<%BANNER%>

Anisotropic Fracture Analysis of the BX-265 Foam Insulation Material Under Mixed-Mode Loading

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 E20110209_AAAABI INGEST_TIME 2011-02-09T06:10:35Z PACKAGE UFE0015560_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES
FILE SIZE 10402 DFID F20110209_AAAENM ORIGIN DEPOSITOR PATH knudsen_e_Page_005.jpg GLOBAL false PRESERVATION BIT MESSAGE_DIGEST ALGORITHM MD5
772d80d4999daf8cf390fffef92f2567
SHA-1
8ff8ba4fce581309ee57974e011f0907a84030ca
87534 F20110209_AAAEOA knudsen_e_Page_019.jpg
5cb7159a11fc29de0773b78199421d9a
73deea2f9edefa4bf2c70acc5d4ae108ac356fbb
79894 F20110209_AAAENN knudsen_e_Page_006.jpg
929141e6373f8adaeccfbf6745f0176b
bb5988e72f978be05a7da283bb27f01814833135
26345 F20110209_AAAEMY knudsen_e_Page_112.pro
f9bcd1dba607fba013a264ee62721acf
32ae9778cd7cbe6cb1362d7a0f93e80fbf019f41
100903 F20110209_AAAEOB knudsen_e_Page_020.jpg
5f65c5bbea38ee11a038ce219b087e55
efb0480e8a7ecfab5cec1b65cc9a8f57741ffbc6
51344 F20110209_AAAENO knudsen_e_Page_007.jpg
edf70f0d1035cdccdb56b125519e7c2b
522498fa31a71ebe44425eca54c0114ac2c1b6cb
33923 F20110209_AAAEMZ knudsen_e_Page_113.pro
452c9803ac8032970ef72a2e5c4ace70
684044d41030e4309b16ace9eb0707d9a6f1b685
80279 F20110209_AAAEOC knudsen_e_Page_021.jpg
3e50d101835e08f49a0f65cf5317fb96
08cfd99a37d32708dbc477f3a86cb9fce0033e84
34658 F20110209_AAAENP knudsen_e_Page_008.jpg
613118943003e251c191819b54d03492
801ecb36b824252aca96ae7c862c8d3249e39d9f
71192 F20110209_AAAEOD knudsen_e_Page_022.jpg
ea6c5d5f4fefd77666d8b0030dcfb574
67d11176aacf6997af18d49c4984c974097d6999
70481 F20110209_AAAENQ knudsen_e_Page_009.jpg
b79b2aefe5dd2d467ebaf2ca1c8afdeb
a2f7bfca83b959d3d1ff628e871730d76218038d
71122 F20110209_AAAEOE knudsen_e_Page_023.jpg
48e015af2a09b820850a8424f8b666c5
f5e781d848cd6e1b3fe5f37dd7f35eeca7761975
88895 F20110209_AAAENR knudsen_e_Page_010.jpg
256ba80c33f2dc32d6ed9a3032aba806
00945a5f674214d8cedc79bc74665e037b14ed3b
68004 F20110209_AAAEOF knudsen_e_Page_024.jpg
fef6978cc239f64faf8730fb0522e4d2
ac774786e5586beea49e8ec8c5e7820fb80d809b
56274 F20110209_AAAENS knudsen_e_Page_011.jpg
21c80298520c649b0c04e5600a591850
b37d6b54e7d6a29ef628f2d670fe376a1fb2b5c8
65653 F20110209_AAAEOG knudsen_e_Page_025.jpg
1501d0f1effdfbef5f5031ca748cf194
cd85d3f9d13880ee01340cb18a09a7cb8b8631d6
79712 F20110209_AAAENT knudsen_e_Page_012.jpg
25f1715af48197dd89601d1621321b05
935fb0caac6f7e11e80512f57d9809981477908c
80047 F20110209_AAAEOH knudsen_e_Page_026.jpg
fd4c21fa4091dfe1bf80e163811cd8f3
ffa045241550c4c9fc2c66885a4488ae637d8aa3
70461 F20110209_AAAENU knudsen_e_Page_013.jpg
9194c821ade9acd317af22055cd485f2
3a01016808ac87e02237cd69df117c4430eb29e7
94028 F20110209_AAAEOI knudsen_e_Page_027.jpg
a9f8086817f4b224d5e0381ed7690b21
66ccfcf0d58ce32510dee7e6ab5cdff1efe66d9b
83791 F20110209_AAAENV knudsen_e_Page_014.jpg
54792d9007c6905f8a24d4ae3620eddb
6896c3f9fc7b6e378ab5b508a2398d447ef0d061
41308 F20110209_AAAEOJ knudsen_e_Page_028.jpg
b8d55dab25d3e1143b418b34b9aa8405
e53cc36ad7ef891ab400f79332879a1e9ea3e9a7
88163 F20110209_AAAENW knudsen_e_Page_015.jpg
8884a8d1ea9a72ca0d8b6c3cd6261146
ecf6dcf2a027d9b3531ffa72f315fcab5bbc4a88
71446 F20110209_AAAEOK knudsen_e_Page_029.jpg
eee014b665fbc27b59ebdbe14f4ffc6c
38d570e85a2d341b25f154e7d4df7b773af75141
83401 F20110209_AAAENX knudsen_e_Page_016.jpg
8ad80971ed7ec89216cafb036fbd325b
f0920b8fc8940ebad9c1d388d6726e8a8ebc169b
62102 F20110209_AAAEOL knudsen_e_Page_030.jpg
7f85c5793fea4f94ba4bac5ffbf87962
ad21119d97e1fdc520d74504313e2a3f8c219c75
95127 F20110209_AAAENY knudsen_e_Page_017.jpg
d2e0827d2138f5d43cf762dc6545d6b4
d3a6e4f13869a15f174044265458e1c74f4cc0f5
74759 F20110209_AAAEPA knudsen_e_Page_050.jpg
ab56faf7374f8346a16a67d8b8a855c0
1877e84c079f8edd11f30798a1334300bfd80993
67151 F20110209_AAAEOM knudsen_e_Page_031.jpg
77b85d1e63297e7623a6eeae9af674c8
ce650a3a5b2e153f9c1f3b69b87433496b7eb358
102881 F20110209_AAAEPB knudsen_e_Page_051.jpg
b44dd69effd77e012e474250f17e94f9
e056d587339f14a046be87d09d6542a7387d7fc9
65726 F20110209_AAAEON knudsen_e_Page_032.jpg
5d67c48963e76a48bf0901b51a75189d
7838203a87b738f01514d6db28aa79877197efdf
8799 F20110209_AAAENZ knudsen_e_Page_018.jpg
87bcea11f2742fce8a9134135ef9446a
57ba9679f6e6cc275628e28058db6eb58630bbe8
67425 F20110209_AAAEPC knudsen_e_Page_052.jpg
1950bae9d50860821c74506b6b781583
9f6832079af11e412149962d9fa883fd08f08c8d
98849 F20110209_AAAEOO knudsen_e_Page_034.jpg
bbd6d31499cedd291e2f25210464def1
1d306bec905cfb37cbf1b34bb5acff6116836255
71839 F20110209_AAAEPD knudsen_e_Page_053.jpg
6f43cc6d366013cc499394ed57cfffe0
1433b17ca7d8f3c0f470610b6c201181b9d00927
87080 F20110209_AAAEOP knudsen_e_Page_035.jpg
eb04de532ef367243d54aa7dcae01379
53266936eb2e5e77fba5ee3e17f806c86b9a9012
90814 F20110209_AAAEPE knudsen_e_Page_054.jpg
3d7799651db99250745f8d3914d799fa
10aa7eb285b80a4123239f8cd2d03f067538c97e
55125 F20110209_AAAEOQ knudsen_e_Page_037.jpg
95721c6bc66bc28d9c56005780c846b3
08d44c5e3f5ee35784541a0a91cabaee0eeb35bb
81130 F20110209_AAAEPF knudsen_e_Page_055.jpg
e92b3b6f2a422891698ee03e80f6c788
6bcdc7082d0d3995b33c68feaff84bfd4199783e
68496 F20110209_AAAEOR knudsen_e_Page_038.jpg
75ed56549707de303876873140172b38
3aa245d9fd950c255b5e0e0008400aa845720f96
44373 F20110209_AAAEPG knudsen_e_Page_056.jpg
87a5127f166f19f988016ce390dc083a
c988910af192b72d0662592af08a11c2091acbf3
66003 F20110209_AAAEOS knudsen_e_Page_039.jpg
dca22d04d829f87edf638f1e5af2123f
f73191b8c22fba2d6b2c6638cd925a272e13a0f6
87648 F20110209_AAAEPH knudsen_e_Page_057.jpg
0becf546faa1dfcb23f20b194ac43f90
d7b4737cf44f267ba70ce66b0f1d2ae1872493c7
55479 F20110209_AAAEOT knudsen_e_Page_040.jpg
04d92b6635da757d5b4112e0de1b70be
1ac261e3f473eb2ad1fa52f9671e48e1765f0f49
81700 F20110209_AAAEPI knudsen_e_Page_058.jpg
e84473f9c93b92c3e5e33c135bdd3fee
6d2fcf6d45c2f486ab31a5556c7908a1b4ba88f8
80655 F20110209_AAAEOU knudsen_e_Page_041.jpg
ddaaa928651212a948718ad181120dcd
a21b07ad8b822d4071e3f1f9b3bf84eb8e711083
82290 F20110209_AAAEPJ knudsen_e_Page_059.jpg
d05c857149265303452462542ee8d26e
6a1efb94681806f5897383adef23eadb7d038228
87168 F20110209_AAAEOV knudsen_e_Page_042.jpg
d017a0ccaf70c5f64c744b81e88119b6
a5cb614831aacfdd8bf56171fc89112429e95dfc
59418 F20110209_AAAEPK knudsen_e_Page_060.jpg
1cbc20113873d6be194d5c98be8b5bf3
eb03192dc6f30a920c4a2163a7439c9896afaedc
73684 F20110209_AAAEOW knudsen_e_Page_044.jpg
3bf29d97915984d660183e0ad8aa76d5
b0cc0e167073d9fdc8050d27ab31a8110b19edcb
44524 F20110209_AAAEPL knudsen_e_Page_061.jpg
9c11c48fa16b018daf673862fd09914c
3a6e627dd6380d7242ffe4339bf78a72e5983d80
86727 F20110209_AAAEOX knudsen_e_Page_046.jpg
bbba37ce3ab0dc0fc96d2dd83691aa9c
0e7224b7059ea6f53ef8ba427cd338b75abdf94f
69946 F20110209_AAAEPM knudsen_e_Page_062.jpg
d0b02c983d533e7cfa9f6e4fb91fbe07
f0e7c95f766c8df00ed6c0bae18802497ee8ed1d
62793 F20110209_AAAEOY knudsen_e_Page_047.jpg
0701c6543ffe95eaab5650f3c4ca81ed
cd9df7f768d02e8bd81c8b7d9c3bd4c1631b57cc
51952 F20110209_AAAEQA knudsen_e_Page_078.jpg
7ead987857888e1d306ead59cadbff40
3cca54dead65180bb37b335f67c2cffbe23f6f3e
73409 F20110209_AAAEPN knudsen_e_Page_063.jpg
28075d703a3f8813b823e67c67df42c8
c49ad92068b2fb5ff175ab61d310ff0096bf35a0
74050 F20110209_AAAEOZ knudsen_e_Page_048.jpg
1ab436ae673a3efb77605a87aa9b9ea7
b8468e0d978ffe44b620752b6a33b2e381da1e6f
66181 F20110209_AAAEQB knudsen_e_Page_079.jpg
b5c971dcbd3e34d04a80ed58f2392d69
aa47d0d88aced66ced3776b5b9b4823956624ba0
56464 F20110209_AAAEPO knudsen_e_Page_064.jpg
a1a7cbcc9e97f34cddf645f1a54ff5ab
d89efb7b09bfdef68ac17e7c18d22d8f44e80718
35220 F20110209_AAAEQC knudsen_e_Page_080.jpg
b22e37fa929c4d1b26ac0518213cb493
8c637f179be098577854ac19d38ab64aa2fa7a48
53380 F20110209_AAAEPP knudsen_e_Page_065.jpg
f21357d64ffd64dfa33b23b8a8053ed2
3e21436fa607f0207f6a87d98d5947df7b9f51fc
85407 F20110209_AAAEQD knudsen_e_Page_081.jpg
d22755659e307aa803a168a132bd3359
1d080fd660e87085f386900089eb10d7f1a178ff
75275 F20110209_AAAEPQ knudsen_e_Page_066.jpg
a79f4754898a8f32bbca822f9efe64f9
e20a0b341585dee4e5f55d67be643f268ca7f41d
31713 F20110209_AAAEQE knudsen_e_Page_082.jpg
2c5232a46986722930d5a8b84825b485
39afbfbbadab87055e273af0d894d5a642b8e008
90576 F20110209_AAAEPR knudsen_e_Page_067.jpg
bfe18f5f704f87a1aa64c31bae01ec11
3bcf26e7e503fdcfe67b795d1f09799afb147409
43611 F20110209_AAAEQF knudsen_e_Page_083.jpg
180d5145c023f7dc1e688b4dfa997be6
d81842fdacf8ce5bfb6f7b44db340dfa8a133f15
78421 F20110209_AAAEPS knudsen_e_Page_069.jpg
8691764627def1409c3f9628efae2f11
5db4340c7ed8683d6ddb42a43403471591f2ecdd
83630 F20110209_AAAEQG knudsen_e_Page_085.jpg
23cc4dcbfcab4bbf4530093ab99b096a
7cf1a56ca38a4e54883162ab7b0a858c25cb1af3
64615 F20110209_AAAEPT knudsen_e_Page_070.jpg
5d03816c4ff95a9806f1102bca868bce
06237f24d51a69c259b22676040815d5c1a2c7be
74862 F20110209_AAAEQH knudsen_e_Page_086.jpg
40836cf418c32a7c4e05d8b42cc863ea
3bb2222440c7bea22d6afb5f61defdad50e4e27b
66808 F20110209_AAAEPU knudsen_e_Page_071.jpg
1bb428148be41f30bd125ce74875521f
8a33229ee3752bc3c5705e93acc5862caf6c85f3
54329 F20110209_AAAEQI knudsen_e_Page_087.jpg
8ca59cf42bdf809533a3a5a934f00ed7
2f2b00c1d477135ac7ec17ac77a66833d6e65bfb
76689 F20110209_AAAEPV knudsen_e_Page_072.jpg
6abe5578afab769fe7e50275f15ac8c5
2746a5427f87ca6724305495115a1f9ce54f9863
52500 F20110209_AAAEQJ knudsen_e_Page_088.jpg
a84f6cefa4d6f6d826b4103542d4a9eb
9e71fbe1ed4efbd02e7536f64143a11a1c5d5424
99893 F20110209_AAAEPW knudsen_e_Page_073.jpg
1e6868d7904feec0d38ee5105263d287
18f0bc6f8db1eebc000adb4e732a9733282cacb1
97040 F20110209_AAAEQK knudsen_e_Page_089.jpg
cb218d9db91563631c27d42f0d68413f
5bf79ec0adac1a4025d432f5c432d7233500b421
95972 F20110209_AAAEPX knudsen_e_Page_074.jpg
b996d7acb2a5a09fa8bdcc069c5d0baa
d3c4167e4395f8ab76b123986cb6f3dd08bb4553
96141 F20110209_AAAEQL knudsen_e_Page_090.jpg
896c885ac05c263cfa7488637d5d7c2c
92a0728e96a1cb0d5d2f7e9433d8e5a7d5814bd7
59107 F20110209_AAAEPY knudsen_e_Page_076.jpg
534c3eddc2b5c500158974856720c6b6
1f506cd7f7eb134531ff08099b1fa5061d851155
96369 F20110209_AAAERA knudsen_e_Page_106.jpg
2d13c11f5890209074836dd8146a35d5
fbc4beb2aefbd61e8f2ecf01a29fdc4c646ae5bb
69896 F20110209_AAAEQM knudsen_e_Page_092.jpg
63376cb97ce37d6290262f902d926def
5420b2c433c1eb98293725140012ba8af359d212
94092 F20110209_AAAEPZ knudsen_e_Page_077.jpg
27daf94eb5f0b1710e02d452b6c87a61
2a7c37fa5400aa202e2e7278bd19a7804ea97ec3
94942 F20110209_AAAERB knudsen_e_Page_108.jpg
2904a14fe15b7fa582134bef3ffbf5ea
e97b0bd67cfc57f5187201941e6d67ae2153a051
70311 F20110209_AAAEQN knudsen_e_Page_093.jpg
f46f39e7981f82e6b602b74a6424166c
db41ccc64fb89114dc34268531777c440cf346ea
125516 F20110209_AAAERC knudsen_e_Page_110.jpg
b2cd4b4e026c31e8e071267391e39f79
22c7561b278657741897b56f7e2918b309420532
65990 F20110209_AAAEQO knudsen_e_Page_094.jpg
09c8aec4e1d2a43fec0df36ded4cdfb8
25c44bacaf2ff0707fdb6d9021c56ecc20edd97c
54742 F20110209_AAAERD knudsen_e_Page_111.jpg
016217d821a35e8cdf82792da9d14be0
afb9764f7acd795ba3f05baae76643ccd93594cb
53301 F20110209_AAAEQP knudsen_e_Page_095.jpg
37bc5fe6193c4c7021af8d645f8218c7
ce622d644749504d5836828d42932b35bba13236
48105 F20110209_AAAERE knudsen_e_Page_112.jpg
94f968adef85d63f5fe94750a94fc641
268b6196d9638a617a87d89403e56efe5d05da36
99137 F20110209_AAAEQQ knudsen_e_Page_096.jpg
e59fbca050e337a4961a6abdfa9ce610
21338ecc5d7a7f323859b3d04bf7521ccd44d6e9
63378 F20110209_AAAERF knudsen_e_Page_113.jpg
784a9f043c8fe476287c0532e81b7b89
840cbd5836a9216c1122b5d22d4144f9f0e02184
97636 F20110209_AAAEQR knudsen_e_Page_097.jpg
80b9ebb63f8cc7d148302289a91c9466
24f5c9a07261d06243bd1fe1b119625008688e64
72029 F20110209_AAAERG knudsen_e_Page_114.jpg
79af3b573662edf37a9012e5f821893b
3511251931d503e2ac86d26251db12a327ff2d00
101779 F20110209_AAAEQS knudsen_e_Page_098.jpg
bd547526f09804c889810753b3649444
f344dfc3527e825e97ebabccba77ed016a7f9fef
102260 F20110209_AAAERH knudsen_e_Page_116.jpg
9a406fe958a9fce8f6bd9952b26574d2
be58a4179c3b6aa8bbfb54c9440f3e7979116612
95733 F20110209_AAAEQT knudsen_e_Page_099.jpg
f2837367873b7b05839069749dd53a97
da859725f68f64dc9c2aa7e5b621973b715fd414
119928 F20110209_AAAERI knudsen_e_Page_117.jpg
ee4b0fb0d7d7c5bd340d2ff4c5670275
7e33714fe1e6ca7b0d87f058cf9278a5e50b328d
74201 F20110209_AAAEQU knudsen_e_Page_100.jpg
f3475cc74fa2336c226b8a18bcd46b47
a4fceb0b431420fc9ebae5e65f12f93dce86b500
111660 F20110209_AAAERJ knudsen_e_Page_118.jpg
363eff851b46a43943615d1aa54b60a6
acf3f54364ea335db77ead2ec0c76fb06b92b315
80045 F20110209_AAAEQV knudsen_e_Page_101.jpg
83a921fe84824a9326552c11dec08b13
9022041e2db973200697118c403e2fc62d8990ab
119820 F20110209_AAAERK knudsen_e_Page_119.jpg
a7e02279416b8968ecda03e18e53376f
632d7d329f965812a31ff23b2f6064181f5e16d2
91069 F20110209_AAAEQW knudsen_e_Page_102.jpg
c4be6444e7e47da0107f088b95f807dd
a7dc5d0c88be764049fb19c8c91f10ac1bbc1285
67186 F20110209_AAAERL knudsen_e_Page_120.jpg
7118cdd918c5e7d23e3bd93114353a1e
92c31ede010816cd2d409b23339f6c3a1e37d6f1
81044 F20110209_AAAEQX knudsen_e_Page_103.jpg
c504784770e458004df4d356cd357d85
d70a25e1a6eaf95a42e523d1e06b2780be8a82ee
856988 F20110209_AAAESA knudsen_e_Page_016.jp2
a4881670ffcd0d270699c0883fe4ba8f
417750bb9c5d3aa78f7b91b8e60bdf401fe3a961
41012 F20110209_AAAERM knudsen_e_Page_121.jpg
4a58dcd0933b706d3159137fa48a3c45
f9d721e051ddb48a4107d45689480a89e8ece568
105441 F20110209_AAAEQY knudsen_e_Page_104.jpg
4bb5a043341dc443ef4f6d063255df15
4510e086fbb66ec1143e5550102608f210fbfb79
1051986 F20110209_AAAESB knudsen_e_Page_017.jp2
d4833cd720be717e9e8df43bab658464
ef90e121cd3e7cf53147c3690c97ef339a2036d9
251010 F20110209_AAAERN knudsen_e_Page_001.jp2
6ab689593917b9de58cad8bf6c55de5d
63b5a732c9e9e0c2dd8148b47751d6e1e289d9eb
58685 F20110209_AAAEQZ knudsen_e_Page_105.jpg
806dc3902f8cf1164539bd211df307c3
7a8e1ffd8d3fe7f2fb2a62c17a18bb2ef18ffe03
70389 F20110209_AAAESC knudsen_e_Page_018.jp2
2e9c52c1c1b3a54530287a170a8a5faa
aa4fc7be5019583dea452de6e709e6b9c9b5a153
25590 F20110209_AAAERO knudsen_e_Page_002.jp2
07f25b1126f961d3c819810314d0d5fd
15a2be4fc6c526d5547cb6d9b4d2f9fbe7c3a385
943423 F20110209_AAAESD knudsen_e_Page_019.jp2
998b399cdd72f772ba13cf0accc7b052
69845a2c46cbef068273a3edf7521d5f8d2447ec
94880 F20110209_AAAERP knudsen_e_Page_003.jp2
4f7261fc954d80fee692d89e06b3d23b
78e2eb615bde4246f4c14b9260850211dc1d5da1
1051985 F20110209_AAAESE knudsen_e_Page_020.jp2
b5a5f1c59bc869af4fe84ca28a0de997
6e068a013b37d976d7705e2a8cbf436b8c272801
967663 F20110209_AAAERQ knudsen_e_Page_004.jp2
20efba058f93c7c33a84db6fc1c27fd9
41bbe90a5cb024292731288dae6dc9b75898d546
841575 F20110209_AAAESF knudsen_e_Page_021.jp2
9227533a67ca211212c2d2e0c85b52ed
6597da05db5e8ceae4a1fecdf408a4fa72175e25
84931 F20110209_AAAERR knudsen_e_Page_005.jp2
b3b40860413508a265b1d1b0de0c865c
25ad751f5ea5f9c2cd583e0e8e5c862d761c3d92
737964 F20110209_AAAESG knudsen_e_Page_022.jp2
94e8ff23020298f48ab78734c0ff007d
be7cf49a85e68321dc09a1db3900121268cebe2f
798918 F20110209_AAAERS knudsen_e_Page_006.jp2
aaf1f8adad5eb1cd5a2586e58b1eaf13
bb7a56f0417f761cfc1d0b55e3cf604103ff8437
755535 F20110209_AAAESH knudsen_e_Page_023.jp2
13d6156d78705f8ed3a374112ef79ef2
599be9aeb202b32f0d27db7c7fa8cda8d07dd63b
502010 F20110209_AAAERT knudsen_e_Page_007.jp2
3c00eff697b662267052b134e23be64f
f470b9f80120b82e4fdf894c7f7a48c8c64443eb
713033 F20110209_AAAESI knudsen_e_Page_024.jp2
89dade031bf7c7ce92290a890b86634b
ca66ce75ec3ac75fbb17ac5446a584af6bb2ced1
721923 F20110209_AAAERU knudsen_e_Page_009.jp2
aea3aca63c2521cadd2453daea2e439d
a8b40017369cd93f9055da05b091d17e942433a5
698251 F20110209_AAAESJ knudsen_e_Page_025.jp2
fca9edd00c7a94195a08e3dba031e4ed
85a6badf3f01a67c56ab2448f16d4041f21ac036
574201 F20110209_AAAERV knudsen_e_Page_011.jp2
9e34f22e9e79cb6f6b88a6df40f1164c
d94d3f51cc168d844c69b062735082432dd03876
877498 F20110209_AAAESK knudsen_e_Page_026.jp2
56824e365f0fc65265291608fec8cba4
51500e8bff4da3c3eda6d72f8703d6c345e06238
862919 F20110209_AAAERW knudsen_e_Page_012.jp2
db308595da3303c7a0fc6166e133cd8b
e14cd3bae240b77f71ab2a9f5b7f39ca68c2e299
1031743 F20110209_AAAESL knudsen_e_Page_027.jp2
e5f9c084c8b079ca13d1685ae9d6e245
6fc20773d8d469785d42f19725851cda4d308d5e
765204 F20110209_AAAERX knudsen_e_Page_013.jp2
49dd3c0ed698ad91586758ba0d8359d1
c6a42dcfb3e7eb7370458b0deb61255906cdbbfd
391049 F20110209_AAAESM knudsen_e_Page_028.jp2
d60f3bcc4a5e57b80ba3c7fab2588742
18ae1690294a44b3956f0c9fcaaee6024da88a49
894597 F20110209_AAAERY knudsen_e_Page_014.jp2
4edeca24e1bcd5d23a60c4a09d98bb4d
772c4a3fad51725368a363c0ef683f9fe3dc329c
794894 F20110209_AAAETA knudsen_e_Page_044.jp2
2ce163742d4bedbc940a6c5fd45388eb
a69aa7d58445c3111cbdc7482f85ba426326b010
743834 F20110209_AAAESN knudsen_e_Page_029.jp2
1d843ced535dbfc6727c9c3b7ed5fde2
4c0e68d90b238961ee8fe62a09880462f2d8574c
968810 F20110209_AAAERZ knudsen_e_Page_015.jp2
4cbe7559e1e91a3531937b5351b633ba
c94c856190f1c705ab072eea760d06176a26623a
701619 F20110209_AAAETB knudsen_e_Page_045.jp2
83fd61bd76b9a516a1976dd479bd4c15
885c780ef5dfe0309937f300bc6dc04637cebf15
668517 F20110209_AAAESO knudsen_e_Page_031.jp2
a8a01c069e0cd013f059227c8882d397
076b62ff714fc793ff20e1e4c196d152c90d44e7
934487 F20110209_AAAETC knudsen_e_Page_046.jp2
cde9720d83a6672908415923e71c0346
df01159e4198fcb111138ef8a8fa6b5e48ce739e
625988 F20110209_AAAESP knudsen_e_Page_032.jp2
4d9a1e09967e8704e4f60b4b2e906ccf
641b6020f14974b812d9ec7f7b1d6820caef99ee
786234 F20110209_AAAETD knudsen_e_Page_048.jp2
e12e918ee9de74082444167a1efa556e
a900bd59721b8bc075be5dee599c35da64ab9779
1051924 F20110209_AAAESQ knudsen_e_Page_034.jp2
4192425606f4e8972e76959abd469c77
f5255139160b34fd1f5d6cf0ce948e1ada46eb02
528020 F20110209_AAAETE knudsen_e_Page_049.jp2
3d52c5558ac61c858c0a588e907db179
51f6253648da9d105d75e0b56562ca3ea9ab7ac6
960964 F20110209_AAAESR knudsen_e_Page_035.jp2
72e609b15d83c4345b6925203eea23d5
0f34f909c4bfdade4cf83ddb4263ceb61f2ad481
793528 F20110209_AAAETF knudsen_e_Page_050.jp2
27cf7998c18d64b154c9f3279883288f
c405c57736febff899508377f33ce976dec223f9
749263 F20110209_AAAESS knudsen_e_Page_036.jp2
4f366c221786600e9c9d8f0a6fe91d0c
aa699a1065ab7c43104c5dc4a95391dda378575e
1051960 F20110209_AAAETG knudsen_e_Page_051.jp2
fe3435f422842ed1bbcabf0aad3c89a9
95cc50a92bf858b85fa21d91f6b1485af79b5dc3
557631 F20110209_AAAEST knudsen_e_Page_037.jp2
40eab1b960855d6af6192dac8445df92
482169e0dc302e71aa0289487e3ae5662af2f831
982711 F20110209_AAAETH knudsen_e_Page_054.jp2
94f62cd6bd0212302c4eb1fbf330318c
b18436b0118f64587235153024e7ac165ceeff58
723390 F20110209_AAAESU knudsen_e_Page_038.jp2
7eb21099160f2555409ab268567602c9
a28fa35ad2d67b520ee96a110f24b27884e8ba7c
880595 F20110209_AAAETI knudsen_e_Page_055.jp2
831e7e7dd0c9c64dbb2d7affc8d831ef
07bfa759691eeb4ae63a7d75a5ab4ef0c201db20
714328 F20110209_AAAESV knudsen_e_Page_039.jp2
05ef5e08362633d031b2f600c59f30a9
400227b8d1ff23cd0c96c8a689b43b91991f2945
431003 F20110209_AAAETJ knudsen_e_Page_056.jp2
5b72a19702d6cf050d30ff2a95c07a62
8cc667daaea63453b142bf6d2ce84f16d34eae1e
553427 F20110209_AAAESW knudsen_e_Page_040.jp2
be536408c347bb9d199f01678b98ac23
83929482ea2b71e08992cee7be9c5de9113b51f1
917743 F20110209_AAAETK knudsen_e_Page_057.jp2
c6adef72a8456d14feb05386a6955847
b5ee0a993f2e61d2fcc981194f90a2228ec92e0f
875677 F20110209_AAAESX knudsen_e_Page_041.jp2
d9ef7fdb4095da2c5b6417a694a6d3f4
eac3cdc5708b95eaa4f55e53f633c47031fa665f
874956 F20110209_AAAETL knudsen_e_Page_058.jp2
a8cdbffe280609a68ebbb5627a07c23f
433f6f1e566f1985db4f30f77e44d462c9015302
1051953 F20110209_AAAEUA knudsen_e_Page_073.jp2
28cfedb7eb6a907ff228c4f2f9de4c7f
8f67ccc32d5438095f889632c4acb691b0549f49
890412 F20110209_AAAETM knudsen_e_Page_059.jp2
91c379acf6ad0ba1f2ba2a53047d6502
32d7afb41b8ba1b6b9ebe54c0a266cf3d67642ef
945032 F20110209_AAAESY knudsen_e_Page_042.jp2
4cd0e54a1477346b617d8b98c2b89b64
52029d078fcc59eed8d6afec84decaa897d72e86
1021414 F20110209_AAAEUB knudsen_e_Page_074.jp2
7e55289320f15ab8f1c8f41a215f0d52
d554460edc431159da2d3cc0ce0fe8ef7ef3d91a
611548 F20110209_AAAETN knudsen_e_Page_060.jp2
8d632eed9f6fbad12879772d6e01f98f
71fc06eb812d82738d8c03b72feb6024608cf412
846008 F20110209_AAAESZ knudsen_e_Page_043.jp2
63a96c5020fc99450a4865949a8a09e9
7e054a17d4555df5e424f4275de2e31d6f0b5b50
481410 F20110209_AAAEUC knudsen_e_Page_075.jp2
578b86c6265737871609ec76bdc304a6
20076602a0ea2a0299a6d3f7ecf10c484c6b2296
413274 F20110209_AAAETO knudsen_e_Page_061.jp2
8913565e74dbafb2cbd3d339bc5b3440
328f863864a624eb484fe2016dbd63a2c8f3b1bb
592674 F20110209_AAAEUD knudsen_e_Page_076.jp2
8182cc4bb036aeae21e40894597cb38a
2ac1cebf53c2f681d12bf8691bcc9f170ba6e170
741431 F20110209_AAAETP knudsen_e_Page_062.jp2
220ce483a835f067ce01c20516afbfc1
07ffda8b954d6ed6bddaf8956a3fd48e1ce6abe8
1017915 F20110209_AAAEUE knudsen_e_Page_077.jp2
40d88d6a49d8ac3adebd288a9090733c
4d5043995bfcf755dd4bf2faeb9a46a0076c6f0a
762076 F20110209_AAAETQ knudsen_e_Page_063.jp2
74bb05df50efed1c4f394d3f3a0e718a
097ff02603c25e85d1acf1cfd25a364df6b0e342
555864 F20110209_AAAEUF knudsen_e_Page_078.jp2
2dfe2faed18b9c7eaaabf711a9d0cdef
c0299de31d519bb472ebca3ea8779bc303d00c63
565574 F20110209_AAAETR knudsen_e_Page_064.jp2
bb8b9b4bfbdabc2c7e36ec8aaeebf52b
a8ecb674b3089639c6eaeb388458c2fcc0a2e580
30555 F20110209_AAAFAA knudsen_e_Page_054.QC.jpg
25e767bcff4e22297965417d46af765e
cfbf325bf93bc356b0b758fbdc7c4bdba8541f3b
667326 F20110209_AAAEUG knudsen_e_Page_079.jp2
51a59a03c16a97fed99dec6a9a631d6e
195885cf42f466ce38306fd98c9c1b2b92df1c3e
496962 F20110209_AAAETS knudsen_e_Page_065.jp2
93aecf88ce7d695170aa54ce029c0e3b
419950fb431581fce6bc8c66618c869f6014a608
26993 F20110209_AAAFAB knudsen_e_Page_014.QC.jpg
c01cf1843044af68165955b5c1038f7e
5ef3bf698109baf7f7a33b6a20d4befd978e44ad
390809 F20110209_AAAEUH knudsen_e_Page_080.jp2
e7a5c71405cbe12dd9b300bc77cee403
7bdd5dc2f6939028fe1106dcd83b7640cca0605b
772237 F20110209_AAAETT knudsen_e_Page_066.jp2
c45910607314df70e059a959b73b2666
e25cbe1e982921b0510bda0738d6fee9fd205aa5
5662 F20110209_AAAFAC knudsen_e_Page_030thm.jpg
09cb58ea57fd15dcb9874bcbdbea442e
854f3f89c566f44f992d1cf8e888f43ab02aeecb
907553 F20110209_AAAEUI knudsen_e_Page_081.jp2
f0a031c0147cf71e3aa88c094dafd88a
b17ad862b1673d1670f2db7f0cb4511cf804885b
976437 F20110209_AAAETU knudsen_e_Page_067.jp2
c431b80bae26ba1578f45ba64582de8b
64f746df5a1feb1abda50a870a7f35ac6f0f24f0
7186 F20110209_AAAFAD knudsen_e_Page_019thm.jpg
e9c4c2d2c75fb498bb40d64e92194c43
43a1ea8d4d09f4379fb3eb68d99b9b7af7ef82cd
298928 F20110209_AAAEUJ knudsen_e_Page_082.jp2
b57d51df82cb41b4676963886caddb78
e65c795962dc5df64b5d9086fb62a562c5a4dcb5
659897 F20110209_AAAETV knudsen_e_Page_068.jp2
5e2846b2f10eb1e2ce09f8ac1d10c8bb
0a5a202e3a8cdb1bad6656682e11d61ed660f592
5297 F20110209_AAAFAE knudsen_e_Page_064thm.jpg
303b7f296a2a3dd13cfc5fde4ebd434c
98e860d66c4f790c156a107a9907ff38228422e5
457541 F20110209_AAAEUK knudsen_e_Page_083.jp2
1a722ee1d8981f9dc2625e2b6f58a117
5ef980603ccc2b789cd52b7e2243bbe5f1e8b8db
830030 F20110209_AAAETW knudsen_e_Page_069.jp2
61eb4f8a5f14d402ce7255327a7434e0
e46ed675f796cccedfec39ba2d3ab70b782222c4
20370 F20110209_AAAFAF knudsen_e_Page_120.QC.jpg
1561357a2d9b7050feafcf3535946c97
1bf4fbe35fc8b2d884175c7a356b3aa88b50c264
853778 F20110209_AAAEUL knudsen_e_Page_085.jp2
f2ce3d59f3dfdc88f5807baccd971ba1
62d98deb0d03dcb75c1ba0658705cb1128981a02
675799 F20110209_AAAETX knudsen_e_Page_070.jp2
5390d356c905fddad03dbf859c55512e
f8aa509922748face889984793ab2cf16ec67e1a
32759 F20110209_AAAFAG knudsen_e_Page_073.QC.jpg
51be0d326c8635db079b10f27d7bac7d
14d627a7e5727f8e081530731e7af56d74aeeb86
781654 F20110209_AAAEVA knudsen_e_Page_101.jp2
497cbe89f39b7a6512635d6814e1ce62
a168737691e3c5b0fda6c37d6786ecbea0614f81
765934 F20110209_AAAEUM knudsen_e_Page_086.jp2
9c525e672fcd547ef5e141fbdd52cb3d
6f7663867d9eac7e17ed0a2e80984deb256a0a75
695362 F20110209_AAAETY knudsen_e_Page_071.jp2
430f7ada6dc5f80b5643585531bb0ba9
bc36ae958eb17d728bae3a0445d444a020a8c749
26729 F20110209_AAAFAH knudsen_e_Page_059.QC.jpg
9fbc3444dc66fdf4132375ab9ec65d06
826aa766d4efaf78b6401bf466281555aa9a68d5
602707 F20110209_AAAEUN knudsen_e_Page_087.jp2
eb8f21a5617faed07d830552137ea03a
6d3373fe7144ffd4de4154c320b91ba91e4a6dc0
813866 F20110209_AAAETZ knudsen_e_Page_072.jp2
5b3183766477b5087e38279446230dba
a8016d3539bf88a7af203039c822d009f3a4d79f
26073 F20110209_AAAFAI knudsen_e_Page_041.QC.jpg
331f03a41288e085b747812739eb1a5c
49d039eb537d45d0d56d4559050ea74429511721
934943 F20110209_AAAEVB knudsen_e_Page_102.jp2
cb1d87868e4d30c693a1a8137a6b2e58
d0b54a1a1a5afadfc4801d1c666b80adb9e19952
571928 F20110209_AAAEUO knudsen_e_Page_088.jp2
8254476729354a2a4181c6aeb2816478
3b1cf4f350e96c9a95b1ffaffbc0d4daeebeb633
6942 F20110209_AAAFAJ knudsen_e_Page_021thm.jpg
b7d69ef0a4b9cdbccab2decd69fb8c1b
e95108190044df5b9ab5f4a8a916b1f4a31c82f3
829123 F20110209_AAAEVC knudsen_e_Page_103.jp2
9b353b8365d539cddd5f5b60473b9bf3
dac95a997e135e80cdebb92b11fc7c1d1997f28f
1004238 F20110209_AAAEUP knudsen_e_Page_089.jp2
5b06da7198222067b4b1207f48d8f894
0b34764c73ec12b3c18bbd98d4811a8ea592fe9e
7522 F20110209_AAAFAK knudsen_e_Page_077thm.jpg
c7764edbfc0089bb96dfcc04a37587e5
fdf355057f5c4133013002d8dea461ff13163dfb
1051958 F20110209_AAAEVD knudsen_e_Page_104.jp2
25bdf257bb3026fec7c53d87cea38ec6
734d51c56e6c7f7f585211db46c9620b19b831f9
1027316 F20110209_AAAEUQ knudsen_e_Page_090.jp2
69c0576a8e7c42e5d7c5364333ce2ff7
84ef1d1f1cc41728f9c071429b9c166292f3a766
580705 F20110209_AAAEVE knudsen_e_Page_105.jp2
141385f234bc480e526da7849f6f5b96
e380caea1b3708eff3f94df8da0e6034e19616c1
701657 F20110209_AAAEUR knudsen_e_Page_092.jp2
918fc1cddc63906713b4bd9925516303
4b7068c5926b126f282b4cd6a9c15ba887fbed93
6220 F20110209_AAAFBA knudsen_e_Page_093thm.jpg
6915554c41b994178dc74f9a6327f95a
07633323a0f7c0174ff3c28494204b1124a2ccb3
18503 F20110209_AAAFAL knudsen_e_Page_076.QC.jpg
0e1b842886dbc3354142ace4cbafeb49
7a6219bd6484fe6c85de6279d10e3d35e0edd1ed
1041558 F20110209_AAAEVF knudsen_e_Page_106.jp2
