous solubility prior to the arrival of the flushing solution (third term of equation 5-1). The mean contribution time, referred to hear as is then given by: flfltflaqtmCCmmm1111 (5-1) Where C aq is the solubility of the contaminant in water and C fl is the solubility of the contaminant in the flushing solution. The main difference between Jawitz et al.

PAGE 155

141 (2005) and the equations presented here is that the second moment of the remedial fluid travel time is not incorporated into the second moment of This modification was introduced based on the experiments outlined below where the first pore volume experienced significantly more spreading than that of a nonreactive tracer due to density contrasts between the remedial fluid (50% ethanol and water mixture) and the resident water. The second moment of still includes the second moment of what would be a nonreactive tracer however. After Jawitz et al. (2005), the BTC for flushing based remediation can be represented using cumulative distribution functions: ))()(())(1()( ptpCtpCfTCrfflrfaqcf (5-2) where the subscript rf is used to denote remedial fluid (note that the subscript fl is used to refer to the contaminant concentration in the flushing solution where the subscript rf is used to denote the concentration of the flushing solution/remedial fluid). The term rftp1 of equation 5-2 is the fraction of streamtubes where the flushing solution has not yet reached the extraction well. This term is multiplied by C aq as it is assumed that these streamtubes are discharging a concentration equal to the aqueous solubility. The term ptprf is the fraction of streamtubes where the flushing solution has reached the extraction well yet the duration of flushing has not been long enough to entirely deplete the streamtube of mass. This term is multiplied by C fl as it is assumed that these streamtubes are discharging a concentration equal to the solubility of the contaminant in the flushing solution. For cases where C fl is much larger than C aq the third term of 5-1 can be dropped with negligible error and 5-2 can be approximated by:

PAGE 156

142 )()()(ptpCfTCflflcf (5-3) Flushing BTCs for a fixed NAPL mass, fixed flushing solution breakthrough, and increasing variability in contribution time are displayed below in figure 5-2. Similar to the aqueous dissolution model, an increase in the second moment of resulting from either increased variability in either the t or distributions, leads to more spreading/ tailing of the enhanced dissolution BTC. S 01234500.20.40.60.81 Flushing Solution m2 = 4.1 m2 = 5 m2 = 8Flushing DurationC/Co Figure 5-2. Flushing BTCs for increasing second moment of Advection Dispersion Model As discussed in chapter 2, the cdfs in equation 5-2 can be replaced with the analytical solution for the advection dispersion equation (equation 2-21) such that 5-2 can now be expressed as:

PAGE 157

143 )()()(TCTCfCTCNAPLrfcflf (5-3) where C rf (T) is the ADE describing the flushing breakthrough (with a retardation factor and Peclet number denoted by R rf and a P rf ) and C NAPL (T) is the ADE (with a retardation factor and Peclet number denoted by R NAPL and a P NAPL ) used in a manner analogous to the distribution to describe variability in the combined effects of the Lagrangian NAPL architecture and the flow field. Similar to equation 2-23, integrating 5-3 from zero to infinity yields: (5-4) NAPLrfNAPLrfRRdTTCTC0)()( which allows for expression of total mass as in equation 5-5 below. NfrfNAPLSKRR (5-5) Flushing BTCs for a fixed NAPL mass, fixed flushing solution breakthrough, and increasing variability in P NAPL are shown below in figure 5-3. Similar to the aqueous dissolution model, an increase in P NAPL leads to an increase in BTC spreading/tailing. Extensions for Dispersive Mixing Under the advection-only assumption, the concentration of the remedial agent discharging from each streamtube is zero until the arrival time of the remedial fluid, upon which there is step increase from zero to the injected concentration (figure 5-1). For example, if a 5% by volume surfactant solution was injected into the source zone, the volume fraction of surfactant discharging from an individual streamtube is assumed to be zero until the remedial fluid travel time, upon which there is a step increase in volume fraction of surfactant from 0% to 5%. The spreading that is observed in the integrated

PAGE 158

144 extraction well BTC under the advection only assumption is attributed solely to variability in the remedial fluid travel times associated with each streamtube. The major 01234567891000.20.40.60.8 Flushing Solution P = 100 P = 10 P = 2 P = 0.5Flushing DurationDimensionless Concentration Figure 5-3. Flushing BTCs for increasing second moment of computational advantage of the advection only approach is that there is no need to incorporate equations relating the solubility of the contaminant to the concentration of the remedial agent; the solubility is either at aqueous solubility (prior to flushing solution arrival time) or at flushing base solubility (after flushing solution arrival time). Any type of dispersive mixing at the leading edge of the remedial fluid front, such that transport at the streamtube scale is no longer characterized by a step increase, causes a separation in the flushing solution and contaminant BTCs relative to the advection only case. This

PAGE 159

145 enhanced separation was observed in the experiments below, requiring modifications of the above equations to incorporate mathematical relationships describing contaminant solubility as a function of remedial fluid concentration. To incorporate the effects of transverse mixing, equation 5-3 was modified as follows. For cosolvent flushing, the relationship between contaminant solubility and the ethanol fraction (f) was assumed to be log-linear(Augustijn and Rao, 1995): (5-5) faqflCfC10)( where is the cosolvency power. The ethanol fraction as a function of flushing duration (f(T)) can be defined using equation 5-6 below. )()(TCfTfrfo (5-6) Where f o is the ethanol fraction injected. A solubility enhancement term describing the normalized increase in contaminant solubility as a function of the ethanol fraction is then defined as: ofTfflcCfTfC1010))(()( (5-7) The flux averaged concentration exiting the source can then be expressed as: )(11010)()(TCCfTCNAPLfTfflco (5-8) Because of the nonlinear nature of the solubility enhancement term, a closed form solution for R NAPL is not available and R NAPL must be determined numerically after defining f(T) to ensure the appropriate mass balance. Similar to 5-5, an equation relating contaminant solubility to surfactant concentration can be defined as in 5-9 below (Pennell et al., 1997):

PAGE 160

146 surfaqsurfflmCCCC )( (5-9) where the subscript surf is used to denote surfactant and m is the slop of a solubilization line. Similar to equation 5-6, the surfactant concentration can be defined as: )()(_TCCTCrfosurfsurf (5-10) where C surf_o is the surfactant concentration in the injected remedial fluid. Similar to 5-7, a solubilization enhancement term can be defined as: osurfaqsurfaqflcsurfmCCTmCCCfTCC_)())(( (5-11) Equation 5-8 can now be defined as: )(1_)()(TComCCTmCCCfTCNAPLsurfaqsurfaqflc (5-12) Experimental Data from two cosolvent flooding experiments were modeled with the source depletion models discussed above. The experiments are reviewed only briefly below as the focus here is on the modeling effluent BTCs. Experimental Setup The same flow cell that was used for the DCA-3 aqueous dissolution experiment (see chapter 3) was used for the cosolvent flooding experiments. The porous media packing was similar to experiments TCE-1 and DCE-1 with discrete layers of 40/50 sand in background 20/30 sand. 15 milliliters of PCE for experiment one and 10 ml of PCE for experiment 2 were injected into the upper portion of the domain using a syringe pump. After PCE injection, several pore volumes of water were flushed through the system to obtain a pre-remediation measurement of contaminant flux. After the pre-remediation

PAGE 161

147 water flood, the flushing solution (50% ethanol/50% water) was injected into the flow cell at a rate of 5ml/min until the effluent PCE concentrations declined to 10% of PCE solubility in the flushing solution (C fl ), at which point water was flushed through the box until all of the ethanol had been removed in order to obtain a post-remediation measurement of mass flux. Experiment 1Experiment 2 Figure 5-4. Initial PCE distribution for experiments 1 and 2.