b1d524a28df7e7394e1f24d7318dbd34
bc544650e4df492bef7ee196df9b08d02e2a006f
719439 F20110209_AAAEUS knudsen_e_Page_093.jp2
694da118f3970322bc5693fee59f8455
3b9d5e895f72bb428f552cf261f6f76593d0fb89
23401 F20110209_AAAFBB knudsen_e_Page_029.QC.jpg
8b1400c185ae4180e9cfa00550eedc8d
a7c7c0176bd2d43fdd7cc7b907690d0b5168c080
24257 F20110209_AAAFAM knudsen_e_Page_086.QC.jpg
605af02f110c9272bb70ca5a327ce661
5e5c33bc92c0c4218fefc33bfa59400424d7baa4
1051915 F20110209_AAAEVG knudsen_e_Page_107.jp2
604f35d1650e7549ea92ecea5d5e8f59
d8234f9e7cdc3ec28f0604d849a7aebfa274d32b
694852 F20110209_AAAEUT knudsen_e_Page_094.jp2
e19fecdd03372f49c28b241b63a9ceb2
b73693b1f13d594dcf0d2e2d85ef54e406f64dad
5290 F20110209_AAAFBC knudsen_e_Page_065thm.jpg
c18b9ed702f0347709abcf4de296a062
75109f390833bfed211ca712c956b43d2127cf9b
584 F20110209_AAAFAN knudsen_e_Page_002thm.jpg
3d7456a0c1b79dc980895c2ea7131dab
1160ae9f115d3746f0aee448fc2c0ae56af53a78
1024611 F20110209_AAAEVH knudsen_e_Page_108.jp2
1e14c081236ad2b0b7cba58604bf6719
4843f4c55c0de508021f67fb2d20af513d964810
540609 F20110209_AAAEUU knudsen_e_Page_095.jp2
412055e1efde1b48bf73cb02289e10b8
d92dc1159b855d18a626200c9a8d87f00b98a8b1
29612 F20110209_AAAFBD knudsen_e_Page_116.QC.jpg
243e5a9011bfeafe33fa8798850434f8
228fa1d621159f010cd66bb5857cace9161cf19d
24464 F20110209_AAAFAO knudsen_e_Page_063.QC.jpg
d250b5918102323d8095afae501d7458
4aa05804df084b66b6da010770c73966713d4134
F20110209_AAAEVI knudsen_e_Page_109.jp2
d4b0b985d9ae8e60ed5a0239842a1bc6
c5484c4606dc2f8613abe30541f68ffbd1dacfec
1045690 F20110209_AAAEUV knudsen_e_Page_096.jp2
3d442c86725026eb9af8dfa8ca88dd87
3fe87b47a170b716a92a55650d1c17719a5cc1d2
4645 F20110209_AAAFBE knudsen_e_Page_083thm.jpg
49c938bdacb3e283ad987c0082d751dc
ecf6b29acc831dfec8ec7c2b9e5f927e64049986
27384 F20110209_AAAFAP knudsen_e_Page_055.QC.jpg
7a3feb5bde471e7ac7f195d7f2904c2c
9e0bd3f8e9ec849b9d3be8f857d80870bde7a488
585414 F20110209_AAAEVJ knudsen_e_Page_111.jp2
d5923f3244746e2cb8da8705f7a3da9c
53a4fd3bc1a7496d0f42a3f8748e6714068fa97c
1051947 F20110209_AAAEUW knudsen_e_Page_097.jp2
335c256ac8974aea8ea98247c2a6ab31
635b587daf63dcd3286046826b9e75fe77377573
2211 F20110209_AAAFBF knudsen_e_Page_115thm.jpg
8618488203cccf93f6b441bdf0f5700d
752156c2edf02ae12c9ce24b8f05f386d072d5f0
32482 F20110209_AAAFAQ knudsen_e_Page_110.QC.jpg
06331f7ae1e5b71231f3a143569e2563
82371b1258c7bf9d4b9fab9eb834a5374bb7bfff
479485 F20110209_AAAEVK knudsen_e_Page_112.jp2
b2209b3e6088e67ba73b930909f39813
7226cf0784b81cf150f8aaf8cb399ddd08c09591
F20110209_AAAEUX knudsen_e_Page_098.jp2
a19cd4849cf28aa1e1e29022acc1a489
e23425770d41fec2e52568fcd690e6130a62c34d
7422 F20110209_AAAFBG knudsen_e_Page_067thm.jpg
56eb57f1303eff23b4b00babc2268c97
824a6f61f02599dec005e7789f6111a574a3ed1b
6263 F20110209_AAAEWA knudsen_e_Page_100thm.jpg
3e48ba517133b872a01d3b9c7ffc92e9
fdb59a62844baa282ebb3cb01eb19319f8cb2f4a
16836 F20110209_AAAFAR knudsen_e_Page_065.QC.jpg
17cbfe40bc8ffe00ee9dd22ae8e5bdbc
dd1264178c03121c5b1511d6a2c47ac67ec37d3d
658436 F20110209_AAAEVL knudsen_e_Page_113.jp2
d43ddc104c074adc8ac05a0681b30a4d
161dcc53cb8b768577f643f841cc46ea55fb163b
1029619 F20110209_AAAEUY knudsen_e_Page_099.jp2
8d810198c37910bd1e70dc195ce16095
64f076802aaf6fa4017f055b833905064281e233
7896 F20110209_AAAFBH knudsen_e_Page_017thm.jpg
1f10fee5762b0f14ada31a5055747246
d8974c2505bb8db0fe48946f3fc44d76d836f0b8
23557 F20110209_AAAEWB knudsen_e_Page_022.QC.jpg
ee4ea3646bb0cd18dc751b444b55830c
ca5e2c69ca72f270ba70d09e6261b17b32f39060
7328 F20110209_AAAFAS knudsen_e_Page_046thm.jpg
1ea5079cab2d011965bdd83e3e4c1058
c6324d31626066a6f6a505fa3d8986cc90860165
729103 F20110209_AAAEVM knudsen_e_Page_114.jp2
534f959e5716a8566a69e64c5930d6b1
75e2ba0e4ab334d81b736f41bbff6666d8e4e2d4
747781 F20110209_AAAEUZ knudsen_e_Page_100.jp2
921a0dcf1c1d21f0899c03a7a3cf24bb
e77094775da075222252865e2c1bf4956c2502d6
39217 F20110209_AAAFBI knudsen_e_Page_109.QC.jpg
e95c11c1a29b1fc3d44dc01755c00ea2
54e56a36966e8d2e0edd230286519ca6451bae0e
30165 F20110209_AAAFAT knudsen_e_Page_090.QC.jpg
f13c533b768b62a9d3dd6bebb5b0d4d5
90a4f83a1c320a097561c9a975a91cb3ac750515
187823 F20110209_AAAEVN knudsen_e_Page_115.jp2
38f35c396b7a60648fab999e15ddd9f4
b3df6e4c2a2fe87328d22d8a17d16efcc2f2ac60
7510 F20110209_AAAFBJ knudsen_e_Page_115.QC.jpg
97e245fc48dc431032e92e4a7364dd87
22bc7587d9547fcc94d67f7f3ea44f514277db81
6315 F20110209_AAAEWC knudsen_e_Page_048thm.jpg
e555bae84082ff524b90d13936d195dc
7525ff883e38501787e309d646a51ff9f86c8e47
8186 F20110209_AAAFAU knudsen_e_Page_020thm.jpg
f6f3389257169f054d534ec5a0f09825
e85d279dfca66585332e66f87668e2b5b2044237
1051971 F20110209_AAAEVO knudsen_e_Page_116.jp2
f253c83b3825ac8b0787feccb494e38a
b2c5d3cb7cc10cb702cc09caad77be6e19d632ed
22183 F20110209_AAAFBK knudsen_e_Page_025.QC.jpg
68fd342fc3e96e8b450cbc33ded68b4d
4ee0d3a78f15dcf78fa95d74b4d0646d42e8d527
6355 F20110209_AAAEWD knudsen_e_Page_045thm.jpg
8693a2f7ef75b5512161b854e48f5a59
760380f558f6e53c8e4d11334a88a13e77120a40
32328 F20110209_AAAFAV knudsen_e_Page_020.QC.jpg
2b976d3fd134561ff9d7abc8ab50c752
4a93bccf2b37bd659f66bb0200bb9de45600e1be
1051982 F20110209_AAAEVP knudsen_e_Page_117.jp2
d74332ea04d17058fb7c1048ca63bd5c
b82bf23efeffafd56c7a7e8d0e77e1fb9934ea26
30232 F20110209_AAAFBL knudsen_e_Page_099.QC.jpg
87ca316c51a473aa799df337ca603ef9
c4532a1524104eac4399456b58f4bbd8a9a16624
6772 F20110209_AAAEWE knudsen_e_Page_063thm.jpg
2724c7ca7ab260ac4e15db48159351d3
555ff8db7b1df5df12a92725a28861ecee8e3a71
22468 F20110209_AAAFAW knudsen_e_Page_048.QC.jpg
8f4d3c62e31d142313d72243122f5b34
1d62706a64482602eb0b9b02ae300e69e6d8693a
1051966 F20110209_AAAEVQ knudsen_e_Page_118.jp2
ec3e34edb29cb172d0b417aab89d7552
06b1b316198b9a9ef9f501e33890ccb39bde3f62
26704 F20110209_AAAFCA knudsen_e_Page_026.QC.jpg
ef4c82fd05993d0b9fc6954ff6106b91
03991de5fdce531d402a704ff91a4f0aa4bd8fda
4443 F20110209_AAAEWF knudsen_e_Page_028thm.jpg
2db66dc4ad684dd220ce38c8df202fe9
f087ea24e342722bf0489286260b27a5b11e5fba
23756 F20110209_AAAFAX knudsen_e_Page_053.QC.jpg
31b4102fd749b29af1fa1fa5da850ebe
a5db0d94146a999c59cb8c14c7296daf5132ebee
F20110209_AAAEVR knudsen_e_Page_119.jp2
a5398e04da3d8162b0395cc682831122
f5891b2704401fc59ad14a940db02345d786af17
31482 F20110209_AAAFCB knudsen_e_Page_074.QC.jpg
0a6c862cba2ad43d26d5f7e46c2a5b71
01f2cbd7c8a8020abeea27e49f04d699c9b2809f
6798 F20110209_AAAFBM knudsen_e_Page_069thm.jpg
3c768fa378c243d93f0a310a80439aef
d9c8f8112c7419d6bd7068117075e3f89615febb
18415 F20110209_AAAEWG knudsen_e_Page_075.QC.jpg
7a4d183a36108fbd3cb823309f41e46a
53b60f5deaff6d721af67e7431f558ede9897626
22141 F20110209_AAAFAY knudsen_e_Page_039.QC.jpg
9107d6c1e3343f84d159bf9063a076e1
03a7bb15824680ee25b8cdb2ff67437e3ca7af8e
719044 F20110209_AAAEVS knudsen_e_Page_120.jp2
4902fe416c63072e6ef59ac81b99b759
344b22fd4b4400bbdc3f138ddca9583a493de66b
4780 F20110209_AAAFCC knudsen_e_Page_112thm.jpg
f3bcf0312ec7344835f9a1cfd6ac9255
105700e21d12cdfb50a0e9db88d6be23221f9555
3332 F20110209_AAAFBN knudsen_e_Page_080thm.jpg
2ab947d29f08b3e98d399d63695ecb67
ff00ba3e8995e1fdc06d43134c5665587e2fbc16
6133 F20110209_AAAEWH knudsen_e_Page_029thm.jpg
c0333737207491dd4981f7a79fe7b8bc
92880f7c3ac2d27fef8b6f7a96b6d060452edd58
22044 F20110209_AAAFAZ knudsen_e_Page_024.QC.jpg
1e08a8c6ec8ab4da720a7f056226ae09
132a29df18a456dc5be239d587a2030f57466d66
418112 F20110209_AAAEVT knudsen_e_Page_121.jp2
de6e032f58270eb541113291557ca22f
52c95523dfa8dd5b37abeac511968cd50b79bcd3
7553 F20110209_AAAFCD knudsen_e_Page_054thm.jpg
b5a6b0fadfb7dcc62c3226a2d55c23ce
8c13f7659aa40d057cec769ec42754b8c54465d9
20471 F20110209_AAAFBO knudsen_e_Page_079.QC.jpg
4ccac77a9ecc9b3192b1bae71de4199b
2b3b6963a9d71eca031554c1b622021bfc579fdd
25363 F20110209_AAAEWI knudsen_e_Page_114.QC.jpg
fb65ecb25ba7bb23482dfc18061a1bba
0bf385ba76b605d501ed76ad9bf092746887c7dd
1581661 F20110209_AAAEVU knudsen_e.pdf
f05b28d1720ec0d48336b60e6077539a
682362e481153688aa93d1807e3c0b97ad4bf6ba
8957 F20110209_AAAFCE knudsen_e_Page_109thm.jpg
9de0be20a7144cacf09dcfbe9ce7b729
67b7b57fa1b29ee928e646f1a02e4b720a0f4588
7852 F20110209_AAAFBP knudsen_e_Page_035thm.jpg
d5b91c6124a073c23d470f722d7cd1f0
1ec5f25ce1fb655617568a75bf91f073c8af5643
13085 F20110209_AAAEWJ knudsen_e_Page_007.QC.jpg
73ada83ab56635834b4536de5f21d0f2
78e80d1e006e6cbb1fc7cf831fa37ad8126844f9
6297 F20110209_AAAEVV knudsen_e_Page_010thm.jpg
dc6666f6d99ce7739f4ad20bcbedfa72
974510fe78a71ed21430f47a6f1c69f2bffdab11
5966 F20110209_AAAFCF knudsen_e_Page_071thm.jpg
e63d263328a06c730de2cae3d0173681
1dfb8501d821a55652fce7513fa41c2b6484be99
25046 F20110209_AAAFBQ knudsen_e_Page_091.QC.jpg
a35d6e7b7d028a4bc6d30c208fde1e9f
7ac0a5a2b0dbee40cb21efe407986124c34f1850
16972 F20110209_AAAEWK knudsen_e_Page_011.QC.jpg
4b5e0ae90c9870a91b9a6728cf3994e4
1859844eeb0e121c0a8288c74bcd18df573b77e6
26578 F20110209_AAAEVW knudsen_e_Page_058.QC.jpg
86a49ba0b2e86d1d8769650952ed78d3
d40f92a08b07059adb4c077628e703958e4b7a60
6049 F20110209_AAAFCG knudsen_e_Page_024thm.jpg
1d5bdfcad3819425bdbb0b9022763fb5
38a87e311caac9625e4ac69f32ef08d291d4f872
17879 F20110209_AAAEXA knudsen_e_Page_105.QC.jpg
3833d3e0cd2e2bdb6bcca0c467119840
765c37f2a1e5a3933a83290642528256e0e988f1
20970 F20110209_AAAFBR knudsen_e_Page_031.QC.jpg
7b614f466415d4f6e6b4463c8ca1c025
65a1d9dde6183ad288cb7c84218735c632b5f5b8
30178 F20110209_AAAEWL knudsen_e_Page_108.QC.jpg
59e6367a31d5d6087b763f9423dd109a
7cb3e8320cd06688d2afc3d576687710210bb9e0
23471 F20110209_AAAEVX knudsen_e_Page_062.QC.jpg
81de5d05a0bbec5c127317cfe8d75456
7d45c24251c668cbef389def908ad84e979f477a
8232 F20110209_AAAFCH knudsen_e_Page_104thm.jpg
668ef0e41e74802e2e223b2eb438ed09
e2536a1f432867f5a4f1e6eb0af5028209656bbb
31497 F20110209_AAAEXB knudsen_e_Page_089.QC.jpg
e9df79bfd1216bbd4a36432caf933cdd
f062aca17d7a55f5c0d89a87b8082ead853a603b
29074 F20110209_AAAFBS knudsen_e_Page_046.QC.jpg
3e11ad268654ec07649390bb79c74dd2
db2dd27ca41d4f24fde4a8c2fe2e7b2589d89629
8375 F20110209_AAAEWM knudsen_e_Page_073thm.jpg
68dfed5f6f3145f7e8a4203650718a03
65136e4dcffd66ac10703faf635cae72c9014014
3447 F20110209_AAAEVY knudsen_e_Page_007thm.jpg
cabc11c30c23cea2d9d522318246ce49
c650c2b77b3d8ffd956f69fa781e9f980d48d2d4
6915 F20110209_AAAFCI knudsen_e_Page_026thm.jpg
17896b392f6ef9f75d7b091dd6bf0c24
23b261061810a4cce096b88ed503bd079e4f7391
6799 F20110209_AAAEXC knudsen_e_Page_016thm.jpg
9bfb130dae832edf83e6e84a39154943
c221a86a8e8f384bb93e39309a880c2ea3e5c2e4
29760 F20110209_AAAFBT knudsen_e_Page_057.QC.jpg
ace352f7b436cbde68f489f9203d0b55
be592d880b27c004541d528e52aabef599de9670
21219 F20110209_AAAEWN knudsen_e_Page_113.QC.jpg
58f019bb28048b72297f21cab95a42ba
d90b1c89221c40ca79a3c9105c87c948ff48c463
8378 F20110209_AAAEVZ knudsen_e_Page_098thm.jpg
02a6f0f7cda05e23b72c6c6182f3ac0b
4e54d28167881c43db091da0d5e2819bfa17e922
32884 F20110209_AAAFCJ knudsen_e_Page_118.QC.jpg
83d07f3e95053a59223bb7dbb881d3ea
f63faf33c5f0cbe260f9ca8a6496298effc4236f
4469 F20110209_AAAFBU knudsen_e_Page_011thm.jpg
53dcf52e7f3202f2042c1fab4b8f3970
26d486c413f21720cf6300ebed167c2a70f71445
5518 F20110209_AAAEWO knudsen_e_Page_060thm.jpg
0fd5aa3137fda77c84f9c53ec38634ba
257cc6b269778db6d6dab21a1cc87698151d65ed
27614 F20110209_AAAFCK knudsen_e_Page_085.QC.jpg
3155c85cd307061d39449e2ad2c2d689
ac1048a680e15a455bb0de928400ebbe2c75184a
10357 F20110209_AAAEXD knudsen_e_Page_082.QC.jpg
73c99d9c3d1e27bcff8dc2cd419c715d
2c8676275db7a0694401be59774b884782191c41
7799 F20110209_AAAFBV knudsen_e_Page_089thm.jpg
11fe6cb38ec048b5e39559dde0268724
19bbdf465e0a11227ce0fbae418b8342a768de88
5184 F20110209_AAAEWP knudsen_e_Page_120thm.jpg
6983ece66d4621dbf98ac774a5439140
57778f1fcc4c3e7f76245fb653990de9dc72e0a5
7587 F20110209_AAAFCL knudsen_e_Page_102thm.jpg
9afa641e8ca29627a29631146256421b
83f37937faf69cd85adfdd51c987dc1e98d3f58b
19866 F20110209_AAAEXE knudsen_e_Page_006.QC.jpg
645e58559b50114ab9311b52a9efb4a6
8679e3f6fd5e43b2e0501069ccd64ec3896b1b6b
F20110209_AAAFBW knudsen_e_Page_108thm.jpg
bfa36ef4495abf11d28a315303f277b0
989c96f0a08cc7f10b536c843513fa3daaffe35f
6835 F20110209_AAAEWQ knudsen_e_Page_101thm.jpg
11af05b2942e55887d21d2978a591410
0a9b5229626c4bca19ae2d03bb224b016901e24c
15954 F20110209_AAAFDA knudsen_e_Page_087.QC.jpg
3d76e24da35d1de851e87ca1894076c7
998462c6e2100fdc41c55bf1cf938bc5c87aacb6
25389 F20110209_AAAFCM knudsen_e_Page_044.QC.jpg
2c0d7cc38996cf13b914a89ed3a9d615
969a3d0df22926b855810db249743ea186288e41
29991 F20110209_AAAEXF knudsen_e_Page_102.QC.jpg
aaad570d63f9bf61e80c1be27a8f2ad3
9cea49fb0d925c6579a49cbe96893ba219ae152f
8044 F20110209_AAAFBX knudsen_e_Page_074thm.jpg
c592f58273c978171b3f8f5b1989b969
69b9f7a3623ddf7350ff9175020d3fa7be17c650
6325 F20110209_AAAEWR knudsen_e_Page_052thm.jpg
87354a501a0a0d19f01686b0a4880e6b
e4168422373294a1916df4a9f7264d719bf09b95
196518 F20110209_AAAFDB UFE0015560_00001.xml FULL
33a820f07ca6b9ddf5c8276fb2ef001b
05d404bd06b8c2a31765f69eb73517885d120486
6560 F20110209_AAAFBY knudsen_e_Page_114thm.jpg
dc49119fada265edc34326ac3c05d94e
d072b09bb0e894d3b3924eb9fc5ac6fae1f860c4
18780 F20110209_AAAEXG knudsen_e_Page_095.QC.jpg
32a3a970b788d284ae480251faf930af
f3a6931fc366e1e87563620b568604ca7db6dace
32747 F20110209_AAAEWS knudsen_e_Page_098.QC.jpg
0bbccb19be1ea597f40ab6cbd2817e82
8262a7fcabbeee0b44911185646374e81f5c4391
8088 F20110209_AAAFDC knudsen_e_Page_001.QC.jpg
f1ac9ca95a9d18e7949e5390a8c2fd49
79de5cc51c938a48e34f92202f7ceb735321b848
7481 F20110209_AAAFCN knudsen_e_Page_004thm.jpg
6f646c3bd52bedd2edadff50ea26eb70
9a8db7202c25b47382070fba1cefa90192c590f4
15568 F20110209_AAAFBZ knudsen_e_Page_088.QC.jpg
cf8419029abd143394986b9061c8d31d
110d7ee91410d865b28d360ec66c3361f9e53efe
26325 F20110209_AAAEXH knudsen_e_Page_016.QC.jpg
2cb47aadf0c4fa993de6dea2cac66c86
144c9deb947ba58d2bd1805408f7f431481b6e49
6971 F20110209_AAAEWT knudsen_e_Page_014thm.jpg
0e5ae7807247ed948dd451000520dd01
24cf71dde4bcc6933239f69d5cc3dc6592e494f5
1178 F20110209_AAAEAA knudsen_e_Page_092.txt
cf344aac2bd236bdf8370d11fd434eda
ab819251c01841354e2a7de180633b36156824d8
10287 F20110209_AAAFDD knudsen_e_Page_008.QC.jpg
ec3b5b5034511f5083f1f028b8e96fe1
c48b766a185aa3e4a3925a98c4be3d1ac54ea5e5
4902 F20110209_AAAFCO knudsen_e_Page_061thm.jpg
a5258202e3339d7331cde756f13b5da0
bf996d2a0196d58d5d8b0a3833cae1e1c38cc4d9
30587 F20110209_AAAEXI knudsen_e_Page_027.QC.jpg
99e3c9544d993f8eef90de7f6e67f246
a449e2fa7afcb436ba4ff3be02bbe6552ef63492
5627 F20110209_AAAEWU knudsen_e_Page_113thm.jpg
e8df8fcaa6093e9e61c82b187292d757
02b33d765cd7b33b3bc7ddad91b2531cbf244d2a
1848 F20110209_AAAEAB knudsen_e_Page_043.txt
6e48c864cb90660aa67867799f1f6e3e
03d191bc6455cdaf657d6e4d47abcd0906ee2b03
25074 F20110209_AAAFDE knudsen_e_Page_012.QC.jpg
09916d94fe8d04240280e022f660c48b
6e4ad28236693fdbbb0589d2ae837876a730ca9c
5150 F20110209_AAAFCP knudsen_e_Page_009thm.jpg
33efc66fdc39b2794b6c2391c463ad49
4b810d77f30f743346a27560c9e2fcdd80b14563
29373 F20110209_AAAEXJ knudsen_e_Page_035.QC.jpg
c10563d98acbeb1ed5b84d5302e48588
c35f2167fe67aed4187b5a6ca6ce742d1c345830
34506 F20110209_AAAEWV knudsen_e_Page_117.QC.jpg
f95fc642a2654eaea4dd707cd2dd27ea
b248f84fb1fa6c4ab5e68ed584ed8ecf7779809f
18951 F20110209_AAAEAC knudsen_e_Page_111.QC.jpg
e9f142f71552833e88730b8a4fd49387
4e00d8f2d54ac89df9600e33fdcb01d9bd64f953
30469 F20110209_AAAFDF knudsen_e_Page_017.QC.jpg
6006f3bc9642fd29bdd5ff83b48a658e
396c6cb90922754f82c3d32772897eeda5d8640d
31189 F20110209_AAAFCQ knudsen_e_Page_097.QC.jpg
b121efbb02d3363d50b5b8507c772824
07c14f57c94f6881ee3a0d67002bc70fb17a939e
5785 F20110209_AAAEXK knudsen_e_Page_025thm.jpg
ebc5db7b5f275e67a99918348d20ebbf
a15dfa654289f0c1fd0bb10f4a6bd03cedcc748c
35468 F20110209_AAAEWW knudsen_e_Page_119.QC.jpg
65c53c2f1de780e7758dd7b92b6c5634
d8685eb64e0531ebb222b2e2968d35f03140812f
149757 F20110209_AAAEAD knudsen_e_Page_109.jpg
858d4d01d60e9e4f5e4afc77cf21cf27
d0b939110520984dd9bc048783572bf52969d1a0
26692 F20110209_AAAFDG knudsen_e_Page_021.QC.jpg
b92e7f3ffdc1dc3ae7bee5201e8d96dc
1801c2366ae5c327b6e175f605c0ec850f6a1454
31820 F20110209_AAAFCR knudsen_e_Page_051.QC.jpg
4cce309e0f00755900a4d5f4a4120ad6
980ba69199c8912b3f433819c0fc4f4c5a9eacc1
20493 F20110209_AAAEXL knudsen_e_Page_068.QC.jpg
4543bdb8dd80a68edf804f256044e85c
0a3366a40e31529384fc72d1eb7b3e4ce3c79aa9
5873 F20110209_AAAEWX knudsen_e_Page_031thm.jpg
7da2ef2daa9ed936367730f0def5aff3
f8ee7d4af309fcec6bc8247fa4e543530dc08475
28226 F20110209_AAAEAE knudsen_e_Page_064.pro
02497b1d11c57d8d670cbc3e9f13d549
d55dbf865d3237c9f27eab20d9bd69011e438a0c
6952 F20110209_AAAEYA knudsen_e_Page_036thm.jpg
3b0fcf1746f91bd366af920945fad0da
72a915e0dd993eebbb1bf5b75ec2412f00683862
23342 F20110209_AAAFDH knudsen_e_Page_023.QC.jpg
65fe69461575cae2c5e50f6baf32f1e9
d70659723ef7d3da4778c1deea96424e0e4dde53
18339 F20110209_AAAFCS knudsen_e_Page_064.QC.jpg
c70a683a4796dd30c5f3342a6b92f69f
9d741c2326400475c7bc7ed05fd894dab50a12be
8172 F20110209_AAAEXM knudsen_e_Page_051thm.jpg
d7f22041642f3241b1a007b0a3a28bf7
98f32b5178b599d104fe0eb0700f2ac3493e772a
30020 F20110209_AAAEWY knudsen_e_Page_067.QC.jpg
18fdfaa04861b592bdb3087805c4e0cc
6563f56643e55b9551389ecabef2c0717bcccd97
31873 F20110209_AAAEAF knudsen_e_Page_041.pro
e1c458bcec9d326ed3bbc390ec3b9158
5a5fde413b08e0b361def7660ada8ba4a78bb9af
4023 F20110209_AAAEYB knudsen_e_Page_005.QC.jpg
7c8f022e30cb8f1706a65196ef2b4183
2d7aab2cfb9d8d4fe19926c00adfa1e042272fb3
14553 F20110209_AAAFDI knudsen_e_Page_028.QC.jpg
a516ea84e5258b06c9c551f516b23a33
1530c8f81bfd6e9e01314b08b0b6797339929e32
25058 F20110209_AAAFCT knudsen_e_Page_100.QC.jpg
39656368a4c7a29b8ee10f6719a08a3d
ecfa4b54ac0b2c01e3e79fdd0d125677e80b5f58
3191 F20110209_AAAEXN knudsen_e_Page_018.QC.jpg
e0b8e9b86ace48cbfeb630c919d5abf9
2a1b66b20db24144e5f86401f0ea191282f38c1a
7173 F20110209_AAAEWZ knudsen_e_Page_081thm.jpg
8e4ddbd38b3eefeacda50b86a88539a6
563352dee4efa36f0d9a3bda08e254caa9a67de4
77127 F20110209_AAAEAG knudsen_e_Page_033.jpg
13c65d554d62941cadaff1413c84b0de
7280fe782e2b278c0d58ef6a4bea7551ea28eabf
13229 F20110209_AAAEYC knudsen_e_Page_121.QC.jpg
795c6db3565cf0d8c8c674854a31779d
820ab66200ce9105bf2a785b46f65047487f6733
33230 F20110209_AAAFDJ knudsen_e_Page_034.QC.jpg
ab971f4616fb8ecf07ed13894155a6a0
394dbc0cac28570e08d2e979e055d596303b64af
3513 F20110209_AAAFCU knudsen_e_Page_121thm.jpg
30244d5dcddbbbd97c584c9981433983
a57b2fab9de38f638c9e4caa229763f3675b0f6f
5137 F20110209_AAAEXO knudsen_e_Page_032thm.jpg
9a1cd12b0e3c4e3fd3e03a2a6529365d
2a44f2ca32111804982c14c889ab173d0cfaab86
720125 F20110209_AAAEAH knudsen_e_Page_053.jp2
8a7603d6fe7fc3330d45db429fdce46a
6b2c5a5641cde29304bd2f122680f5ee6e68278f
7170 F20110209_AAAEYD knudsen_e_Page_059thm.jpg
0493b40ebb0d4aa1422e957ed6283d8e
7c944e537c436ccfe18d354584ca1d9bc7249a6d
F20110209_AAAFDK knudsen_e_Page_045.QC.jpg
31a7a61d72e253b346b7dbb745804707
ab98d0f267d0c9a53726910a590ea8a1d64d8cb0
28081 F20110209_AAAFCV knudsen_e_Page_081.QC.jpg
f0c5a7c727f0d33dcae89c1c711c6220
5d3ae7ef168325929b969e73f48a729538aa6f5e
7055 F20110209_AAAEXP knudsen_e_Page_042thm.jpg
f6115c78a1f54c2ad96668a419c45d45
fa9899bb13e015d7c8d92e98b418288ab30189fb
F20110209_AAAEAI knudsen_e_Page_072.QC.jpg
58cea0f9f616d58879597de6e16bef97
8111010f658e4ce8e5c0f7a951faf1271c5e8ad8
17928 F20110209_AAAFDL knudsen_e_Page_049.QC.jpg
f7f7ab231010f4884240141381352df3
070a0f56647620008f5305e0391c104db290beec
5637 F20110209_AAAFCW knudsen_e_Page_079thm.jpg
257f266b8320ba64b4ae95c02529db32
52b6d64b07139b3fbac6d6fb0d4399c0eeda5f68
6316 F20110209_AAAEXQ knudsen_e_Page_062thm.jpg
f13448f47b4fff181c5e02f1e335750d
87915d513257d8afc6462401a86bf67e8011ea6b
27984 F20110209_AAAEAJ knudsen_e_Page_103.QC.jpg
257945e554b3e7c73d5c748ce030e75d
89cba4f61826b8e48b6f2613e64c3adeea5b6b2e
6224 F20110209_AAAEYE knudsen_e_Page_023thm.jpg
022c1c05a59ff6d93a777b230ffb1185
b4a3ae57faacb76cf56966fa14e812d64775e421
6634 F20110209_AAAFEA knudsen_e_Page_044thm.jpg
ec2af8673d6e6d9472dcd609a1c8d1dd
d203506f4d3d829e8cd0399a68e141faa1a6fa6a
19809 F20110209_AAAFDM knudsen_e_Page_060.QC.jpg
98d84876bd2dc82fa0f97b568f17a8e6
6b6f7f7bc5b525f6180326928bd55bdb2c22cf88
4422 F20110209_AAAFCX knudsen_e_Page_087thm.jpg
e0f0ff4cd4f7ceadf5deb5835b5f5e1d
9ed41c74beff48efec8899465a4851326248068e
14056 F20110209_AAAEXR knudsen_e_Page_056.QC.jpg
d64e2498d378404558a8aebf67de4fcf
b775bf6709b6c902eb0de4d04a0fb505d655e54b
81025 F20110209_AAAEAK knudsen_e_Page_043.jpg
f15cd130dfaa4ce2d1a1f7dc4a3ca7c7
e19bcaffbd8dd1c40afa87abb45eedd430b0755f
5057 F20110209_AAAEYF knudsen_e_Page_037thm.jpg
856c7e44c4cb9d826839f28e4305eff9
87808a26d92c22dc8b9483ea831067f4ed290bac
6608 F20110209_AAAFEB knudsen_e_Page_050thm.jpg
3472da67bc2061bf194406cf52911625
a35c22329b5dd9e966aec90634ba37d70e8bb32d
21046 F20110209_AAAFDN knudsen_e_Page_070.QC.jpg
00b60056e43762257b0591c37564dbd9
4713900703c42f0b48212308bd5f9ad54bffee96
1142 F20110209_AAAFCY knudsen_e_Page_003thm.jpg
c64ff022dbbc1928964e91e61a7028cd
7832e27d4989b5ed24e0c1a47b7d07bbb54bc129
23880 F20110209_AAAEXS knudsen_e_Page_033.QC.jpg
f0a2b3b76aa6a86941d18b0f37cff7e0
232d7f89cb166f926b636180b1834ad4ed8f86ee
7604 F20110209_AAAEAL knudsen_e_Page_116thm.jpg
1877d7da86391c20ccbe2e31bd73fc31
a3c113609f95bc684efff86f590e0b925ff51100
19295 F20110209_AAAEYG knudsen_e_Page_040.QC.jpg
e527d9ffe46dd5c748191adad3884b2e
a1c921c2a492c92e6af7aed9d480b81dc6182655
6802 F20110209_AAAFEC knudsen_e_Page_053thm.jpg
89043927a45b4a1c3d9120431e3556c7
ed5abde892de5d94b943dc331502e99df5fae8d1
25315 F20110209_AAAFCZ knudsen_e_Page_043.QC.jpg
f867e972be747bcee31cc868e7dccea1
778aab2ff3126c53f525f8e403c109db572bfe27
6950 F20110209_AAAEXT knudsen_e_Page_103thm.jpg
819619ef2fabf7e8fa911656e5adcaf0
d9183f2c909ffcbc59b818424653e2ee5b250ccb
8423998 F20110209_AAAEBA knudsen_e_Page_002.tif
c51fb4e092a5abd871f7119cd409b603
61efb61fbc6e0027fc98df8b567c076ee7114baf
23210 F20110209_AAAEYH knudsen_e_Page_101.QC.jpg
ccf65cdb55bc7c2c775a25893b737f4e
75005d621381999131d1ac0a410ffe72018d81fb
4376 F20110209_AAAFED knudsen_e_Page_056thm.jpg
dbe903da38d9c79e392638de29972a95
cf3933deb85543443f87606ef8944a912f832c4a
30613 F20110209_AAAFDO knudsen_e_Page_077.QC.jpg
8d069b6e048fab983f76293fc5bb57b0
1e974dbc4fccaa7b76d31e8bf63f428e5bbac862
5384 F20110209_AAAEXU knudsen_e_Page_040thm.jpg
5cd7b5910e9f299e2995250e4348c8a0
86f00ddb3d900ab786f436bcdc6dc5decea0fba4
F20110209_AAAEBB knudsen_e_Page_003.tif
eb4a2934d0651a293046b888f5093f0b
67bf1cf1aaf657a496a6800bc568a34eb1ec5fcf
1993 F20110209_AAAEAM knudsen_e_Page_096.txt
1851342c147c6bb18f73ebc42275a09b
f1fb6c5935edb43cbb3cf649d4c146ce726bf285
28708 F20110209_AAAEYI knudsen_e_Page_042.QC.jpg
a9293592e57235445ac930498c36f7b7
fd6dc474f26a607b61fff11cf4a46f1418e310d8
6410 F20110209_AAAFEE knudsen_e_Page_066thm.jpg
fa081c40fa1b82c198f61dd589f39cba
13d1c10bb26fcef389383938d3487d42a4a17c32
17205 F20110209_AAAFDP knudsen_e_Page_078.QC.jpg
890fad84332c23a066f8a67d8c54621c
6483e1035837489833613bbd9b4cd93312c4465e
F20110209_AAAEBC knudsen_e_Page_004.tif
63a0f64d7a627293e5181d402edcbb27
42d49204ff54c485bbf19a799a0d56ad635a7568
F20110209_AAAEAN knudsen_e_Page_056.tif
0d99f7814c04418c3e63c0cb297d2f2a
ce2eb9fe982760d838f844f442fb7ad73b740e6c
28441 F20110209_AAAEYJ knudsen_e_Page_019.QC.jpg
3070ef2f02bbc09c5dc8379925dcb36b
8660b41d7fc16681932aa24534d2d3c2c6f8bbd1
5266 F20110209_AAAEXV knudsen_e_Page_075thm.jpg
e062cb6e167577106f154f897c112580
e3ac5fe151312ea08062ecdaa1a66b1abba462c3
6677 F20110209_AAAFEF knudsen_e_Page_072thm.jpg
c85428e3fbc8f238610b69ccda1ad883
982c33b00cbbd707cb3bb23473f55f7e43145516
15002 F20110209_AAAFDQ knudsen_e_Page_083.QC.jpg
4b63ef2ca8cde624a66b33bd02a0c629
f416ec65b857595ee64e43f4a325a220ecdf0652
F20110209_AAAEBD knudsen_e_Page_005.tif
5c0c002ebaafacbd0820a6f1acf2ce1c
c1f6c4ab1be6a28599333ad61b5685a2e3757f55
4932 F20110209_AAAEAO knudsen_e_Page_095thm.jpg
20c247af2a55d5a33634f696d4de814d
f07d674647b0aa0651ecbdb804b67e85639ff66a
7286 F20110209_AAAEYK knudsen_e_Page_057thm.jpg
1ee729873c26b09bf4d3637e93f113ac
7bde26d2c46324e879c44877cc55253e16385c80
20527 F20110209_AAAEXW knudsen_e_Page_047.QC.jpg
8b25d838d27871b2332109ef6aaf8fd9
1bc521b7b5bba475202b20fc1c9776743375643a
5196 F20110209_AAAFEG knudsen_e_Page_078thm.jpg
d29af0740aa61dd2b0fb7e4c0d74448d
1255d03aa80e57cd133b5de90e88c95de65e787f
17779 F20110209_AAAFDR knudsen_e_Page_084.QC.jpg
e4239f4d95e9213925393c144ee2e013
2907682cf325cad0620381e4c6e9f5084d9e1535
F20110209_AAAEBE knudsen_e_Page_006.tif
4a9fce350a10c493ed21166026220516
0722693d3fd345e07f230e3fb4d2429d98f31826
1849 F20110209_AAAEAP knudsen_e_Page_099.txt
d97b7b167a099049f3c7edb5955ea90d
f955969803dda53a967243feee9adf0c1bfba6ad
6462 F20110209_AAAEZA knudsen_e_Page_086thm.jpg
a3c6ab8250717a2ffc91ceb3473245a8
6096d64cc25192b324a3eb92def63916296eaf9d
7879 F20110209_AAAEYL knudsen_e_Page_106thm.jpg
45e2e724b44120fe7134b1e58ec33509
c551ba9e41cea09ca11fb35319b022f63e85e91f
871 F20110209_AAAEXX knudsen_e_Page_018thm.jpg
28401e914cc4caad50d458f60fbfdd1c
e9125e00beada76fa278bfd6362529f7bb409d4d
7194 F20110209_AAAFEH knudsen_e_Page_085thm.jpg
d3edcbddf8576e19ae25a372f7bb0d00
95ced6f7d3acd7bedc2c9006107bc320f524e6be
23714 F20110209_AAAFDS knudsen_e_Page_092.QC.jpg
73d92640b877bac4591dfab4e5bc009c
ba6317a805ec4176a34e4a3c03e58682575694ae
F20110209_AAAEBF knudsen_e_Page_009.tif
5ebe7e5103b31a7a2949196d9a3aa29e
b636ce2d5ac87f7b8d7b3112f9dc355d6b1b3e7d
F20110209_AAAEAQ knudsen_e_Page_072.tif
40a982372a62e09de822724177e743d0
0befc2255e275d953564e854569f793600688cd3
30677 F20110209_AAAEZB knudsen_e_Page_106.QC.jpg
b82ab7f684fdc25de83a2ffadc9ab8d3
aa2b0b0379dc1211e916ec036634463eaa3c4f33
6299 F20110209_AAAEYM knudsen_e_Page_094thm.jpg
68174a71d5e84d4fdfe476b3a6d3fbdd
7202e8ba07f9685eb132251ea6cfdcdadf792ff1
2982 F20110209_AAAEXY knudsen_e_Page_003.QC.jpg
8ca36e4c5b34609670dd44228f24f449
b852c068f02216eecee2b045119ec0f06a9ba4b9
6635 F20110209_AAAFEI knudsen_e_Page_091thm.jpg