PAGE 162

148 Initial Architectures The initial distribution of PCE following cessation of the migration phase is shown in figure 5-4. The initial architectures were consistent with the experiments detailed in chapter 3. The darker areas in figure 5-4 are the lower permeability lenses. Prf= 20.5Rrf= 1.43Prf= 5.71Rrf= 1.02 Figure 5-5. Model fit of ethanol BTCs for experiments 1 (top graph) and 2.

PAGE 163

149 Modeling of Effluent BTCs The first step in modeling of the cosolvent flood was to fit the ethanol breakthrough with the C rf function. Model fits, along with the optimized values of P rf and R rf are displayed in figure 5-5. The fraction of streamtubes contaminated (f c ) was estimated based on the pre-remediation flux-averaged concentration, which was approximately 100 mg/L. A PCE solubility of 150 mg/L was assumed, such that f c was equal to 0.66. A cosolvency power of 3.5 for ethanol and PCE was determined based on data from Cho (2001). The model fit for experiment 1 is shown below in figure 5-6 along with the optimized values for P NAPL and R NAPL As shown in figure 5-6, the model provides an excellent fit of the observed data with an R 2 value greater than 99%. PNAPL = 4.0RNAPL = 4.3 Figure 5-6. Model fit of experiment 1 data.

PAGE 164

150 The equivalent natural gradient (or pump and treat) source depletion for experiment 1, calculated with equations 2-1 and 2-21 using the optimized value of P NAPL and a retardation factor calculated with 2-23, is shown below in 5-7. Figure 5-7. Equivalent aqueous dissolution source depletion for experiment 1 The same process outlined above was used to fit the data from experiment 2. The initial flux-averaged concentration of experiment 2 was approximately 130 mg/L, such that f c was 0.87. The model fit and optimized parameters are shown below in figure 5-8. The model again provides an excellent fit of the observed data with an R 2 value greater than 99%. The equivalent natural gradient (or pump and treat) source depletion for experiment 2 is shown below in figure 5-9.

PAGE 165

151 PNAPL = 2.7RNAPL = 3.0 Figure 5-8. Model fit of experiment 2 data. Figure 5-9. Equivalent aqueous dissolution source depletion for box 2.

PAGE 166

152 Remedial Endpoints One of the primary advantages of utilizing solubilization-based remedial technologies and orienting arrays of injection and extraction wells normal to the mean direction of groundwater flow is that after the flushing solution has completely broken through, the mathematics describing the remedial process and that of natural flowing groundwater are essentially identical. The implication of this is that the equivalent natural gradient source discharge can be predicted by scaling the flux-averaged contaminant concentration in the extraction well. For experiments 1 and 2, cosolvent injection was terminated once the effluent concentration reached 10% of its maximum solubility in the cosolvent solution (10,000 mg/L). As shown in figure 5-10 below, the natural gradient flux-averaged concentration once all of the cosolvent had been flushed out of the system was below 10% of aqueous solubility (assumed 150 mg/L). This suggests that orienting injection and extraction well arrays normal to the mean direction of groundwater flow provides an effective way to decide when to terminate remediation. Terminating the enhanced solubilization portion of the flood (cosolvent or surfactant injection) once the scaled concentration in the extraction well has declined to the target source discharge still leaves an additional pore volume of flushing solution in the source zone to account for discrepancies in mass transfer rate coefficients and residence time. Conclusions and Suggestions for Further Research The work above suggests that similar to aqueous dissolution, the simplified enhanced dissolution models can serve as computationally efficient analogs to more sophisticated numerical simulators. The models are ideal for exploring uncertainty in key input parameters, such as NAPL architecture and velocity variability. The models of Jawitz et al. (2005) were modified to incorporate experimentally observed loglinear

PAGE 167

153 01020304050607080901001101201300510152025 01020304050607080901001101201301401501601700510152025 Experiment 1 Pore VolumesConcentration (mg/L) Experiment 2 Scaled concentration (C/Cs) at transition to water flood Figure 5-10. Pre and post flux-averaged concentrations in experiments 1 and 2. effects thought to result from mixing processes at the leading edge of the injected remedial fluid front. The above work would benefit from simulations of cosolvent or surfactant based remediation and additional experiments with which to compare the simplified models. Similar to the aqueous dissolution models, a more in depth analysis of the implications of neglecting dispersion, remedial fluid properties, relative permeability, and mass transfer rate coefficients is an area for further research. The difficulty in comparing UTCHEM surfactant simulations with equation 5-12 is that UTCHEM uses a ternary phase diagram to determine contaminant solubility (Delshad et al., 1996), such that the solubility of the contaminant is dependent on the local NAPL saturation in addition to the surfactant concentration. The MALVOR model

PAGE 168

154 (Dekker and Abriola, 2000b; Lemke and Abriola, 2003; Lemke et al., 2004b) and a modified MT3D/MODFLOW code (Saenton et al., 2002) contain formulations similar to 5-9 and are perhaps more appropriate for evaluating the above models. Table 5-1. Fitted model parameters from UTCHEM surfactant simulations. SetSimulationmPNAPLRNAPL210.72108.81.7220.6681.41.6310.562.9330.413.23.2410.661.61.9420.53.62.7 Despite the fact that UTCHEM does not employ a relationship similar to 5-9, equation 5-12 was fit to surfactant flood simulations conducted on two spills from each of simulation sets 2, 3, and 4 for comparison purposes. Parameters describing the ternary phase diagram were adapted from a UTCHEM example file previously developed to model the column data of Mayer et al. (1999). Because the ternary phase diagram describes the relationships between NAPL solubility and the local surfactant concentration and not equation 5-9, the parameter m was treated as a fitting parameter. Table 5-1 displays the results from the curve fitting process. Figure 5-11 displays the model fits. As shown in figure 5-11, the curve fits are reasonable despite the fact that UTCHEM does not include a relationship similar to equation 5-9 to model surfactant induced solubility enhancement. Similar to the aqueous dissolution simulation, simulation 1 of set 4 exhibits a bimodal behavior with two distinct portions of the contaminant BTC, one associated with the residual component and another associated with the late term pooling portion.

PAGE 169

155 Model Fit (surfactant breakthrough)UTCHEM SimulationModel Fit (TCE breakthrough)UTCHEM simulation Set 2; Sim1 Set 2; Sim1Set 3; Sim1Set 3; Sim3Set 4; Sim1Set 4; Sim2Pore VolumesC/Cs Figure 5-11. Model fits of UTCHEM surfactant floods.

PAGE 170

CHAPTER 6 SYNTHESIS: A FLUX BASED FRAMEWORK FOR MANAGING DNAPL CONTAMINATED SITES Defining Efficiency and Favorability Three Definitions of Efficiency Prior to conducting any type of contaminated site remediation, it is prudent to define the desired outcome, or the objectives of the remedial effort. Ideally, these objectives should be quantifiable, such that upon completion of remediation the employed technology can be evaluated for its effectiveness in terms of meeting the defined performance objectives. A recent EPA report (EPA, 2003) grouped source zone remediation performance metrics into three categories. Type I metrics are measurements that can be reliably acquired and are commonly used such as the total mass of DNAPL removed from the subsurface and changes in resident groundwater concentrations. Type II metrics are metrics that can sometimes be measured, but are not in wide use such as remaining DNAPL mass and DNAPL architecture. Type III metrics are metrics that are theoretically possible and under development such as mass flux and mass discharge. The above performance metrics can be generalized as metrics relating to resident groundwater concentrations in the source zone, metrics relating to contaminant mass, and metrics relating to contaminant flux or source discharge. As discussed in the opening chapter, a remedial objective of reducing local resident groundwater concentrations below the regulatory MCL is almost always, without question, technologically infeasible. 156

PAGE 171

157 When utilizing mass removal based performance metrics, an efficiency assessment for a particular technology would quantify the mass removed per unit work input (e.g., financial expenditures, number of pore volumes, etc. etc.). As an example consider three hypothetical source zones, source zone A, B, and C, which are characterized by sequentially increasing degrees of spatial variability in NAPL architecture (figure 6-1) (Note that the curves below were generated using the ADE model outlined in chapter 2). Implementing a pump and treat system and considering only mass reduction efficiency, source zone A would be considered the most favorable, yielding the largest reduction in mass for a given work input (figure 6-1). Cumulative Work InputCumulative Mass Removed Source Zone ASource Zone BSource Zone C Figure 6-1. Comparison of source zones A, B, & C using mass reduction efficiency.

PAGE 172

158 If the ultimate goal is a reduction in risk however, the cumulative mass removed tells us little about risk reduction unless it can be correlated in someway to the expected reduction in mass discharge. An alternative definition of efficiency to consider then is the reduction in source discharge for a given reduction in mass. Using this definition of efficiency, source zone C would now be considered the most favorable where source zone A would now be considered least favorable (figure 6-2); which is the opposite conclusion one would make when considering mass reduction efficiency. Because cumulative mass removed is the time integral of source discharge, the most favorable systems in terms of mass removed per unit input of work are, by definition, the least favorable in terms of source discharge reduction per unit reduction in mass. Cumulative Reduction in MassCumulative Reduction in Source DischargeSource Zone ASource Zone BSource Zone C Figure 6-2. Comparison of source zones A, B, & C using reduction in source discharge per reduction in mass.

PAGE 173

159 From the perspective of a site manager who is primarily concerned with risk reduction, the definition of efficiency which is most applicable is the reduction is source discharge per unit input of work. Using this definition of efficiency, the question of which source zone is more favorable becomes time dependant (figure 6-3). At early stages, the more heterogeneous source zone C is more favorable, realizing a larger reduction in source discharge per unit work input. At later stages, as the high flux producing mass is preferentially removed and source zone C enters into its later term tailing stage, source zone A becomes more favorable, realizing a larger reduction in source discharge per unit work input. Cumulative Work InputCumulative Reduction in Source DischargeSource Zone ASource Zone BSource Zone C Figure 6-3. Comparison of source zones A, B, & C using reduction in source discharge efficiency.

PAGE 174

160 The implication of the time dependency of the source discharge efficiency metric is that for increased aging (i.e. natural gradient dissolution occurring after the initial contamination event and before remediation), the relationship between mass and source discharge becomes less favorable for highly heterogeneous systems and more favorable for relatively homogenous systems. As source zone heterogeneity increases, more mass is shifted to the late term tailing portion of the natural gradient (or pump and treat) dissolution profile, such that there is a large initial decrease in source discharge for a relatively small reduction in mass (figure 6-2; figure 6-3). Relatively early in the aging process of highly heterogeneous systems, the high flux producing mass is consumed by the natural gradient dissolution process, such that the relationship between mass and source discharge becomes less favorable with aging (figure 6-4). The opposite is true for the more homogenous source zone, source zone A; with increased aging the initial stage of constant source discharge is consumed by the natural gradient process, such that the relationship between mass and flux becomes more favorable with aging (figure 6-4). For source zone B, which for no aging is defined by an approximate 1:1 relationship between mass and source discharge, aging has minimal impact on the relationship between mass and flux. Not considering time scale dependency and expressing mass and source discharge reduction in a dimensionless form, as in figure 6-4, becomes extremely limiting when considering the implications of aging on source zone remediation. From figure 6-4, the relationships between reductions in mass and source discharge moves closer to the 1:1 line with increased aging. Assuming that source zone C remains a good candidate for remediation after it has aged appreciably based solely on the fact that the relationship

PAGE 175

161 Cumulative Reduction in MassSource Zone CSource Zone BSource Zone ACumulative Reduction in Source Discharge Increased AgingIncreased Aging Figure 6-4. Aging effects on relationships between reduction in mass and reductions in source discharge for source zone A, B, & C.

PAGE 176

162 between mass and flux is still favorable would be a mistake. Considering aging effects on source discharge efficiency (source discharge reduction per unit work input) one would come to a completely different conclusion regarding the appropriateness of source zone remediation. The scenarios considered in figure 6-4 are plotted in terms of source discharge efficiency below in figure 6-5. Considering source discharge efficiency, it is apparent that source zone C becomes highly unfavorable relatively early in the aging process as all of the high flux producing mass is removed and the system enters into its late term tailing stage where there is little reduction in source discharge per unit input of work. The trending of the mass reduction/source discharge reduction relationships towards the 1:1 line is primarily a property of the ADE and the lognormal functions, lacking a physical meaning that is readily deducible from currently available literature. Mathematically, this suggests that the latter portion of the ADE and lognormal BTCs are defined by exponential decay. Recall from chapter 2 that a 1:1 relationship between mass and source discharge is defined by an exponential source discharge profile where the source discharge (S D ) is defined by: oDoMTSoDDeSTS)( (6-1) where S Do is the initial source discharge [M/T] and M o is the initial mass [M]. Equation 6-1 is useful for demonstrating that a 1:1 relationship between mass and source discharge can still be characterized by highly unfavorable source discharge reduction efficiency if the ratio of the initial source discharge to the initial mass is very high. A suite of source discharge profiles are shown in figure 6-6 for a fixed initial mass and alternate initial source discharges. In cases where the ratio of initial source discharge to the initial mass is

PAGE 177

163 Cumulative Pore VolumesCumulative Reduction in Source DischargeSource Zone ASource Zone BSource Zone C Increased Aging Increased Aging Increased Aging Figure 6-5. Aging effects on source discharge efficiency for source zone A, B, & C.