472a7009c8f050650d362ee9078d0471
1029458f2eff08b76e4a60d1cfc01c5865c4a864
21902 F20110209_AAAFDT knudsen_e_Page_093.QC.jpg
001a5433c9134310f1aae69d33cec984
6d514c98948b543262775fbd327d4622cdb05a17
F20110209_AAAEBG knudsen_e_Page_010.tif
8235d67a8c9a15cc1060dbe47380bc8f
25f94790e2dc0406f82aa312f472437137c9a76d
62555 F20110209_AAAEAR knudsen_e_Page_068.jpg
6bae6a87fbf831dd2e63d8202d659c10
e2e35e0c1e23c09d49c62d930ceecc5193e6aa76
7761 F20110209_AAAEZC knudsen_e_Page_015thm.jpg
6371c5afd9b6bef2a094db96c3f9d78e
7848490a5cfba2b4f37f9188410a2bc564749137
5195 F20110209_AAAEYN knudsen_e_Page_068thm.jpg
b50c0a69a3249dc4f55c25d3aec3a44b
09a1144972927e763378480d546938ddf46e1847
1505 F20110209_AAAEXZ knudsen_e_Page_002.QC.jpg
31e69ec38aa73c7ffa6119c91b1bd501
fae1e7836c5eafc4ca5377b7dcbd8a8939c4f75f
7948 F20110209_AAAFEJ knudsen_e_Page_096thm.jpg
5aa6475b0aefe2b9723e9a768a6e0413
49317cac1acd2b155f363916f1b34fa6138764ce
22734 F20110209_AAAFDU knudsen_e_Page_094.QC.jpg
c21b4eb83d6ca9db81c77805de998082
8f12ffb7c8c19d70aaed5bb9019a67b838d0685e
F20110209_AAAEBH knudsen_e_Page_011.tif
f9ade89058fce960129923315f218c44
5616680731c15e3e349ce4b2415c87c4b2c107d6
29096 F20110209_AAAEAS knudsen_e_Page_004.QC.jpg
d284099ca934e47159702508268d7a2b
b93d9ef384e6922884aebf03dd91a0aac2011c8c
7846 F20110209_AAAEZD knudsen_e_Page_097thm.jpg
728e49d2df4bd1141d744060114dd32e
3f0a1b006b0dbfcdd2e6f2a83235ef76d67a3491
17288 F20110209_AAAEYO knudsen_e_Page_112.QC.jpg
b7997156a4040fe579d0563212e9a795
164ac3834dde47bf72b50985db66c19a2f7151d1
8365 F20110209_AAAFEK knudsen_e_Page_117thm.jpg
008563b837e0ce492d96bf83af74416e
6241f2a42381bc264141681f0058f817cc55f27e
32064 F20110209_AAAFDV knudsen_e_Page_107.QC.jpg
fccc68ca6969178dfe71daa960e00fc2
ab230ee84fa1c66db8efa9b7f533408dec42b046
30130 F20110209_AAAEAT knudsen_e_Page_030.pro
030c5d4c383683ce71578c29c278f23d
10f5aa9ecb9c6ca6b28e1ff41aa3e85dc2fb9713
8391 F20110209_AAAEZE knudsen_e_Page_118thm.jpg
983e67aeff7adf05b5108360c617b1a3
1b9257492aa17cbfd187a40e03b58996905a9842
5005 F20110209_AAAEYP knudsen_e_Page_049thm.jpg
4b18791b7caa96b8b28418f201e2c5af
a8469cd60331c880e1dd27873bea2d06c2a8ccf4
F20110209_AAAEBI knudsen_e_Page_012.tif
c659aa80352dc552c9b468ab7786478a
aa94c9a5be0af1dd4560e120ea4bdc4c911b560f
8850 F20110209_AAAFEL knudsen_e_Page_119thm.jpg
b7ba64ccc37f22ec352de565d3572c34
1d888e2a62ca2bc7fa843283a737de286d8c09e8
2731 F20110209_AAAFDW knudsen_e_Page_008thm.jpg
cdb8caade575c8659864bc5dd8a18881
ecfc2093ec165f58863f6c427a4cb5d3db91c123
49904 F20110209_AAAEAU knudsen_e_Page_098.pro
90a920bfb7d80215585f9ec19150142e
91904da13677f44fee4cdade5a11f0763aa9ddc3
23580 F20110209_AAAEYQ knudsen_e_Page_052.QC.jpg
27080fa2f4aff28e8fbaf1d307a2122c
3b1da94d249fbd46aa34035a400ffc8b330b268b
F20110209_AAAEBJ knudsen_e_Page_013.tif
f284ab4327705d4f48f9cb4f1fb2da0e
e958f40f8fe14dab60454e0b5471a4cdc0ccad20
6548 F20110209_AAAFDX knudsen_e_Page_022thm.jpg
23c1d115b79232e99cab7420584fcf4d
3fa4ca77995025dcb95cc8c37a65806b05f01add
F20110209_AAAEAV knudsen_e_Page_065.tif
b86182f2706a7112926b40de9f623a6a
0a76843a38339872becd154fae291d6ccf6664bb
5019 F20110209_AAAEZF knudsen_e_Page_105thm.jpg
5f1fc5f3619f1e615ec7ad429319e269
c0392a713748e87a32304a6cc8225871e62ad9cd
3395 F20110209_AAAEYR knudsen_e_Page_082thm.jpg
f02566f65f7c21edfa1e88b0f8038d5c
ceee64b7129739392005ca2fd3fe3b7381c0407a
F20110209_AAAEBK knudsen_e_Page_014.tif
7534c3ceca5f110c52e46f27035124c8
f4102691cee32409921e6e6e7e19336773349e71
6205 F20110209_AAAFDY knudsen_e_Page_038thm.jpg
2645fc0d5076e79ef957bff3956c8879
9a09093affa872c3d906c7c5adb397ae2179d72d
F20110209_AAAEAW knudsen_e_Page_088thm.jpg
3686706246868f19811f0d6d4e8b3b3f
de0d73f18a5f8c74846ce9cfc8cab1eddeb6c692
1104 F20110209_AAAEZG knudsen_e_Page_005thm.jpg
b039e64ba6055fb7dec44d8975547fde
1ae8fbf8af13ad8648f3453358e4fc30530b1648
5764 F20110209_AAAEYS knudsen_e_Page_076thm.jpg
9baecf8556e3cad7bd27423f4cbbd56f
3f85e18e02b1c787064f91c574bf6d841b2157f7
F20110209_AAAEBL knudsen_e_Page_015.tif
9d41b5e10bfb1155e966433f76a692d9
6aeaa0f15063ce1089528000cf767ea0390be2ad
7246 F20110209_AAAFDZ knudsen_e_Page_041thm.jpg
fd5781a79ffad2cf4b5d6b537e3a50ec
25951690c215849d8efa3dbd0bee075b34a1c3a7
143635 F20110209_AAAEAX UFE0015560_00001.mets
3510d40dfa97cfee0454135af1f02aff
5a929098ec54b54a3de8b8470e518e9feb8cad18
7552 F20110209_AAAEZH knudsen_e_Page_110thm.jpg
3b6fc878bed3d6436b777ec6dd22c72a
b826a9729270e539c6383f0e84064933e975d800
8200 F20110209_AAAEYT knudsen_e_Page_107thm.jpg
480e7ad2a6cb12dd02cb3904250a62be
cb6ef8eb488e46acb00b7fbe2fa7275a5e7324ec
F20110209_AAAECA knudsen_e_Page_030.tif
74257859c6a85c1e0b3f081e1d6f4969
266441a64d94fd94367396d551ce0ff4c6efdc77
F20110209_AAAEBM knudsen_e_Page_016.tif
5ab03350c046330c719f396249d2132a
4352984a06dc84372fc4c094d7e0b3e1e16126bd
4824 F20110209_AAAEZI knudsen_e_Page_006thm.jpg
cc2396a746ba1b3290a93198965ea47c
894cad8071705eb5107a5843745296ee7426921b
19425 F20110209_AAAEYU knudsen_e_Page_032.QC.jpg
0d4026694c78bf9d1b6ba8add9c98f41
b31a981ad0af1f9bd892de2440978952344f8037
F20110209_AAAECB knudsen_e_Page_031.tif
a108578a36bf316f7ea1d1fee0d2a51c
dc5f53e79bc3e20f82a52a356bf070579d543bba
23225 F20110209_AAAEZJ knudsen_e_Page_013.QC.jpg
8c924951c1338ede338bdac2e01e8f9d
c48e439f63a1a506d6a1d21c9c9fed26b102b952
20490 F20110209_AAAEYV knudsen_e_Page_009.QC.jpg
f83f020b06af15490dbb0b17cea9e8d4
223b695a44c326500c1a09f60208d8c86a09ccbf
F20110209_AAAECC knudsen_e_Page_032.tif
e84b334e4e447637ae81b6bf70f28ada
ac8d1878c3ef62173810e38dbbf85d0a84fe918c
F20110209_AAAEBN knudsen_e_Page_017.tif
dbd45c2696f55e06ce53b65f0b793730
b44c5986d97805dee9a4b6dba4aef0420fd16e5c
6311 F20110209_AAAEZK knudsen_e_Page_033thm.jpg
fdc405ef91e8eb8ce81f9e67bfdea8b0
63276ca2943c6934ac01d22700e9fbfde514f6cc
15526 F20110209_AAAEYW knudsen_e_Page_061.QC.jpg
d08d80565ae68afa48c5640798755b25
29ae0a9baa8d1213dac7dd562ec9c68a17140bf3
F20110209_AAAECD knudsen_e_Page_033.tif
58f7d58bd1babc9e386e5baab721343e
adb6918f035fcc42cccaf610481327b5285c1fce
F20110209_AAAEBO knudsen_e_Page_018.tif
b5de181a29aa86e9d0fe580497a63ab0
9251afa67c1f7db9b75a834e763c9cb51c38e29a
21823 F20110209_AAAEZL knudsen_e_Page_071.QC.jpg
86598c5fd99f5e055c0c5ffc3acab72f
5a5264b18b4cacbcd220dcdb1518210fb5e357aa
18603 F20110209_AAAEYX knudsen_e_Page_037.QC.jpg
40d2edd186626e65653d003c16372e0e
d86b18892cc982c15fa1a2d7fd50346682a18202
F20110209_AAAECE knudsen_e_Page_034.tif
40f096ce3b6089ec89bb10a19c4bc173
17888340f9dff2d2970a5c5b14e1157152806aa5
F20110209_AAAEBP knudsen_e_Page_019.tif
2df60b60470b863b2b68ab8b958b909d
bdaed35972f3890168df3ff0a91f41d64d2870d0
8023 F20110209_AAAEZM knudsen_e_Page_034thm.jpg
1e1da567545cef4233fad32d0bc87ba4
dc730dddb28eb8375c8ba93742f2410d9ff8341c
22420 F20110209_AAAEYY knudsen_e_Page_038.QC.jpg
8289387b1cbe6dc26e9d04beca77edc3
b37af9fabff0b011bf557431a15a7cfdacc89b68
F20110209_AAAECF knudsen_e_Page_035.tif
d7b9f14a75f5fe6febc806aefbae200c
33e8f9abb5ee5c5b881176102444f0eb47e0ede3
F20110209_AAAEBQ knudsen_e_Page_020.tif
8697fe3d04b774d447ab886882d19795
7bd99d9823b2e69e65e854e104fd318be3f75dcc
26472 F20110209_AAAEZN knudsen_e_Page_069.QC.jpg
49e3b34a9c6133c63a728f71c10f9b5d
ebeda420175c51ae8928761e743dba3cd376262c
6664 F20110209_AAAEYZ knudsen_e_Page_012thm.jpg
b0edadeb5a72fffe34b87ba7ee26610f
09cb9ccfd0f1c295aac8c66223979e4756799c44
F20110209_AAAECG knudsen_e_Page_036.tif
194081d1cba42aefb30617afcd89b1d7
3b7e623c651268e08f9a2e42021a25ec4ff91ddf
F20110209_AAAEBR knudsen_e_Page_021.tif
aae9bdcb23fdcefa27ab9eacedf25c7e
cf5f063cba6e4e30b248e66b43fd2d7173208e5e
28253 F20110209_AAAEZO knudsen_e_Page_015.QC.jpg
e4c144d9a381dcb6906ab17c8a1d4e8b
b1351b3aa3caabf54f6439b348e3eb953c66c3e0
F20110209_AAAECH knudsen_e_Page_037.tif
15922c30393cb332b5e9332802b3741a
71a13a145024d09ca1d5dfbfc3eb89787531f3fb
F20110209_AAAEBS knudsen_e_Page_022.tif
8236009e62119370db7e62f0157eec4c
5f0c9bd7bed053d5d56bfc4b66edc2a3945f4541
24830 F20110209_AAAEZP knudsen_e_Page_050.QC.jpg
0af8b26fa45185353fe6b853d5d1a27e
cd3aff5e8cad6907e35920bbf13bcf25bfd1e452
F20110209_AAAECI knudsen_e_Page_038.tif
f6674ff38215890d3899605ea85da9d7
2b0c5b89ef8f4cc8445927c09c35c4b48f2de2d4
F20110209_AAAEBT knudsen_e_Page_023.tif
af0ac4f82d0b147792a0fb0c24559ea1
dffdd936791ef72eaf32a5c81adcf0e63e6ee6f6
31385 F20110209_AAAEZQ knudsen_e_Page_096.QC.jpg
6a13fd69f72ba1ccc57e51261960b426
3e9669e69aa124d9a69044e10caf69efba557ec7
F20110209_AAAECJ knudsen_e_Page_040.tif
a84f72dd725afb8b3fabde8cae4d68f5
b18214dcddcdb44c181a98d5371e55ad3b1e6db9
F20110209_AAAEBU knudsen_e_Page_024.tif
eb8b4e683b0299fe13e5f55ddf7ac623
5f093c790ba83ee6357d4dbc071693d52bcdeced
8089 F20110209_AAAEZR knudsen_e_Page_090thm.jpg
cf05a10797e2675f37c7adc5c2febb26
ab88fff16d6391d2989f5ab8c22ceb88010c2d5b
F20110209_AAAECK knudsen_e_Page_041.tif
091bf0128f56e3b65a5dabc470a377bf
da405c9b5ff9815f3fbdbd1d8675a9dad92243a6
F20110209_AAAEBV knudsen_e_Page_025.tif
0b28210bef5e1a6afd398f50ed17e76d
e35e7b62138b5e18cf475bcb3ebfe1b40c36928b
5006 F20110209_AAAEZS knudsen_e_Page_111thm.jpg
cc8c81e3bf1b99a755423972cc88a9d6
7b66ab89dc0c29640f044db63ead0920bda7479b
F20110209_AAAECL knudsen_e_Page_042.tif
defd1d7310abd3ebe55a79e797acf891
ef8f90b61512b8d7bac62646bc87134bd82e485f
F20110209_AAAEBW knudsen_e_Page_026.tif
fb4b9469d2c934a7f7b52c9f9fc1c445
e0db993fa7f0539476df31810fab81fdc1f7c08a
5716 F20110209_AAAEZT knudsen_e_Page_039thm.jpg
23a67c1a2cc0b85a5187c1c7dfdda8ea
d90c0b8585118b2756dfc969968889176434f94d
F20110209_AAAEDA knudsen_e_Page_059.tif
26730502563e8b516990048c9ca25ef4
eb8167d841bf40c8ca230831e9d449a8e53f66f7
F20110209_AAAECM knudsen_e_Page_043.tif
ef9d2e3eeb2a7ca2480764c8ac9ca63d
ba9522c9fd0a4fe0e7bf0c5999d59d34812ac838
F20110209_AAAEBX knudsen_e_Page_027.tif
036a25d92f1ab5c628c861234449bce2
f752f66ba2fa083cf3cce586dcc8f89a6626bcdc
5203 F20110209_AAAEZU knudsen_e_Page_084thm.jpg
1687c442b8fc370deef9d2ba1bd611d8
13e2faea21fc46df0f7fdb8a88fdc074751ab49b
F20110209_AAAEDB knudsen_e_Page_060.tif
5c988c22f733a94763fca948551d2ee6
99a70688c2c549cc180169d3c5e8b97b9a04d7ba
F20110209_AAAECN knudsen_e_Page_044.tif
a7fc57fd85f51dc9bebf0454efea5c48
89e40679f5b1649ffea5c525199befd4e63d7526
F20110209_AAAEBY knudsen_e_Page_028.tif
3fc259fe4c3a06d01c7c63c82ffa902c
a898512f0f3401215fc64510313f62038068f822
33447 F20110209_AAAEZV knudsen_e_Page_104.QC.jpg
0ed3deb56efc75533ad19fe9d5eee070
2d18bb7e2ab31301585daff65af5ef69c1637c89
F20110209_AAAEDC knudsen_e_Page_061.tif
c35ac3879bfbb08b462b40e25f0e96a1
205c4fb245c0ce5c97b8814333c44b88ecdfe91f
F20110209_AAAEBZ knudsen_e_Page_029.tif
01c04aa3a939ef95c42aeb2feb824f38
e991b5767eaba90a95f9d116827352934adbbdd6
7945 F20110209_AAAEZW knudsen_e_Page_027thm.jpg
394b9755658551efec0618a1bf38c997
e6751e518425f9e8dec5eef921f49cb160c7db4e
F20110209_AAAEDD knudsen_e_Page_062.tif
6bf4a7ed62fd877204246bec72e6f0ec
ad1070fd2b5c6ad4324686e7c8b1014c63bf2305
F20110209_AAAECO knudsen_e_Page_045.tif
dc6f35fcc8888314e8a350c254a488f0
34926f136c8d769ce935104d5456cdfb8f27b654
6814 F20110209_AAAEZX knudsen_e_Page_058thm.jpg
8199bf79f24f8f2bb7c04b3fbf49bee2
8535492a7d17e87f305253b22f94a3f4559297f7
F20110209_AAAEDE knudsen_e_Page_063.tif
16bb562222d3ec393dc20705dc56ecb6
1318be52248bc6a47698eaed73de48a99190bb02
F20110209_AAAECP knudsen_e_Page_046.tif
ac361c9b88ad575098e742359f150305
082bad58111d88cca93b410d0d84f5d16d2eb638
5803 F20110209_AAAEZY knudsen_e_Page_013thm.jpg
0bb7edf01c00aa528f3abcb5d821806a
27cab6b67a0de8b3bc98b56d424d4659cbb6cf9d
F20110209_AAAEDF knudsen_e_Page_064.tif
8cd93eff1270b4d59048efbe6fb9ac9d
2af0724cfede08d6d8961431153d092e7bf2cbb2
F20110209_AAAECQ knudsen_e_Page_047.tif
09e6e2dce9b6e1866de846ab29657f5e
668bc6a56087c45819e8ac668521145215e6e4d7
5781 F20110209_AAAEZZ knudsen_e_Page_070thm.jpg
5a8f273082f1e464327afb4ae44f4856
791d9db6e2812529fcda940fae0175f72efb5428
F20110209_AAAEDG knudsen_e_Page_066.tif
19189097c93af2f6fe2960c8b9a21877
1338823c03798bcc88f4e3d80992bd06b4a314b4
27728 F20110209_AAADYA knudsen_e_Page_039.pro
e93d1e283dd267b0129fc93f9862bb5b
d137cf42ea57220fd2349aab87a62d26686d099d
F20110209_AAAECR knudsen_e_Page_048.tif
54c96c3e9966ff4f0d7191c52ca95f5c
c98d7d0a870ce44866312d896426705c8d11c180
F20110209_AAAEDH knudsen_e_Page_067.tif
2ee1bdcdc0f9bde4439f31144d40a797
87a2fdb10cc8606d687b4f9c7107119af83d85c5
52924 F20110209_AAADYB knudsen_e_Page_075.jpg
a0f168c671bd549d4d5481111358fcb5
24abc609e56a2d31d18f86e4783a71bfe267f2eb
F20110209_AAAECS knudsen_e_Page_049.tif
d6ba2515b3c94eff405007c277d54e56
e25216a154b837a3e656c5d3fb3988d2ce730377
F20110209_AAAEDI knudsen_e_Page_068.tif
992e433fc650f894c8efa82486400818
737493f70f1690c37de694ce0df5a6d2e3341e95
33544 F20110209_AAADYC knudsen_e_Page_035.pro
61fbac5a6b661214e28ffa6f569feb03
c9d105b5bee212db08794a5fa50a8763498f1745
F20110209_AAAECT knudsen_e_Page_050.tif
dd566be037b691c0370ba5320b40b8dc
55cd58cca51685f69133c4738cce7491dd0c3a1c
F20110209_AAAEDJ knudsen_e_Page_069.tif
1494eae81672607419e5fdbe9494d8ff
741bd91fd314ed9d90c273dd8e0266234e6de2a0
1384 F20110209_AAADYD knudsen_e_Page_091.txt
cc2979cef013d804fabfa2d9767fe778
7d602408fdc605ec28d08754760ab69235a164a9
F20110209_AAAECU knudsen_e_Page_051.tif
47c6ccceebd572663d92377a8bb2d86b
4dc13a20b5ec4269fa68ff2f3da0e3398d9a1b4f
F20110209_AAAEDK knudsen_e_Page_070.tif
f7bbef1d45bca70f09bf68ef45c62a36
4af594ed371ac7939e1c61c0d04f7dc1ff5cc81c
21228 F20110209_AAADYE knudsen_e_Page_115.jpg
8b6138b14e0d451b01afeaaf1f773100
13277fe465a7702a2e9a915dae7996aafdbc67ef
F20110209_AAAECV knudsen_e_Page_053.tif
132113681e5d06898e4382cf89d3396b
a6766ef3005458c05da161e265ece9b59365dc66
F20110209_AAAEDL knudsen_e_Page_073.tif
bf38667b64708a19252aa57d3ff3b40b
15eab58daa8679dea18f3b74a40ee2fb850eceaa
F20110209_AAAECW knudsen_e_Page_054.tif
7cb00a83341fbe94a13c4ace237a0c32
40bc73a8136f647086f516e95223c342a6a902af
F20110209_AAAEDM knudsen_e_Page_074.tif
3f8d3c13c478f9132d88ba0a4c6771c6
b226c8df7081bca257fb032eec67766e0c8c1001
F20110209_AAADYF knudsen_e_Page_001.tif
569128c041a0c98e78bf70c41e03e195
ebdfe751ffb293166a0cb8b3beb7bad77fe26858
F20110209_AAAECX knudsen_e_Page_055.tif
c2bf160d18905a454dcf1b26748f1f63
9377e2a6d08e45eee03b8706f7dd4d9c8562815a
F20110209_AAAEEA knudsen_e_Page_088.tif
e8627b8fa28188a4fc825a3f8bdcbeb3
496feb13b79e6a5a65a92613ab5b37a260e7b8cb
F20110209_AAAEDN knudsen_e_Page_075.tif
81154e5c0b19ba40a96ef9b82d6684d1
17c3b10efd928f0701bdcf63b0f9a19b40879c3f
624576 F20110209_AAADYG knudsen_e_Page_047.jp2
da417275b01614d436577d984549bd1a
16530cfb7339f1603fa5c7bc53526b2245a92bbc
F20110209_AAAECY knudsen_e_Page_057.tif
e35885f008fdacd2ce1cb59182fe5078
b6e696d499ae8d1db8839c708fca5e456a4bc7c1
F20110209_AAADXS knudsen_e_Page_115.tif
1fc8658fbd36de28383f5c6f7a675a8b
d32cee067274e11e53ec8bc8b39b9369ccb751a6
F20110209_AAAEEB knudsen_e_Page_089.tif
778f7bb7aa871aaa1c3b25632163aee4
976acf907f9e60617e43ee3edf9550dfa891c871
F20110209_AAAEDO knudsen_e_Page_076.tif
a213f36601f8f079bca10d6b11ac4c83
77a86fbc6a5cda3eb44dcdf73d5d8cf142f5d1cc
F20110209_AAADYH knudsen_e_Page_008.tif
894f9983815135116e95de68627f7d49
ba99444cb050c7d3b133aba96477e8cc9b9d7811
F20110209_AAAECZ knudsen_e_Page_058.tif
f46213d3b33644b7d116833ea85f5df3
123ab85347484ba01124f1d8ba41fc7aa7f24c02
2192 F20110209_AAADXT knudsen_e_Page_051.txt
85b25ddfb6690032268988d454b06c6d
79444ba38975916c420aa67ffdf857d1a6d7a5a8
F20110209_AAAEEC knudsen_e_Page_090.tif
78e3cb5c9094e5f553129b33f0cd3a23
42f965f18454fe2405826d402f0f1d843236044f
49791 F20110209_AAADYI knudsen_e_Page_104.pro
8a7cf6ed411c058539da3c6a52dc11cd
46d2ee679eba46fb09d329b0e0bd839824dd430c
25019 F20110209_AAADXU knudsen_e_Page_010.QC.jpg
8b5f8373589ef6171fb70e3814906e1d
98efd9bc23f1be7c9a9aa5d621b3c57c309ea7d7
F20110209_AAAEED knudsen_e_Page_091.tif
342a0455c8baa6297d14d4e85f1d808e
a6454a11469a479f10af44614c96867e28cd34df
F20110209_AAAEDP knudsen_e_Page_077.tif
d944f1f61074b139d643d5b98cae1e99
54ae5c37f649fe9039f6c0cba38e1837a7f8e25b
2323 F20110209_AAADYJ knudsen_e_Page_009.txt
17c2c4c485f8f3f1bd637b6cf073c979
ac6a4116ac89b19979713773ebdc090733ea80bf
76153 F20110209_AAADXV knudsen_e_Page_036.jpg
b80bb533821b5dbfd583767443f721e2
8cba94fdbd20901f7887ec91238b84f1cd39d7ea
F20110209_AAAEEE knudsen_e_Page_092.tif
36c1e64505c01cc36097e0fff6bcecde
4da73430ed0f73964e08b672124d4449438c12f7
F20110209_AAAEDQ knudsen_e_Page_078.tif
89286e9f64c9195aa098a3e34bee5e4a
6b5e939e70dd4b82d65beaf61a437a6fe44ef43d
3813 F20110209_AAADYK knudsen_e_Page_005.pro
52f55c1b83417c6eb8b28b11283e5b2e
7275f6596eba84345fceb8e3e4294fc807f2ea38
6932 F20110209_AAADXW knudsen_e_Page_043thm.jpg
a02aa4cba4b4596319d679ca293b464c
f2df2966e6b32f0fa7f97f37bcd902325e5b7ad4
F20110209_AAAEEF knudsen_e_Page_093.tif
ae536bfc109bec31a3be82ea76d88e8b
636362e09abda4df6265a5c1ec968c926a85fab7
53670 F20110209_AAADZA knudsen_e_Page_049.jpg
4aacfcfa09d2aac96d32590b6439bfb7
08a996ab9a08affaafb18c013fc6a18e9ce4fce8
F20110209_AAAEDR knudsen_e_Page_079.tif
c8387b948fa2c3bae7e8490f98666710
3f58bac2a7f463ac3133e182b3aeb83eba398035
F20110209_AAADYL knudsen_e_Page_007.tif
753bf1b67a506a73e48a3ed27e9c1e74
8910a830fa46f03391edb5fe22edc3277598c7d6
1991 F20110209_AAADXX knudsen_e_Page_098.txt
0e5770de19e2d8d63464ba9065a70f73
ee9baa8538ba1f3a0d4a0de1f9bc7d88915dc156
F20110209_AAAEEG knudsen_e_Page_094.tif
e3fa4bf23ce7b9bdbbb8eb585029002b
79d4fee069a6a470217bbe8c0969b03b4e312dff
339619 F20110209_AAADZB knudsen_e_Page_008.jp2
208777c57ae35ae9e9e7e14f4fb9d830
aad2e6bc704236b6e07540067f89d082f3880a1d
F20110209_AAAEDS knudsen_e_Page_080.tif
641b23f28d57c4f8353ac0299e7103e8
79068b4f75d66b53c66bec98d9080bdc7b98cb13
42769 F20110209_AAADYM knudsen_e_Page_055.pro
92282065fb56fc8a327d43f62061ca3d
3254f7d0e93a9363c8f70957ee53127c3de642a9
232 F20110209_AAADXY knudsen_e_Page_003.txt
bf706d19b42365f4629c54b238fafd44
56567520f10fe367ccdc6dcc0baa4da22bd981a2
F20110209_AAAEEH knudsen_e_Page_095.tif
1e4d438e6f8e18e351e21d38a685e4ef
50268a164007d2cfa07fa8177660ae6350d337d1
99940 F20110209_AAADZC knudsen_e_Page_107.jpg
a83051abd27fba8340b6c2299a031633
4216224ff11f73c413f955d6536bb6b469fff566
F20110209_AAAEDT knudsen_e_Page_081.tif
1329719ebef68260971b339a83176f25
795a6c8b55f43000d4d0687105755a4a431f55b8
905001 F20110209_AAADYN knudsen_e_Page_010.jp2
46e315a882abb85ccf227ec081c28b1a
689cb7c5960990ee91fc10c13698046d6bed8777
F20110209_AAADXZ knudsen_e_Page_052.tif
59ce65e25d9747bc79f93cb58db00efe
d82074b3de611d4c5fab63b904dc895487fe7297
F20110209_AAAEEI knudsen_e_Page_096.tif
48d8944be6ac78f6bd6cddbd606e4570
b7da14bea9c2a9eafc40a20b5aa41ce0f01cf613
5448 F20110209_AAADZD knudsen_e_Page_080.pro
011cde20a768cd039e89a3be040df995
ffa94f552a5e988f0131f46620eef73285b9cb5d
F20110209_AAAEDU knudsen_e_Page_082.tif
1e97c58fddbd9f8d607b968ca8230eec
98b02721bee31bdfa4f27feabced6c9da7ba9125
47674 F20110209_AAADYO knudsen_e_Page_017.pro
52644d59749b61fdbe8ed7468e2dada3
8d6bb7333f5b88d33b86f15e04721a7dd6540505
F20110209_AAAEEJ knudsen_e_Page_097.tif
aa4a814ba95e901027309c32f6cbfa26
028728468ef7a892e1fbc57402752c9210ea7239
8041 F20110209_AAADZE knudsen_e_Page_099thm.jpg
320f44d8276eb3b89b2c4da97a05bb52
1a1251ce1a9718d97bbbf6acc19a6c154a2ff288
F20110209_AAAEDV knudsen_e_Page_083.tif
036aa057fcd18a9ee9fb9f88cbd46600
be4f204ac33e161698e3391ab3b1a395dcfcba54
21516 F20110209_AAADYP knudsen_e_Page_030.QC.jpg
73a75c5157bb7df37a61fdcf3f2d4833
bc4a9defe843b1942f78095b7a9967089689ce02
F20110209_AAAEEK knudsen_e_Page_098.tif
dcb955d8808800ff17e0b7a7dcb3aadd
ba899f83ed0967261d6521e9cb960b678565bae0
72804 F20110209_AAADZF knudsen_e_Page_091.jpg
df5d3ecc17635fcbe47010ff756010c5
2e7b972c8ba0db37b2db688b4bf8312e784e5bbd
F20110209_AAAEDW knudsen_e_Page_084.tif
da614c06e155632c4a897afd9c4096ac
83add11b399b6c0105452da8d98e98a7e01fd0be
34907 F20110209_AAADYQ knudsen_e_Page_024.pro
4352b606c570903117b759bf117355c0
3e667905f4964587d0f0d9ff52d230a8b884f5fb
F20110209_AAAEEL knudsen_e_Page_099.tif
b12cab511a51ff5fe55717a3de519ff2
1780a0881b3764de7f9270eab08482d6692b8751
F20110209_AAAEDX knudsen_e_Page_085.tif
471c771f8d398f941f250561fd182113
6c6a98ebabb84184339e5c0375e776a8fa4442c0
F20110209_AAADYR knudsen_e_Page_106.tif
080589941a8d8eee32b694e7461e7a48
2185933921b284d576fe51aac923a00c0eb94ff9
F20110209_AAAEFA knudsen_e_Page_116.tif
f10d6a38ffb9e9644fac36c93d4e48cd
da5c2a2d5cf33f9525d5e47822a672e17e7484a8
F20110209_AAAEEM knudsen_e_Page_100.tif
4bc0b08a4d2b1f8066482b42e311e72a
b7c39b643ff7d897599a42e6ef3270db7cf48ae4
10109 F20110209_AAADZG knudsen_e_Page_028.pro
572ec7835ed8ef3ac41d22c29d177a85
1d4bc6050f6fccd8ecdd95f14d5fd61bf943df69
F20110209_AAAEDY knudsen_e_Page_086.tif
ee8daff11b705d9ad0d28fed5b1322ec
345f996d59f5bb33bf43b769428b88e032b4fbc6
35244 F20110209_AAADYS knudsen_e_Page_044.pro
e93dd6410c8e62906d1e2785abc8b7d8
1944a9c6693699461fa3f6c7c1f1f879f5004b2f
F20110209_AAAEFB knudsen_e_Page_117.tif
3d6bc29195694364a2e7d23456265b6b
5e886c059a290b7b577c2674adef1b1bc60b6313
F20110209_AAAEEN knudsen_e_Page_101.tif
b85531ee7761bb101c397d342cbe1911
c4b46779e9622b5faaf4e27e06e34af495f36adc
F20110209_AAAEDZ knudsen_e_Page_087.tif
09cdb06cf129a65d556acfb78714e2a7
444b2bfe7ee2d6f6316dd97f412f7e6dd4cda290
25709 F20110209_AAADYT knudsen_e_Page_036.QC.jpg
842dfdeb4d388d6d5039372c002793ec
6ca0d69f3c50ba91e32eb2df072877f1a781e2e1
F20110209_AAAEFC knudsen_e_Page_118.tif
975262231c1eed84f22b42add1994a69
9bd46538e3fc9496f2656af6020730d6629a5c9e
F20110209_AAAEEO knudsen_e_Page_102.tif
c2c6a7ad88ae87f0fe8fb94a0fa61155
a5b52fd0441d9d5e839aa1062a6cb5b88a2a9e50
F20110209_AAADZH knudsen_e_Page_039.tif
1c418944c6ded929f843e913b4f56e9a
cad2d4e45ad29eb08a4cef894294b2db3c717b58
53126 F20110209_AAADYU knudsen_e_Page_084.jpg
90485380d911e475ce687f804ca9133d
1fbe4ad086d5dfed5fec9a1582cf261066ef005f
F20110209_AAAEFD knudsen_e_Page_119.tif
ae483843f4eefa4808003a12e0d0f245
8474b2b72d5d57c390719a12b9cdf81aba1e21fc
F20110209_AAAEEP knudsen_e_Page_103.tif
4ccad874fee725cc2d5d665064840785
cefe9baeb28121b9aa1ffaecc88e7883355fccc2
F20110209_AAADZI knudsen_e_Page_071.tif
380bf4fa7caad7c6a45c5a42d82eb484
5f55660e072756a80a193356ddf0142535c0b62b
50294 F20110209_AAADYV knudsen_e_Page_073.pro
13f581b27568603b3330f4cf05996cb5
ba179b45dcc1e3c32aba43cb45c3a55df971b0c6
F20110209_AAAEFE knudsen_e_Page_120.tif
9b7d77ae2a45ea940de2adef7ccedbb2
a8dd2032fcafabde0e07cfaf3baf2fc24311e231
11525 F20110209_AAADZJ knudsen_e_Page_080.QC.jpg
e459f4802e65b44bc8a8c667a105a063
2bf1903462f16f9a9f3a5bd1fa1eafad0d5a142f
634813 F20110209_AAADYW knudsen_e_Page_030.jp2
814dc6a077fe8c0a3399d7ef1abea4dd
3fddabf969781a49dcc6d2a5ff85e937d0bbe8fa
F20110209_AAAEFF knudsen_e_Page_121.tif
13e69ca22c6b618d0a14e6bdb064b089
59265edc2d51a11f6339e522881a58cc34961762
F20110209_AAAEEQ knudsen_e_Page_104.tif
69e768c6ba4f5003e1aa3b85d9a72937
f0085c86c528f04eae7b13c59d955a92c1b3c372
1850 F20110209_AAADZK knudsen_e_Page_046.txt
a9082c6d489cd15e5bcf8c2a37914e8d
e13faa658eeb35f04529328660a7e1fdffc24cd8
1738 F20110209_AAADYX knudsen_e_Page_042.txt
5379d2336934baf5497794fd96eb5a0f
dad26a9ac0815513f006f10b70c79dc2675e163d
467 F20110209_AAAEFG knudsen_e_Page_001.txt
f2204ae7b9b5f9ab884e6d9e0e70ab8a
ec118083105cf74d5fed30eb4e20731691efe9be
F20110209_AAAEER knudsen_e_Page_105.tif
7ba2812334ff9b24bebe824cb2abecd7
707f3073097a328814fec72bc639b0273404bb21
5843 F20110209_AAADZL knudsen_e_Page_047thm.jpg
fcbf5fa86c5c7529b22533e50f5d1a23
907ea6f3b8e55280199817a14eeda455a4450574
1800 F20110209_AAADYY knudsen_e_Page_077.txt
730f5ceb6425f8f856b0f27ba379ce86
1743d660633d600712c3278517232e9a059b093d
107 F20110209_AAAEFH knudsen_e_Page_002.txt
f880c7f9579ae5e2327bb3eaef753fec
96eef45bddd472b696a590fd0c7b8ec75a132e5e
F20110209_AAAEES knudsen_e_Page_107.tif
43e1c1099b994cb5593c9357fc682946
188dce02c269cc8ffbef16409dc75ccaa8715cbd
717177 F20110209_AAADZM knudsen_e_Page_091.jp2
ff661035dd091bec95525a60f16c5767
5c5c2d1fd11fdd4131d1b64e930c880db61e9936
2652 F20110209_AAADYZ knudsen_e_Page_110.txt
bf46241c3b487ce70674d66521cc8d7c
52311e045deb5e98b067fc25087521273927830b
1757 F20110209_AAAEFI knudsen_e_Page_004.txt
fabafd470f36aff8e64de0d0a5afd65c
6a107c99e923f3e8a7dd84b4c07a1360f836fed5
F20110209_AAAEET knudsen_e_Page_108.tif
8e6483bc85d4dc7ac9e489faf7ec5c7d
29684387353a40119fb6e676e06acced585f7c6c
2894 F20110209_AAADZN knudsen_e_Page_010.txt
0a22fdaa149480052dedcd1ea88a272d
814ed71b5bb280c938c868f7907438ddd1caaa09
155 F20110209_AAAEFJ knudsen_e_Page_005.txt
86f6e29d8639c88fc22338e587006141
e2833d9fb9b9ae61eb9743708597ed2d1e23caff
F20110209_AAAEEU knudsen_e_Page_109.tif
b57a1b0f9cca9c91112fffec88eb692b
c0310a08b13c6fa6cac87223c2cffad91db85f22
447 F20110209_AAADZO knudsen_e_Page_082.txt
d795f90f8e705510d737420cea384f6f
f605fc1d67f78caeb7db89c5139a6c5406516ccb
2736 F20110209_AAAEFK knudsen_e_Page_006.txt
c3a0bb481bd1f63e9ca1cc5694604e30
b9cdbaed056f1204c32f1b9d4281327cc344bf29
F20110209_AAAEEV knudsen_e_Page_110.tif
54a2fabb381d4d72114f2738a69b7c4b
682a2e9f36517b474492edf46cf749f9ac4ca506
5947 F20110209_AAADZP knudsen_e_Page_092thm.jpg
e954ec1bc7465af64aec5067ac1cda86
acf8ddd93102c7a10315dd026cc7099cfae1a200
1613 F20110209_AAAEFL knudsen_e_Page_007.txt
904ba8612c397e253907063f54848cba
4180abcca14edbf6d393a14b9dd308b04318fc7b
F20110209_AAAEEW knudsen_e_Page_111.tif
a3564475d6c387bd3493f3664d2984ad
ea3070daaa3e92a5bf51c26f02bcc0081b8047a6
31960 F20110209_AAADZQ knudsen_e_Page_038.pro
39fbb70106d8c056f2437e54d1a4ba81
339252b5752fcce41c514367c73637403356591b
1567 F20110209_AAAEGA knudsen_e_Page_024.txt
24d9a7a0f0a33ef99dd03213c2a4418c
957ec2d914bbe4e992d8158c61091fe31accb1f0
1141 F20110209_AAAEFM knudsen_e_Page_008.txt
ce25b68a338fbb63ffd3f903340dcb6b
0b16882b5e570aa633569c7c0cd1b0806ff15bbb
F20110209_AAAEEX knudsen_e_Page_112.tif
a22235b83063292dcd906adb25e70437
a16773446d59c1ac49213a6a6969e6f1ea411377
6929 F20110209_AAADZR knudsen_e_Page_055thm.jpg
6801cac663df87cfcd755bf7c4f4e27a