PAGE 178

164 Source DischargeCumulative Pore Volumes Figure 6-6. Source discharge profiles for a fixed initial mass and alternative initial source discharge. very low, the source discharge profile closely resembles that of a step functionthat is the least favorable mass reduction/source discharge reduction relationship. This is essentially true for all relationships between mass and source dischargewhen the ratio of the initial source discharge to the initial mass is very low, there is minimal discrepancy in the source discharge efficiencies regardless of what the mass reduction/source discharge reduction relationship is. Combining Three Efficiency Definitions The dimensionless plots of reduction in source discharge vs. reduction in mass are limiting in that they do not consider time scale dependency and secondly, they are dimensionlessthat is they are divorced from the site specific target for source discharge reduction and absolute values of source discharge reduction. Clearly, if the source discharge is already below the site specific target, an expensive source zone remediation effort is unnecessary. By incorporating tie lines of equivalent work input (e.g. pore volumes; financial expenditures) and the target reduction in source discharge, the plot

PAGE 179

165 displayed in figure 6-2 now encompasses the three definitions of efficiency (mass reduction/work input; source discharge reduction/mass reduction; source discharge reduction/work input) and is connected tangentially to information relating to the site specific target reduction in source discharge (contaminant fate and transport, exposure pathways) (figure 6-7). Instead of incorporating the site specific target for source discharge reduction as in figure 6-7, figure 6-2 could be expressed in a dimensional form (i.e. source discharge and mass reduction take on actual values) with the target source discharge denoted on the y-axis in some fashion. Figure 6-7. Figure 6-2 modified to include net work input and site specific flux-based performance metric. Flux Based Characterization Source Identification (Pre-Reactor Based Characterization) The utilization of tracers for the parameterization of the source depletion models is expected to be implemented in the later stages of site characterization where a source has

PAGE 180

166 been identified and preliminary contaminant flux measurements have indicated that the source discharge is large enough to warrant potential remediation. The source zone during tracer analysis is idealized as functioning in an analogous manner to a chemical reactor, with inflows and outflows to the reactor controlled and analyzed for relevant chemical constituents on site. There are of course several preliminary, less cost intensive steps prior to the establishment of reactor-like conditions. For discussion purposed I will group all of the traditional phase I & II type characterization (document (prior use) review, soil coring, groundwater samples, etc. etc.) into pre-reactor characterization. Transitional Reactor-Based Characterization: Mass Flux as a Remedial Screening Tool and Qualitative Estimator of Aging The decision to move from pre-reactor characterization to more cost intensive reactor-based characterization should be determined based on the perceived level of risk posed by a given site, which, as discussed previously, involves an estimate of source discharge coupled with predictive contaminant fate and transport analysis to determine exposure pathways and potential contaminant receptors. Mass flux measurements should function not just as a remedial performance assessment tool, but as an integral component in the decision to conduct source zone remediation and in the remedial design process. Technologies such as the passive flux meter (PFM) (Hatfield et al., 2004) which are capable of providing flux-based measurements without reactor-based infrastructure and without generating large volumes of sampling derived waste, are ideal for preliminary flux measurements and should function as a transitional characterization technology between pre-reactor characterization and reactor-based characterization. Aside from an estimate of source discharge, the PFM also provides a qualitative estimate of aging, which is a critical component to consider when deciding whether to

PAGE 181

167 implement a source zone remediation technology. Figure 6-8 below displays the source discharge profile and down gradient flux planes for an arbitrary source zone (simulation conducted using UTCHEMsee chapter 2). After the initial spill, the flux plane is well connected along the initial NAPL migration pathway. As the source zone ages, the residual component is consumed by natural gradient dissolution, such that the flux plane becomes discontinuous, characterized by isolated hot spots associated with up-gradient pools. Utilized in this way, spatially resolved flux information functions as a valuable screening tool, especially for flushing based remediation where successful remediation is dependent upon the same flux generation mechanisms as natural gradient dissolution (e.g. hydrodynamic accessibility to the NAPL interface). Put in another way, high flux measurements at the down-gradient flux measurement plane connote up-gradient flux producing mass that can be effectively targeted and removed with flushing based technologies. Discontinuous, lower flux measurements connote isolated pools which are likely to be significantly more difficult to remediate, requiring several additional pore volumes to achieve appreciable reductions in flux, which perhaps suggests implementation of a more passive technology with a lower flux assimilation capacity, such as enhanced bioremediation. In relation to predicting remedial performance it is important to emphasize the difference between spatially resolved mass flux measurements and spatially integrated source discharge measurements. A highly aged source zone that spans a large cross sectional area could have a high source discharge yet be characterized by low point flux measurements. High flux and not high source discharge indicates up-gradient

PAGE 182

168 hydrodynamically accessible mass that can be effectively targeted with flushing based remedial technologies. 010203040506070809010001020304050Dissolution Time (Yrs) Source Discharge (g/day) Mixed length residual Laterally Ext. pooling pools 40% initial mass20% initial flux Figure 6-8. Source discharge profile and down-gradient flux planes at selected time steps. Reactor-Based Characterization If the transitional flux measurement is high enough to suggest potential source zone remediation, the next step should be the establishment of reactor-like conditions. The establishment of reactor-like conditions, such that the inflow and outflow to the source zone are controlled, provides immediate source stabilization where continued contaminant mass discharge from the source zone to the dissolved plume is arrested. It also provides an integrated measurement of source discharge (Einerson and Mackay, 2001) in addition to the PFM. The question of whether the integrated measurement provides a more accurate estimate of source discharge, such that it should serve as the

PAGE 183

169 final step in the decision process of whether source zone remediation is warranted, is dependant upon the spatial variability of the flux plane. If the spatial variability is high enough where the PFM misses hot spots that are between sampling locations, the integrated measurement is likely to provide a more accurate measurement of the total source discharge. I have not evaluated this question in any systematic fashion here. If it is eventually established that the PFM and integrated steady state measurement are comparable, then moving immediately to the reactor-based characterization discussed below, as opposed to first obtaining an integrated measurement of source discharge, may be appropriate. Of critical importance in remedial design is the successful delivery of remedial constituents to the NAPL interface. In the flux-based paradigm, the delivery of remedial fluids to the high flux producing components of the source zone is especially important. The mapping of flux producing mass, a priori, in terms of its location in relation to injection and extraction wells can be accomplished through utilizing the spatially resolved flux information from the PFM and assuming the high flux producing mass is immediately up gradient, or, perhaps more accurately, injecting a unique nonreactive tracer into each injection well and using the tracer response at the extraction wells, together with the contaminant concentration of interest, to map the location of high flux producing mass. The utilization of spatially resolved flux information for flux mapping requires two key assumptions (figure 6-9). The first assumption is that the flux producing mass is immediately up-gradient. The second assumption is that remedial fluids can be focused towards the high flux producing mass by injecting fluids immediately up-gradient and extracting immediately down-gradient (figure 6-9). The above assumptions

PAGE 184

170 are likely less tenable as heterogeneity in the porous media increases, increasing the likelihood for more tortuous pathways between injection and extraction wells. Trace flux back to source directly up-gradient?Assumption 1: Flux producing mass is immediately up-gradientAssumption 2: Pairs of injection and extraction wells are in direct communication with each other High flux producing NAPL mass injectorsextractors Figure 6-9. Key assumptions required when using spatially resolved flux information for flux mapping. A more rigorous flux mapping approach involves the injection of a unique nonreactive tracer into each injection well. The injection of unique tracers allows for a measurement of communication between injection and extraction wells and a pseudo flux measurement based on the contaminant concentration measured in the extraction well (figure 6-10). Based on the extraction well data depicted in figure 6-10, the majority of the flux is discharging out of extraction well 2 (EW2)which is in good communication with injection well 2 (IW2), such that the preferential injection and withdrawal into and out of IW2 and EW2 is likely to lead to an increase in flux reduction efficiency. Likewise, there appears to be virtually no flux producing mass in communication with

PAGE 185

171 EW3, suggesting that minimal, if any remedial fluids should be injected and withdrawn into and out of IW3 and EW3. A key assumption to the above statement is that the remedial fluid travels on similar paths as the nonreactive tracer. The influence of remedial fluid properties on the applicability of using nonreactive tracers to map communication between injection and extraction wells is an area for further research. 0510152025300246 05101520250246 0510152025300246EW1 Unique tracers used for mapping communication between injector and extractor pairs (after initial baseline flux measurement) PCEIW1 tracerIW2 tracerIW3 tracerUse NAPL EW concentration for pseudo flux measurementC(mg/L)Pore VolumesEW2EW3 injection wells extraction wells Figure 6-10. Use of unique nonreactive tracers for flux mapping. In addition to the injection of unique nonreactive tracers, partitioning tracers can also be injected in order to provide an estimate of the total NAPL mass (Jin et al. 1995) and to parameterize the source depletion models outlined in chapter 2. The question of whether or not to inject partitioning tracers once reactor like conditions have been established is primarily an economic one. Based on the arguments outlined above, flushing based remediation should be conducted based on high flux measurements, which

PAGE 186

172 connote up-gradient high flux producing mass which can be effectively targeted. Information derived from partitioning tracers would help to more accurately estimate the number of pore volumes of flushing solution required to meet the source discharge reduction objective. Instead of utilizing partitioning tracers, uncertainty relating to the NAPL mass and spatial distribution could be incorporated into the design process by allowing for additional flushing solution. Again, this is primarily an economic question where the cost of conducting a partitioning tracer test must be evaluated in terms of whether the enhanced prediction capabilities of the remedial process provided by the tracers allow for savings in the volume of flushing solution purchased which would offset the cost of the tracer test. Remedial Endpoints and Site Closure An integrated measurement of source discharge, as outlined above, would of course require that the extraction wells be oriented in the mean direction of groundwater flow. It is suggested that orienting the well configuration in the mean direction of groundwater flow and implementing flushing based remediation significantly simplifies the decision of when to terminate remediation. After complete remedial fluid breakthrough, such that the concentration of the remedial agent (e.g. surfactant or cosolvent) extracted equals that injected, the mathematics describing the remedial process are identical to those describing natural gradient dissolution. The implication of this is that the concentration of the contaminant in the extraction well during remediation can be used to predict the natural gradient source discharge by accounting for discrepancies in mass transfer rate coefficients, solubility, and velocity between remediation and natural gradient conditions. It is suggested that terminating remedial fluid injection and transitioning to water injection once the extracted contaminant concentration is equal to the scaled site specific

PAGE 187

173 source discharge target is a conservative definition of a remedial endpoint as there is still an entire pore volume of flushing solution in the source zone that must be extracted. It is also suggested that in a mass discharge type framework, it is appropriate to maintain reactor like conditions (source stabilization) until the system returns to natural gradient conditions and it is demonstrated that the source discharge has been reduced below the target level. Figure 6-7 as a Conceptual Site Model and Design Tool The characterization outlined above is sufficient to parameterize figure 6-7 using the techniques outlined in subsequent chapters. Figure 6-7 can function as a useful conceptual site model with which to evaluate remedial alternatives. The different mass reduction/source discharge lines previously defined as representing different source zones with different degrees of NAPL architecture variability can be used instead to represent uncertainty in the mass reduction/source discharge relationship at a single site. As an example, consider the following scenarios displayed in figure 6-11. In the first scenario, a large increase in plume attenuation is possible with the implementation of a relatively passive technology. With the large increase in attenuation capacity, the site specific target for source discharge reduction is substantially lowered such that the time scale of plume management is small enough where it becomes the more economical remedial alternative. In the second case, the capacity for enhancing attenuation is limited, such that the current source discharge is well above the target level. Plume management, if implemented, would need to be in place for a substantial time frame. In addition, high flux measurements indicate accessible mass that can be effectively targeted and removed with flushing based remedial technologies. In this case, source zone remediation is preferable.

PAGE 188

174 Possible to significantly increase Plume Attenuation Capacity Such that Plume Management is more cost effective Limited Potential for Increasing Plume Attenuation Capacity Such that Source Zone Remediation is more cost effective site specific target reduction in source discharge Figure 6-11. Conceptual example of using figure 6-5 in the remedial design process. Conclusions In this chapter I have attempted to outline a general flux-based framework for addressing DNAPL contaminated sites. Critical to this framework is that flux based measurements function not just as remedial performance assessment tools, but as integral components of the decision to conduct source zone remediation and in the actual design of source zone remediation. The models and parameterization techniques outlined in the previous chapters are designed to fit into this framework.