17de1870d15f2aba79784b38d523e94c1c05b4ea
1535 F20110209_AAAEGB knudsen_e_Page_025.txt
2377bca3a8170d497c9cf518d56f84ec
b6935847a88f280d2d68d131e843af16daabc5a0
1644 F20110209_AAAEFN knudsen_e_Page_011.txt
60248b494fc8903011b006da9c4e12c5
8e9c376165e11efb7f8400c0d026bf0e3c9520c3
F20110209_AAAEEY knudsen_e_Page_113.tif
2f440e685635e7c25d85ae657df01bde
fc48e6a9be3936224406ad253e8ab1d17a16a1fc
699106 F20110209_AAADZS knudsen_e_Page_052.jp2
1ab19ba45089aab3599fa6999c568e90
309aaecd905512182268d189c154e73324cc4613
1761 F20110209_AAAEGC knudsen_e_Page_026.txt
e15c9b4e90e9881bc304c3539f818af7
f37d948b42598645a74f8cef39f689d3a23eb490
1667 F20110209_AAAEFO knudsen_e_Page_012.txt
27327d08c38348bdda904da67dd3dcde
fab7a57c3d1e084cc2f2fd61bf28a98bfbe58b8e
F20110209_AAAEEZ knudsen_e_Page_114.tif
d9aac6f9e5414b53f79825ab039f6e71
8aebc3c8b16c25204f6c0b824e70dbef17f7a194
68785 F20110209_AAADZT knudsen_e_Page_045.jpg
042c2a8801e025b852ab4722ea8a9b7b
8415c785af025a7dbe10fc575aa40771c4df95aa
1851 F20110209_AAAEGD knudsen_e_Page_027.txt
5956bedafbecfb49a8d867c03f8a4a20
09fafdb680fb1669ed56333627e320b5f557c2dd
1373 F20110209_AAAEFP knudsen_e_Page_013.txt
4066d486f191bcc9b8de10adbc65aba3
4f67b1afdcec8c3c27ffc5310c7ee629ae49ed96
504617 F20110209_AAADZU knudsen_e_Page_084.jp2
898749cb159a6b7f85b7ecdf29959ed6
f353ad685be618db80c3f568f189f6a9e7823432
436 F20110209_AAAEGE knudsen_e_Page_028.txt
35c0c7a3ad1f929fb7a90292e10275af
4e681ac7e0535d9e1b30fae20a3d3b4540f41b16
1712 F20110209_AAAEFQ knudsen_e_Page_014.txt
e135b86d65416a51ea71651ac4255c22
d43eded343b3e63a874c0e2616f53ad07d58a365
2318 F20110209_AAADZV knudsen_e_Page_001thm.jpg
3cac173889dd3033727da9107e698afd
0e8f6b97657d8379fbd34d3c1b3dc37f286d4be0
794184 F20110209_AAADZW knudsen_e_Page_033.jp2
9eefe26af304159113b5cb2fc9fd139e
d8ee620c65948cd22bfcac25439a5e302bae27e6
1124 F20110209_AAAEGF knudsen_e_Page_029.txt
2608cb1d03683fa149549184931a933b
326d331a650d306aa688c008d42599cb92976c62
1360 F20110209_AAAEFR knudsen_e_Page_015.txt
fa1cbd5bcfc2b8293aab08f4b18cbf7f
16757d7ca3153e91185dd36b0098f1b728a32caf
24202 F20110209_AAADZX knudsen_e_Page_066.QC.jpg
21bdda2068631069b9c430a12fa9e725
a5e948896f676cac682934704be40d283d43b79c
1452 F20110209_AAAEGG knudsen_e_Page_030.txt
59925e160ba6fb768428f2a84a209a6d
6017b6f9b56f36fe5e0713c78e7d3bf981c06952
1794 F20110209_AAAEFS knudsen_e_Page_016.txt
d1e3cdec6b87acf1fe1621940e7683da
35122e2ac52ff51854f37e9dbe98b10e59212fe0
1379 F20110209_AAADZY knudsen_e_Page_033.txt
8c9cf80e353f3c87bd0e1c25546eb523
e0beedfc2d6ce317e1aec3a3433a91420ad5c549
1364 F20110209_AAAEGH knudsen_e_Page_031.txt
b3f182617c4b75e8cf90bceeb3edd19b
e09b4fab30a85292109e9883d54ea11258b78265
1902 F20110209_AAAEFT knudsen_e_Page_017.txt
ec07ee39dae4dfe565ad9536d7e8d396
73f205c4f9912ba585d430b7d8473c3ebacfef8e
1051950 F20110209_AAADZZ knudsen_e_Page_110.jp2
fa4472be1c665e8ba4c26f187e66c2d2
e90db3dad53f4aa8e4af9676ae62189aed4a8953
729 F20110209_AAAEGI knudsen_e_Page_032.txt
0a6620638338b16c5dd48f2e6a720f0d
b71cf4b4968cb2642ad105c1b44b447ae8497b82
177 F20110209_AAAEFU knudsen_e_Page_018.txt
c253f87db337e11d5f1bb80685e8e5d7
aeffae2679a7d1943643edafe12d336c5b743926
1999 F20110209_AAAEGJ knudsen_e_Page_034.txt
b35fdec86123b7495f038cb124ac6485
34c6894ac7bae3438ac2f6ea7290b5c04d84baa9
1722 F20110209_AAAEFV knudsen_e_Page_019.txt
acbabba9d4a2d631a2ea82bafea8e0e0
55d7b837bb2a48fe159bdf96b76abeb57e42701b
1369 F20110209_AAAEGK knudsen_e_Page_035.txt
d6119ec1928264c70aa517370345bc6f
aa596aa0178c39952f86482627e0141db476a5cd
1984 F20110209_AAAEFW knudsen_e_Page_020.txt
069435751cb70638803cb0ac145841b6
f08e15eaa7e7fd87d2ee1115ea782ebe25ba68ed
1454 F20110209_AAAEGL knudsen_e_Page_036.txt
16e3b0e2d3ac926413b7f3c8c662dc5f
3e59e3815b26c42dfc03f34dd54e0f4fd3773b83
1283 F20110209_AAAEFX knudsen_e_Page_021.txt
5dc56c5913ca7cce0b23649089c8f422
c6e95ceaf54924cf8e47a8b2a287403c82029165
1855 F20110209_AAAEHA knudsen_e_Page_055.txt
ee5614ecdb6566c584e0473a1ac6cfc7
72f714e4b088cb6ae5900bd12af2810ea7b5fdff
1063 F20110209_AAAEGM knudsen_e_Page_037.txt
d4a60fbd15689ad9dc275c8126069130
d81766d4f47e23e825cceb7bed2a9bce09e4c927
1650 F20110209_AAAEFY knudsen_e_Page_022.txt
3505e45952ebdf39cc11ccfa4ed1ee66
8f8bd5e66612523dcafa49b6dedd1cef99ef7b5e
849 F20110209_AAAEHB knudsen_e_Page_056.txt
c071dabc9f9539b99131e799c8d44acd
d583c7e129a65460bdb40b3a7dae977ef0aede6a
1350 F20110209_AAAEGN knudsen_e_Page_038.txt
57bc7b910603d0a02322a3b12e9c0e24
411a98ddc1607fe1f74b777b5febb729c7d235fe
1703 F20110209_AAAEFZ knudsen_e_Page_023.txt
2c327bddc921eac72526408252b9a51b
dd1bade6c3e8052338dcdfd0eaf5c1192ea9f9ec
1925 F20110209_AAAEHC knudsen_e_Page_057.txt
08e54869e5ac73ee7d3b065342d39592
a1b978dd3b4f9751e3a2b1b90ed57a8cb34eb736
1138 F20110209_AAAEGO knudsen_e_Page_039.txt
27730b229bf79796b0978ae3f1305cb7
0810e99ca483d49d114f4a612a03a2eb2d752382
1763 F20110209_AAAEHD knudsen_e_Page_058.txt
66d9580b22195529542a94c4dd394741
7900e675a19c9a5cb6d805a25369c91dd1796f6a
1046 F20110209_AAAEGP knudsen_e_Page_040.txt
7a2bf60f7d1e8cda5878753da399ea24
cd9652e4b12da1ccc84d3888098583486d5f5800
1646 F20110209_AAAEHE knudsen_e_Page_059.txt
befd4299a437fcaaa1c8f361a6b54120
30b7859c76f0348e5202810e3a3b07f031032c8b
1326 F20110209_AAAEGQ knudsen_e_Page_041.txt
294be64b24d0400663db83cccedfcc19
8115dc72457c579d9392bf92bd5b7c01489c9a2e
F20110209_AAAEHF knudsen_e_Page_060.txt
ec37b8fe328803fcc6978e457adfdb2a
2932bcbd43c3c51fab5422b09e171055f8f48f3c
1552 F20110209_AAAEGR knudsen_e_Page_044.txt
51cba2dad0ae8817ac8ef1b9ce656586
f8734d1d9e0e2b68d95d7b5c49820a9f15bc09b3
851 F20110209_AAAEHG knudsen_e_Page_061.txt
8c77d20b659a36c66947d6e95dba0b3f
9a1b2472d2453b3430878745a0368ad2d2618705
1592 F20110209_AAAEHH knudsen_e_Page_062.txt
53472795f2561cc5c67ca3c33a648e8e
73c25937c4cf91260f3729a6bfc6cf0123b41737
1555 F20110209_AAAEGS knudsen_e_Page_045.txt
6709bf592a5a54886d448480f39a20a7
e426608625d00fa586b4b3360d308b799847db5c
1833 F20110209_AAAEHI knudsen_e_Page_063.txt
d4fc55fac12e4b728c0a77584abedd35
0f6b5eaf20e460e4562429efe7b1c472a2e6cbf4
1266 F20110209_AAAEGT knudsen_e_Page_047.txt
1481376504670359ecda71901f28e685
7f109fca1fffcd75ebf60a3ad9ae54c9cd1a8354
1412 F20110209_AAAEHJ knudsen_e_Page_064.txt
d71898459f3b1499aa8113f198dc4e23
a0edb90a488671d8672f0286ef019abf99928559
1688 F20110209_AAAEGU knudsen_e_Page_048.txt
0258bbf13b391f87ec7d1968622951d5
0e07998bd7dc17db46749cf4f8727fc9c2ffeb5c
1293 F20110209_AAAEHK knudsen_e_Page_065.txt
772be80252ae0b20cef03fe59ccede8a
f29ef6c914f2c3e107fbf8619f799a735afff7d0
1268 F20110209_AAAEGV knudsen_e_Page_049.txt
7c444d1a4416b3e0197eb0232027fefa
3ce5db68028f6247f462468833a5ef5e0676fe7a
1959 F20110209_AAAEHL knudsen_e_Page_066.txt
cc9ade1a16778d474f91e1cd08d12b87
c8eb22ccdb25c55b873330ca75ccde9d3d468f45
F20110209_AAAEGW knudsen_e_Page_050.txt
5be0d94a1069684f193c421a231658e3
3c4d6d6f9e88367ecdc4ab3d9b61c227aa4012da
414 F20110209_AAAEIA knudsen_e_Page_083.txt
94997aa8f7c14cac57648d95ac4d156b
1a3fe7f419c7a97bc6e1887aff3d2af9c5ca08dd
1949 F20110209_AAAEHM knudsen_e_Page_067.txt
e9ee490a226c53f105d1d1c3c0d8c928
3b5021771cd1506b885e0bd385985698d167fd06
1436 F20110209_AAAEGX knudsen_e_Page_052.txt
5f5489d539b80fd4c3c8791dd60cc94e
cf19bfc1f79c507a2efa243edcd3ec26ea4c721f
694 F20110209_AAAEIB knudsen_e_Page_084.txt
ab128ca14ddd0dece573065e99603c67
39c3c584fd6d5c0adbcc9dc4bd304adb58b787eb
1174 F20110209_AAAEHN knudsen_e_Page_068.txt
d3a1f4fb18f962ea837c48e75e22201d
8f4257b996b6bc549458f95ccb540c1a851ee882
1399 F20110209_AAAEGY knudsen_e_Page_053.txt
10024281013760952a4d16ca73be31b8
547cdd3c1609550d0ddb61fa1d300574e444fd80
1512 F20110209_AAAEIC knudsen_e_Page_085.txt
ff5cfaf17f0a59fdd2ee698222cdf22d
06c83e92ce8b0bb59199a27de8f6a859c5e59505
1717 F20110209_AAAEHO knudsen_e_Page_069.txt
056e2c0ee62773065b0a9f0fd660646e
bd4ba3047a48a97ead14cfb192b76bc728e768bc
1897 F20110209_AAAEGZ knudsen_e_Page_054.txt
8027ba7ba13c1022515f984284f44d3f
80557ab418012fe0ef98a655c52597898caf29c3
1513 F20110209_AAAEID knudsen_e_Page_086.txt
d4b100ab42fa7578e11b22ca25c6c3d9
91c41f073963dda922538a955bb14a20657246aa
1258 F20110209_AAAEHP knudsen_e_Page_070.txt
63b9c53c0d84bbf71dcaab84dcc24641
e45cee63d891f98d86fd45b980539e03245d8a0e
517 F20110209_AAAEIE knudsen_e_Page_087.txt
b1c60f71083a9647b9cb66bdefda499b
9fb7af351d5db6ccb37898cb99e276c38080f140
1367 F20110209_AAAEHQ knudsen_e_Page_071.txt
c9f084bf0a966dfb1c922bfb81fb7d2e
748c034211dbe455380e8349b863e66410fd54a1
521 F20110209_AAAEIF knudsen_e_Page_088.txt
528cb1eba53443e0de2df11bd2df6f50
d37f0b9f204d8a8fa64c5fb3baa43faa641ab778
1504 F20110209_AAAEHR knudsen_e_Page_072.txt
0579e0add90be9abaff60c1b34299bef
35c245ece44a930dbc485fd4acbb94c587687f0d
1939 F20110209_AAAEIG knudsen_e_Page_089.txt
f6dd4c692ae560c69b07f271b1e1420a
6d0ec88883fa16e8d9cabd7eb86c27bfcc15420c
1985 F20110209_AAAEHS knudsen_e_Page_073.txt
52d4ff56268d45392ec8d15fc3b9d257
69b5530a09595fb18f624bc4bcf4d9be401e71d9
1527 F20110209_AAAEIH knudsen_e_Page_090.txt
b52e629f059d3c88e9b6475f9a60b101
e154366494bcf7e608187976880a5d9f6a676e78
1885 F20110209_AAAEII knudsen_e_Page_093.txt
596eb90ff9b16233ba2d1c936b9af27f
da449cd52c2a715416d77e5a74836e0a8ba6c6d9
1879 F20110209_AAAEHT knudsen_e_Page_074.txt
9b172ee85fb98d8a16fcd7c8a0711190
49a9347789c56697ee1f0b3a170cb6949e9e4ad7
1274 F20110209_AAAEIJ knudsen_e_Page_094.txt
feb00dc4fa024bd0718c0dd129ba190b
c0323b3b753ee65aa0dc29a26bee354a694e1c7f
588 F20110209_AAAEHU knudsen_e_Page_075.txt
895b0c53ed83863967b0a05b79a87a47
0a76d0f83bae13ace1cdf43e49e879ba4bfb257a
965 F20110209_AAAEIK knudsen_e_Page_095.txt
b656416b6849787a23a7747324f269b6
121d4be408ca88fb80ea762ab37535bb67112b0b
F20110209_AAAEHV knudsen_e_Page_076.txt
6d8c29fedb156f490bbed747d429af29
9e3d97833e05e13b7e6e34949730668556e5c2f5
1924 F20110209_AAAEIL knudsen_e_Page_097.txt
5695943ac794ba905978d6b94998ed0c
a8a463a2980d278ad8207cb0906b0645ddb09716
609 F20110209_AAAEHW knudsen_e_Page_078.txt
350523c47799a83132df6dc1816c7301
dbe1815c505785959fae8111511ebd224b1d4612
1354 F20110209_AAAEIM knudsen_e_Page_100.txt
e4ed912fa5c14d34b4a6ef7852aa1c2a
9d0642551dfde18e5f0026afd3429c89b49481b3
1255 F20110209_AAAEHX knudsen_e_Page_079.txt
5e29629c117ddb899f8a04c56aa759f7
4f5245ef7ba14a8f761c7a30e743f7e9aa3fa66d
525 F20110209_AAAEJA knudsen_e_Page_115.txt
51c60dd507f7e81682f68a11652374ec
0ff8580460aae00a1270a42571bc09c0bc34a51e
990 F20110209_AAAEIN knudsen_e_Page_101.txt
3fcaf1b82ff5c0328207e92759929538
6cb1e2ab677d175c38fa6288c5fdd0ad2423d587
271 F20110209_AAAEHY knudsen_e_Page_080.txt
e418a192048018fd71ce1f2c21324a51
0989bc95708cf203116966efc620bb8bcd0b0d6c
2004 F20110209_AAAEJB knudsen_e_Page_116.txt
a764560b4ad2594debf72ffb29798d8f
11e69fa675ec12a46ff352ff93d68a4f0a9815b1
F20110209_AAAEIO knudsen_e_Page_102.txt
fd04b2fc46c76059d552c08733f211eb
b3a071cfe6f3d88053fa67a93326ef9c8715acbc
1728 F20110209_AAAEHZ knudsen_e_Page_081.txt
0402cc99aa143f8cf825a342dea9f733
d903b2d15a4abd6493385c1dd3a502d2821ad83a
2269 F20110209_AAAEJC knudsen_e_Page_117.txt
283b978229a7dbd2be350806db035870
7dcb09ef0e9b26839f314c21ed84a168e0ec669d
1755 F20110209_AAAEIP knudsen_e_Page_103.txt
ffe10903afb2f7ec4ad57c0a8837f08d
ddba531a944f0520e59219d8c1c0edfa39b76c7f
2263 F20110209_AAAEJD knudsen_e_Page_118.txt
fe67e076d720aadba00840fc1b05c55c
15415224c3e867fe9d159d9cabf51f6b70769077
2062 F20110209_AAAEIQ knudsen_e_Page_104.txt
d1aa337854c351f525ea9835489a934b
969ab9e8d93e616a1103b03d25ab5275a2a22f08
2467 F20110209_AAAEJE knudsen_e_Page_119.txt
ee40907c0cd127dae4c06fa11c0dc234
29dbb4ed5d55846a27e4bb2add796d8f4de97878
646 F20110209_AAAEIR knudsen_e_Page_105.txt
0ea6cdc3dd26eb470fdee8e84d7cc7a3
a6604cc1de727508160444f9d8c926cc1b8bc9fa
1336 F20110209_AAAEJF knudsen_e_Page_120.txt
4450679be9905359324b8047e90c77ad
6721cdedfbcadf96920eefe3dd2bbfb5ff486d4c
1940 F20110209_AAAEIS knudsen_e_Page_106.txt
b7f024936937c8b6f07b6250943429ff
a6705f8d8e6edc23ad877b901cc197a82fbe68a3
766 F20110209_AAAEJG knudsen_e_Page_121.txt
333b73ed18fc1ad52899f676e6372859
e1a7693e056647b0918ba8368f6260ef28ed0598
1958 F20110209_AAAEIT knudsen_e_Page_107.txt
656bb8112fed6c55e587dcd98322d3f6
aa1bbead61dc301fc58e6e9169d581e9b95c9e74
8357 F20110209_AAAEJH knudsen_e_Page_001.pro
666cfbf276048e0142ce1df9e3c857dc
a0d0db6add59b94fedc2d3353332adf0ea2c3c7b
1098 F20110209_AAAEJI knudsen_e_Page_002.pro
348a875a2c33ce9610af902a8d325d3f
368ab2562853a275404147f9e43178441c0e26b7
1871 F20110209_AAAEIU knudsen_e_Page_108.txt
0a686617e2ceab91d46af35d2d755b36
d78f497dfb178fa34fef9178251823f77d71a732
4437 F20110209_AAAEJJ knudsen_e_Page_003.pro
ace8706f8af17be47c3c64d667362924
04296b61fe2395fc06f1d24a6c96e3311601e510
3228 F20110209_AAAEIV knudsen_e_Page_109.txt
637c8a8c2cbfeda2923ba279375fcc4a
da1012d172a057f56fe1ef7bc128247e09bedc7a
43417 F20110209_AAAEJK knudsen_e_Page_004.pro
c9772166d43937ff53198821ee9b622b
93be419bb1e6d091b6360990b07715365fddbcbb
1335 F20110209_AAAEIW knudsen_e_Page_111.txt
874769a79b23d415dfe2fb5867d46d85
9391fc606ceff18ee488d47ac831a242b1cf6dac
67930 F20110209_AAAEJL knudsen_e_Page_006.pro
2cb5fb458232efc63a253215f48ef666
7fb0c92c359c2d49e9f95a7bcc9516d124c5f572
1410 F20110209_AAAEIX knudsen_e_Page_112.txt
c1f8a057691c77f83438d3558e2d7dc4
383a4363836d77aaab591b839222251554a9627e
37190 F20110209_AAAEKA knudsen_e_Page_022.pro
e9860efc81fd8c85615bbe8a0beca529
0417eb36c4e3ca7efb69a6d1b40b7e00f42e98c5
39880 F20110209_AAAEJM knudsen_e_Page_007.pro
f1c9dbc744cc9912a28de1355f6c102d
5079d72c9dfea8f1b7214dceba17e9b2f8018c35
1543 F20110209_AAAEIY knudsen_e_Page_113.txt
edab2ccf7bfdd4a062d6036aee245959
1afa3bc48f7261fda260589f81cd914a5895e085
38487 F20110209_AAAEKB knudsen_e_Page_023.pro
c213a3c00811fe98e4b18f3cf0903d79
c3862f0ee98a57b2124ac2cfd739e781dcf37b2c
26679 F20110209_AAAEJN knudsen_e_Page_008.pro
0e1ba3ea0e5a7c63cd98c06ed5dbc129
7e11fccd667b635f929ad460ce3e2b5cb7afb6e9
1770 F20110209_AAAEIZ knudsen_e_Page_114.txt
cabde132706e218adfabf64b2009e543
4d9ed266000c85a5d16823bb45b1bb51f69d1b5c
33069 F20110209_AAAEKC knudsen_e_Page_025.pro
4d49282de4ef2d64dba52e030cd3305e
d60599f9b3a2599d00f431f99bcb56fcf3ad9b54
57940 F20110209_AAAEJO knudsen_e_Page_009.pro
5a8161fee741c93209aff4a338f2c213
37d4393563e4658c02d2932e48d09915b05d271f
40994 F20110209_AAAEKD knudsen_e_Page_026.pro
f4b77f92ae2299b734ca55e84c02ca1f
4bcf60ee79830a2d75cadc5e259500dd75c06fab
73851 F20110209_AAAEJP knudsen_e_Page_010.pro
7aea8aa5c1e0aade44fcb5bfd017e569
042f2717b7664563e6143594dfb71c6407465a6c
46619 F20110209_AAAEKE knudsen_e_Page_027.pro
8e7399cf7bbb880d509bde0ca2a01056
837bab74113bb33b38180a4692a7fa80594218dc
41178 F20110209_AAAEJQ knudsen_e_Page_011.pro
8cc217301be5db95a06b17d104957bfe
f66103e47fc63fe0a9b6d26b8150260b7cae776e
26901 F20110209_AAAEKF knudsen_e_Page_029.pro
cf0724a55f4a7f827b78c309b159f9d8
2ac2a8854ce37206a90a3eb75a5bbae61466ab2e
37676 F20110209_AAAEJR knudsen_e_Page_012.pro
8a73fb7f3fd20d453e6f3454f60302c0
78ac4e54e884ad7501f95f65b99444b5efab4255
25366 F20110209_AAAEKG knudsen_e_Page_031.pro
6ac67cbca687e7865448a835ee6cbf80
c3082ea9e19b0e05a1cd3e645ded80c14074b524
34444 F20110209_AAAEJS knudsen_e_Page_013.pro
67c5700a72c8e4f995c941ada0b15d52
2833e1439a0a9c75a29e7a2e8da7c185de4a34c6
14298 F20110209_AAAEKH knudsen_e_Page_032.pro
4fd8df7b9b335561843e0d552bdd19ff
6c68090cab61f5d760aeffcb14bd5738118be16f
40329 F20110209_AAAEJT knudsen_e_Page_014.pro
35b05140171a63a5a6baa53a48073b28
eef31380db429d73a60675cd80c59aed8063c7d7
30088 F20110209_AAAEKI knudsen_e_Page_033.pro
59ef8e0757d0ea063bb84bc5c55e0e37
ac15c73e56685203c5642e4ee37c25e6d19b57dd
33004 F20110209_AAAEJU knudsen_e_Page_015.pro
953f1a43ac4b6ce2f930bcd3b4ff12c6
cd3ae26bce73db915bf3089a20f3623c593a709d
49929 F20110209_AAAEKJ knudsen_e_Page_034.pro
6d29d7eb3349ba6643c3240d215e65be
b1738e92f25c8ab2bbf33713a4647fcb61b0c03d
32008 F20110209_AAAEKK knudsen_e_Page_036.pro
ae70281bd7a5d481465630f418e8a355
358fc7ee0a30e65e6c14e89391c06e9272014767
38059 F20110209_AAAEJV knudsen_e_Page_016.pro
fc1528c4add3e6e9198ad68a8e560f37
c619469f96289f60f91d32fb47c95cdb9ec5040d
22799 F20110209_AAAEKL knudsen_e_Page_037.pro
59dd102bda5c98bd0ef486b973d13a75
92a2b747f21f93a302b8b00deee5eca19e0af678
3201 F20110209_AAAEJW knudsen_e_Page_018.pro
a9ce2a678d4ae07385a5aaa861eefd5e
7916677eead7d177cf31f95d49905a7ccd3fa7f0
40134 F20110209_AAAELA knudsen_e_Page_057.pro
96f843ed73d320a1ca68a51e7ed054a0
c80b00e866dad42e9a47cf7c63cf6b8d0e9f59c5
23472 F20110209_AAAEKM knudsen_e_Page_040.pro
901e8e2e179305f452facce89c427ba6
772e90c166415e3406d1fd38bdc8130da8192144
41360 F20110209_AAAEJX knudsen_e_Page_019.pro
add60cd303e46d957dffb155702149a1
77469dd77f883ee4fad18d4fa653c2a748510441
40680 F20110209_AAAELB knudsen_e_Page_058.pro
ea8311b19d23b1f43ded55c6e5496024
92c182fe1f55e00030be4c8ad1fdd41831245286
43624 F20110209_AAAEKN knudsen_e_Page_042.pro
82296b07db87a93330385e0724b51758
3110eaa796ff0e473a929c05fc11e71c0bc5e192
48727 F20110209_AAAEJY knudsen_e_Page_020.pro
1a9dc7dc8f6ee32925fea861f2036187
2dda9c50ca63d536055db3d7a5cbefaf5d5d65fd
41115 F20110209_AAAEKO knudsen_e_Page_043.pro
334be861048346cb482c1d58d2efb4ea
6ede7d847ba5aba1830555973648ea2eb5d6435a
31809 F20110209_AAAEJZ knudsen_e_Page_021.pro
176ae25a75425ae1d1cbf7a8b3cecd9c
5c86860053d9c1edb16a1e60b4427b45f2941bae
F20110209_AAAELC knudsen_e_Page_059.pro
2e220cb5ffbe7739b60f3bd5edd963b4
32db26571ad0d0135bce8f567dca1e9b77cfd77b
30883 F20110209_AAAEKP knudsen_e_Page_045.pro
3748cc7b99f011b95e20010dce3165cf
4cdfae3ae588d8c27b8dcc4d8d0b1bb4dc75ac09
29807 F20110209_AAAELD knudsen_e_Page_060.pro
311770fb7d1cf7210bafe7132bd1f30f
fa0de7844bdeeadfd2bf801b54f2605b2268db1d
43853 F20110209_AAAEKQ knudsen_e_Page_046.pro
303a80075fdeff2a43f08658b1086bce
a69d62c0cd36f487cbfb93f59a3546309ec1a6bc
17901 F20110209_AAAELE knudsen_e_Page_061.pro
b6cd4d9b4385760d9b8a7acf9793c549
8acbce8738ea8a3e7ba579ab6033fb4ff05b96e5
27395 F20110209_AAAEKR knudsen_e_Page_047.pro
3fd8f10cfb078bec644a169bd560329c
bfe37e302f402f0863e6d26a74276867fed976b2
34995 F20110209_AAAELF knudsen_e_Page_062.pro
0ae1d56d9890e50ee9a33c6b04150bde
d09535d099bf87440d62442817ec4737ad34242b
38837 F20110209_AAAEKS knudsen_e_Page_048.pro
8216b113fbf6b14d9ecd72ba12c711a2
53420b15c248d4782ccaf588ff149195867147a6
38767 F20110209_AAAELG knudsen_e_Page_063.pro
ae05f0153e0f84451bd473c9cd3f6af4
a5091ffcc3d582d5e34bd068c12d8737f4f34cea
27209 F20110209_AAAEKT knudsen_e_Page_049.pro
2d1875727ac7d9ba50068ce86aa3d9bf
7d5a3520862cdd29fc8bca5a0177d2eeb347cf2f
24236 F20110209_AAAELH knudsen_e_Page_065.pro
206e7a12fac04f8278e8d0cc6ddd60a8
072dd9557620f64b7ff21eb74534f57c64fe9e03
36941 F20110209_AAAEKU knudsen_e_Page_050.pro
dcb44ee2e03d2b38a81c1ab1ef47f666
97661c34bf18a101208dddfa24a29f061e9fedaf
42005 F20110209_AAAELI knudsen_e_Page_066.pro
97ed6a3e3c36863fb15c0de63a3a4659
48f3ec78df665c27c2fc226423521cccb456acf3
49604 F20110209_AAAEKV knudsen_e_Page_051.pro
7a3032e58bd844437162f6102e653c3f
bf075fc7d9197795cf556ec9468e812852939af4
45896 F20110209_AAAELJ knudsen_e_Page_067.pro
3a14a6c30eb2584b22cf9bfb7e28e084
f74cc9c4b6bab85d337607e0ba9999d6ede5f9c4
29240 F20110209_AAAELK knudsen_e_Page_068.pro
ceb8d997819c71b47beb4d90c1bd882f
372dfc49d2d814322b1f556e05db905b27171bb6
32645 F20110209_AAAEKW knudsen_e_Page_052.pro
517a7b1fa8c6e60affd69f678505d6f1
83d40c22f838bd31be5ab1e7993d88c7feac34b6
38602 F20110209_AAAELL knudsen_e_Page_069.pro
83761519d508fd21c9b70108c9b0da61
0681113a6f88e88194a5f2309c6afc04a220ec76
F20110209_AAAEKX knudsen_e_Page_053.pro
ebf5ba6c6199ee22310af02f27e7692b
92fe810dde285c543eaea4a7e3e28564ad266573
35258 F20110209_AAAEMA knudsen_e_Page_086.pro
62dafe862772328683c1113b94ec9e3e
92e50eae072c1e5bedf24f87a942b96dca1cc9cf
29453 F20110209_AAAELM knudsen_e_Page_070.pro
b63821956162daca3c74e97a7fdbefa7
6262e416167fef8f8d2bc45651b6273d712421c9
46149 F20110209_AAAEKY knudsen_e_Page_054.pro
9683bf63c5ab4720bf9d59b4ea8d2694
7f619be1a5f7e0c1cc63bb3957b0e1355c5730ae
11931 F20110209_AAAEMB knudsen_e_Page_087.pro
6d5b7abb6d8b552a8fd7a136a4a8caf9
b4b2fe12ac20dba6dae555717e711f52ae2ef775
32485 F20110209_AAAELN knudsen_e_Page_071.pro
351ee41235229b5138046c9a86b189bd
5fc1fdc902679b0ddd1ab4b4f27d8921bc0d8032
18132 F20110209_AAAEKZ knudsen_e_Page_056.pro
147662780c066f7f4124461e51cb4802
723884dc739ee7b248bbb381ea4fb9290db3262d
12342 F20110209_AAAEMC knudsen_e_Page_088.pro
856ad4afade17a0ae2470e035112b650
5c29e65dff2307033c0c502148a4b346526915ec
36188 F20110209_AAAELO knudsen_e_Page_072.pro
0aff771f18a4c726ffa403191365f6ac
47e02f6eb3a58b21980ebe1af9edde1260a4af95
46302 F20110209_AAAEMD knudsen_e_Page_089.pro
58ce39f452c02a71bd7a0e894de1f0a5
83bbb680150baf7345aefd1d5f46268d229b23b0
47555 F20110209_AAAELP knudsen_e_Page_074.pro
1600268e9f60c0c4195c1e30f34aa753
de0b0bc25792ed08454f05b4e8ba96950feea28e
38046 F20110209_AAAEME knudsen_e_Page_090.pro
1213058311671b48b9851e68be49c6d2
d507192ead3633e6070380f9aedfca3804605c42
12824 F20110209_AAAELQ knudsen_e_Page_075.pro
a47e2159469245fde5ae2b9a3d1e9956
16f764799c6c2c3a8801d95a1c18d46f13eb5630
30174 F20110209_AAAEMF knudsen_e_Page_091.pro
b0e8badeb1cd1842f4aa42963bc0363c
4885122d93d13fa46bbb17a30051c04d2a29c754
31261 F20110209_AAAELR knudsen_e_Page_076.pro
7b900bb1e770ffd281ff9b4044c4e03b
1956799ff987f9ce05a96bfb932ee1ac0bcab9dd
27934 F20110209_AAAEMG knudsen_e_Page_092.pro
a47e97fc5903563af55b2321fa6fa279
6e7790c677c9d213984b7c09751430e6a4bedf99
44729 F20110209_AAAELS knudsen_e_Page_077.pro
a8d514462805e557e484859d3cea32e1
408b514b7c4f42bd9698276f53fac9214fd04b74
40040 F20110209_AAAEMH knudsen_e_Page_093.pro
06f7859c1b9916f4694e7c291ccd6226
7e2a550594c553d3e821c39e3732cf9fe12362d0
15114 F20110209_AAAELT knudsen_e_Page_078.pro
bd3c9ab2cd8ae3f668f987cafb80365b
7dc35a54d2273fd406b368ee1a9a0c6087bead83
29494 F20110209_AAAEMI knudsen_e_Page_094.pro
489278d0ff0211365fad7aa95693729c
8e6e151ae072b2e6f70eab826a435ec74effbc57
30116 F20110209_AAAELU knudsen_e_Page_079.pro
986184608c576fc05c5a9c393a087b72
7d0f2a8b7e23df40b5dc222ebd3c2e2082921015
22895 F20110209_AAAEMJ knudsen_e_Page_095.pro
cb8e6114cf5e3b7bc6c0ad1cb670f741
71d1415d06cafd0540188df04190289398c9a4f3
41437 F20110209_AAAELV knudsen_e_Page_081.pro
03e537c43edec25f66bb61f8f7471165
77f4966657eccad6d259d078fb8193741235235d
48234 F20110209_AAAEMK knudsen_e_Page_096.pro
c7f99bdcd79b125c6013e9cdc49311c3
13990b0f78da3b7abe76aedb197509d824daac60
10084 F20110209_AAAELW knudsen_e_Page_082.pro
1b37023b8f9cb71d624339ea40ef58a5
17f5ff9680631659c40eee5d8e86d9fef179eeb5
48658 F20110209_AAAEML knudsen_e_Page_097.pro
02ecfb82a0b971ebb691fe1c776f231b
e1894436374194e9c08e20466bc2b06ed4f10610
35734 F20110209_AAAENA knudsen_e_Page_114.pro
a41afefe774c3a634daf5fca213c9be7
96e6c36ea185a3246b7c1762fc14d995c5383108
46582 F20110209_AAAEMM knudsen_e_Page_099.pro
5af8fa1ef9e355815b6b9444e13a5b8c
256b753be82366458aaec2b64b8c6c94e1bcf6f1
9574 F20110209_AAAELX knudsen_e_Page_083.pro
35ef3884a30a426118c280f51a7d8121
6f6c923fef44eb63d5ad621916a63408daa459bd
8853 F20110209_AAAENB knudsen_e_Page_115.pro
59cda23f38bc87c9c7b78c826ccb7591
d29061f3ce4e37bda689ef6f39c7dc1a3c02a506
32805 F20110209_AAAEMN knudsen_e_Page_100.pro
ae8f26c8a407577656037fd8b908a1a7
303905a1e6835efdc9f01d2149de40bc1a25c08d
F20110209_AAAELY knudsen_e_Page_084.pro
738a6296f25ce7626413cbe7a1a194ab
40601b91b2018d71777f7c62dd930e35cc3e7b85
49218 F20110209_AAAENC knudsen_e_Page_116.pro
3d3d2a3eee52b1f55236da71381c7f47
07fba57b26144682fb674ea8ef543ab79e91a4ec
17778 F20110209_AAAEMO knudsen_e_Page_101.pro
2cc71a21c9c2e47713f38b83037280a3
ca8691c280660f3fd569fbf7c29b2c929771d1d7
35380 F20110209_AAAELZ knudsen_e_Page_085.pro
503b4cae1efe82a8280a7997290f0bcf
969bdb43168f17549ee27f5f437b1a77d8f303a8
56034 F20110209_AAAEND knudsen_e_Page_117.pro
0fa333ffbd9c12e4edfb6544f475f364
d89abe451ecdf438e7488c0bc355567c16508863
33754 F20110209_AAAEMP knudsen_e_Page_102.pro
cab2fbbb5f63294dbe75484137addeb4
6f6c6f2c2500e51728d10b6fa9ca31a1560a59c9
55754 F20110209_AAAENE knudsen_e_Page_118.pro
fe1d4f1fc16d76736c39ed49d220d558
48326188e9d625956eb64fd68bc12b6cb973e048
29754 F20110209_AAAEMQ knudsen_e_Page_103.pro
98d81af8e16dbda24d25960e023e0e72
f0fe758eafa5ed29f4c4e6ed29ff5283948f293a
61130 F20110209_AAAENF knudsen_e_Page_119.pro
136c39831a192dfcf0c35bae739def04
e4a4b2186498802214277d2e5c73f91006011bcb
F20110209_AAAEMR knudsen_e_Page_105.pro
257fc61861b18eb83e1998c03ee39e5d
7591cfaf68c75f683d7b4f4feba6f2b4d7b79fb3
31687 F20110209_AAAENG knudsen_e_Page_120.pro
e05deffc6ad2cffbb72953d9f433ad02
73fa83333948eea78f5ef143a423326d3a001003
48078 F20110209_AAAEMS knudsen_e_Page_106.pro
6b9cd9a7ded279a97a00622b789126eb
26f8c1665e201c418afa18cc6acef389301f7637
18112 F20110209_AAAENH knudsen_e_Page_121.pro
027639f1374270bc773e04177e244211
4b2c4b0c20c85f7c0a4afac95825da1c5d0a5198
49601 F20110209_AAAEMT knudsen_e_Page_107.pro
85862128cbb60d9f687629e37504a3b6
a030a7434898113e68d7542e80ab58f1fb61fe38
26654 F20110209_AAAENI knudsen_e_Page_001.jpg
f8d608cc62963a345bca9be7648a5ac5
79a6abf1e6a834abd2f8453543d93ed0b05b5a4d
45304 F20110209_AAAEMU knudsen_e_Page_108.pro
dafa4af84f5caace58aa4bf0a35855c5
9a75c9a99e205b52a9553ae559c2abeb010b46b9
4506 F20110209_AAAENJ knudsen_e_Page_002.jpg
5c7b9c7876d3a5516c1fb0edca00184e
6f643c6b198768395495ecbb111db50523f0c36d
79600 F20110209_AAAEMV knudsen_e_Page_109.pro
70d91d7de4a15965c81d417d45e542d3
16359ca32e89366ecb3e1bf36f7a78366c2b2163
12185 F20110209_AAAENK knudsen_e_Page_003.jpg
3ffb2c44e52243918dc6a949e2fe2f4a
64376f748650d6406b3e604f91e73f1ed8bd56a4
65951 F20110209_AAAEMW knudsen_e_Page_110.pro
5299f3b96b18f2252a034e3401aa8dcd
bfc33a642978b1213459e26cb1d3b27a78c2b79c
89682 F20110209_AAAENL knudsen_e_Page_004.jpg
2fc563656d568978a4c98dc4d8081336
5dd20a708100b5488cae0943ce9bb28928fe83b9
26233 F20110209_AAAEMX knudsen_e_Page_111.pro
f303e58b04474ed80f79692aec3c4329
41fb35e2c9bba5d3dc6737111c00b850eef33cdb