PAGE 189

LIST OF REFERENCES Annable, M. D., P. S. C. Rao, K. Hatfield, W. D. Graham, A. L. Wood, and C. G. Enfield, Partitioning tracers for measuring residual NAPL: Field-scale test results, J Environ Eng-Asce, 124(6), 498-503, 1998. Augustijn, D. C. M., and P. S. C. Rao, Enhanced removal of organic contaminants by solvent flushing, Acs Sym Ser, 607, 224-236, 1995. Berglund, S., Aquifer remediation by pumping: A model for stochastic-advective transport with nonaqueous phase liquid dissolution, Water Resour Res, 33(4), 649-661, 1997. Berglund, S., and A. Fiori, Influence of transverse mixing on the breakthrough of sorbing solute in a heterogeneous aquifer, Water Resources Research, 33(3), 399-405, 1997. Bradford, S. A., K. M. Rathfelder, J. Lang, and L. M. Abriola, Entrapment and dissolution of DNAPLs in heterogeneous porous media, J Contam Hydrol, 67(1-4), 133-157, 2003. Bradford, S. A., R. A. Vendlinski, and L. M. Abriola, The entrapment and long-term dissolution of tetrachloroethylene in fractional wettability porous media, Water Resour Res, 35(10), 2955-2964, 1999. Brooks, M. C., M. D. Annable, P. S. C. Rao, K. Hatfield, J. W. Jawitz, W. R. Wise, A. L. Wood, and C. G. Enfield, Controlled release, blind tests of DNAPL characterization using partitioning tracers, J Contam Hydrol, 59(3-4), 187-210, 2002. Brusseau, M. L., N. T. Nelson, M. Oostrom, Z. H. Zhang, G. R. Johnson, and T. W. Wietsma, Influence of heterogeneity and sampling method on aqueous concentrations associated with NAPL dissolution, Environ Sci Technol, 34(17), 3657-3664, 2000. Brusseau, M. L., Z. H. Zhang, N. T. Nelson, R. B. Cain, G. R. Tick, and M. Oostrom, Dissolution of nonuniformly distributed immiscible liquid: Intermediate-scale experiments and mathematical modeling, Environ Sci Technol, 36(5), 1033-1041, 2002. 175

PAGE 190

176 Cain, R. B., G. R. Johnson, J. E. McCray, W. J. Blanford, and M. L. Brusseau, Partitioning tracer tests for evaluating remediation performance, Ground Water, 38(5), 752-761, 2000. Cho, J., Characterization of spatial NAPL distribution, mass transfer and the effect of cosolvent and surfactant residuals on estimating NAPL saturation using tracer techniques, 2001. Christ, J. A., L. D. Lemke, and L. M. Abriola, Comparison of two-dimensional and three-dimensional simulations of dense nonaqueous phase liquids (DNAPLs): Migration and entrapment in a nonuniform permeability field, Water Resour Res, 41(1), 102-117, 2005. Cirpka, O. A., E. O. Frind, and R. Helmig, Numerical simulation of biodegradation controlled by transverse mixing, Journal of Contaminant Hydrology, 40(2), 159-182, 1999. Cirpka, O. A., and P. K. Kitanidis, An advective-dispersive stream tube approach for the transfer of conservative-tracer data to reactive transport, Water Resources Research, 36(5), 1209-1220, 2000. Cvetkovic, V., G. Dagan, and H. Cheng, Contaminant transport in aquifers with spatially variable hydraulic and sorption properties, P Roy Soc Lond a Mat, 454(1976), 2173-2207, 1998. Dagan, G., Flow and transport in porous formations, Springer-Verlag, Berlin ; New York, 1989. Dekker, T. J., and L. M. Abriola, The influence of field-scale heterogeneity on the infiltration and entrapment of dense nonaqueous phase liquids in saturated formations, J Contam Hydrol, 42(2-4), 187-218, 2000a. Dekker, T. J., and L. M. Abriola, The influence of field-scale heterogeneity on the surfactant-enhanced remediation of entrapped nonaqueous phase liquids, J Contam Hydrol, 42(2-4), 219-251, 2000b. Delshad, M., G. A. Pope, and K. Sepehrnoori, A compositional simulator for modeling surfactant enhanced aquifer remediation .1. Formulation, J Contam Hydrol, 23(4), 303-327, 1996. Eberhardt, C., and P. Grathwohl, Time scales of organic contaminant dissolution from complex source zones: coal tar pools vs. blobs, J Contam Hydrol, 59(1-2), 45-66, 2002. Einarson, M. D., and D. M. Mackay, Predicting impacts of groundwater contamination, Environ Sci Technol, 35(3), 66a-73a, 2001.

PAGE 191

177 Environmental Protection Agency (EPA), The DNAPL Remediation Challenge: Is There a Case for Source Depletion?, EPA/600/R-03/143, E.P.A., 2003. Falta, R. W., Modeling sub-grid-block-scale dense nonaqueous phase liquid (DNAPL) pool dissolution using a dual-domain approach, Water Resour Res, 39(12), -, 2003. Falta, R. W., C. M. Lee, S. E. Brame, E. Roeder, J. T. Coates, C. Wright, A. L. Wood, and C. G. Enfield, Field test of high molecular weight alcohol flushing for subsurface nonaqueous phase liquid remediation, Water Resour Res, 35(7), 2095-2108, 1999. Fiori, A., Finite Peclet extensions of Dagan's solutions to transport in anisotropic heterogeneous formations, Water Resources Research, 32(1), 193-198, 1996. Gelhar, L. W., Stochastic subsurface hydrology, Prentice-Hall, Englewood Cliffs, N.J., 1993. Geller, J. T., and J. R. Hunt, Mass-Transfer from Nonaqueous Phase Organic Liquids in Water-Saturated Porous-Media, Water Resour Res, 29(4), 833-845, 1993. Gerhard, J. I., and B. H. Kueper, Capillary pressure characteristics necessary for simulating DNAPL infiltration, redistribution, and immobilization in saturated porous media, Water Resour Res, 39(8), 210-220, 2003a. Gerhard, J. I., and B. H. Kueper, Influence of constitutive model parameters on the predicted migration of DNAPL in heterogeneous porous media, Water Resour Res, 39(10), 220-230, 2003b. Gerhard, J. I., and B. H. Kueper, Relative permeability characteristics necessary for simulating DNAPL infiltration, redistribution, and immobilization in saturated porous media, Water Resour Res, 39(8), 230-245, 2003c. Ginn, T. R., Stochastic-convective transport with nonlinear reactions and mixing: finite streamtube ensemble formulation for multicomponent reaction systems with intra-streamtube dispersion, Journal of Contaminant Hydrology, 47(1), 1-28, 2001. Hatfield, K., M. Annable, J. H. Cho, P. S. C. Rao, and H. Klammler, A direct passive method for measuring water and contaminant fluxes in porous media, J Contam Hydrol, 75(3-4), 155-181, 2004. Illangasekare, T. H., J. L. Ramsey, K. H. Jensen, and M. B. Butts, Experimental-Study of Movement and Distribution of Dense Organic Contaminants in Heterogeneous Aquifers, J Contam Hydrol, 20(1-2), 1-25, 1995.