PAGE 1

ANISOTROPIC FRACTURE ANALYSIS OF THE BX-265 FOAM INSULATION MATERIAL UNDER MIXE D-MODE LOADING By ERIK KNUDSEN 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 2006

PAGE 2

Copyright 2006 by Erik Knudsen

PAGE 3

This thesis is dedicated to my parents, fr iends, family, lab mates and everyone else who was there for me when I needed them the most. I will never forget you!

PAGE 4

iv ACKNOWLEDGMENTS First and foremost I would like to thank my parents for their unwavering love and support. My advisor, Dr. Nagaraj Arakere, ha s been in my corner since I first came to graduate school and his contacts at NASA pa ved the way for the two degrees I now hold in engineering. In the summer of 2002 I acco mpanied Dr. Arakere to Huntsville as part of the Summer Faculty Fellowship Program During our ten week stay, I met many experts and scientists who work at Marshall, particul arly Dr. Greg Swanson, Jeff Rayburn, Greg Duke, Dr. Gilda Battista, Stan Oliver, Preston McGill, Doug Wells, Dr. Eric Poole, Dr. Philip Allen and Denise Duncan (in a later trip), and many, many others. They have helped me grow as a student and a person, and I am forever indebted. After my trip to NASA, approximately one year later, I earned a NASA Fellowship to sponsor this work. Ed Adams of the E ducation Directorate at Marshall deserves special praise here, and I thank him for bl essing me with the opportunity to, once more, work with NASA. I shared an on office Dr. Philip Allen and one could not ask for a more pleasant person to work with and learn from. The Cornell Fracture Group was also a tr emendous help in my quest to better understand and learn the principl es of fracture mechanics. Drs. Paul Wash Wawryznek and Bruce Carter are two of the nicest and most patient people I have ever known and I owe them many thanks for taking the time to help me. Last, but certainly not least, I wish to thank my labmat es over the years: Srikant, Sangeet, Shadab, Niraj, Shannon, Yogen, Jeff, TJ, and George. Without their insight,

PAGE 5

v jokes, and patience I would not be where I am today. I now consider many of them to be my best friends and I hope we can stay in touch.

PAGE 6

vi TABLE OF CONTENTS page ACKNOWLEDGMENTS.................................................................................................iv LIST OF TABLES...........................................................................................................viii LIST OF FIGURES...........................................................................................................ix ABSTRACT......................................................................................................................xii CHAPTER 1 BACKGROUND..........................................................................................................1 Problem Motivation......................................................................................................1 Program Scope and Focus............................................................................................2 2 FOAM TESTING AND MATER IAL CLASSIFICATION........................................6 Cellular Solid Background Information.......................................................................6 Initial Foam Material Testing.......................................................................................7 Determination of Elastic Constant s and Material Classification...........................7 Fracture Testing for the BX-265 Foam Material.................................................20 Additional Testing: Divot Test Specimens.................................................................24 Summary.....................................................................................................................28 3 AN INTRODUCTION TO FRACTURE MECHANICS...........................................30 Material Definitions....................................................................................................30 Isotropic Fracture Mechanics.....................................................................................31 Linear Elastic Fracture Mechan ics for Anisotropic Materials....................................33 Determining the Stress Intensity Factor K ...............................................................37 FRANC3D Next Generation Software.......................................................................38 Computing SIFs Using Displacement Correlation..............................................45 Effects of Temperature on the SIF Solution...............................................................52 Summary.....................................................................................................................54 4 CRACK TURNING THEORY AND FINITE ELEMENT MODELING.................56 Crack Turning Theory................................................................................................56

PAGE 7

vii FE Modeling of the M(T) Specimen..........................................................................64 5 RESULTS AND DISCUSSION.................................................................................68 Effect of Material Orient ation on the SIF Solution....................................................68 Crack Turning Predictions..........................................................................................76 M(T) Specimen Fabrication and Determ ination of Local Knit Line Axes.........77 Local Crack Tip Computations...........................................................................78 Influence of Thermal Loads on the SIF Solution.......................................................85 Summary.....................................................................................................................93 6 CONCLUSIONS AND RECOMMENDATIONS FOR FUTURE WORK..............95 Conclusions.................................................................................................................95 Recommendations for Future Work...........................................................................96 APPENDIX GENERAL SOLUTION FOR THE PLANE PROBLEM OF A LINEAR ELASTIC ANISOTROPIC BODY..............................................................................................98 LIST OF REFERENCES.................................................................................................103 BIOGRAPHICAL SKETCH...........................................................................................108

PAGE 8

viii LIST OF TABLES Table page 2-1 Measured material prope rty values from experiment...............................................18 3-1 Various methods for determining the SIF................................................................38 4-1 Direction cosines......................................................................................................66 5-1 Direction cosines for the cas e one material orientation...........................................73 5-2 KI for cases 0-8, 0 degree inclination......................................................................76 5-3 KI for cases 9-16, 0 degree crack inclination...........................................................76 5-4 Measured and predicted tu rning angles (in degrees)................................................83 5-5 Summary of mid-plane KI-KII.................................................................................91 5-6 Summary of mid-plane KIII.....................................................................................91

PAGE 9

ix LIST OF FIGURES Figure page 1-1 Foam loss during Shuttle ascension...........................................................................2 1-2 Spray-on foam insulation (SOFI)...............................................................................3 2-1 Idealized foam microstructure....................................................................................8 2-2 Square specimen tensile test used to determine the Poisson ratio............................15 2-3 Foam test specimens.................................................................................................15 2-4 Summary of stre ss-strain data .................................................................................16 2-5 Coordinate system for the tr ansversely isotropic material.......................................17 2-6 Youngs moduli vs. temperature..............................................................................18 2-7 Shear moduli vs. temperature...................................................................................19 2-8 Coefficient of thermal expansion vs. temperature...................................................19 2-9 Thermal conductivity vs. temperature......................................................................20 2-10 From left to right: M(T), C(T), and SNE(B) fracture test specimens......................22 2-11 Summary K-values for va rious fracture specimens.................................................23 2-12 Clip gauge on C(T) specimen used to measure CMOD...........................................24 2-13 Typical load vs CMOD curve for a brittle material.................................................25 2-14 Divot test panel.........................................................................................................26 2-15 2-D view of divot test...............................................................................................27 2-16 2-D view of test panel after heat flux has been applied...........................................27 2-17 Foam blow out from divot test.................................................................................28 3-1 Coordinate system us ed in equation (3-1)................................................................31

PAGE 10

x 3-2 From left to right: mode I (opening), mode II (shearing), mode III (tearing).........32 3-3 Path of the J-integral................................................................................................39 3-5 Definition of Irwins crack closure integral.............................................................43 3-6 Correlation points typically used for displacement correlation...............................46 3-7 Triangular quarter point elements............................................................................48 3-8 Isoparametric element and degenerated element with mid-side nodes moved to quarter point locations..............................................................................................48 4-1 Definition of crack turning angle.............................................................................57 4-2 Elliptical relationship for T...................................................................................58 4-3 Definition of hoop stress..........................................................................................59 4-4 Orthotropic toughness values...................................................................................62 4-5 Definition of the a and n vectors..............................................................................62 4-7 M(T) model created in ANSYS...............................................................................65 4-8 Sample transformation.............................................................................................66 4-9 Fractured M(T) specimen: the dotted lin e is added to show the location of the knit line.....................................................................................................................67 5-1 Definition of crack inclination angle........................................................................69 5-2 Definition of crack front distance.............................................................................69 5-3 M(T) model built in ANSYS....................................................................................70 5-4 Definition of material orientation.............................................................................70 5-5 Top view of the cones shown in Figure 5-4.............................................................71 5-6 Hypothetical material orie ntation relative to the ET................................................71 5-7 Definition of case one material or ientation for the M(T) FE model........................72 5-8 Mode I SIF for cases 1-4; 0 degree crack inclination..............................................74 5-9 Mode I SIF for cases 1-4; 30 degree crack inclination.............................................74 5-10 Mode II SIF for cases 1-4; 30 degree crack inclination...........................................75

PAGE 11

xi 5-11 Mode III SIF for cases 1-4; 30 degree crack inclination..........................................75 5-12 From left to right: front, left, righ t, and rear sides of the M(T) specimen...............77 5-13 Determination of knit line plane...............................................................................78 5-14 M(T) model with inclined crack...............................................................................79 5-15 Cartesian and cylindrical stresses.............................................................................81 5-16 Crack turning angle..................................................................................................82 5-17 Application of thermal loads....................................................................................87 5-18 Room temperature KI-KIII (F 3D) vs. KI-KIII (DC) with T1 = 0oF and T2 = 100oF........................................................................................................................88 5-19 Room temperature KI (F3D) vs. KI (DC) with T1 = -200oF and T2 = 100oF...........88 5-20 = 30 degrees: room temperature KI (F3D) vs. KI (DC) with T1 = 0oF and T2 = 100oF........................................................................................................................89 5-21 = 30 degrees: room temperature KI (F3D) vs. KI (DC) with T1 = -200oF and T2 = 100oF................................................................................................................90 5-22 Temperature distribut ion along the crack front........................................................92 5-23 KI-KIII vs. normalized crack front distan ce; TG2 applied in thickness direction...92

PAGE 12

xii Abstract of Dissertation Pres ented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy ANISOTROPIC FRACTURE ANALYSIS OF THE BX-265 FOAM INSULATION MATERIAL UNDER MIXE D-MODE LOADING By Erik Knudsen August 2006 Chair: Nagaraj K. Arakere Major Department: Mechanic al and Aerospace Engineering The National Aeronautics and Space Ad ministration (NASA) uses a closed-cell polyurethane foam to insulate the external tank (ET) which cont ains the liquid oxygen and hydrogen for the Space Shuttle main engine s. This is a type of spray-on foam insulation (SOFI), similar to the material used to insulate attics in residential construction. In February of 2003, the Shuttle Columbia suffered a catastrophic accident during re-entry. Debris from the ET impac ting the Shuttles thermal protection tiles during liftoff is believed to have caused the Space Shuttle Columbia failure during reentry. NASA engineers are very interested in unde rstanding the processes that govern the breakup/fracture of this complex material from the shuttle ET. The foam is anisotropic in nature and the required stress and fracture mech anics analysis must include the effects of the direction dependence on mate rial properties. Over smoot h, flat areas of the ET the foam can be sprayed down in a very uniform fashion. However, near bolts and fitting

PAGE 13

xiii points it is possible for voids and other defects to be present after the foam is applied. Also, the orientation of the foam, as it rises from non-uniform surfaces, can be significantly different from the orientation over typical acreage sprays. NASA believes that air present in these small voids is liquefied and then turned into a gas during liftoff. While the Shuttle is ascending to space, the pressure in these cavities can become large enough to force a subsurface crack toward the exterior of the tank, thus freeing portions of foam insulation. As a first step toward understanding the fracture mechanics of this complex material, a general theoretical and numerical framework is presented for computing stress intensity factors (SIFs), under mixed-mode loading conditions, taking into account the material anisotropy. The effects of mate rial orientation and mode mixity on the anisotropic SIF solution are analyzed. Cr ack turning predictions under mixed mode loading are presented. Furthermore, the in fluence of temperature gradients on the SIF solution is studied, in view of the thermal gradient present thro ugh the foam thickness on the ET. The results presented represent a quantitative basis for evaluating the strength and fracture properties of anisotropi c BX-265 foam insulation material.

PAGE 14

1 CHAPTER 1 BACKGROUND Problem Motivation On February 1st, 2003, the Space Shuttle Columbia suffered a catastrophic failure during re-entry. NASA has conduc ted an exhaustive investiga tion of the failure and the consensus now is the breakup was caused by a segment of BX-265 foam insulation, roughly the size of a suitcase, striking the wing during liftoff. The foam impact, in prior Shuttle launches, has been known to cause imp act damage to the thermal protection tiles, but was not considered to be a serious probl em until the Columbia disaster. The damage to the thermal protection tiles on the leading edge of the wing is thought to have allowed gases to penetrate the wing during re-entry, triggering a cascading se ries of catastrophic events that led to the loss of the Shuttle. It is believed that when the foam insulation is applied to the external tank (ET) that voids are created in certain areas where geomet ric discontinuities such as bolts, flanges, and fittings are encountered. Since the tank is filled with liquid oxygen and hydrogen the foam is exposed to cryogenic temperatures at the ETs surface. The air inside these voids can be liquefied and nitrogen can conde nse into these small cavities. During liftoff the outer surface of the foam is exposed to ae rodynamic heating. This heating raises the temperature of the liquid nitrogen, turning it into a gas. The pressure difference associated with the formation of the gas can cause pieces of foam to be blown out during liftoff( see Figure1-1).

PAGE 15

2 Figure1-1. Foam loss during Shuttle ascension Program Scope and Focus Prior to the Columbia disaster, the foam that insulates the tank was not understood at a fundamental level. Quantities like the el astic moduli, and thermal properties such as coefficient of thermal expansion are needed for a meaningful failure analysis of this material. This prompted NASA to initiate an intensive testing program to ascertain various properties at cryogenic, room, and elevated temperatur es. The pictures in Figure 1-1 show the foam breaking off areas near fit ting points. At these locations, insulation is sprayed around various parts, such as bolts, a nd it is here that th e foams orientation, relative to the tank, can change by quite a bit. For typical ac reage sprays, there is very little difference between the local coordinate ax es of the foam and those of the substrate (Figure 1-2). Pictures of the Shuttle during liftoff, su ch as those shown in Figure 1-1, show us that large pieces of foam are not really breaking away over smooth, open, areas of the tank. The problem seems to be confined to areas where the foam is applied over fitting points, bolts, and other areas where various pa rts can protrude the su rface of the ET.

PAGE 16

3 Figure 1-2. Spray-on foam insulation (SOFI) Right away we notice that tw o potential problems come to mi nd at these locations: it is possible for voids to be created when the foam is being laid down, and the orientation of the foam can change by a large degree. The material tests conducted by NASA have show n that this type of foam is brittle. Perhaps the best way to analyze sharp, crack-lik e, defects for this sort of material is through linear elastic fracture mechanics (LEF M). For isotropic materials, the near-tip stresses are completely characterized by K the stress intensity factor (SIF). The analytical K-solutions for anisotropic materi als take a similar form, but have additional terms that come from the various elastic moduli. The formulations of near-tip stresses for isotropic materials are well established and are now incorporated in many finite element (FE) software packages. Some K-solutions for certain kinds of direction-dependent materials started appearing in the 1960s and mo re generalized solutions were presented in the 1980s. Since many engineeri ng materials are not isotropic (such as rolled aluminum

PAGE 17

4 in aircraft fuselages and singl e-crystal superalloys used in high performance gas turbine engines), more attention is being paid to ward the implementation of anisotropic Ksolutions in various FE schemes. Still, many of the most popular FE software packages do not allow accurate computations of SI Fs for simulations involving anisotropic materials. Foams are inherently anisotropic, and NASA, after extensive testing, has classified this material as transversely isotropic. Su ch materials also have fracture characteristics that are direction dependent and both of th ese traits must be accounted for in the analytical and numerical analyses. Alt hough anisotropic K-soluti ons are not readily available, the Cornell Fracture Group has deve loped a software package, FRANC3D, that simplifies the meshing process necessary for FE modeling of crack-like defects in elastic solids and also computes a K-solution for anisotropic materials. Concurrently, an alternative method for computing the SIFs, based on the work of Hoenig (1982), for anisotropic materials has been employed and extended to account for thermal stresses and strains. It is hoped that the present study will enhance NASAs understanding of this material, since the K-solutions used in th is analysis account fo r both the directiondependent properties and the impact of therma l stresses and strains. The following points summarize the scope of this study in a broad sense: 1. Through the use of the middle tension, M(T), specimen and the FE method a systematic rotation of material propertie s and its impact on the K-solution is examined 2. The impact of mode mixity on the K-solution is studied 3. Crack turning predictions that take into account both the anisotropic stiffness and fracture properties are presented

PAGE 18

5 4. The effect of a thermal gradient through the M(T) specimen and its impact on the K-solution are presented

PAGE 19

6 CHAPTER 2 FOAM TESTING AND MATER IAL CLASSIFICATION Cellular Solid Background Information Cellular solids are generally comprised into two sets of structur al categories: 1) honeycomb layouts where the microstructure is comprised of two-dimensional arrays of polygons that fill a planar area and 2) foam s, the three-dimensional counterpart to honeycombs where the cells take the form of polyhedra. Within Gibson and Ashbys (1988) definitive text on cellular solids, we see how many types of materials can be foamed: metals, glasses, and ceramics can be fabr icated into cells that take on a variety of shapes and sizes. Among the most common t ypes of foam are polymers which are used for many applications including insulation fo r residential homes or disposable coffee cups. Polymer foams are usually very light and they are popular in industries where weight is of prime importance, such as in the aerospace industry. Foams are typically made by bubbling a gas into the liquid monomer or hot polymer. The gas can be introduced mechan ically by stirring or by using a blowing agent. Physical blowing agents are norma lly gases like nitrogen or carbon dioxide. Chemical blowing agents are special additive s that decompose when they are heated or react in such a fashion where gas is released. The bubbles gr ow and eventually settle and then, as the liquid is cooling down, become solid. In general, polymer foams have considerab ly lower strength a nd density than solid metals. However, their low thermal conductivity coupled with their much lighter weight

PAGE 20

7 make them an ideal insulator for the ET. Table 2-1 (Gibson and Ashby, 1988) lists several properties of polymer foams and true solids (solid metals and ceramics). Table 2-1. Properties of solids and polymer foams Density (lb/in3) Conductivity (BTU/hr in F)Youngs modulus (psi) Solids 3.6 x 10-2 3.6 x 10-1 4.8 x 10-3 48 3 30 x 106 Polymer foams 3.6 x 10-4 4.8 x 10-3 3 x 103 Initial Foam Material Testing The Columbia disaster forced NASA and a ll those associated with the application of the foam insulation material to the ET to see it in an entirely diffe rent light. Numerous tests were conducted at cryogenic, room, and elevated temperatures to ascertain the properties of this material. In addition to st andard torsion and tension tests to obtain the elastic moduli, various tests we re performed to evaluate the failure characteristics, such as the plane strain fracture toughness, KIc. While these tests are an important first step in understanding how the foam behaves on a funda mental level, NASA also took extra steps to examine how the foam sheds from the tank through a specialized test covered in a later section. Determination of Elastic Constant s and Material Classification Some of the initial work toward understa nding the mechanical properties of foams took places in the 1960s and 1970s. Gent and Thomas (1959), and Patel and Finnie (1970) among others were among the first to expl ore the properties of cellular materials. The model developed by Gibson and Ashby (1982) however, is perhaps the most widely accepted way to analyze the propert ies of cellular solids. Thei r analysis is rooted in a unit cell comprised of a network of interconnected struts (Figur e 2-1) The equations that arise from this mathematical treatment are in terms of the strut dimensions, which in turn

PAGE 21

8 Figure 2-1: Idealized foam microstructure allow some material properties to be related to the relative density The relative density is the density of cellular material, *, divided by the density of the solid from which the cell walls are made, s. Generally, foams can be divided into two main categories: those with open or closed cells. By closed we mean material s where each cell is partitioned from its neighbor by membrane-like faces. Within open -celled structures the individual members, struts say, are individually connected. With Gibson and Ashbys (1982) model, many of the equations that determine various properties, such as the elastic moduli or the fracture toughness are derived considering the c onfiguration of the cell walls. Understanding the mechanical and metallur gical aspects of these struts at a microscopic level is very important. After all, it is these struts that transmit the applied mechanical loads and transfer heat energy. But from a practical standpoint, models that contain these microscopic parameters are not very useful to the engineering community. The engineer may not have access to the equi pment and facilities necessary to measure

PAGE 22

9 things like cell wall thickness. Generally, the analyst, or engineer, may only be privy to a quantity like the relative dens ity. This parameter is one of the most important, and useful, concepts that help define and understa nd the properties of cellu lar materials. The relative density can be related to the dimens ions of the cell walls. For honeycombs we have 1*st C l (2-1) where t is the cell wall thickness and the l is the edge-length. For open-cell foams 2 2*st C l (2-2) Finally, for the foams of the closed-cell variety we have 3*st C l (2-3) The C terms are numerical constants that de pend on the cell shape, normally taking a value of unity. As mentioned beforehand, certain open-ce lled foams can be modeled as a network of beams. Using an idealized array of connected beams, the theory in many solid mechanics texts such as Timoshenko and G oodier (1982) is adequate to determine the deflections, strains, and stress es. It can be shown that the Youngs modulus for the opencelled foam is given by 1 4* s CEI E l (2-4) where E* is the modulus of the foam material. The relative density is related to the dimensions, t and l as shown in (2-2). The second moment of the area, I can also be related to the cell wall dimensions

PAGE 23

10 4 I t (2-5) Now (2-4) can be re-written as 2 1**ssE C E (2-6) Using the same reasoning, we can write down the relationship for the shear moduli 2 2**ssC E (2-7) Lastly, we need *, the Poisson ratio, which is defined as the ratio between the lateral and axial strains. The strains are proportional to the bending deflecti on and their ratio is constant. Thus, depends only on the cell wall geometry. We can show this for isotropic foams where s, the shear modulus is 21sE (2-8) Using (2-8) along with (2-6) and (2-7) we have 1 3 2*1 2 C C C (2-9) The above analysis is for open-celled foam s. When examining closed-cell foams, the derivations are more complex because most foams are made from a liquid and surface tension draws the material to the edges which might leave only thin membrane across the cell. For sufficiently thin membranes, the opencelled formulas can be applied. But if an appreciable quantity of foam is present in these membranes, the amount, or fraction, of this material contributes to the stiffness. There are three potential contributions that sum to the total stiffness of a closed-cell mate rial. The aforementioned cell walls make a contribution when they are stretched. The s econd contribution is present due to a fluid,

PAGE 24

11 usually air, trapped within th e cells, and finally the cell wa lls, and beams, can also be bent when a load is applied. The contribution from the stretching is de rived by considering how the applied load bends and stretches the cell face. The cell edge bending is proportional to S 2 where S is the stiffness of the cell edge, and S EsI/l3, and is the displacement that arises from the applied load. The contribution from the stretching of the face is proportional to Es 2Vf. The Vf term is the volume of solid in the cell face and /l and Vf l2tf. The thickness of the edges a nd faces are denoted as te and tf, respectively. In the linearelastic regime, we can define the work done by the applied force as 2 2 31 2s s fEI FElt ll (2-10) Equation (2-10) can be re-wri tten when we note that I t4 e and E* (F/l2)/( /l) which yields 4 4* '' f e st t E Ell (2-11) For many foams the edge and face thickness is related to the relative density by the following empirical relations 1/2 1/2* 1.41 0.93f s e st l t l (2-12) If we substitute (2-12) into (2-11) we obtain the ratio of the elastic moduli in terms of the relative density and volume fraction of material in the cell edges is

PAGE 25

12 2 11*** '1 s ssE CC E (2-13) where , , , C1, and C1are all constants of proportionality. The above equations take care of the contribution from the stretching of the cell walls. The next contribution comes from accounting for the fl uid trapped in the cell walls, E*g. From Gent and Thomas (1963) we have 012* * 1g sp E (2-14) where p0 is the initial gas pressure. Gibson a nd Ashby (1988) note that if this pressure happens to be equal to the atmospheric pressure this contribution is small. Lastly, shear modulus for the closed-cell foam is shown below 2 22*** '1 s ssG CC E (2-15) Finally the last contribution to consider comes from the bending on the cell struts, and that analysis is identical to the ope n-celled formulation presented earlier. We note that the above relations are for foams where the cell walls are equiaxed. Most man-made foams are anisotropic. When this particular foam is sprayed down, foaming agents cause it to rise and the cells are stretched in th e rise direction. Cell shape, then, can significantly impact the material properties. The shape anisotropy ratio, R, is 1 12 2 1 13 3L R L L R L (2-16)

PAGE 26

13 where 1L, 2L, and, 3L are the lengths of principal cell dimensions and 1L is the largest of the three. Following a similar analysis for the bending stresses within the beams for the isotropic lattice Gibson and Ashby (1988) go ont o define a quantity called the Youngs modulus anisotropy ratio for open-celled foams as 2 3 3 1* 2 *(1(1/))E R ER (2-17) For closed cells an additional term 2 1 1(1/)R R appears in (2-17). For the anisotropic shear modulus, we have 31 12* 2 *1G GR (2-18) The Poisson ratio is once again the ratio of late ral and normal strains and, just like in the equiaxed case, is independent of the relative density. As such this elastic constant is dependent solely on the cell geometry. Un fortunately, for the anisotropic foams, employing the same type of dimensional analysis used to determine the elastic moduli for the Poisson ratio will not offer any insight in to its dependence on cell wall geometry and Gibson and Ashby (1988) do not di scuss this in their text. So there appear to be two ways to try and determine the properties of the foam insulation material. The properties can be es timated using the relative density and shape anisotropy ratio or they can be experimenta lly evaluated. The BX-265 foam contains closed cells, so the equations presented earlie r for this particular cell wall geometry along with corrections necessary to account for the anisotropy could be utilized for evaluating various material properties in te rms of the relative density.

PAGE 27

14 Evaluating the all-important relative density though, may not be as straightforward as it seems. While it is not difficult to measure the density of the foamed material (one simply weighs the sample and divide it by the measured volume), getting a handle on the un-foamed density for this particular kind of polyurethane foam is not so simple unless the exact same foam is available in the literature so mewhere. Also, when NASA was preparing various foam test panels (from wh ich the test specimens are machined) they determined the density could vary within the specimen. Also, there are several ways in which the insulation can be sprayed down. The foam can be applied via a mechanical process where a machine sprays it down, or it is laid down by hand using special equipment that delivers the insulation to the testing surface (normally an aluminum plate). Thus, since the density could change th roughout the test panel, and even vary depending on the process by which it is spra yed, NASA developed a test program where a vast database of information was to be established for this particular foam material. It was decided that standard test procedures for evaluating both elastic properties and fracture response were to be employed, in stead of estimating the properties based on relations that require the relative density to be known beforehand. Figures 2-2 and 2-3 display some of the various test specimens used in various tension and torsion tests. A su mmary of typical stress-strain da ta is shown in Figure 2-4. Within this chart the results of several test s performed with different foam orientations (denoted as local rise, spray, and axial) at a few temperatur es (RT denotes room temperature and LN2 is the temper ature of liquid nitrogen).

PAGE 28

15 Figure 2-2. Square specimen tensile te st used to determine the Poisson ratio As expected the colder the temperature, the more stiff the foam becomes and in general the foam fractures with little deformation. It should be pointed out that the foam is not a material that behaves exactly like a generic isotropic steel alloy. Most engineering Figure 2-3. Foam test specimens

PAGE 29

16 Figure 2-4. Summary of stress-strain data metals, when placed in a uniaxial tension te st, have a linear rela tionship between stress and strain up until the yield point. Cellular materials generally have a non-linear relationship between stress and strain and the constitutive relations defined in this chapter assume that this material has a linear stre ss strain relationship because a model that accurately describes the response for fo am has not yet been developed. The aforementioned tension a nd torsion tests have led NASA to classify the foam as a transversely isotropic material. Th ese materials are sometimes called hexagonal materials and five independent elastic constants relate the stresses to the strains in the constitutive matrix. The 11-22 plane in Figure 2-5 is a plane of isotropy. Source: personal communication, Doug Wells (MSFC) Summary of Typical Tensile Stress Strain Curves BX-265 / Category 40 20 40 60 80 100 120 00.020.040.060.080.10.120.14 Strain (in/in)Stress (psi) 33 (Local Rise) @ RT Axial & Spray @ RT Spray @ LN2 Axial @ LN2 33 (Local Rise) @ LN2

PAGE 30

17 Figure 2-5. Coordinate system for th e transversely isotropic material Using the coordinate axes from Figure 2-4 we can define the constitutive relations for a transversely isotropic material. The Young s moduli in the 11 and 22 directions are defined as being equal as are the shear moduli in the 33, or rise, direction. The other shear modulus, G12, is determined through a relation with E11 and 12: G12 = 2(1+ 12)/E11. Table 2-1 lists the room temperature values of the elastic constants for the orientation shown in Figure 2-5. Since one of the objectives of this study is to analyze the effect of temperature, or thermal loads, on the K-solution the material properties at vari ous temperatures are needed for a meaningful analysis. NASA ev aluated the foams elastic moduli, thermal conductivity, and coefficient of thermal expansion. 13 12 123 23 21 11 11 123 22 22 3132 33 33 123 23 23 13 13 23 12 12 13 121 000 1 000 1 000 1 00000 1 00000 1 00000 EEE EEE EEE G G G (2-19) where the five independent elastic constants are 33 22 11

PAGE 31

18 112233 2313 1331 1221 31; EEE GG EE (2-20) Table 2-1. Measured material property values from experiment E11 = 950 psi 12 = 0.45 G12 = 328 psi E22 = 950 psi 31 = 0.3 G23 = 231 psi E33 = 2400 p si 1 3 = 0.3 G31= 231 p si These values are plotted in Figures 2-6 through 2-9. The tests performed by NASA indicated that the Poisso n ratios did not vary substantiall y with temperature. As of this writing, only one value for k, the thermal conductiv ity, is available. It is possible for that value to vary with direction as well as temperat ure. It will be assume d, however, that k is isotropic, or has no direction dependency. Young's moduli0 1000 2000 3000 4000 5000 6000 -550-450-350-250-150-5050150250 temperature (F)psi E11 E22 E33 Figure 2-6. Youngs moduli vs. temperature Source: personal communication, Doug Wells (MSFC)

PAGE 32

19 Shear moduli0 100 200 300 400 500 600 -500-400-300-200-1000100200300 temperature (F)psi G12 G13 G23 Figure 2-7. Shear moduli vs. temperature Secant coefficient of thermal expansion0 0.00002 0.00004 0.00006 0.00008 0.0001 0.00012 0.00014 0.00016 -500-400-300-200-1000100200300 temperature (F)in/in/F as11 as22 as33 Figure 2-8. Coefficient of th ermal expansion vs. temperature Source: personal communication, Doug Wells (MSFC)

PAGE 33

20 Thermal conductivity0 0.0002 0.0004 0.0006 0.0008 0.001 0.0012 0.0014 0.0016 0.0018 -500-400-300-200-1000100200300 temperature (F)BTU/hr ft F k Figure 2-9. Thermal conductivity vs. temperature Fracture Testing for the BX-265 Foam Material While many foam materials do not exhibit linear-elastic response before yielding, brittle foams normally exhibit linear-elastic beha vior (while in tension) up until fracture takes place. Thus, we would like to use LEFM c oncepts to compute the near-tip stresses. These analytical expressions assume that we are dealing with a continuum. So if the crack is very small, cell size could influen ce these computations. Brittle fracture of foams has been studied by Fowlkes (1974), McIntyre and Anderton (1978), Morgan et al. (1981), Maiti et al. (1984), and Huang and Gibson (1991) among others. Perhaps the most widely accepted theory for the br ittle facture of foam is the Maiti et al. (1984) study. They present a relation to com pute the toughness for brittle foams as 1/23/2 8*/cfssKICl (2-21)

PAGE 34

21 where C8 is a constant of proportionality and fs is the failure strength of the struts. Equation (2-21) can also be re-written as 3/2 8'*/cCssKICKI (2-22) where KICs is the toughness of the solid material Equation (2-22) assumes that KICs is proportional to fsl1/2 and once more the desired property is related to the relative density. The concern, now, is at what length scale is such an equation valid. Huang and Gibson (1991) analyze the frac ture toughness of brittle honeycombs. The primary aim of their paper is to examin e the effect of short cracks and determine some sort of correction factor for (2-21). They go on to show that cell size does indeed impact the fracture toughness calculations. Using fractographic analysis, Brezny and Green (1991) study short cracks in oxidized carbon foams. Their results dovetail with Huang and Gibsons (1991) in that cell si ze can influence the fracture response. However, in both studies short cracks are used in the various experiments. By short, we mean crack lengths less than an or der of magnitude greate r than the average cell size. But in Huang and Gibsons (1991) paper, the well-known relations based on continuum assumptions seem to be valid as long as the crack length is an order of magnitude larger than the cell size. Thus, if one is considering cracks much greater than the size of the cell la ttice, it seams reason able to model the foam as a continuum. To that end NASA conducted numerous single-edge bending SEN(B), compact tension C(T), and middle tension M(T) tests to examine how the foam fractures under load and used the regular, continuum-based relations to determine fracture properties, such as the plane strain toughness. Pictures of these test specimens are shown in Figure 2-10. Since this material is anisotropic, each time the fracture toughness is calculated,

PAGE 35

22 the materials orientation is supposed to be reported along with it. Figure 2-11 summarizes some of the tested geometries at various temperatures (l eft and right sides of the vertical lines), for a few orientations, denoted by the labels along the bottommost horizontal axis. Figure 2-10. From left to right: M(T), C(T), and SNE(B) fract ure test specimens While NASA utilized many geometries to test the fracture toughness, the C(T) test was the most extensively tested specimen used to estimate a toughness. The plane strain fracture toughness is usually ob tained with slow loading rate s (Barsom and Rolfe, 1999). By plane strain, we mean thick plates, or test specimens, with deep cracks. The foam is brittle and fracture is sudden with little or no stable crack growth. Wells (1966) suggested that the fracture behavior near th e crack tip can be characterized by crack tip opening displacement (CTOD). He showed that CTOD is analogous to the crack extension force (sometimes referred to as Gc in the literature) and CTOD could be related to the plain strain fracture toughness-KIc. If the material is brittle and the subsequent load vs crack mouth opening displacement (CMOD) curve (Figure 2-12) is linear-elastic and the C(T) specimen meets

PAGE 36

23 the standards mandated by the ASTM, KIc can be calculated from this test method and this approach is that NASA used to calcula te the fracture toughness for the foam. In order to determine KIc, a preliminary fracture toughness, KQ, is determined by the following relationship (Anderson, 1991). Where PQ (Figure 2-13) is maximum load appl ied to the C(T) specimen before failure, B is the specimen thickness, W is th e width of the specimen measured from the centerline of the pin holes to the rear of the specimen, and f(a/W) is a dimensionless polynomial function. The ASTM E399 standard denotes which function to use depending on what specimen one is using. If the a/W ratio and B (thickness) are correct per ASTM procedure and as long as Pmax is 1.10PQ then KQ = KIc. Figure 2-11. Summary K-values for various fracture specimens BX265 / Geometry and Constraint Overview / CAT 40.0 5.0 10.0 15.0 20.0 25.0 30.0K (psi(in)0.5) SEN(B)(A-S)M(T)(A-S)C(T)(A-S)M(T)(33-S)C(T)(33-S) RT -320F RT -320F RT -320F RT -320F RT -320F 0.1 < a/W < 0.5

PAGE 37

24 (/)Q QP KfaW BW (2-5) Figure 2-12. Clip gauge on C(T) specimen used to measure CMOD Additional Testing: Divot Test Specimens While NASA was conducting tests to evaluate the material properties at various temperatures, additional experiments were devised to examine how and why the foam becomes detached from the tank. Gibson and Ashby (1988) discuss the design of sandwich panels with foam cores. Thes e types of structural members are normally comprised of two, stiff, skins separated by a lightweight core. These parts are utilized heavily by the aircraft industry, particularly in applications where resistance to bending and buckling loads are important, such as heli copter rotor blades. One mode of failure for these types of structural members is d ecohesion of the adhesive bond between the skin and the core. CMOD Clip

PAGE 38

25 Figure 2-13. Typical load vs CM OD curve for a brittle material Often the epoxy bond is stronger than the core material itself. So if the interface between the skin and core is defect-free, delamination is usually not a concern. The situation changes if there are defects within th e interface, however. This type of analysis is complicated by the difference in elastic constants between skin, adhesives and core. The delamination described by Gibson and As hby can be likened to weakening the bond between the core and the outer skin. When the strength of the bond becomes too weak, the foam core and skin can peal apart. The failure mode that NASA is encountering, however, is more of an explos ive and sudden blowout of foam from the ET. Since the foam loss seems to be the great est near areas where the surface of the tank is somewhat uneven (from bolts and fittings), NASA believes that voids created when the foam is first being sprayed down are the primary reason why the foam is able to break off. These are indeed defects between the ET, or skin, and the foam, or core material. To model this phenomenon, NASA devised an experiment to examine how CMOD Load, P Pmax = PQ

PAGE 39

26 various voids within foam test panels coul d contribute to large pieces of foam being blown out under certain conditi ons. Rectangular test panels like the one shown in Figure 2-14 are used for the divot testing. Figure 2-14. Divot test panel In this experiment a cylindrical bore is mach ined into the panel. At the top of that bore, a sharp notch is created to simulate a crack, or defect, near this void. Liquid nitrogen is then pumped into the bored hole and a heat flux is applied across the other side of the foam panel. Two-dimensional sc hematics of this process are shown in Figures 2-15 and 2-16. The applied flux warms the liquid nitrogen and eventually a phase change takes place. The experiment takes place in a thermo-vacuum chamber to simulate the environment the foam is exposed to during lift off. As the Shuttle ascends to space, the atmospheric pressure is dropping, but the gaseous N2 applies pressure on the walls of the void and the crack faces. With enough force, the flaw turns and propagates toward the surface.