PAGE 192

178 Imhoff, P. T., S. N. Gleyzer, J. F. Mcbride, L. A. Vancho, I. Okuda, and C. T. Miller, Cosolvent Enhanced Remediation of Residual Reuse Nonaqueous Phase Liquids-Experimental Investigation, Environ Sci Technol, 29(8), 1966-1976, 1995. Imhoff, P. T., P. R. Jaffe, and G. F. Pinder, An Experimental-Study of Complete Dissolution of a Nonaqueous Phase Liquid in Saturated Porous-Media, Water Resour Res, 30(2), 307-320, 1994. Jawitz, J. W., M. D. Annable, C. J. Clark, and S. Puranik, Inline gas chromatographic tracer analysis: An alternative to conventional sampling and laboratory analysis for partitioning tracer tests, Instrum Sci Technol, 30(4), 415-426, 2002. Jawitz, J. W., M. D. Annable, G. G. Demmy, and P. S. C. Rao, Estimating nonaqueous phase liquid spatial variability using partitioning tracer higher temporal moments, Water Resour Res, 39(7), 112-115, 2003a. Jawitz, J. W., M. D. Annable, and P. S. C. Rao, Miscible fluid displacement stability in unconfined porous media: Two-dimensional flow experiments and simulations, J Contam Hydrol, 31(3-4), 211-230, 1998a. Jawitz, J. W., M. D. Annable, P. S. C. Rao, and R. D. Rhue, Field implementation of a Winsor type I surfactant/alcohol mixture for in situ solubilization of a complex LNAPL as a single phase microemulsion, Environ Sci Technol, 32(4), 523-530, 1998b. Jawitz, J. W., D. P. Dai, P. S. C. Rao, M. D. Annable, and R. D. Rhue, Rate-limited solubilization of multicomponent nonaqueous-phase liquids by flushing with cosolvents and surfactants: Modeling data from laboratory and field experiments, Environ Sci Technol, 37(9), 1983-1991, 2003b. Jawitz, J. W., R. K. Sillan, M. D. Annable, P. S. C. Rao, and K. Warner, In situ alcohol flushing of a DNAPL source zone at a dry cleaner site, Environ Sci Technol, 34(17), 3722-3729, 2000. Jensen, K. H., K. Bitsch, and P. L. Bjerg, Large-Scale Dispersion Experiments in a Sandy Aquifer in Denmark Observed Tracer Movements and Numerical-Analyses, Water Resour Res, 29(3), 673-696, 1993. Jin, M. Q., M. Delshad, V. Dwarakanath, D. C. Mckinney, G. A. Pope, K. Sepehrnoori, C. E. Tilburg, and R. E. Jackson, Partitioning Tracer Test for Detection, Estimation, and Remediation Performance Assessment of Subsurface Nonaqueous Phase Liquids, Water Resour Res, 31(5), 1201-1211, 1995. Jury, W. A., and K. Roth, Transfer Functions and Solute Movement through Soil, Berkhauser Verlag, 1990.

PAGE 193

179 Kaluarachichi, J. J., and J. C. Parker, Multiphase Flow with a Simplified Model for Oil Entrapment, Journal of Contaminant Transport, 5, 349, 1990. Kapoor, V., L. W. Gelhar, and F. MirallesWilhelm, Bimolecular second-order reactions in spatially varying flows: Segregation induced scale-dependent transformation rates, Water Resources Research, 33(4), 527-536, 1997. Kapoor, V., C. T. Jafvert, and D. A. Lyn, Experimental study of a bimolecular reaction in Poiseuille flow, Water Resources Research, 34(8), 1997-2004, 1998. Klenk, I. D., and P. Grathwohl, Transverse vertical dispersion in groundwater and the capillary fringe, J Contam Hydrol, 58(1-2), 111-128, 2002. Kueper, B. H., and J. I. Gerhard, Variability of point source infiltration rates for two-phase flow in heterogeneous porous media, Water Resour Res, 31(12), 2971-2980, 1995. Kueper, B. H., D. Redman, R. C. Starr, S. Reitsma, and M. Mah, A Field Experiment to Study the Behavior of Tetrachloroethylene Below the Water-Table Spatial-Distribution of Residual and Pooled Dnapl, Ground Water, 31(5), 756-766, 1993. Kuhn, T. S., The structure of scientific revolutions, University of Chicago Press, Chicago, 1970. Lemke, L. D., and L. M. Abriola, Predicting DNAPL entrapment and recovery: the influence of hydraulic property correlation, Stoch Env Res Risk A, 17(6), 408-418, 2003. Lemke, L. D., L. M. Abriola, and P. Goovaerts, Dense nonaqueous phase liquid (DNAPL) source zone characterization: Influence of hydraulic property correlation on predictions of DNAPL infiltration and entrapment, Water Resour Res, 40(1), 78-94, 2004a. Lemke, L. D., L. M. Abriola, and J. R. Lang, Influence of hydraulic property correlation on predicted dense nonaqueous phase liquid source zone architecture, mass recovery and contaminant flux, Water Resour Res, 40(12), 119-127, 2004b. Mackay, D. M., P. V. Roberts, and J. A. Cherry, Transport of Organic Contaminants in Groundwater, Environ Sci Technol, 19(5), 384-392, 1985. Mayer, A. S., and C. T. Miller, The influence of mass transfer characteristics and porous media heterogeneity on nonaqueous phase dissolution, Water Resour Res, 32(6), 1551-1567, 1996.

PAGE 194

180 Mayer, A. S., L. R. Zhong, and G. A. Pope, Measurement of mass-transfer rates for surfactant-enhanced solubilization of nonaqueous phase liquids, Environ Sci Technol, 33(17), 2965-2972, 1999. Meinardus, H. W., V. Dwarakanath, J. Ewing, G. J. Hirasaki, R. E. Jackson, M. Jin, J. S. Ginn, J. T. Londergan, C. A. Miller, and G. A. Pope, Performance assessment of NAPL remediation in heterogeneous alluvium, J Contam Hydrol, 54(3-4), 173-193, 2002. Miller, C. T., M. M. Poiriermcneill, and A. S. Mayer, Dissolution of Trapped Nonaqueous Phase Liquids Mass-Transfer Characteristics, Water Resour Res, 26(11), 2783-2796, 1990. Nambi, I. M., and S. E. Powers, NAPL dissolution in heterogeneous systems: an experimental investigation in a simple heterogeneous system, J Contam Hydrol, 44(2), 161-184, 2000. Nelson, N. T., and M. L. Brusseau, Field study of the partitioning tracer method for detection of dense nonaqueous phase liquid in a trichloroethene-contaminated aquifer, Environ Sci Technol, 30(9), 2859-2863, 1996. NRC, Alternatives for ground water cleanup, National Academy Press, Washington, D.C., 1994. Oostrom, M., C. Hofstee, R. C. Walker, and J. H. Dane, Movement and remediation of trichloroethylene in a saturated heterogeneous porous medium 1. Spill behavior and initial dissolution, J Contam Hydrol, 37(1-2), 159-178, 1999. Parker, J. C., R. J. Lenhard, and T. Kuppusamy, A Parametric Model for Constitutive Properties Governing Multiphase Flow in Porous-Media, Water Resour Res, 23(4), 618-624, 1987. Parker, J. C., and E. Park, Modeling field-scale dense nonaqueous phase liquid dissolution kinetics in heterogeneous aquifers, Water Resour Res, 40(5), 247-264, 2004. Pennell, K. D., A. M. Adinolfi, L. M. Abriola, and M. S. Diallo, Solubilization of dodecane, tetrachloroethylene, and 1,2-dichlorobenzene in micellar solutions of ethoxylated nonionic surfactants, Environ Sci Technol, 31(5), 1382-1389, 1997. Powers, S. E., L. M. Abriola, and W. J. Weber, An Experimental Investigation of Nonaqueous Phase Liquid Dissolution in Saturated Subsurface Systems Steady-State Mass-Transfer Rates, Water Resour Res, 28(10), 2691-2705, 1992.

PAGE 195

181 Powers, S. E., L. M. Abriola, and W. J. Weber, An Experimental Investigation of Nonaqueous Phase Liquid Dissolution in Saturated Subsurface Systems Transient Mass-Transfer Rates, Water Resour Res, 30(2), 321-332, 1994. Powers, S. E., I. M. Nambi, and G. W. Curry, Non-aqueous phase liquid dissolution in heterogeneous systems: Mechanisms and a local equilibrium modeling approach, Water Resour Res, 34(12), 3293-3302, 1998. Rao, P. S. C., M. D. Annable, R. K. Sillan, D. P. Dai, K. Hatfield, W. D. Graham, A. L. Wood, and C. G. Enfield, Field-scale evaluation of in situ cosolvent flushing for enhanced aquifer remediation, Water Resour Res, 33(12), 2673-2686, 1997. Rao, P. S. C., and J. W. Jawitz, Comment on "Steady state mass transfer from single-component dense nonaqueous phase liquids in uniform flow fields" by T.C. Sale and D.B. McWhorter, Water Resour Res, 39(3), 142-157, 2003. Rao, P. S. C., J. W. Jawitz, C. G. Enfield, R. W. Falta, M. D. Annable, and A. L. Wood, Technology integration for contaminated site remediation: Cleanup goals and performance metrics., in Ground Water Quality 2001, vol., pp. 410-412, Sheffield, U.K., 2001. Rathfelder, K., and L. M. Abriola, The influence of capillarity in numerical modeling of organic liquid redistribution in two-phase systems, Adv Water Resour, 21(2), 159-170, 1998. Russo, D., Stochastic-Analysis of Simulated Vadose Zone Solute Transport in a Vertical Cross-Section of Heterogeneous Soil During Nonsteady Water-Flow, Water Resour Res, 27(3), 267-283, 1991. Saenton, S., T. H. Illangasekare, K. Soga, and T. A. Saba, Effects of source zone heterogeneity on surfactant-enhanced NAPL dissolution and resulting remediation end-points, J Contam Hydrol, 59(1-2), 27-44, 2002. Sale, T. C., and D. B. McWhorter, Steady state mass transfer from single-component dense nonaqueous phase liquids in uniform flow fields, Water Resour Res, 37(2), 393-404, 2001. Sardin, M., D. Schweich, F. J. Leij, and M. T. Vangenuchten, Modeling the Nonequilibrium Transport of Linearly Interacting Solutes in Porous-Media a Review, Water Resour Res, 27(9), 2287-2307, 1991. Schroth, M. H., S. J. Ahearn, J. S. Selker, and J. D. Istok, Characterization of miller-similar silica sands for laboratory hydrologic studies, Soil Sci Soc Am J, 60(5), 1331-1339, 1996. Schwille, F., Dense chlorinated solvents in porous and fractured media : model experiments, Lewis Publishers, Chelsea, MI, 1988.

PAGE 196

182 Sillan, R. K., M. D. Annable, P. S. C. Rao, D. P. Dai, K. Hatfield, and W. D. Graham, Evaluation of in situ cosolvent flushing dynamics using a network of spatially distributed multilevel samplers, Water Resour Res, 34(9), 2191-2202, 1998. Soerens, T. S., D. A. Sabatini, and J. H. Harwell, Effects of flow bypassing and nonuniform NAPL distribution on the mass transfer characteristics of NAPL dissolution, Water Resour Res, 34(7), 1657-1673, 1998. Soga, K., J. W. E. Page, and T. H. Illangasekare, A review of NAPL source zone remediation efficiency and the mass flux approach, J Hazard Mater, 110(1-3), 13-27, 2004. Taylor, T. P., K. D. Pennell, L. M. Abriola, and J. H. Dane, Surfactant enhanced recovery of tetrachloroethylene from a porous medium containing low permeability lenses 1. Experimental studies, J Contam Hydrol, 48(3-4), 325-350, 2001. Tompson, A. F. B., R. Ababou, and L. W. Gelhar, Implementation of the 3-Dimensional Turning Bands Random Field Generator, Water Resour Res, 25(10), 2227-2243, 1989. Travis, C. C., and C. B. Doty, Can Contaminated Aquifers at Superfund Sites Be Remediated, Environ Sci Technol, 24(10), 1464-1466, 1990. Walker, R. C., C. Hofstee, J. H. Dane, and W. E. Hill, Surfactant enhanced removal of PCE in a nominally two-dimensional, saturated, stratified porous medium, J Contam Hydrol, 34(1-2), 17-30, 1998. Zhu, J. T., and J. F. Sykes, Simple screening models of NAPL dissolution in the subsurface, J Contam Hydrol, 72(1-4), 245-258, 2004.

PAGE 197

BIOGRAPHICAL SKETCH Adrian Fure grew up on the south shore of Lake Superior in Marquette, Michigan. After accepting a hockey scholarship to Michigan Technological University he somewhat arbitrarily decided to study environmental engineering based on his enjoyment of the outdoors. In his third year of study, the first year of actual environmental engineering coursework, he had the good fortune of taking Marty Auers Surface Water Quality Modeling class. It was in this class that he found his passion for environmental engineering, deciding he would attend graduate school. Upon graduating from Michigan Tech he accepted an Alumni Fellowship to the University of Florida to pursue his doctoral degree under the auspices of fellow Michigan Tech alumni Dr. Mike Annable and renowned flag football offensive guru Dr. Jim Jawitz. 183