PAGE 40

27 Figure 2-15. 2-D view of divot test In Figure 2-17, we see the aftermath of one of these experiments. A frustum shaped piece of foam has been blown out and the crater left behind resembles a divot, similar to the removal of a small piece of turf after a golf shot has been played. It is here that predicting the crack turning angle could have a practical application. Figure 2-16. 2-D view of test pane l after heat flux has been applied For a given void and flaw size, along with the applied tractions, it might be possible to calculate this angle, denoted as c in Figure 2-16, by calculating near-tip stresses and applying a turning theory such as the maxi mum tangential stress criterion. The concept of the crack turning angle and methods to comput e it will be covered in a later chapter.

PAGE 41

28 Figure 2-17. Foam blow out from divot test Summary The pertinent theory (Gibson and Ashby 1988) regarding cellular solids is presented and various formulas are availabl e depending on what t ype (open or closecelled) of foam one is analyzing. Formulae for anisotropic foams are presented as well. The elastic constants, for both isotropic a nd anisotropic constant s can be estimated through the relations presented in the early s ections of this chapter or via experiment. While there is extensive literature available on foams, NASA decided to perform tests to evaluate the elastic constants and fracture properties at various temperatures. This enabled NASA to have a large collection of experimental data for several material orientations, over a wide range of geometries. Once the tests were conducted, determin ing the fracture properties can be done using the well-known empirical re lations for standard test specimens. One of the main assumptions behind these equations is that th ey are to be applied to a continuum. The foam is a porous material and it has been shown that cell size can impact fracture properties if one is conducting analyses with short cracks. However, continuum

PAGE 42

29 assumptions are acceptable as long as the crack is an order of magnitude longer than the cell length. Lastly, additional and highly specialized te sts were performed to investigate how the foam is able to free itself from the tank. NASA believes that voids near fittings and bolts allow liquefied air to collect in these cav ities prior to liftoff. During liftoff, the air rushing over the outside surface of the insulation warms the liquid nitrogen which converts it into a gas. The pressure, while sm all, appears to be sufficient to drive a crack toward the exterior and this seems to accoun t for the foam loss outlined in chapter one. An idealized case of this process is examined through the divot test. These experiments entail machining oval or cylindric al-shaped voids into a foam test panel and pumping in liquid nitrogen. The cryogen is heated when a flux is applied to the top part of the foam panel. At the top of the void, a sharp notch is inserted and when the nitrogen is turned into a gas, the pressures exerted on the void walls and crack faces are sufficient to drive a crack toward the surf ace. It is here that a dete rmination of the crack turning angle could be of some use to NASA and ther e a few theories preval ent in the literature on this topic. Most of them require the evaluation of near-tip stresses and/or strains. Since this an anisotropic material, the gene ral equations for isotropic materials are not applicable in most cases. In the next chap ter a detailed discussion covering linear, elastic, anisotropic fracture mechanics is co vered and these concepts are applied in chapter four when crack turning theory is presented.

PAGE 43

30 CHAPTER 3 AN INTRODUCTION TO FRACTURE MECHANICS Material Definitions For local crack tip calculations there are ma terial definitions, or classifications, that need to be defined before moving on with discussions on how to compute near-tip stresses for materials that are anisotropic, or have direction-depe ndent properties. In general, most texts on fracture mechanics focu s on materials that are isotropic. Here the constitutive matrix contains three elastic co nstants, two of which are independent, E and For isotropic materials, there are an infini te number of planes of material symmetry and the strains and stresses are rela ted to each other via equation (3-1). 1000 1000 1000 2100 210 21 x x xx yy yy z z zz yz yz x z xz x y xyEEE EE E E E symE (3-1) Many other types of materials can be char acterized by the number of planes of internal symmetry (Dieter 1976). Cubic materi als, for example, have three independent material parameters and nine pl anes of symmetry. Materials with three internal planes of symmetry are known as orthotropic materials and many engineering metals fall into this category, such as rolled aluminum. A special class of orthotropic materials has a single plane of symmetry that is also isotropic. These are known as transversely isotropic materials and the constitutive relations for this material were defined in the previous

PAGE 44

31 chapter. Monoclinic materials possess a si ngle plane of material symmetry, but the behavior in that plane is not isotropic. Being cognizant of the constitutive relations for the particular material in question (be it isotropic or fully anisotropic) is importa nt because many of the initial derivations of near-tip stress fields for an isotropic materials tried to take advantage of material symmetry to decouple in-plane, and out-of-plane displaceme nts, which, in turn, makes the mathematical formulati on a little less complicated. Isotropic Fracture Mechanics For isotropic materials, the near-tip stress fields have been analyzed by Westergaard (1939), Sneddon (1946), Williams (1957), and Irwin (1960) among others. These solutions pertain to specific cracked conf igurations with applied far-field loads and are governed by a single parameter, K, the stress intensity factor (Anderson, 1991). (,)()higher order terms 2 ijijK rf r (3-1) Figure 3-1. Coordinate sy stem used in equation (3-1) The coordinate system for equation (3-1 ) is shown in Figure 3-1, where the fij term is a trigonometric function. We note the Greek lette r, theta, is normally used to define the angle with the horizontal in the polar coordina te system presented above, but theta will be

PAGE 45

32 used later on to define the material orientation. A different variable name is selected to avoid confusion. The key points about (3 -1) are that the near-tip stresses have no dependence on the elastic constants and the first term is proportional to 1/ r and, therefore, the stresses approach infinity as r is made smaller and smaller. Also, as r approaches zero the higher order terms in (3 -1) get smaller or remain at some finite value. Thus, in many texts and papers the analysis is restricted to small values of r such that the higher order terms are neglected. There is typically a roman numeral written next to the stress intensity factor to denote the mode of loading and Figure 3-2 displays the three modes of deformation. Figure 3-2. From left to right: mode I ( opening), mode II (shearing), mode III (tearing) While the stresses are proportional to 1/ r, the displacements are related to r-equation (3-2). 22 x 22 y s zcos -1+2sinsin +1+2cos0 2222 u 1r u=sin +1-2cos-cos -1-2sin0 22 2222 u 00sin 2 KI KII KIII (3-2)

PAGE 46

33 where = 3-4 for plane strain and = (3)/(1+ ) for plane stress and s is the shear modulus. Linear Elastic Fracture Mechan ics for Anisotropic Materials In general, the equations in the previous section cannot be applied to this problem directly because the material properties are direction-dependent. The analytical work geared at developing the neartip stress fields for elastic anisotropic materials has been examined by Sih et al. (1965), Bowie and Freese (1972), Barnett and Asaro (1972), Bogy (1972), Tupholme (1974), Kuo and Bogy (1974) Rathod (1979), Hoenig (1982), and Dhondt (2002) among other researchers. The important work of Sih et al. is one of the most cited references for determining the stress fields near the crack tip for anisot ropic materials. Following Westergaards use of stress functions and complex variables to determine near-tip stresses for isotropic materials, Sih et al. use a similar approach for materials with direction-dependent properties. Whenever stress f unctions are utilized to solve this kind of problem, there is normally a corresponding charac teristic equation (a deta iled explanation of the characteristic equation can be found in Appe ndix A). For the most general case, the order of the characteristic equation is six. In Sih et al. s paper the material in question is monoclinic and the crack is resting in the symm etry plane. The constitutive relations for monoclinic materials are s hown in equation (3-3). 11121316 222326 3336 4445 55 6600 00 00 0 0xx yy zz yzyz x zxz x yxySSSS SSS SS SS S symS (3-3)

PAGE 47

34 This configuration is advantageous because the in-plane and out-of-plane displacements become decoupled, which in turn reduces the order of the characteristic equation to four. 22 122121 12211221 12 12211221 12 121 ReRe0 1111 ReRe0 1 Re 2xx yy xy xz yzQQQQ QQQQ r 12 121212 3 3 3111 Re0 00Re 1 00Re KI KII QQQQ KIII Q Q (3-4) 31323633 zzxxyyxySSSS (3-5) where the elastic constants in compliance form are defined as iijjS (3-6) 1222112211 1212 1222112211 1212 3 4534411 ReRe0 211 ReRe0 00Rex y zpQpQpQpQ u KI r uqQqQqQqQKII u KIII Q CC (3-7) The roots 1, 2, and 3 for this particular formulation come from the special case for monoclinic materials listed in Appendix A. 2 111612'''jjj p SSS (3-8) 122622'''jjjqSSS (3-9) Sih et al. enforce a plane strain condition and the reduced compliance matrix becomes

PAGE 48

35 33 33 ij ijijSS SS S (3-10) cossiniiQ (3-11) The C45 and C44 terms in (3-7) are the elastic constants which are in stiffness form iijjC (3-12) where 1 ijijCS (3-13) The aforementioned decoupling of the anti -plane displacement is now evident in equation (3-7). The uz displacement does not depend on KI and KII. This is similar to the isotropic case in equation (3-2); the z component of displacement only depends on KIII. When analyzing the foam problem for NAS A, a crack, or defect, can be at any orientation relative to the insulation. In th at sense, it does not seem very practical to employ the Sih et al. solution unless we know the crack always lies in the symmetry plane. More general solutions, ones that do not require a decoupli ng of the anti-plane displacements, will be used instead. The asymptotic solution derived by Hoenig is the one we will adopt to get the neartip stress fields. Materials with virtually a ny orientation, irrespective of the location of the crack front, can be modeled with this me thod. Hoenig uses the pioneering work of Lekhnitksii (1963) to derive the stresses and displacements given below by using complex variables and stress functions to deri ve a coupled set of differential equations whose solution entails determining the roots of the ensuing sixth order characteristic

PAGE 49

36 equation. We notice that K once again arises as a scalar multiplier just as it did for the isotropic solution. 21 3 1 1 3 1 1 3 1 1 3 1 11 Re 2cossin 1 Re 2cossin 1 Re 2cossin 1 Re 2cossin 1 Re 2cossiniijj x i i ijj y i i iiijj xz i i iijj yz i i iijj xy iNK r NK r NK r NK r NK r 3 1 i (3-14) 313234353633 zzxxyyyzxzxySSSSSS (3-15) 3 1 12 Recossiniijjllj jr umNK (3-16) Here an important distinction is made be tween the general case and Sih et als formulation. Notice in (3-16) how the mode I-III SIFs are to be summed over all three modes of displacement. Thus, all three components of displacement are dependent on KI-KIII. 2 11116121514 2224 2212625 4244 3414645''''' '' ''' '' '''iiii iii ii iii iimSSSSS SS mSSS SS mSSS (3-17)

PAGE 50

37 The i are the roots that occur in conjugate pa irs and they arise from the sixth order polynomial, equation (3-18). The coordinate sy stem used in equations (3-14) and (3-16) is the same as the one used in (3-1). 2 4230 lll (3-18) Where 2 2554544 32 3151456254624 432 41116126626222 222 lSSS lSSSSSS lSSSSSS The N matrix in equation (3-14) is defined as 123 123111ijN (3-20) Where i is defined as l3(i)/ l2(i). There are two situations, also called degenerate cases, where this solution is invalid, however. If anti-plane shear and pl ane strain displacements becomes decoupledfor monoclinic materials, say-Hoenigs form ulation falls apart. He does present an alternate solution for monoclinic materials which turns out to be consistent with the Sih et al solution mentioned earlier. His solution also encounters difficulties if the crack is lying in a plane that is isotropic. Th e N-matrix, equation (3-20), becomes singular because 1 and 2 = i. However, we have near-tip solutions for the isotropic cases. These are shown in equati ons (3-1) and (3-2). Determining the Stress Intensity Factor K In the above sections, it is necessary to develop solutions for determining the stress intensity factor, K. Over the past 50 years, numerous methods have been developed to

PAGE 51

38 determine this parameter. For simple geometries, a handbook (Tad a et al, 1973) of Ksolutions could be consulted. Alaibadi a nd Rooke (1991) present an excellent summary of the literature on the subject of numerical ev aluations of SIFs. They divide many of the available methods into three main categor ies, denoted as stages in Table 3-1. Table 3-1. Various methods for determining the SIF Stage 1 Stage 2 Stage 3 Handbooks Superposition Collocating, or mapping Stress concentration Integral transform Stress distribution Body force method Greens function Method of lines Approximate weight function Finite element method Compounding Boundary element Alternating technique As one might have guessed, the procedures listed in the Stage 3 category are the most accurate. Since computers have become less expensive and more powerful in recent years, many of the current studies geared to ward K-solutions incorporate the use of the finite element (FE) method or boundary elem ent (BE) techniques. Boundary element methods are particularly attractive because the dimensionality of the numerical problem is reduced, i.e. 2-D problems involve discre tizing the line-boundary of the domain and for 3-D problems, just the surface of the domain is discretized. This analysis will call on the FE method to help compute K, and there are a few ways of determining it with this method, such as: extrapolation of near-tip stress and displacement fields, Rices Jintegral (1968), strain ener gy, and the virtual crack extension method. FRANC3D Next Generation Software The Cornell Fracture Group has developed software capable of creating crack meshes and computing an anisotropic K-solu tion. This particular method of solution would fall under Stage 3 in Table 3-1, and in this case Rices J-contour integral along

PAGE 52

39 with displacements output from FE software will be used to determine K along the crack front. This software is very versatile in that it is not restricted to analyzing any one kind of material; K-solutions can be generated fo r isotropic and fully anisotropic materials alike. Since many FE models were built and the subsequent K-solutions computed with this software, it would be prudent to discuss the pertinent theo ry (Banks-Sills et al, 2005) that governs this particular way of calcula ting K. Rices J inte gral, a conservation integral that measures the energy flux into th e crack tip, is equivalent to the strain energy release rate (denoted as G) for small scale yielding. Furthermore, it can be shown for a plane problem that only has KI and KII that 22()GJKIKII (3-21) Where = (1-2)/E for plane strain and = 1/E for plane stress. One can see that the Jintegral equation, as is cannot give us the SIFs on an individual basis. i iu JWdyTds x (3-22) with W = (1/2)ijij, also known as the stra in energy density, T = ijnj and ui is the displacement vector. The path, is defined in Figure 3-3 Figure 3-3. Path of the J-integral

PAGE 53

40 The J-integral was initially formulated for 2-D problems, however since this material has properties that ar e direction dependent, our analys is requires the use of 3-D FE models. The J-integral, then, needs to be modified so it can be used in integrations that take place over volu mes. Following Li et al (1985), th e J-integral can be rewritten as ds x q W x u Jj j i ij 1 1 (3-23) In this formulation, the crack is undergoing a virtual extension, and only a small portion of the crack front is being a dvanced. When using the formulation from the Li et al paper, the J-integral is now evaluated in a cylindrical domain centered over the crack front. Figure 3-4. Cylindrical do main and virtual crack advance for a 3-D crack front Following Figure 3-4 we can define increm ental area of virtual crack advance as max q LAaqsds (3-24) The definition of J in equation (3-22) requires to be very small. It is difficult to compute stresses and strains about a vanishingl y small path about the crack tip. Instead,

PAGE 54

41 area integrals (Moran and Shih, 1987) are us ed (for 2-D problems) because they are easier to implement numerically. So, the q term in (3-23) is simply a mathematical tool that enables us to recast the J-integr al into a slightly more usable form. This virtual extension method is the mode rn-day approach to solve computational fracture mechanics problems. One of the main advantages of such a method is the ability to compute the energy release rates across th e crack front for 3-D problems. Since the energy release rates can vary over the front that implies K can also change with respect to crack front position as well q t tA J ds s q ds s q s J J ) ( ) ( ) ( (3-25) where qt(s) is the crack tip perturbation. Rices J-integral does not allo w us to look at the SIFs on an individual basis for a mixed mode crack problem. This hurdle wa s first surpassed by Ishikawa (1980) where he decomposes the stress, strain, and disp lacement fields into symmetric and antisymmetric parts. This allows him to use th e J-integral to comput e the Ks individually. Ishikawas method, though, has not been exte nded for anisotropic materials. Yau et al. (1980) uses the M-integral developed by Freund (1978) to determine KI and KII, individually, by using the idea that two separate and inde pendent equilibrium states (denoted by superscripts (1) and (2)) for a linea r elastic, isotropic material can be related to a third equilibrium state, denoted with no superscript. That is to say the stresses, strains, displacements, and SIFs from these sepa rate equilibrium states can be superposed (1)(2)(1)(2) (1)(2)(1)(2) (1)(2)(1)(2) ijijij ijijij iiiKIKIKI KIIKIIKII uuuKIIIKIIIKIII (3-26)

PAGE 55

42 The J-integral can now take the form (1)(2)(1,2)JJJM (3-27) The M(1,2) term is known as the M-inte gral initially derived by Freund and it can be recast in terms of the aforementioned equilibrium states (2)(1) (1,2)(1,2)(2)(1,2) 1, 11 ii ijijj juu q MW x xx (3-28) This integral describes the interaction between the two equilibrium states, and sometimes the M-integral is called the interacti on integral in the literature. The first step in the procedure to determine the SIFs is to define the strain energy release rate using Irwins (1957) crack closure in tegral. Consider the crack in Figure 3-5. Now a compressive, or closure stress, is applied in such a fashion to clamp the crack faces down along the length -r. The work it takes to perform this closure can be related to the energy release rate, G. Irwin goes on to use the force displacement curve and equate that to the work required to close th e crack. Substitutions are made for near-tip displacements and stresses and eventually an integral expression is obtained that relates the energy release rate to the stress intensity factor. In a similar manner, Hoenig (via Equa tion 3-30) provides us an analytical expression for the energy release for general an isotropy in terms of the SIFs and material constants. Irwins work was initially developed for brittl e, isotropic materials; his expression for the energy release rate includ es the isotropic Poi sson ratio and the Young and shear moduli. 0 01 lim(,0)(,) jjGrurdr (3-29)

PAGE 56

43 Figure 3-5. Definition of Irw ins crack closure integral Now, if the stress and displacement fiel ds are substituted in (3-29), we have 111 2131 Im()Im()Im() 2iijjiijjiijjGJKImNKKIImNKKIIImNK (3-30) When we use the relations for SIFs in (3-26), we get (1)(2)1(1)(2)1 21 (1)(2)1 3()Im()()Im() 1 2 ()Im()iijjiijj iijjKIKImNKKIIKIImNK GJ KIIIKIIImNK (3-31) The terms in (3-27) have bars over them because a certain definition of J is used in (323). Let us define an alternate form of (327) where J is not normalized with respect to the extended area as (1)(2)(1,2)JJJM (3-32)

PAGE 57

44 Now we substitute (3-31) into (3-32) and for clarity the individual terms of that equation are listed below (1)(1)1(1)(1)1(1)(1)1(1) 213 (2)(2)1(2)(2)1(2)(1)1(2) 2131 Im()Im()Im() 2 1 Im()Im()Im() 2iijjiijjiijj iijjiijjiijjJKImNKKIImNKKIIImNK JKImNKKIImNKKIIImNK (3-33) (1,2)(1)1(2)(2)1(1) 1 22 2 (1)1(2)(2)1(1) 11 (1)1(2)(2)1(2) 33ImIm ImIm ImImiijjiijj iijjiijj iijjiijjMKImNKKImNK KIImNKKIImNK KIIImNKKIIImNK (3-34) It is now evident how the decomposition of displacements, strains, and stresses into separate equilibrium states is helpful. One can see that we have two definitions of the Mintegral, equations (3-28) and (3-34). If they are equated, the individual SIFs (KI(1), KII(1), KIII(1)) can be computed. But before that step, we must define auxiliary solutions for the second equilibrium state, de noted by superscripts (2a, 2b, 2c). The terms with the (1) superscript are going be computed numerically via FE software. The three auxiliary solutions are define d as: 2a, 2b, and 2c. For case 2a, KI(2a) = 1 and KII(2a) = KIII(2a) = 0. For case 2b, KII(2b) = 1 and KI(2b) = KIII(2b) = 0, and for case 2c, KIII (2c) = 1 and KI(2c) = KII(2c) = 0. Since the SIFs are being prescribed beforehand, the stresses, for those equilibrium states, can be computed from (3-14). Now the two relations for the M-integral are equated (1)1(2) 2 (2)1(1) (2)(1) 2 (1)(2)(1,2) (1)1(2) 1 1 11 (2)1(1) 1 (1)1(2) 3 (2)1 3Im() Im() Im() 1 2 Im() Im() Im(iijj iijj ii ijijj iijj j q iijj iijj iijKImNK KImNK uu q Wds KIImNK xxx A KIImNK KIIImNK KIIImNK (2))j (3-35)

PAGE 58

45 From (3-35) we can assemble a system of equations that contai n the three unknown SIFs we are after 11111 (1)(1,2) 2111223123 11111(1)(1 2211123213 (1) 11111 23311332332ImImIm Im2ImIm ImIm2Ima iiiiiiiiii q iiiiiiiiii iiiiiiiiiimNmNmNmNmN K IMA mNmNmNmNmNKIIM KIII mNmNmNmNmN ,2) (1,2) b q c q A M A (3-36) Finally, we note that (3-36) i nvolved the extraction of stresse s near the crack tip. These stresses are taken from the Gauss point locati ons within the elements that encompass the crack front. Also, the integral in (3-36) is evaluated numerically using Gauss quadrature. Computing SIFs Using Displacement Correlation Displacement correlation (DC) is perhap s the oldest and simplest method to compute SIFs via FE displacements. Chan et al (1970) were among the first group of researchers to employ such a method to obtain SI Fs this way. In general, the correlation point is selected where the crack tip displacements are the larg est so the relative error in displacements is small. Another advantag e in using such a method is the SIFs are separated into the three modes defined in Figure 3-2. In it s most basic form, crack tip opening displacement (CTOD) can be used to determine the SIF by equating the displacements at the points in questi on to some analytical solution. From equation (3-2) with = and Figure 3-6 for a plan e strain problem dealing with an isotropic material, we have 2 2 1ybyasuu KI r (3-37) 2 2 1xbxasuu KII r (3-38)

PAGE 59

46 where s, again, is the shear modulus determined by E/2(1+ ) and = 3-4 for plane strain We note that uy and ux are the opening and sliding m odes of displacement. The above method is sometimes called a onepoint matching techni que since it involves extracting displacements from one point near the crack tip. The downside of such a Figure 3-6. Correlation poi nts typically used for displacement correlation method is that very refined meshes are needed for accurate results. Improved accuracy is obtained through the use of a special elem ent specifically designed to capture the singularity present at the crack tip. From equation (3-1), we see how the stresses have a 1/ r singularity at the crack tip. For accurate computations, we would like our FE models to also have this singularity. Quarter point elements are one way to do this. The first attempts at modeling cracks were done so using quadril ateral elements with the mid-side nodes moved to quarter point locations (Barsoum, 1976 ) When this is done the desired strain singularity is achieved. Howeve r, one downside of using this type of element is that the singularity is only present on the edges that contain the quarter point. If triangularshaped elements are used instead, the singul arity exists both along the edges and within the element (Anderson, 1991). In Figure 3-7, we see how an eight node quadrilateral element can be transformed to a triangle and then cert ain nodes are moved to the quart er point locations. Let us

PAGE 60

47 derive how the required strain singularity is obtained along th e edge of the quadrilateral element with the mid-side nodes moved to th e quarter point locations. Consider the bottom edge of the element on the right in Fi gure 3-8. Nodes one, five, and two lie along this boundary. In order to show how the stra in singularity comes about, we will need to list the shape functions for this particular element. Th e shape functions (Logan, 2002) for the isoparametric element at the corner nodes are 1 111 4iiiiiN (3-39) where i is the number of the shape functions at th e corresponding node (i.e. N1 is the shape function for node 1) and 1,1,1,1 (1,2,3,4) 1,1,1,1 (1,2,3,4)i ii i (3-40) We also have shape functions for the mid-side nodes 2 21 11 1,1 (5,7) 2 1 11 1,1 (6,8) 2iii iiiNti Ni (3-41) The global origin is placed at the corner of the element. We will n eed the shape functions at nodes one, two, and five. Setting = -1 we have 1 2 2 51 1 2 1 1 2 1 N N N (3-42)

PAGE 61

48 Figure 3-7. Triangular quarter point elements Figure 3-8. Isoparametric element and dege nerated element with mid-side nodes moved to quarter point locations The global x coordinate is related to th e isoparametric coordinate system via ii x Nx (3-43) where the repeated index is an implied summation. Setting x1 = 0, x2 = L, and x5 = L/4 we have 21 11 24 L xL (3-44)

PAGE 62

49 We can now solve for 21 x L (3-45) In the same way that we can relate the globa l x coordinate to the shape functions, we can also determine the displacements using the shape functions iiuNu (3-46) Again determining the displacements along th e edge containing nodes one, two, and five we write 2 12511 111 22 uuuu (3-47) Now we can substitute (3-45) into (3-47) to obtain 1251 12221224 2 xxxxxx uuuu LLLLLL (3-48) We note that the displacement shown in (3-48) has a x term present. This is consistent with equation (3-2) where the displacements vary with r. We need to differentiate equation (3-48) to obtain the strain in the x direction 12513411424 22xu uuu xLLL xLxLxL (3-49) Finally, it is shown in (3-49) how the strain disp lays the desired singularity (x-1/2). One can see why the quarter point element ha s been the standard for modeling cracks for the past 30 years. No special programming, or internal tampering, with the FE code is required. Any FE program that carries quadr atic, or higher order, elements can support these special ones with the nodes moved to the quarter points.

PAGE 63

50 Let us now return to our discussion of using displacement correlation to extract SIFs from FE models. If quarter point elemen ts are used, equations (3-37) and (3-38) are modified because r = L/4 4 2 1ybyasuu KI L (3-50) 4 2 1xbxasuu KII L (3-51) The SIFs extraction via the DC method need not be restricted to just one set of nodes, or correlation points. One can also consider th e displacements over the whole element. The form of these equations depends on the shape functions of the elements. Ingraffea and Manu (1980) derive a DC extraction method usi ng 3-D elements. This formulation is particularly useful to the present study si nce 3-D simulations are used to model the fracture response of the foam material. Ingraffea and Manu use a special labeling convention shown in Figure 3-10 for a collapse d 20 node brick element (Figure 3-9), also known as a 15 node wedge. Figure 3-9. 20 node brick and 15 node wedge elements The shape functions for a 20 node brick el ement are listed below. The local ( , ) coordinate system is placed at the center of the brick element. For the corner nodes at positions i = 1,2 and i, i, i = 1

PAGE 64

51 111 2 8iii iiiiN (3-52) For the nodes at positions i = 17, 18, 19, 20 and i = 0, i = i = 2111 4ii iN (3-53) At i = 10, 12, 14, 16 and i = i = 0, i = 2111 4ii iN (3-54) Finally, for the nodes at positions i = 9, 11, 13, 15 and i = i = i = 0 2111 4ii iN (3-55) For anisotropic materials, Saouma and Siki otis (1986) extend Ingraffeas work for anisotropic materials. They use the displacement relations developed by Sih et al A similar procedure is followed in this study, but instead of using displacement equations from Sih et als paper, the relations in (3-1 6) are to be employed. The equations to determine the anisotropic Ks, using Hoenig s displacement equations, are as follows 12 KI KIIBA L KIII (3-58) where [B] is 1 1 1 2 1 3cossin []Recossin cossinjjlj jjlj jjljmN BmN mN (3-59)

PAGE 65

52 Figure 3-10. Collapsed 20-node brick element ''''''''' 2 ''' ''''''''' 2 '''1 22224444 2 1 22 2 1 22224444 2 1 22 2 2BCEFBCEFDBCEFBcEF FCDFCD BCEFBCEFDBCEFBcEF FCDFCD Bvvvvvvvvvvvvvvvvv vvvvvv uuuuuuuuuuuuuuuuu uuuuuu w ''''''''' 2 '''1 2224444 2 1 22 2CEFBCEFDBCEFBcEF FCDFCDwwwwwwwwwwwwwwww wwwwww (3-57) Once again, the repeated indices are an implied summation. The terms in the B matrix only depend on the roots of the sixt h order characteristic equation. Effects of Temperature on the SIF Solution In chapter two it is pointed out that temp erature plays no small role in causing the foam to break off from the ET. For example, when Shuttle is resting on the launching

PAGE 66

53 pad, one side of the foam is exposed to the temperature of li quid hydrogen and oxygen; roughly -300oF. The other side of the foam is exposed to air, perhaps at 75 oF prior to liftoff. Over typical acreage sp rays, the foam is approximately three inches thick. This small thickness coupled with a large thermal gradient suggests that thermal strains could be significant. As shown with the divot tes ting in chapter two, even though there are no applied far-field mechanical loads, the therma l gradient seems to be sufficient enough to impart stresses capable of driving a crack to ward the surface of the ET resulting in a loss of material. To account for thermal strains, we simply a dd this term to our constitutive matrix. Following Hyer (1998), the stress-stra in relations can be written as 1111 11 111213 22 2222 212223 33 313233 3333 44 23 23 55 13 13 66 12 12(,) 000 (,) 000 000 (,) 00 0Th ref Th ref Th refTT SSS TT SSS SSS TT S S symS (3-60) where the total strain is { 11 22 33 23 13 12}T, Th 11(T, Tref) is the free thermal strain in the 11-direction, and Tref is the reference temperature. The mechanical strains are 111111 22 2222 33 3333 23 23 13 13 1212(,) (,) (,)mechTh ref mech Th ref mech Th ref mech mech mechTT TT TT (3-61) or, stated more concisely the total strain is total = mech + thermal (3-62)

PAGE 67

54 If the thermal expansion is linear with the te mperature change, we can rewrite (3-60) as 11111 111213 22222 212223 33333 313233 44 23 23 55 13 13 66 12 12000 000 000 00 0 T SSS T SSS T SSS S S symS (3-63) The i term is the coefficient of thermal expansion and T is the difference of temperature between the state of inte rest and the reference state. The above equations allow us to compute th ermal strains in the numerical models. Incorporating thermal and mechanical loads will involve two sepa rate steps because a thermal gradient implies that a temperature distribution must be obt ained before solving the model that contains both the thermal a nd mechanical boundary conditions. The first step is to solve the conduction problem usi ng the thermal conductivity coefficient listed in chapter two. The nodal temperature distributi on is saved for later use. In a subsequent run, now with mechanical loads, the nodal temperatures are carried over as body forces along with any applied mechanical loads. Summary Summing up, the analytical expressions from Hoenigs near-tip solution is presented. His equations depend solely on the roots of the characteri stic equation and the stress intensity factor, K Hoenigs solution is not as well known as Sih et als paper and as such the vast majority of studies dealing with near-tip solutions for cracked anisotropic bodies use the Sih et al formulae to compute the stresses. This is acceptable as long as one realizes those particular equations are valid for cert ain configurations where a decoupling of in-plane and out-o f-plane displacements takes pla ce. Hoenigs solution is

PAGE 68

55 more general and his formulation of the energy release rate is the one chosen to compute the SIFs within the FRANC3D software. The DC technique presented also uses Hoenigs equations, but only the ones that pertain to displacements. These two methods, interaction integral and DC, are presented to show how the mode I-III SIFs can be computed. These fo rmulations are different in a few ways. Computing K with the J-integral is inherent ly more accurate because only a moderate level of mesh refinement is needed for th e solution. However, the implementation of such a method is very involved. Alternativel y, the DC approach is much easier to apply but the accuracy of the solution does depend on the level of mesh refinement. Finally, the relations dealing with handling thermal strains are presented. These equations assume an expansion that is linear with temperature change. Handling thermal loads is straightforward by first solving the conduction problem to obtain the nodal temperature distribution. With that, those values are forwarded as body forces that are applied alongside the mechanical loads.

PAGE 69

56 CHAPTER 4 CRACK TURNING THEORY AND FINITE ELEMENT MODELING Crack Turning Theory There are three theories preval ent in the literature to pr edict incipient crack turning angles for isotropic materials: maximum energy release rate, minimum strain energy density, and maximum hoop stre ss. All of these theories are based on LEFM concepts and were initially deve loped from experiments involving brittle plates. The turning theory proposed by Hussain et al. (1974) seeks the direction where the energy release rate, G will be a maximum. The minimum strain energy density theory (Sih, 1974) postulates crack turni ng in the direction where the st rain energy density is at a minimum. The maximum hoop stress theo ry proposed by Erdogan and Sih (1963) predicts crack propagation in the direction where, defined in equation (4-1), is a maximum relative to the crack tip. Crack turn ing, in this study, denotes the incipient turning angle. We define the crack turning angle, c, below in Figure 4-1. For isotropic materials, the ne ar-tip stresses (in the r, polar coordinate system) have been derived by Williams (1957). If the derivative with respect to theta of equation (4-1) is taken and set to zero, the critical angle, c, can be determined; equation (4-4). 213 coscossin 222 2 KIKII r (4-1) 213 cos1sinsin2tan 2222 2rKIKIIKII r (4-2)

PAGE 70

57 Figure 4-1. Definition of crack turning angle 1 cossin3cos1 2 22rKIKII r (4-3) 2 1118() 2tan 4()cKIIKI KIIKI (4-4) The maximum hoop stress theory is the easies t to implement and th at is perhaps the main reason why it is widely used for turni ng angle predictions. The minimum strain energy density criterion is also very popular and there is some debate (Gdoutos, 1984) as to which one is superior. Some might argue that it is not correct to use just a single component of stress to predict the incipient turning angle, whereas a quantity such as strain energy density seems to be more co mprehensive as all components of stress and strain are included. Ma ccagno and Knott (1989, 1992) study mixed-mode facture behavior of PMMA and lightly tempered steel alloys and in both cases the crack turning angles are well approximated by the maximum tangential stress theory. There have been some topical investigations for crack tu rning within materials that have direction dependent prope rties. Buzcek and Herakovi ch (1985) pred ict the crack

PAGE 71

58 extension angle for orthotropic ma terials and some of the recent work in this area mirrors their ideas. They assume the tensile strength, T of the orthotropic material varies with direction, Since it is difficult to measure the strength of a material in all possible orientations, they assume that T = f( , XT, YT) where is the angle in a polar coordinate system, is the material orientation and XT and YT are the strengths of the material in the axial and fiber directions respectively (they are analyzing orthotropic composites). The equation for T, also denoted as a fracture resistance parameter, must be independent of if the material is isotropic. The lo west order function th at satisfies these requirements is an ellipse. This f unction is plotted in Figure 4-2 in (T, ) polar coordinates. Now T can be expressed as 22sincosTTTXY (4-5) Figure 4-2. Elliptical relationship for T The crack will turn, according to Buzek a nd Herakovich, in the direction where the ratio of the hoop stress to T is a maximu m. The hoop stress, sometimes called the tangential stress (or ), is defined in Figure 4-3. Consider a point P and a vector

PAGE 72

59 connecting that point to the crack tip. Th e hoop stress is normal to this vector that connects P to the crack tip. Figure 4-3. Definition of hoop stress Mixed mode crack propagation in anisotr opic materials is also investigated by Saouma et al. (1987). The near-tip stre sses are obtained via Sih et al. (1965) and in their analysis the maximum tangential stress theory is used to predict the angle of propagation. Saouma et al. use a slightly different criterion for cr ack turning, but it is of similar flavor to what Buzek and Herakovich propose. Instead of using strength as a fracture resistance parameter, the plane strain fracture toughness is utilized. Boone et al. 1987) use FE models to study cr ack propagation in orthotropic materials. The idea of maximizing the ratio of the hoop stress to some strength parameter is once again utilized to pred ict the crack turning angle. Chen (1999) analyzes crack propagation in aircraft fuselage s. He, too, makes use of a ratio of the hoop stress to the plane strain fracture toughness. In this case, the material in question has isotropic elastic moduli, but the fracture properties are dire ction dependent. A similar analysis is conducted by Pettit (2000). He examines crack turning in rolled aluminum, which again, has isotropic stiffness properties but an isotropic fracture characteristics.

PAGE 73

60 Carloni et al. (2003) and Nobile and Carloni ( 2005) consider incipient crack turning for an orthotropic pl ate under biaxial loading. Th e former researchers utilize essentially the same fracture resistance pa rameter as Buzek and Herakovich, whereas the latter study uses a slightly m odified version of the Saouma et al. turning criterion. Carloni et al. and Nobile and Carloni also call on complex variables and stress functions in their formulations for near-tip stresses. Both studies, however, make assumptions that reduce the order of the characte ristic equation that arises in such derivations (see chapter three), down to order four. While Pettits analysis is for a material with isotropic sti ffness properties, he generalizes the fracture resistance parameter to three dimensions. His analysis accounts for an arbitrary location of the crack with respect to the mate rial. Thus, we would like to use the method to obtain stress inte nsity factors given by Banks-Sills et al. (2005) in conjunction with Pettits formul ation for a fracture resistance parameter. This approach is a bit more generalized than the aforementioned studies because their work seems to be confined to specific configurations that us ually decouple the in-p lane and anti-plane displacements. Pettits method for determining the fracture resistance is formulat ed for orthotropic materials which possess cubic symmetry, i.e. three orthogonal planes of symmetry and within each plane lay two principal toughness values resulting in six principal toughness values: K12, K21, K23, K32, K13, and K31. The first number in that nomenclature is the vector normal to the crack plane and the s econd number corresponds to the direction of propagation. It is difficult to obtain a toughness for all possibl e orientations for materials that exhibit some form direction dependence. Essentially we are interpolating between

PAGE 74

61 the six principal toughness values for the orthotro pic material. This gives us an estimate of the toughness in any direc tion we desire. Consider Fi gure 4-4 and the material orientation as shown. What we see in that figure is block of foam with various M(T) specimens oriented within it. If the M(T) specimens are aligned with the principal axes of the large block of foam, th ere are six possible orientati ons and with each orientation there is a corresponding toughness. Pettits interpolation equation, then, requi res six unique toughness values for the orientations shown in Figure 44. In our case, only thr ee toughness values are needed, however. In Figure 4-4, consider th e M(T) specimens that have the K31 and K32 labels. Here the load is applied in the 3-directi on and the crack is propa gating in the 1 and 2directions, respectively. Since the 3-direction is the rise di rection and the 1-2 plane is a plane of isotropy, it is re asonable to assume that K31 = K32. That is to say, the crack is resting in the plane of the knit line for thos e two cases. For the other four cases, the fracture plane is normal to the knit line plane. Thus, K21 = K12 and K13 = K23. So the total number of required toughness values ha s been reduced to three and NASA did conduct tests, at room temp erature, to obtain them: K32 = 22.4 psi in, K12 = 17.4 psi in, and K23 = 19.5 psi in. Pettits equation that will determine the e ffective K needs severa l variables besides the six principal toughness values. Remember the goal here is to try and predict the direction of propagation. We can convenientl y describe this direction with a vector, denoted as a. That vector lies in a plane tangent to the fractu re surface, whose normal vector will be defined as n. These two vectors are shown in Figure 4-5.

PAGE 75

62 Figure 4-4. Orthot ropic toughness values Figure 4-5. Definition of the a and n vectors To develop the equation for the effective K, we first define the trace angles of the vector a makes with the principal planes. These are 3 12 123 231tan tan tan a aa aaa (4-6) The subscript in the theta term denotes the axis normal to the principal plane. Using equation (4-5) we can write a 2D in terpolation function of the form

PAGE 76

63 22cossinkkkikkjkKKK (4-7) We also observe the following trigonometric identities 2 21 22 2 21 22costan sintan bc cbc bb cbc (4-8) and the components of unit vector a must satisfy the relation 222 1231 aaa (4-9) Pettit goes on to define equa tions that can be interpreted as the fracture resistance components of a in the principal planes. Consider Figure 4-4 once more where the 1-axis is the loading direction. To estimate an effective K on this plane, we will interpolate between two toughness values, K12 and K13. Using (4-7) through (4-9) we can write K1 as 22 1122133 2 11 () 1 KaKaKa a (4-10) In a similar fashion we can determine K2 and K3 22 2233211 2 2 22 3311322 2 31 () 1 1 () 1 KaKaKa a KaKaKa a (4-11) Finally, Pettit sums the relations in (4-10) th rough (4-11). The effective K, denoted as Kp, can now be written as 2 22222 1 112233122133 2 1 2 2 2222 3 2 233211311322 22 23(a,n) 1 11pn KKnKnKnKaKa a n n K aKaKaKa aa (4-12)

PAGE 77

64 FE Modeling of the M(T) Specimen While the importance of crack turning was made evident by th e divot test, this experiment is highly specialized and is not recognized as a standard fracture test specimen. A more conventional way of anal yzing crack turning can be performed with the middle tension, M(T), test specimen. Material orientation, mode mixity, and boundary conditions can all be carefully controlled with the M(T) specimen. The idea behind creating FE models of M(T) specimens is two-fold. One of the objectives of this study is to examine the e ffect of mode mixity on the K-solution, or stated differently, as the material orientat ion is changed how does this impact KI-KIII along the crack front? Also, several foam M(T) specimens were fabricated and the resulting crack turning angles measured. Nu merical predictions, ba sed on Ks extracted from the FE models are used to make pr edictions via the maximum hoop stress theory discussed earlier. M(T) FE models are built using ANSYS FE software. Crack meshing is handled by FRANC3D, a FE software package devel oped by the Cornell Fracture Group. Figure 4-6 shows a schematic of the M(T) specime n and Figure 4-7 displays one of the FE models used in this analysis. The FE mode l has a unit applied pre ssure at the top and bottom faces and is simply supported. The model is comprised of entirely of SOLID92 and SOLID95 elements (ANSYS, 1999) which are used to make 3D models and have the capability to handle anisotr opic material properties. The model is twelve inches high, five-and-a-half inches wide, and an inch-and-ha lf thick. The boxed in area in Figure 4-7 highlights the densely meshed region containing the crack.

PAGE 78

65 Figure 4-6. Schematic of M(T) specimen Figure 4-7. M(T) m odel created in ANSYS Since the foam has direction dependent prope rties, this needs to be accounted for in the FE models. This is best done by using di rection cosines. Using Figure 4-8, let us look at simple transformation of axes to define a new material orient ation. Consider two sets of axes: x, y, z and x, y, z. Initially the both sets of axes are aligned and then a rotation, anti-clockwise (of angle zeta) about the x axis, takes place. Table 4-1 lists the 12 5.5 1.5

PAGE 79

66 direction cosines, i.e. the co sines of the angles between the primed and unprimed axes. Here 1 is the cosine of the angl e between the x and x axis, 2 is the cosine of the angle between the y and x axis, and so on. We w ill examine several material orientations with different initial crack inclinations. In this way, mode mixity is introduced into the problem and we also examine the effect of material orientation on the predicted propagation path. Figure 4-8. Sample transformation Table 4-1. Direction cosines x y z x 1 1 1 y 2 2 2 z 3 3 3 As mentioned in an earlier section, for typi cal acreage sprays there is little offset between the material and substrate coordinate systems except when the foam is applied near fittings and/or bolts where the foam ca n to rise at angles as high as 30 degrees relative to the ET. Twelve M(T) fracture te st samples with varyi ng material orientation and crack inclination were fabr icated and tested. A broken M(T) specimen is shown in Figure 4-9 and a comparison between the numer ical prediction and the measured turning

PAGE 80

67 angle, as well as how the foam M(T) specimens are fabricated, will be made in chapter five. Figure 4-9. Fractured M(T) specimen: the dotte d line is added to show the location of the knit line

PAGE 81

68 CHAPTER 5 RESULTS AND DISCUSSION Effect of Material Orientation on the SIF Solution On most sections of the ET, the foam is sprayed on in such a fashion that it rises normal to the surface. In that sense, there is very little offset between the tank (substrate) coordinate system and the foam (material) coordinate system. Also, when modeling these specimens, we should note that the curvatur e of the ET is rather large; some 28 ft in diameter and our specimens tend to include sm all defects (usually two inches). Thus, it seems reasonable to assume that curvature e ffects are minimal. But when the spraying process takes place near a bolt or fitting point the foam must be applied around, and over top of, these parts. In this situation it is possible to have the foam rising at different angles relative to the surface. The anisotropic nature of the foam require s 3-D FE models to be employed so the SIFs can be evaluated along the crack front. To that end, the effect of material orientation on the K-solution will be examined within models that have both straight and diagonal cracks. Before moving on with this discussion, let us define a few important terms that we will use extensively throughout th is chapter. The crack inclination angle, is the angle the crack makes with the horizon tal as shown in Figure 5-1. M(T) models use though-cracks, or flaws that are placed through the thickness of the specimen, or model, in question. The crack front distance defined as the length of the crack in the thickness direction, shown in Figure 5-2.

PAGE 82

69 Figure 5-1. Definition of crack inclination angle To analyze the effect of material orie ntation of the SIF solution, 17 material orientation cases are defined; see Figures 54 and 5-5. M(T) models are constructed using ANSYS FE software and the flaw is in serted using FRANC3D Next Generation. Figure 5-2. Definition of crack front distance

PAGE 83

70 A unit pressure is applied to the simply s upported M(T) models and in all cases, the crack length is two inches. A picture of the ANSYS model (with dimensions) is shown below. The material orientat ions are also input into the FE models along with the loads 12 5.5 1.5 12 5.5 1.5 Figure 5-3. M(T) model built in ANSYS Figure 5-4. Definition of material orientation z y x 3 2 1 445o 8 5 6 7 10 11 12 13 9 14 15 16 30o

PAGE 84

71 Figure 5-5. Top view of the cones shown in Figure 5-4 and boundary conditions. There are many differ ent orientations at which the foam can rise relative to the tank. But when the foam is sprayed down, at some locations, the rise direction relative to the tank can be as high as thirty degrees. An example of this happening is shown in Figure 5-5. Figure 5-6. Hypothetical material orientation relative to the ET One can see how the foam can rise at angl e when it sprayed out the bolt shown in the above in Figure 5-6. As such we would like to have a reasonable t olerance that the rise 9 13 7 6 8 5 4 2 1 3 0 10 11 12 14 15 16 45

PAGE 85

72 direction can fall into. This w ould give rise to a range, or set, of possible orientations that could occur when the foam is being applied. Si nce the foam can rise at thirty degrees, it might be interesting to examine the possibility of the foam rising at another, larger, angle; 45 degrees, for example. Thus we now have two possible cases to examine; a rise angle that is typical near fitting points and bolts, and one that is slightly larger. The material orientations used in this study can be visualized by setting up two concentric cones. The points on the rims of the cones denote where the z-axis (local rise direction) will be for each material case. Let us consider an example. Point one in Figure 5-4 lies on the inner, 30 degree, cone. To define the material axes for this case, a single transformation of 30 degr ees, denoted by zeta, in Figure 5-7 is performed. This can be likened to the knit lines being orient ed within the M(T) specimen shown in Figure 5-7. Table 5-1 lists the dire ction cosines, i.e. the cosine s of the angles, between the primed and unprimed axes. Here 1 is the cosine of the angl e between the x and x axis, 2 is the cosine of the angle betw een the y and x axis, and so on. Figure 5-7. Definition of case one mate rial orientation for the M(T) FE model

PAGE 86

73 Table 5-1. Direction cosines for the case one material orientation x y z x 1 = 1 1 = 0 1 = 0 y 2 = 0 2 = 0.866 2 = 0.5 z 3 = 0 3 = -0.5 3 = 0.866 Figure 5-8 is a plot of the mode I SIF against normalized crack front distance for cases 14. The mode I SIF for the case zero orientation, the case where the material and substrate axes are coincident, is also plotted to make a compar ison. Figure 5-8 pertains to a straight crack that is para llel to the horizontal (no in clination). Even though this material is anisotropic, for this geometry, KII and KIII remain small compared to KI. This is not always the case, however, particular ly when the crack is inclined. To examine the effect of mode mixity, th e crack in the M(T) model is inclined by 30 degrees. All cases are rerun and in Figures 5-9 through 5-11, the mode I-III SIF is plotted versus the normalized crack front distance for cases one through four. As expected, when the crack is inclined KI decreases, a nd KII-KIII increase. This is consistent with isotropic assumptions. For the configur ation shown in Figure 6-1, th e KI and KII are related to or the crack inclination angle (Anderson, 1991) 2cos sincos KIa KIIa (5-1) We can see that KII is zero when = 0 and that KI will decrease for increasing values of

PAGE 87

74 Figure 5-8. Mode I SIF for cases 1-4; 0 degree crack inclination Figure 5-9. Mode I SIF for cases 1-4; 30 degree cr ack inclination

PAGE 88

75 Figure 5-10. Mode II SIF for cases 1-4; 30 degree crack inclination Figure 5-11. Mode III SIF for cases 1-4; 30 degree crack inclination

PAGE 89

76 The mode I SIF for = 0 is listed in tables 5-2 and 5-3. Here we can see the variation of the mode I SIF as the material orientation is changed. For these cases, the ones that resulted in the largest KI are: 11 and 15 (2.27 psi in). When you compare that number to the KI for case zero, the orientation for t ypical acreage sprays, the largest KI is just 14% higher. This is an interes ting result in that even if we l ean the z-axis by quite a bit, the resulting mode I SIF is not much great er than the mode I SIF for the case 0 orientation. However, it is po ssible to orient this material in an infinite number of ways and it could be possible to obtain a KI value fo r an arbitrary orientation that is greater than the largest value determined in this section (2.27 psi in). Table 5-2. KI for cases 08, 0 degree inclination Case 0 1 2 3 4 5 6 7 8 KI (psi in) 1.99 2.01 2.14 2.252.142.012.142.25 2.14 Table 5-3. KI for cases 9-16, 0 degree crack inclination Case 9 10 11 12 13 14 15 16 KI (p si in ) 2.07 2.192.272.192.072.192.272.19 Crack Turning Predictions We can numerically predict the crack turn ing angles using the maximum tangential stress criteria since we have SIFs from FE models and the near-tip stress solutions from Hoenig (1982). As discussed in chapter four we will use a ratio of the tangential stress, to Kp for turning angle estimations. Let us denote the ratio of /Kp as R and where R is a maximum, that is the directi on of predicted propagation. To make a comparison, NASA has donated several broken M(T) specimens with various material orientations. The turning angles within these specimens will be measured and compared to their numerical counterparts.

PAGE 90

77 M(T) Specimen Fabrication and Determ ination of Local Knit Line Axes Fabrication of the M(T) specimens involve s spraying the foam on a metal plate, it rises and eventually a rind forms. When the foam has settled, the next layer is applied until the desired thickness is achie ved. From here, the foam is cut, or machined, from this parent block so the knit lines are running in the desired directions. One of the foam samples is shown in Figures 5-12. Three M(T) foam test specimens are analyzed in this section. These samples have various orientations a nd crack inclinations ( values). The first step in this analysis is to determine what the material orientation is for each specimen and we do this by measuring Figure 5-12. From left to right: front, left, right and rear sides of the M(T) specimen the angles of the knit lines on the specimens surface. Consider Figure 5-13. We need to determine the orientation of the x, y and z vectors in order to define the material orientation. This information is used to define the element c oordinate system in ANSYS. The intersection of two planes is determ ined by taking the cr oss-product of their normal vectors. If one wants the trace angl e on a particular face of the M(T) specimen, the dot product can be used. In our case, th e normal to the knit line plane is not known beforehand. Instead we will measure two angles, and and use them to define the knit

PAGE 91

78 line plane shown in Figure 5-13. What we are interested in, is the trace, or intersection, the knit line plane makes on the x and y = 0 faces. Along the x = 0 face, we need the angle of the knit line measured from the horiz ontal, or y-axis. In this case the knit lines on the x = 0 face are relatively straight; is approximately zero. On the y = 0 face the angle, is also measured relative to the horizontal, or the x-axis. Thus, the traces on the y and x = 0 faces define the x and y material axes respectively. Using the cross product, the axis normal to knit line plane is determined, i.e. z = x y. Figure 5-13. Determination of knit line plane Local Crack Tip Computations For crack tip computations i nvolving stresses, there are a few subtleties that should be addressed in this section because the materi al classifications discussed in chapter three are in a global sense. When looking at a foam M(T) specimen, one would say it is indeed a transversely isotropic material. However, the post-processing of near-tip

PAGE 92

79 stresses requires material properties rotated into a local crack front coordinate system. Let us consider an example. First, assume we have the case one orientation; a material rotation ( = 30o) takes place about the global x axis shown in Figure 5-13. If the inclined crack is oriented in such a fashion so that it rests in the plane of symmetry, the out-of-plane and in-plane displacements become decoupled and this is a so-called degenerate case described by Hoenig. His fo rmulation cannot be applied for this special case. As described in chapter three, when this situation arises (crack lying in a symmetry plane) the usual isotropic near-tip formulas are utilized. Figure 5-14. M(T) model with inclined crack Now set = 15 degrees. The crack is no longer resting in the symmetry plane. Let us denote this as a general case and a rotati on is needed to set the properties in the local crack front system. The way the constitutiv e matrices are transformed depends on how they are defined. If they are in compliance form

PAGE 93

80 S (5-2) then the transformation formulas (Ti ng, 1996) for the elastic constants are 'T SSSQSQ (5-2) where S is the matrix of elastic constants in the rotated system and 222 111111111 222 222222222 222 333333333 232323232323232223 313131313131313131 121212121212121212222 222 222SQ (5-3) The order of the terms in and (5-2) and (5 -3) is based on the convention commonly used by the composites and elasticity community wher e the stresses and strains are ordered as { x, y, z, yz, xz, xy}T. If the constitutive matrix is in stiffness form C (5-4) there is an alternate definition of the [QS] matrix 222 111111111 222 222222222 222 333333333 232323232323232223 313131313131313131 121212121212121212222 222 222CQ (5-5) The transformed elastic constants in sti ffness form are therefore determined by '[][][]T CCCQCQ (5-6) The two [Q] matrices are related 1T SCQQ (5-7)

PAGE 94

81 Equations (5-2) or (5-6) allow us to transform the material properties from one Cartesian coordinate system to another. The only info rmation we need for such a transformation is the direction cosines of the angles between th e coordinate axes where we start, or the system where the material properties are initia lly defined, and the system where we want to set the properties. The incipient turning angles analyzed in this program use the maximum tangential stress theory. In order to apply it, the first thing we need is the hoop, or component shown in Figure 5-15. The near-tip stress es derived by Hoenig are in a (x, y, z) Cartesian coordinate system. To obtain the hoop stress, we transform axes, in this case a rotation anti-clockwise about the z-axis by degrees. This is done using standard transformation equations (Boresi et al 1993). Figure 5-15. Cartesian and cylindrical stresses Calculating is done via 222 222222222222 x xyyzzyzzxxy (5-4) Here the 2, 2, 2 terms are the direction cosines of the angles between the old and new axes

PAGE 95

82 Now we can calculate the hoop stress at any r and of our choosing. The next step is to divide the hoop stress by the effective K, Kp. We do this because not only are the elastic constants direction dependent, but the fr acture properties are as well. The crack is predicted to turn in the direction where R is a maximum. The crack turning angle, defined as c is shown in Figure 5-16. Here we have a close-up view of a M(T) specimen containing an inclined crack. The inclined crack offset by angle, and the direction of incipient turning is denoted by the dashed line. Figure 5-16. Crack turning angle Let us pause briefly and quickly summarize the procedure needed to compute neartip hoop stresses and the pr edicted turning angle Enter material orientation (denoted by cas e number in Figure 5-2) into a FE model coordinate system

PAGE 96

83 Run the model and extract the mode I-III SIFs via FRANC3D Rotate material properties into a local crack front coordinate system Compute near-tip stresses using Hoenigs formulas Transform yy to via the transformation equati ons found in Boresi et al Divide yy by Kp and determine where this ratio is a maximum to predict the turning angle, c Table 5-4. Measured and predicte d turning angles (in degrees) c (measured) c (predicted) c (isotropic prediction) specimen A 27 32 43 specimen B 9.7 21 27 specimen C 12 19 27 For the three test specimens, one of them had a crack inclination of 30 degrees (specimen A). The other two specimens (B a nd C) had crack inclin ations of 10 degrees. All three specimens have different material or ientations. Table 5-4 lists the measured and predicted incipient turn ing angles. The SIFs can vary al ong the crack front and since the free surface can influence the numerical K-solu tion, the measured and predicted angles are computed at the mid-plane. For comparis ons sake, isotropic predictions for the same M(T) geometries are also tabulated. The isotropic angle is the critical angle, c, from equation (4-4). The experimental tests were carried out at 75oF and as such, the elastic properties at that temperature are us ed to predict the turning angle. The turning angle was measur ed using a Brown and Shar pe coordinate measuring machine. These devices are used to inspect various geometries such as cylinder heads or gear boxes. This instrument uses a probe that physically touches the work piece and the points are recorded digitally. In this case, we wish to use the points to define the fracture planes. The angle between these planes is th e crack turning angle. Within each specimen

PAGE 97

84 are four fracture surfaces. The fracture angle was measured at a location near the middle and an average of the four values is taken and listed in Table 5-4. Specimen A had the best agreement with experiment whereas the other two predictions fall somewhere between the isot ropic prediction and actual values. Even though both the direction dependence of both th e fracture toughness a nd near-tip stresses are accounted for in these predictions, the results presented here still vary by large amount. One possible source for the difference is how the material orientation is defined and modeled. When tasked to predict the turning angle, NASA fu rnished us with the three test specimens but the orientations were not specified beforehand. These particular specimens were initially fabri cated to analyze mode mixity, not the effects of material orientation. As such, the or ientations of the knit lines vary throughout the specimens. When defining the knit line plane, the knit li nes (denoted with black dots in Figure 5-12) closest to the crack were used to define the orient ation, even though the material orientation can vary over the specimen. Based on the results seen here, it seems that perhaps the crack is seeking a pure mode I condition, or that the crack is seeki ng a direction that is normal to the highest principal stress. In such cases, if a crack, for example, is orientated at 30 degrees relative to the horizontal, the crack will turn -30 degr ees to it is running straight across or in a direction normal to the maximum principal stress. However, only three specimens are analyzed in this section and to some extent, the material orientation can be more carefully controlled. Obviously more samples would have to be an alyzed to determine if the cracks favor a direction that is norm al to the highest principal stress.

PAGE 98

85 Influence of Thermal Loads on the SIF Solution FRANC3D is very versatile in that it can generate different types of crack meshes and also compute a K-solution for both isotropi c and fully anisotropic materials but at the time of this writing, it is unable to handle proble ms with a thermal gradient. This is main reason why the DC technique outlined in chapter three is utilized. To examine the effect of a thermal gradient throughout the M(T) specimen we will examine the case two orientation with a straight and inclined crack. This varies the amount of mode mixity and two different sets of temperatures are appl ied to the top and bottom faces of the M(T) specimen shown in Figure 5-17. In the figures that follow the primary ai m is to see how the SIFs change when temperatures are applied to the M(T) speci men. Since the Ks are evaluated using a different method, it is useful to compare th e results of these two approaches when the same set of boundary conditions is applied. We have results tabulated for several cases at room temperature. Selecting one of them, case 1 for example, we can compare the SIF results for a straight crack. The mid-pl ane KI at room temperature is 2.01 psi in. Using the DC method we obtain a value of 1.8 psi in, a difference of appr oximately 10%. The quality and density of the mesh does impact the alternate solution presented here. In Banks-Sills et al. (2005) study a similar routine is used to check the M-integral computations. Recall that the interaction, or M, integral is used to compute the anisotropic K-solution in FR ANC3D. Excellent agreement is shown between their various schemes, but much denser meshes are utilized within their FE models. In this study it is possible to refine th e mesh, perhaps by changing the location of the quarter point elements by some degree, but this may not be feasible because there is a

PAGE 99

86 node and element limitation within this FE software. Banks-Sills et al. use 2-D models and that allows one to change the mesh de nsity extensively without adding a lot of computational cost in terms of phys ical model size or run time. Consider Figure 5-17 with = 0 degrees (straight crack). Here T1 = 0oF and T2 = 100oF. The same unit pressure is also applied. All SIF data from FRANC3D is at room temperature (75oF). The DC method used to compute th e SIFs is able to determine these parameters at any temperature. To examine the influence of thermal loads on the Ksolution, the mode I-III SIF for case two at both room temperature and with a thermal gradient is plotted in Figure 5-18. From Table 5-2, the room temperature mid-plane value for KI is approximately 2.14 psi in. With the prescribed thermal boundary conditions, the mid-plane KI drops to 1.44 psi in; a 33% reduction. Also, in a few of the figures, some odd ac tivity takes place near the free surface. When looking at Figure 5-18, and the line that pertains the to KI, for the DC solution, the third point drops down to approximately 0.946 psi in. This happens again in the rest of the figures that pertain to th e DC solution. One possible ex planation is that the free surface along with the boundary conditions is influencing the displacement results at that point. Sometimes quadratic elements are prone to spurious oscillations. Within Dhondts (2002) paper on computing mixed-mode anis otropic SIFs, he pe rforms a smoothing operation via an averaging technique to remove the unwanted oscillations. Likewise Man et al (1993) encounter a si milar problem when using boundary elements to model a contact problem. Their solu tion involves removing the hi gher order elements in the

PAGE 100

87 problem area, and replacing them with linea r elements instead. No such averaging technique or element modi fication is employed here. There is a marked change in KI, and the other SIFs (KII and KIII) also had fairly substantial changes for this particular orienta tion and geometry. In this case KIII remains small across the crack front. But KII increases by quite a bit, just like KI does. KII at the mid-plane is roughly -0.04 psi in at room temperature. W ith the addition of temperature to this model, KII rises to 0.315 psi in. Granted for this geometry and set of loads, KII and KIII still remain small when compared to KI. Let us now consider a case where we have a larger gradient applied to the M(T) model in Figure 5-17. The next set of results for this material orientation are for T1 = -200oF and T2 = 100oF. In Figure 5-19 we see how the larger ther mal gradient impacts the mode I SIF. With these boundary conditions, KI, at the mid-plane, is 1.78 psi in. But this time, KII and KIII both show a fairly large increase after th is gradient is applied. Here KII at the mid-plane reaches -0.122 psi in and KIII = 0.355 psi in. Figure 5-17. Applicatio n of thermal loads

PAGE 101

88 KI-KIII vs normalized crack front distance = 0 degrees-0.5 0 0.5 1 1.5 2 2.5 00.20.40.60.81 normalized crack front distance psi in KI (DC) KI (F3D) KII (DC) KII (F3D) KIII (DC) KIII (F3D) Figure 5-18. Room temperature KI-KIII (F3D) vs. KI-KIII (DC) with T1 = 0oF and T2 = 100oF KI-KIII vs normalized crack front distance = 0 degrees-1.5 -1 -0.5 0 0.5 1 1.5 2 2.5 00.20.40.60.81 normalized crack front distancepsi in KI (DC) KI (F3D) KII (DC) KII (F3D) KIII (DC) KIII (F3D) Figure 5-19. Room temperature KI (F3D) vs. KI (DC) with T1 = -200oF and T2 = 100oF

PAGE 102

89 We see in the above simulations that a dding a thermal gradient can impact the mode I-III SIFs. In the first section of th is chapter, we looked at SIFs for various orientations for straight a nd diagonal cracks alike. When the crack is straight ( = 0), at room temperature, the KII and KIII are sma ll compared to KI. Again using the same geometry, when thermal loads are applied to the top and bottom faces of the M(T) model, KII and KIII are still small co mpared to KI, but they do exhibit a change from the room temperature reference state. As we have s een with the room temperature results, KII and KIII become more significant when the crack is inclined. This introduces a degree of load asymmetry, or mode mixity. To furthe r explore the influence of mode mixity along with the thermal gradient, let us consider the case two orientati on but with a 30 degree inclined crack. KI-KIII vs normalized crack front distance = 30 degrees-0.7 -0.2 0.3 0.8 1.3 1.8 00.20.40.60.81 normalized crack front distancepsi in KI (DC) KI (F3D) KII (DC) KII (F3D) KIII (DC) KIII (F3D) Figure 5-20. = 30 degrees: room temperature KI (F3D) vs. KI (DC) with T1 = 0oF and T2 = 100oF

PAGE 103

90 Taking a look at KI at the mid-plane in Figure 5-20, KI goes from 1.37 psi in at room temperature to 0.978 psi in when the thermal gradient is present. KII and KIII are 0.545 and 0.188 psi in at the mid-plane, respectively. Our final set of plots is again for case two with a thirty degr ee inclined crack, but now T1 = -200oF and T2 = 100oF. KI at the mid-plane is approximately 1.74 psi in, and KII and KIII are -0.455 psi in 0.0394 psi in, respectively. KI-KIII vs normalized crack front distance = 30 degrees-1.75 -1.25 -0.75 -0.25 0.25 0.75 1.25 1.75 2.25 00.20.40.60.81normalized crack front distance psi in KI (DC) KI (F3D) KII (DC) KII (F3D) KIII (DC) KIII (F3D) Figure 5-21. = 30 degrees: room temperature KI (F3D) vs. KI (DC) with T1 = -200oF and T2 = 100oF A final summary of the mid-plane values shown in the tables below. The abbreviations are as follows RT = room temperature, TG1 (T1 = 0oF and T2 = 100oF), TG2 (T1 = -200oF and T2 = 100oF). The values are taken at the mid-plane. The units for KI-KIII are psi in.

PAGE 104

91 Table 5-5. Summary of mid-plane KI-KII KI (RT) KI (TG1) KI (TG2) KII (RT) KII (TG1) KII (TG2) Case 2 ( = 0) 2.13 1.44 1.78 -0.038 -0.315 -0.122 Case 2 ( = 30) 1.37 0.978 1.74 1.39 0.545 -0.455 There are many more ways to apply boundary conditions to this problem. In the above analysis, since the specimen is 12 inches long, the temperature is different than room temperature, but still fairly uniform across the crack front. Table 5-6. Summary of mid-plane KIII KIII (RT) KIII (TG1) KIII (TG2) Case 2 ( = 0 ) 0.049 0.2580.289 Case 2 ( = 30 ) -0.242 0.2060.0394 The information presented in chapter tw o allows an estimation of the elastic constants over wide range of temperatures. Perhaps in some instances, the flaw can be six inches, or more, away from the surface of th e ET. In that case, given the foams very low thermal conductivity, it is po ssible to have a temperatur e over the crack front that does not vary substantially. But over t ypical acreage sprays the thickness of the insulation is roughly thre e inches. That is to say, ther e certainly are c onditions where the temperature across the crack front will not be uniform. To investigate this scenario let us re-use the same M(T) model as before but now the gradient will applied to th rough the thickness, or 1.5 inch dimension. Once again, the case two orientation along with a 30 degree inclined crack. The TG2 boundary conditions are applied and the resulting temp erature distribution al ong the crack front is plotted in Figure 5-22. Here the temperature varies appr eciably. To analyze this problem, the elastic constants are evaluated usi ng Figures 2-5 and 2-6. With those values the sixth order characteristic equation is solv ed at each point, and those roots are used in equation in (3-58) to obtain the corresponding Ks along the front.

PAGE 105

92 Crack front temperature (F) Case 2 orientation; = 30 degrees -250 -200 -150 -100 -50 0 50 100 150 00.20.40.60.81 normalized crack front distancetemperature (F) Crack front temperature (F) Figure 5-22. Temperature dist ribution along the crack front KI-KIII vs normalized crack front distance = 30 degrees-1.5 -1 -0.5 0 0.5 1 1.5 2 2.5 00.20.40.60.81 normalized crack front distancepsi in KI (DC) KII (DC) KIII (DC) Figure 5-23. KI-KIII vs. normalized crack front distance; TG2 applied in thickness direction

PAGE 106

93 The results of this model are shown in Fi gure 5-23. Since the temperatures changes with respect to crack front distance, the elas tic constants also vary which impact the Ksolution. In some parts of the plot, KII is la rger than KI, but the mode I SIF dominates in the middle portion of the crack front. At the mid-plane, KI = 1.92 psi in. Comparing this to the mid-plane value in Figure 5-21, KI = 1.74 psi in; the value increased by approximately 10%. Summary The effect of material orientation on th e anisotropic K-solution is examined. Seventeen material cases are examined and wh en the crack is straight we do not see a large difference in the mode I SIF even for or ientations that depart quite a bit from the norm. The SIFs are also computed for an incl ined crack. From the isotropic relations in (5-1) we see that KI should decrease while the remaining modes will increase as the gets larger. This trend is evid ent in Figures 5-9 through 5-11. Crack turning, at room temperature, was also estimated using both anisotropic and isotropic turning assumptions. In all cases the isotropic calculations predicted a higher turning angle than seen with in the three samples presente d here. Along with isotropic predictions, crack turning estimations that account for the direction dependence of the elastic constants and fracture toughness. Here the predictions are somewhat closer to the experimental results, but there is still a sizable difference. Perhaps one reason why this is so, is because when these specimens were fa bricated the material orientation was not carefully controlled. This makes the determ ination of the knit line orientation somewhat difficult and this can certainly influence the results because nearly all of the terms in Hoenigs equations depend on the elastic constants.

PAGE 107

94 Finally, the effect of temper ature is analyzed for the case two orientation. Several different sets of temperature boundary c onditions are applied along with the unit pressure. This analysis is conducted for both a st raight and inclined crack It is here that an important distinction can be made between the SIFs ev aluated at room temperature and those done so with the DC method with the applied te mperatures. This method, while simple to use, is not without its defici encies. Mid-plane results were compared for the case 1 orientation and a di fference of 10% between the mode I SIF is encountered. Thus, the numbers presented here should not be taken as absolutes and better agreement might be obtained by using more refined mode ls. Within the realm of the material orientations examined in this chapter, the m ode I SIF did not vary a great deal, but when there is a thermal gradient present, significan t changes in KI are encountered. The mode II and III SIFs also saw marked changes wh en the crack is no longer horizontal. In particular KIII, which is relatively small comp ared to KI and KII, for all of the room temperature tests showed larg e increases particularly when the second set of thermal boundary conditions are applied. In our last example, the thermal gradient is applied through the thickness where this dimension is much smaller than the height. Such a large gradient over a small thickness (1.5 inches) should translate in to a thermal gradient that va ries over the crack front. In this case the gradient does indeed change substantially and, unlike the prior thermal analysis, one can no longer assume a uniform temperature and prope rties along the crack tip. Thus, at each point the constants are re -evaluated and the resulting Ks are plotted against the normalized crack front distance. In this scenario, KI is the largest at the midplane.

PAGE 108

95 CHAPTER 6 CONCLUSIONS AND RECOMMENDATIONS FOR FUTURE WORK Conclusions Initial investigations of analytical solu tions for cracked anisotropic elastic solids have existed since the 1960s and the overwhelm ing majority of literature on this subject assumes the crack lies in a sy mmetry plane where anti-plane decoupling of displacements takes place. Relatively few studies take on the issue of general anisotropy where no assumptions are made to simplify the mathema tical treatment of such a complex problem. In this study, two such methods are presente d to provide a solution to obtain near-tip stresses for a cracked anisotropic elastic body. The first of which uses Hoenigs (1982) derivation of an energy release rate for genera l anisotropy. This is the preferred choice for the computation of SIFs because of the inherent accuracy in this method. As such, the Cornell Fracture Group developed a software package, FRANC3D, which is able to not only create crack meshes but also compute an anisotropic K-solu tion for a variety of material orientations: isotropy, m onoclinic, and general anisotropy. Alternatively, using the DC method, another approach to determining the anisotropic K-solution is deve loped to not only provide anot her general approach as far as material orientation goes but also one that can pr ovide a solution for problems containing both mechanical and thermal load s. FRANC3D does provi de a very versatile solution. But only structures subjected to isot hermal conditions can be analyzed with this software. However, the problem that NASA is encountering likely invo lves the effects of a thermal gradient within the foam insulation. It was intended that these methods would

PAGE 109

96 provide NASA with a more complete and funda mental understanding with regard to the fracture response of this material. The va rious conclusions are summarized below: 1. Numerical studies conducted at room temperature ove r a range of material orientations showed us that the mode I SIF did not depart substantially from the solution that arises from the orientation (case 0) which is present over the vast majority of the tank. The maximum diffe rence in the mode I SIFs between the prescribed orientations and the cas e 0 scenario was shown to be 14%. 2. The specialized divot tes ting conducted by NASA gave so me insight as to how subsurface crack, perhaps emanating from sharp void, could propagate toward the surface. It is here that crack turning an alysis could be useful. While the physics, loading, and geometry are quite complicated there are several th eories available in the literature that are app licable to this problem sin ce it is one involving brittle fracture. The fracture toughness and elastic co nstants are direction dependent in this type of foam material. As such, crack turning predictions that account for both of these characteristics were performed on th ree M(T) specimens provided by NASA. One of the three analytical predictions had reasonable agreement and while the remaining two had a large difference relativ e to the experimental value, they are closer than what isotropic assumptions would predict. 3. Temperature effects play an important role in this study. This is among the main reasons why NASA chose to initiate an exha ustive series of tests over range of temperatures. A large body of literature is available on foams, covering their applications and ways to estimate elastic a nd fracture response, but few studies deal with the harsh environment that this foam is subjected to. Since the range of temperatures involved in this problem is quite large, the effects of thermal gradients on the anisotropic K-solution were studied. Several variables come into play with this type of analysis, namely the am ount of mode mixity, specimen geometry, material orientation, and boundary conditi ons (be it thermal, mechanical, or a combination of both). Many of the above variables were manipulated and the DC method developed was able to produce an anisotropic SIF soluti on that showed how the individual Ks varied ove r the crack front. It was shown that the SIFs could change dramatically depending on the level of mode mixity and the manner in which the thermal boundary conditions are applied. The results obtained in this study represent a quantitative basis for determini ng the response of this material for a given set of boundary conditions using the computational methods outlined in earlier chapters. Recommendations for Future Work One of the primary aims of this st udy was to help NASA acquire a more fundamental understanding of the fracture res ponse of the BX-265 foam insulation. To some extent that was achieved as made evident by the conclusions made in the previous

PAGE 110

97 section. Nevertheless, much remains to further our understanding of such a complex material. What follows is a list of recommended work for future study: 1. Various models exist in the literature to evaluate the material properties and fracture response for foam materials. Such estima tions come from considering the geometry of the cells that comprise the cellular mate rial. While the density does vary over any given test specimen, or volume of foam, pe rhaps a wide range of measurements could be taken so an average can be made. This enables us to compute the relative density and this dimensionless parameter is quite us eful in estimating the elastic properties, even for anisotropic materials. 2. The crack turning investigation presented he re is far from comprehensive. A more complete testing program invol ving many different orientati ons and crack inclinations should be developed so better criteria can be developed for more accurate predictions. Within the three test samples studied, it a ppears that the crack is seeking a path normal to the maximum principal stress. Bu t with only three specimens on hand, it is difficult to generalize results. 3. While this study made use of simple and st andardized geometries, the problem NASA is faced with is one that involves a rather complicated geometry. Divot models can be built using FE software and crack turning analysis similar to what was done in this program can be performed. This should provide NASA with a very comprehensive representation of the actual problem since anisotropy and temperature effects can be accounted for. Many different geometries and loads could be applied and the resulting turning angles coul d be compared to experime nt. This could lead to additional refinements being made to the crack turning theories for anisotropic materials. 4. While the analysis in this study is confined to foam, there is no reason why it cannot be extended to other engineering materials th at exhibit direction de pendent properties. For example, perhaps crack turning calcula tions can be performed on various types of aluminum, or other metals where the crysta l orientation influences the behavior. Additional work can be done to modify the various theories, crack turning for example, to account for a plastic zone in front of the crack tip. Plasticity is neglected in this study because the foam is so brit tle, but many engineering materials are quite a bit more ductile which means that a process, or plastic zone, needs to be accounted for.

PAGE 111

98 APPENDIX GENERAL SOLUTION FOR THE PLANE PROBLEM OF A LINEAR ELASTIC ANISOTROPIC BODY Following Lehkhnitskii (1963), Sih et al (1965), Ting (1996) and Cherepanov (1979) the characteristic equation described in chapter three will be derived here. The ensuing discussion is for a plane probl em, meaning the stresses, strains, and displacements have no dependence on the z coor dinate. We start by listing the equations of equilibrium 0 0 0xyyxyyz xxzxyyxxy (A-1) Since this a plane problem, the reduced compliance matrix becomes 33 33'ij ijijSS SS S The strains and displacements are related to each other 0 xyz xyxzyzuvw xyz uvww yxxy (A-2) Finally we have the compatibility equations 22 2 22 yxyyz xxz y xxyyx (A-3) Two arbitrary stress functions (U and F) are introduced as a means to solve the equilibrium equations

PAGE 112

99 22 22 2 xy xyxz yzUU y x UF x yy F x (A-4) If we substitute (A-4) into the compatibility relations we have a system of equations in terms of U and F 43 320 0LULF LULF (A-5) Where 44444 4222612661611 432234 3333 3242546145615 3223 222 2554544 22'2'2''2'' '''''' '2''LSSSSSS x xyxyxyy LSSSSSS xxyxyy LSSS xxyy (A-6) Through equation (A-5) we can eliminate one of the stress functions, F for example, and it can be shown that 2 4230LLLU (A-7) We note that the stress functions are in terms of the x and y coordinates and the 2 423LLLU operator is of order of order six. The solution to this problem will use a composite variable, z such that (,)()UxyFz (A-8) zxy (A-9)

PAGE 113

100 Before moving on with this discussion, let us define the term characteristic equation with a simple example. Consider, for a moment, a linear seco nd order homogeneous equation such as '''0yayby (A-10) where a and b are constants, one method (Kre yszig, 1962) used to solve this type of differential equation is to assume the solution takes the form x y e (A-11) with first and second derivatives 2' '' x x y eye (A-12) If we substitute the derivates into equations (A-12) we get 20ab (A-13) This equation is called the characteristic, or auxiliary, equation w hose solution is readily determined by the quadratic formula. Here we ll will obtain two roots th at can be real and distinct, complex conjugates, or double real. The above pro cedure is not just valid for second order linear differential equations, but ca n be applied to another equation of order n. Continuing on, we now substitute (A -8) into (A-7) which results in 2 243lll (A-14) where 2 2554545 32 3151456254624 432 4111612662622'2'' '''''' '2'2''2'' lSSS lSSSSSS lSSSSSS (A-15)

PAGE 114

101 The coefficients in (A-14) are real a nd the corresponding roots are going to be complex and occur in conjugate pairs. The general case of the plane problem makes no assumptions that decouple the out-of-plane and in-plane displacements. This decoupling can take place, depending on the type of materi al one is analyzing. Sih et al consider monoclinic materials. In this special cas e, Cherepanov derives an alternate solution where a fourth order character istic equation is developed 432 111312332322'2'2''2''0 SSSSSS (A-16) 2 554544'2''0SSS (A-17) The roots 1, 2 and their conjugates are determin ed from the first equation and 3 and its conjugate is obtained by the la tter. Finally, the near-tip solution for such a material is 22 122121 12211221 12 12211221 11 ReRe 1111 ReRe 1 Re 2xx yy xy zx yzKIKII QQQQ KIKII QQQQ KI r 212 12121212 3 3 3111 Re Re 1 Re KII QQQQ KIII Q KIII Q (A-18) 31323633zzxxyyxySSSA (A-19) 2 111612'''jjj p SSS (A-20) 122622'''jjjqSSS (A-21) cossiniiQ (A-22)

PAGE 115

102 1222112211 1212 1 21222112211 1212 3 3 4534411 ReRe 211 ReRe Re KIpQpQKIIpQpQ u r uKIqQqQKIIqQqQ u Q KIII CC (A-23) where C45 and C44 are the elastic constants in stiffness form C (A-24) where 1 ijijCS (A-25)

PAGE 116

103 LIST OF REFERENCES Aliabadi and Rooke D.P. Numerical Fracture Mechanics, Boston: Computational Mechanics Publications, 1991, pp. 45. Anderson T.L. Fracture Mechanics, Boca Raton, FL: CRC Press, 1991, pp. 70, 120-121, 430-431. ANSYS Elements Reference, ANSYS Rel ease 5.6; ANSYS, Inc. November 1999. Banks-Sills L., Hershkovitz I., Wawrzynek P., Eliasi R., and Ingraffea T., Methods for Calculating Stress Intensity Factors in An isotropic Materials: Part 1-Z = 0 is a Symmetric Plane, Engineering Fracture Mechanics, 2005, Vol. 72, pp. 2328-2358. Barnett D.M., and Asaro R.J., The Fracture Mechanics of Slit-Like Cracks in Anisotropic Elastic Media, Journal of th e Mechanics and Physics of Solids, 1972, Vol. 20, pp. 353-366. Barsom, and Rolfe S. Fracture and Fatigue Control in Structures: Applications of Fracture Mechanics, Woburn, MA: Butterworth-Hei nemann, 1999, pp. 55, 77, 8081. Barsoum R.S., On the Use of Isoparametric Finite Elements in Linear Fracture Mechanics, International Journal for Nu merical Methods in Engineering, 1976, Vol. 10, pp. 25-37. Bowie O.L., and Freese C.E., Central Crack in Plane Orthotropic Rectangular Sheet, Engineering Fracture Mechanics, 1972, Vol. 8, pp. 49-58. Bogy D.B., The Plane Solution for Anisot ropic Elastic Wedges Under Normal and Shear Traction, Journal of Applied Mechanics, 1972,Vol. 39, pp. 1103-1109. Boone T.J., Wawrzynek P.A., and Ingraffea A.R., Finite Element Modeling of Fracture Propagation in Orthotropic Ma terials, Engineering Fract ure Mechanics, 1987, Vol. 26, pp. 185-201. Boresi A.P., Schmidt R.J., Sidebottom O.M. Advanced Mechanics of Materials, New York: John Wiley and Sons, Inc., 1993, pp. 36. Brezny R., and Green D.J., Factors Controlli ng Fracture Resistance of Brittle Cellular Materials, Journal of the American Ce ramics Society, 1991, Vol. 78, 1061-1065.

PAGE 117

104 Buczek M.B., and Herakovich C.T., A Normal Stress Criterion for Crack Extension Direction in Orthotropic Co mposite Materials, Journal of Composite Materials, 1985, Vol. 19, pp. 544-553. Carloni C., Piva A., and Viola E., An Altern ative Complex Variable Formulation for an Inclined Crack in an Orthotropic Medium , Engineering Fracture Mechanics, 2003, Vol. 70, pp. 2033-2058. Chan S.K., Tuba I.S., and Wilson W.K., O n the Finite Element Method in Linear Fracture Mechanics, Engin eering Fracture Mechanics, 1970, Vol. 2, pp. 1-17. Cherepanov G.P. Mechanics of Brittle Fracture, New York: McGraw-Hill, 1979, pp. 7178. Chen C.S., Crack Growth Simulation and Re sidual Strength Predic tion in Thin Shell Structures, Ph. D Thesis, Cornell University, 1999. Dhondt G., Mixed-Mode K-Calculations fo r Anisotropic Materials, Engineering Fracture Mechanics, 2002, Vol. 69, pp. 909-922. Dieter G.E. Mechanical Metallurgy, New York: McGraw-Hi ll Inc., 1976, pp. 55-64. Erdogan F., and Sih G.C., On the Extensi on of Plates Under Plane Loading and Transverse shear, Journal of Basic E ngineering, 1963, Vol. 85, pp. 519-527. Fowlkes C.W., Fracture Toughness Tests of a Rigid Polyurethane Foam, International Journal of Fracture, 1974, Vol. 10, pp. 99-108. Freund L.B., Stress Intensity Factor Calcul ations Based on a Conservation Integral, International Journal of Solids and Structures 1978, Vol. 14, pp. 241-250. Gent A.N., and Thomas A.G., The Deforma tion of Foamed Elastic Materials, Journal of Applied Polymer Science, 1959, Vol. 1, pp. 107-113. Gent A.N., and Thomas A.G., Mechanics of Foamed Elastic Materials, Rubber and Chemistry, 1963, Vol. 36, pp. 597-610. Gibson L.J., and Ashby M., The Mechanics of Three-Dimensional Cellular Solids, Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences, 1982, Vol. A 382, pp. 43-59. Gibson L.J., and Ashby, M. Cellular Solids, Elmsford, NY: Pergamon Press Inc., 1988, pp. 76-81, 241-256. Gdoutos E.E. Problems of Mixed Mode Crack Propagation, The Hague, The Netherlands: Martinus Nijho ff Publishers, 1984, pp. 2.

PAGE 118

105 Hoenig A., Near Tip Behaviour of a Crack in a Plane Anisotropic Elastic Body, Engineering Fracture Mechanics, 1982, Vol. 16, pp. 393-403. Huang J.S., and Gibson L.J., Fracture Toughness of Brittle Honeycombs, Acta Metallurgica Materialia, 1991, Vol. 39, pp. 1617-1626. Hussain M.A., Pu S.L., and Underwood J.H., Strain Energy Release Rate for a Crack Under Combined Mode I and Mode II, Frac ture Analysis ASTM STP, 1974, Vol. 560, pp. 2-28. Hyer M.W. Stress Analysis of Fiber-Rein forced Composite Materials, New York: McGraw-Hill Inc., 1998, pp. 69-75. Ingraffea A.R., and Manu C., Stress Intensity Factor Computation in Three Dimensions with Quarter-Point Elements, International Journal for Numerical Methods in Engineering, 1980, Vol. 15, 1427-1445. Irwin G.R., Analysis of Stresses and Strains near the End of a Crack Traversing a Plate, Journal of Applied Mechanic s, 1957, Vol. 24, pp. 361-364. Irwin G.R., Fracture Mechanics, In: G oodier J.N., and Hoff N.J., (editors), Structural Mechanics: Proceedings of First Naval Symposium, Pergamon Press, 1960, pp.557. Ishikawa H., A Finite Element Analysis of Stress Intensity Factors for Combined Tensile and Shear Loading by Only a Vi rtual Crack Extensi on, International Journal of Fracture, 1980, Vol. 16, pp. 243-246. Ko W.L., Deformations of Foamed Elasto mers, Journal of Cellular Plastics, 1965, Vol.1, pp. 45-50. Kuo M.G., and Bogy D.B., Plane Solutions for the Displacement and TractionDisplacement Problem for Anisotropic Elastic Wedges, Journal of Applied Mechanics, 1974, Vol. 41, pp. 197-203. Kreyszig E. Advanced Engineering Mathematics, New York: John Wiley and Sons Inc., 1962, pp. 105-105, 136. Lekhnitskii S.G. Theory of Elasticity of an Anisotropic Body, San Francisco: HoldenDay, Inc., 1963, pp. 3, 104-110, 117-119, 122-123. Li F.Z., Shih C.F., and Needleman A., A Comparison of Met hods for Calculating Energy Release Rates, Engineering Fract ure Mechanics, 1985, Vol. 21, pp. 405421. Logan D.L. A First Course in the Finite Element Method, Pacific Grove, CA: Wadsworth Group, 2002, pp. 413-414, 433-434.

PAGE 119

106 Maccagno T.M., and Knott J.F., The Fracture Behaviour of PMMA in Mixed Modes I and II, Engineering Fracture Mechanics, 1989, Vol. 34, pp. 65-86. Maccagno, T.M., and Knott J.F, The Low Temp erature Brittle Fracture Behaviour of Steel in Mixed Modes I and II, Engineer ing Fracture Mechanics, 1992, Vol. 38, pp. 111-128. Maiti, S.K., Ashby M.F., and Gibson L.J., Fracture Toughness of Brittle Cellular Solids, Scripta Mettalurgica, 1984, Vol. 18, pp. 213-217. Man K.W., Aliabadi M.H., and Rooke D.P ., BEM Frictional Contact Analysis: Modeling Considerations, Engineering Analysis with Boundary Elements, 1993, Vol. 11, pp. 77-85. McIntyre A., and Anderton G.E., Fracture Properties of a Rigid Polyurethane Foam Over a Range of Densities, Polymer, 1979, Vol. 20, pp. 247-253. Moran B., and Shih C.F., A General Treat ment of Crack-Tip Contour Integrals, International Journal of Fracture, 1987, Vol. 35, pp. 295-310. Morgan J.S., Wood J.L., and Bradt R.C., Ce ll Size Effects on the Strength of Foamed Glass, Materials Science and Engi neering, 1981, Vol. 47, pp. 37-42. Nobile L., and Carloni C., Fr acture Analysis for Orthotropic Cracked Plates, Journal of Composite Structures, 2005, Vol. 68, pp. 285-293. Patel M.R., and Finnie I., Str uctural Features and Mechan ical Properties of Rigid Cellular Plastics, Journal of Materials, 1970, Vol.5, pp. 909-932. Pettit R., Crack Turning and Arrest in Integr ally Stiffened Aircraft Structures, Ph. D Thesis, Cornell University, 2000. Rice J.R., A Path Independent Integral and the Approximate Analysis of Strain Concentration by Notches and Cracks, J ournal of Applied Mechanics, 1968, Vol. 35, pp. 379-386. Rathod H.T., A Study of Asymmetrically Load ed Griffith Crack in a Plane Anisotropic Elastic Body, Engineering Fracture M echanics, 1979, Vol. 11, pp. 87-93. Saouma V.E., and Sikiotis E.S., Stress Inte nsity Factors in Anis otropic Bodies Using Singular Isoparametric Elements, Engineer ing Fracture Mechanics, 1986, Vol. 25, 00. 115-121. Sih G.C., Paris P.C., and Irwin G.R., On Cr acks in Rectilinearly Anisotropic Bodies, Engineering Fracture Mechanics, 1965, Vo1. 1, pp. 189-203. Sih G.C., Strain Energy Density Factor Applied to Mixed Mode Crack Problems, International Journal of Fracture, 1974, Vol. 10, pp. 305-321.

PAGE 120

107 Sneddon IN., The Distribution of Stress in the Neighbourhood of a Cr ack in an Elastic Solid, Proceedings, Royal Society of London, 1946, Vol. A-187, pp. 229-260. Tada, H., Paris, P., and Irwin, G. The Stress Analysis of Cracks Handbook, Hellerton, PA: Del Research Corp., 1973. Ting T.C. Anisotropic Elasticity: Theory and Applications, Oxford: Oxford University Press, 1996, pp. 54-55, 118-133. Timoshenko S.P., and Goodier J.N. Theory of Elasticity, Tokyo: McGraw-Hill, 1982, pp. 354-380. Tupholme G.E., A Study of Cracks in Orthotro pic Crystals Using Di slocation Layers, Journal of Engineering Mathematics, 1974, Vol. 8, pp. 57-69. Wells A., Unstable Crack Propagation in Metals-Cleavage and Fracture, Cranfield Crack Propagation Symposium, 1961, Vol. 1, pp. 210. Westergaard H.M., Bearing Pressures on Cracks, Journal of Applied Mechancis, 1939, Vol. 6, pp. 49-53. Williams M.L., On the Stress Distribution at th e Base of a Stationary Crack, Journal of Applied Mechanics, 1957, Vol. 24, pp. 109-114. Yau J.F., Wang S.S., and Corten H.T., A Mixed-Mode Crack Anal ysis of Isotropic Solids Using Conservation Laws of Elasti city, Transactions of the ASME, 1980, Vol. 47, pp. 335-341.

PAGE 121

108 BIOGRAPHICAL SKETCH Erik Knudsen was born in West Chester, PA, on February 18th, 1979. At the age of four his family moved to Naples, FL, wher e he spent the remai nder of his childhood and graduated from Lely High School in 1997. He earned a B.S. in mechanical engineering from Villanova in the Spring of 2001 and returned to Florida for his graduate work. Under the tutelage of Dr. Nagaraj Arakere he earned a M.S. in mechanical engineering in the Fall of 2003 working on a NASA related pr oject. Upon completion of the doctorate from the University of Florida, he would lik e to become a full-time teacher at the college level or perhaps retain a positi on in the aerospace industry.


Permanent Link: http://ufdc.ufl.edu/UFE0015560/00001

Material Information

Title: Anisotropic Fracture Analysis of the BX-265 Foam Insulation Material Under Mixed-Mode Loading
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: UFE0015560:00001

Permanent Link: http://ufdc.ufl.edu/UFE0015560/00001

Material Information

Title: Anisotropic Fracture Analysis of the BX-265 Foam Insulation Material Under Mixed-Mode Loading
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: UFE0015560:00001


This item has the following downloads:


Full Text












ANISOTROPIC FRACTURE ANALYSIS OF THE BX-265 FOAM INSULATION
MATERIAL UNDER MIXED-MODE LOADING















By

ERIK KNUDSEN


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA


2006

































Copyright 2006

by

Erik Knudsen

































This thesis is dedicated to my parents, friends, family, lab mates and everyone else who
was there for me when I needed them the most. I will never forget you!















ACKNOWLEDGMENTS

First and foremost I would like to thank my parents for their unwavering love and

support. My advisor, Dr. Nagaraj Arakere, has been in my corner since I first came to

graduate school and his contacts at NASA paved the way for the two degrees I now hold

in engineering. In the summer of 2002 I accompanied Dr. Arakere to Huntsville as part

of the Summer Faculty Fellowship Program. During our ten week stay, I met many

experts and scientists who work at Marshall, particularly Dr. Greg Swanson, Jeff

Raybum, Greg Duke, Dr. Gilda Battista, Stan Oliver, Preston McGill, Doug Wells, Dr.

Eric Poole, Dr. Philip Allen and Denise Duncan (in a later trip), and many, many others.

They have helped me grow as a student and a person, and I am forever indebted.

After my trip to NASA, approximately one year later, I earned a NASA Fellowship

to sponsor this work. Ed Adams of the Education Directorate at Marshall deserves

special praise here, and I thank him for blessing me with the opportunity to, once more,

work with NASA. I shared an on office Dr. Philip Allen and one could not ask for a

more pleasant person to work with and learn from.

The Comell Fracture Group was also a tremendous help in my quest to better

understand and learn the principles of fracture mechanics. Drs. Paul "Wash" Wawryznek

and Bruce Carter are two of the nicest and most patient people I have ever known and I

owe them many thanks for taking the time to help me.

Last, but certainly not least, I wish to thank my labmates over the years: Srikant,

Sangeet, Shadab, Niraj, Shannon, Yogen, Jeff, TJ, and George. Without their insight,









jokes, and patience I would not be where I am today. I now consider many of them to be

my best friends and I hope we can stay in touch.
















TABLE OF CONTENTS



A C K N O W L E D G M E N T S ................................................................................................. iv

LIST OF TABLES .... ...... .. ........................... ........... .................... .. viii

LIST OF FIGURES ......... ......................... ...... ........ ............ ix

ABSTRACT .............. ..................... .......... .............. xii

CHAPTER

1 BA CK GR OU N D ......... ....................... .... ........ ..... ............................ .

Problem M motivation ......... .. ...... ...... ......... .. ....... .......... ...............
Program Scope and Focus ................................................. ............................... 2

2 FOAM TESTING AND MATERIAL CLASSIFICATION ...............................6

Cellular Solid B background Inform action ............................................ .....................6
Initial F oam M material T testing ................................................................ ....................7
Determination of Elastic Constants and Material Classification..........................7
Fracture Testing for the BX-265 Foam Material......................................20
Additional Testing: Divot Test Specim ens.............................................................. 24
Sum m ary ...................................... ................. ................. .......... 28

3 AN INTRODUCTION TO FRACTURE MECHANICS..........................................30

M material D efin ition s ......................................................................... ..................... 3 0
Isotropic Fracture M echanics ...................... ....................... ................... 31
Linear Elastic Fracture Mechanics for Anisotropic Materials............................... 33
Determining the Stress Intensity Factor-K............... ..... ............... 37
FRANC3D Next Generation Software ............................................ ............... 38
Computing SIFs Using Displacement Correlation............................................45
Effects of Temperature on the SIF Solution...........................................................52
Sum m ary ...................................... ................. ................. .......... 54

4 CRACK TURNING THEORY AND FINITE ELEMENT MODELING ...............56

C rack T u rning T h eory ...................................................................... ................ .. 56









FE M odeling of the M (T) Specim en ................................. ...................................... 64

5 RESULTS AND DISCU SSION ........................................... .......................... 68

Effect of Material Orientation on the SIF Solution ............................... ...............68
Crack Turning Predictions ....... ................ .... ..... ... ....... ...................76
M(T) Specimen Fabrication and Determination of Local Knit Line Axes .........77
Local Crack Tip Com putations ........................................ ....................... 78
Influence of Thermal Loads on the SIF Solution ....................................... .......... 85
Sum m ary ............... .... ..... ......... ......................................93

6 CONCLUSIONS AND RECOMMENDATIONS FOR FUTURE WORK ..............95

C o n c lu sio n s............................... .......................................................................... 9 5
Recom m endations for Future W ork ........................................ ....................... 96

APPENDIX

GENERAL SOLUTION FOR THE PLANE PROBLEM OF A LINEAR ELASTIC
A N ISO TR O PIC B O D Y ................................................................... .....................98

L IST O F R E FE R E N C E S ........................................................................ ................... 103

BIOGRAPHICAL SKETCH ............................................................. ............... 108
















LIST OF TABLES

Table page

2-1 Measured material property values from experiment ........................ ...........18

3-1 Various m ethods for determ ining the SIF ..................................... .................38

4-1 Direction cosines .......... ....... ........ .................... ................... 66

5-1 Direction cosines for the case one material orientation .......................................73

5-2 KI for cases 0-8, 0 degree inclination ........................................... ............... 76

5-3 KI for cases 9-16, 0 degree crack inclination ................ .................................. 76

5-4 Measured and predicted turning angles (in degrees)................... ... .............83

5-5 Sum m ary of m id-plane K I-K II......... ............... ........................... .. ...................91

5-6 Sum m ary of m id-plane K ill......... ............... .................................. ............... 91
















LIST OF FIGURES

Figure page

1-1 Foam loss during Shuttle ascension ........................................ ........ ............... 2

1-2 Spray-on foam insulation (SO FI)................................................................... ......3

2-1 Idealized foam microstructure...................................................... ........................ 8

2-2 Square specimen tensile test used to determine the Poisson ratio..........................15

2-3 F oam test specim ens........... ............................................................ ...... .... ... ... 15

2-4 Summary of stress-strain data .............................................. 16

2-5 Coordinate system for the transversely isotropic material............................... 17

2-6 Young's moduli vs. temperature .................... ........................... 18

2-7 Shear m oduli vs. tem perature...................... ................................. ............... 19

2-8 Coefficient of thermal expansion vs. temperature ...........................................19

2-9 Thermal conductivity vs. temperature....... .......... ................... ..............20

2-10 From left to right: M(T), C(T), and SNE(B) fracture test specimens ....................22

2-11 Summary K-values for various fracture specimens ...........................................23

2-12 Clip gauge on C(T) specimen used to measure CMOD .....................................24

2-13 Typical load vs CMOD curve for a brittle material ...........................................25

2 -14 D iv ot test p an el............ .... ........................................................................... .. ...... .. 2 6

2-15 2-D view of divot test................................................................. ............... 27

2-16 2-D view of test panel after heat flux has been applied .......................................27

2-17 Foam blow out from divot test ............................... ............... 28

3-1 Coordinate system used in equation (3-1)................................... ............... 31









3-2 From left to right: mode I (opening), mode II (shearing), mode III (tearing) .........32

3-3 Path of the J-integral .............................................. .... .... .. ........ .... 39

3-5 D definition of Irwin's crack closure integral ............................ .............................43

3-6 Correlation points typically used for displacement correlation ............................46

3-7 Triangular quarter point elem ents ........................................ ........................ 48

3-8 Isoparametric element and degenerated element with mid-side nodes moved to
quarter point locations .............................................................................48

4-1 D definition of crack turning angle ........................................ ......................... 57

4-2 Elliptical relationship for T ............................................................................58

4-3 D definition of hoop stress ................................... .... ............................ ...............59

4-4 Orthotropic toughness values ........................................................ ............. 62

4-5 Definition of the a and n vectors ................................................... ............... 62

4-7 M (T) m odel created in AN SY S ........................................ .......................... 65

4-8 Sam ple transform action ................................................. ................................ 66

4-9 Fractured M(T) specimen: the dotted line is added to show the location of the
k n it lin e ............................................................................ 6 7

5-1 Definition of crack inclination angle.................................................................. 69

5-2 D definition of crack front distance....................................... .......................... 69

5-3 M (T) m odel built in A N SY S ........................................................................ .. .... 70

5-4 D definition of m material orientation..........................................................................70

5-5 Top view of the cones shown in Figure 5-4......................................................71

5-6 Hypothetical material orientation relative to the ET..............................................71

5-7 Definition of case one material orientation for the M(T) FE model ......................72

5-8 Mode I SIF for cases 1-4; 0 degree crack inclination ...........................................74

5-9 Mode I SIF for cases 1-4; 30 degree crack inclination..................... ...............74

5-10 Mode II SIF for cases 1-4; 30 degree crack inclination .......................................75









5-11 Mode III SIF for cases 1-4; 30 degree crack inclination.............................75

5-12 From left to right: front, left, right, and rear sides of the M(T) specimen..............77

5-13 D eterm nation of knit line plane..............................................................................78

5-14 M (T) m odel with inclined crack ..................................................... ............. 79

5-15 Cartesian and cylindrical stresses................................ ................... ...... ........ 81

5-16 Crack turning angle .................. ............................... ..... .. ............ 82

5-17 Application of therm al loads .............................................................................87

5-18 Room temperature KI-KIII (F3D) vs. KI-KIII (DC) with T1 = 0F and T2 =
10 0 F ............................................................................... 8 8

5-19 Room temperature KI (F3D) vs. KI (DC) with T1 = -2000F and T2 = 100F..........88

5-20 (p = 30 degrees: room temperature KI (F3D) vs. KI (DC) with T1 = 0F and T2 =
10 0 F ............................................................................... 8 9

5-21 (p = 30 degrees: room temperature KI (F3D) vs. KI (DC) with T1 = -2000F and
T 2 = 10 0 F ........................................................................ 9 0

5-22 Temperature distribution along the crack front..................................................... 92

5-23 KI-KIII vs. normalized crack front distance; TG2 applied in thickness direction...92















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

ANISOTROPIC FRACTURE ANALYSIS OF THE BX-265 FOAM INSULATION
MATERIAL UNDER MIXED-MODE LOADING
By

Erik Knudsen

August 2006

Chair: Nagaraj K. Arakere
Major Department: Mechanical and Aerospace Engineering

The National Aeronautics and Space Administration (NASA) uses a closed-cell

polyurethane foam to insulate the external tank (ET) which contains the liquid oxygen

and hydrogen for the Space Shuttle main engines. This is a type of spray-on foam

insulation (SOFI), similar to the material used to insulate attics in residential

construction. In February of 2003, the Shuttle Columbia suffered a catastrophic accident

during re-entry. Debris from the ET impacting the Shuttle's thermal protection tiles

during liftoff is believed to have caused the Space Shuttle Columbia failure during re-

entry.

NASA engineers are very interested in understanding the processes that govern the

breakup/fracture of this complex material from the shuttle ET. The foam is anisotropic in

nature and the required stress and fracture mechanics analysis must include the effects of

the direction dependence on material properties. Over smooth, flat areas of the ET the

foam can be sprayed down in a very uniform fashion. However, near bolts and fitting









points it is possible for voids and other defects to be present after the foam is applied.

Also, the orientation of the foam, as it rises from non-uniform surfaces, can be

significantly different from the orientation over typical acreage sprays. NASA believes

that air present in these small voids is liquefied and then turned into a gas during liftoff.

While the Shuttle is ascending to space, the pressure in these cavities can become large

enough to force a subsurface crack toward the exterior of the tank, thus freeing portions

of foam insulation.

As a first step toward understanding the fracture mechanics of this complex

material, a general theoretical and numerical framework is presented for computing stress

intensity factors (SIFs), under mixed-mode loading conditions, taking into account the

material anisotropy. The effects of material orientation and mode mixity on the

anisotropic SIF solution are analyzed. Crack turning predictions under mixed mode

loading are presented. Furthermore, the influence of temperature gradients on the SIF

solution is studied, in view of the thermal gradient present through the foam thickness on

the ET. The results presented represent a quantitative basis for evaluating the strength

and fracture properties of anisotropic BX-265 foam insulation material.














CHAPTER 1
BACKGROUND

Problem Motivation

On February 1st, 2003, the Space Shuttle Columbia suffered a catastrophic failure

during re-entry. NASA has conducted an exhaustive investigation of the failure and the

consensus now is the breakup was caused by a segment of BX-265 foam insulation,

roughly the size of a suitcase, striking the wing during liftoff. The foam impact, in prior

Shuttle launches, has been known to cause impact damage to the thermal protection tiles,

but was not considered to be a serious problem until the Columbia disaster. The damage

to the thermal protection tiles on the leading edge of the wing is thought to have allowed

gases to penetrate the wing during re-entry, triggering a cascading series of catastrophic

events that led to the loss of the Shuttle.

It is believed that when the foam insulation is applied to the external tank (ET) that

voids are created in certain areas where geometric discontinuities such as bolts, flanges,

and fittings are encountered. Since the tank is filled with liquid oxygen and hydrogen

the foam is exposed to cryogenic temperatures at the ET's surface. The air inside these

voids can be liquefied and nitrogen can condense into these small cavities. During liftoff

the outer surface of the foam is exposed to aerodynamic heating. This heating raises the

temperature of the liquid nitrogen, turning it into a gas. The pressure difference

associated with the formation of the gas can cause pieces of foam to be blown out during

liftoff( see Figure 1-1).


























Figurel-1. Foam loss during Shuttle ascension

Program Scope and Focus

Prior to the Columbia disaster, the foam that insulates the tank was not understood

at a fundamental level. Quantities like the elastic moduli, and thermal properties such as

coefficient of thermal expansion are needed for a meaningful failure analysis of this

material. This prompted NASA to initiate an intensive testing program to ascertain

various properties at cryogenic, room, and elevated temperatures. The pictures in Figure

1-1 show the foam breaking off areas near fitting points. At these locations, insulation is

sprayed around various parts, such as bolts, and it is here that the foam's orientation,

relative to the tank, can change by quite a bit. For typical acreage sprays, there is very

little difference between the local coordinate axes of the foam and those of the substrate

(Figure 1-2).

Pictures of the Shuttle during liftoff, such as those shown in Figure 1-1, show us

that large pieces of foam are not really breaking away over smooth, open, areas of the

tank. The problem seems to be confined to areas where the foam is applied over fitting

points, bolts, and other areas where various parts can protrude the surface of the ET.











Spray
Direction


-" 11
Material
coordinates


Knitines -
Knitlines Note: For typical acreage sprays,

SubstrateS hoop
Substrate H
Coordinates A=axial
R=radial
If theta and sub state coordinates are aligned


Figure 1-2. Spray-on foam insulation (SOFI)

Right away we notice that two potential problems come to mind at these locations: it is

possible for voids to be created when the foam is being laid down, and the orientation of

the foam can change by a large degree.

The material tests conducted by NASA have shown that this type of foam is brittle.

Perhaps the best way to analyze sharp, crack-like, defects for this sort of material is

through linear elastic fracture mechanics (LEFM). For isotropic materials, the near-tip

stresses are completely characterized by K-the stress intensity factor (SIF). The

analytical K-solutions for anisotropic materials take a similar form, but have additional

terms that come from the various elastic moduli. The formulations of near-tip stresses for

isotropic materials are well established and are now incorporated in many finite element

(FE) software packages. Some K-solutions for certain kinds of direction-dependent

materials started appearing in the 1960s and more generalized solutions were presented in

the 1980s. Since many engineering materials are not isotropic (such as rolled aluminum









in aircraft fuselages and single-crystal superalloys used in high performance gas turbine

engines), more attention is being paid toward the implementation of anisotropic K-

solutions in various FE schemes. Still, many of the most popular FE software packages

do not allow accurate computations of SIFs for simulations involving anisotropic

materials.

Foams are inherently anisotropic, and NASA, after extensive testing, has classified

this material as transversely isotropic. Such materials also have fracture characteristics

that are direction dependent and both of these traits must be accounted for in the

analytical and numerical analyses. Although anisotropic K-solutions are not readily

available, the Cornell Fracture Group has developed a software package, FRANC3D, that

simplifies the meshing process necessary for FE modeling of crack-like defects in elastic

solids and also computes a K-solution for anisotropic materials. Concurrently, an

alternative method for computing the SIFs, based on the work of Hoenig (1982), for

anisotropic materials has been employed and extended to account for thermal stresses and

strains.

It is hoped that the present study will enhance NASA's understanding of this

material, since the K-solutions used in this analysis account for both the direction-

dependent properties and the impact of thermal stresses and strains. The following points

summarize the scope of this study in a broad sense:

1. Through the use of the middle tension, M(T), specimen and the FE method a
systematic rotation of material properties and its impact on the K-solution is
examined

2. The impact of mode mixity on the K-solution is studied

3. Crack turning predictions that take into account both the anisotropic stiffness and
fracture properties are presented






5


4. The effect of a thermal gradient through the M(T) specimen and its impact on the
K-solution are presented














CHAPTER 2
FOAM TESTING AND MATERIAL CLASSIFICATION

Cellular Solid Background Information

Cellular solids are generally comprised into two sets of structural categories: 1)

honeycomb layouts where the microstructure is comprised of two-dimensional arrays of

polygons that fill a planar area and 2) foams, the three-dimensional counterpart to

honeycombs where the cells take the form of polyhedra. Within Gibson and Ashby's

(1988) definitive text on cellular solids, we see how many types of materials can be

foamed: metals, glasses, and ceramics can be fabricated into cells that take on a variety of

shapes and sizes. Among the most common types of foam are polymers which are used

for many applications including insulation for residential homes or disposable coffee

cups. Polymer foams are usually very light and they are popular in industries where

weight is of prime importance, such as in the aerospace industry.

Foams are typically made by bubbling a gas into the liquid monomer or hot

polymer. The gas can be introduced mechanically by stirring or by using a blowing

agent. Physical blowing agents are normally gases like nitrogen or carbon dioxide.

Chemical blowing agents are special additives that decompose when they are heated or

react in such a fashion where gas is released. The bubbles grow and eventually settle and

then, as the liquid is cooling down, become solid.

In general, polymer foams have considerably lower strength and density than solid

metals. However, their low thermal conductivity coupled with their much lighter weight









make them an ideal insulator for the ET. Table 2-1 (Gibson and Ashby, 1988) lists

several properties of polymer foams and true solids (solid metals and ceramics).

Table 2-1. Properties of solids and polymer foams

Density (lb/in3) Conductivity (BTU/hr in F) Young's modulus (psi)
Solids 3.6 x 10-2- 3.6 x 10- 4.8 x 10 48 3 30 x 106
Polymer 3.6 x 10-4 4.8 x 103 3 x 103
foams

Initial Foam Material Testing

The Columbia disaster forced NASA and all those associated with the application

of the foam insulation material to the ET to see it in an entirely different light. Numerous

tests were conducted at cryogenic, room, and elevated temperatures to ascertain the

properties of this material. In addition to standard torsion and tension tests to obtain the

elastic moduli, various tests were performed to evaluate the failure characteristics, such

as the plane strain fracture toughness, KIc. While these tests are an important first step in

understanding how the foam behaves on a fundamental level, NASA also took extra steps

to examine how the foam sheds from the tank through a specialized test covered in a later

section.

Determination of Elastic Constants and Material Classification

Some of the initial work toward understanding the mechanical properties of foams

took places in the 1960s and 1970s. Gent and Thomas (1959), and Patel and Finnie

(1970) among others were among the first to explore the properties of cellular materials.

The model developed by Gibson and Ashby (1982), however, is perhaps the most widely

accepted way to analyze the properties of cellular solids. Their analysis is rooted in a

unit cell comprised of a network of interconnected struts (Figure 2-1) The equations that

arise from this mathematical treatment are in terms of the strut dimensions, which in turn




























Figure 2-1: Idealized foam microstructure

allow some material properties to be related to the relative density. The relative density

is the density of cellular material, p*, divided by the density of the solid from which the

cell walls are made, ps.

Generally, foams can be divided into two main categories: those with open or

closed cells. By closed we mean materials where each cell is partitioned from its

neighbor by membrane-like faces. Within open-celled structures the individual members,

struts say, are individually connected. With Gibson and Ashby's (1982) model, many of

the equations that determine various properties, such as the elastic moduli or the fracture

toughness are derived considering the configuration of the cell walls.

Understanding the mechanical and metallurgical aspects of these struts at a

microscopic level is very important. After all, it is these struts that transmit the applied

mechanical loads and transfer heat energy. But from a practical standpoint, models that

contain these microscopic parameters are not very useful to the engineering community.

The engineer may not have access to the equipment and facilities necessary to measure









things like cell wall thickness. Generally, the analyst, or engineer, may only be privy to a

quantity like the relative density. This parameter is one of the most important, and

useful, concepts that help define and understand the properties of cellular materials. The

relative density can be related to the dimensions of the cell walls. For honeycombs we

have

p* t
=c (2-1)
P, I

where t is the cell wall thickness and the 1 is the edge-length. For open-cell foams

-= C2 (2-2)
P, 1

Finally, for the foams of the closed-cell variety we have

p* t
=C3 (2-3)
P, I

The C terms are numerical constants that depend on the cell shape, normally taking a

value of unity.

As mentioned beforehand, certain open-celled foams can be modeled as a network

of beams. Using an idealized array of connected beams, the theory in many solid

mechanics texts such as Timoshenko and Goodier (1982) is adequate to determine the

deflections, strains, and stresses. It can be shown that the Young's modulus for the open-

celled foam is given by


E* C1EJ (2-4)
E 14

where E* is the modulus of the foam material. The relative density is related to the

dimensions, t and 1 as shown in (2-2). The second moment of the area, I, can also be

related to the cell wall dimensions









I oc t4 (2-5)

Now (2-4) can be re-written as

E C,P (2-6)


Using the same reasoning, we can write down the relationship for the shear moduli

P =C2 (2-7)


Lastly, we need v*, the Poisson ratio, which is defined as the ratio between the lateral and

axial strains. The strains are proportional to the bending deflection and their ratio is

constant. Thus, v* depends only on the cell wall geometry. We can show this for

isotropic foams where [as, the shear modulus is

E
p- ) (2-8)
2(1+ v)

Using (2-8) along with (2-6) and (2-7) we have

v* C= 1 = C (2-9)
2C,

The above analysis is for open-celled foams. When examining closed-cell foams,

the derivations are more complex because most foams are made from a liquid and surface

tension draws the material to the edges which might leave only thin membrane across the

cell. For sufficiently thin membranes, the open-celled formulas can be applied. But if an

appreciable quantity of foam is present in these membranes, the amount, or fraction, of

this material contributes to the stiffness. There are three potential contributions that sum

to the total stiffness of a closed-cell material. The aforementioned cell walls make a

contribution when they are stretched. The second contribution is present due to a fluid,









usually air, trapped within the cells, and finally the cell walls, and beams, can also be

bent when a load is applied.

The contribution from the stretching is derived by considering how the applied load

bends and stretches the cell face. The cell edge bending is proportional to 12S62 where S

is the stiffness of the cell edge, and S o Esl/13, and 6 is the displacement that arises from

the applied load. The contribution from the stretching of the face is proportional to

/2EsE2Vf. The Vf term is the volume of solid in the cell face and P oc 6/1 and Vf o 12tf.

The thickness of the edges and faces are denoted as te and tf, respectively. In the linear-

elastic regime, we can define the work done by the applied force as

1 aE I32 CL J12
I- 13 +E. 12tI (2-10)
2 /f

Equation (2-10) can be re-written when we note that I t4e and E* ocx (F/12)/(6/1) which

yields


=a' +/f' (2-11)
E, I4 1

For many foams the edge and face thickness is related to the relative density by the

following empirical relations

t- =1.4(1 ~b)p*
1 p,
(2-12)

t 0.9301/2 0P*
I P

If we substitute (2-12) into (2-11) we obtain the ratio of the elastic moduli in terms of the

relative density and volume fraction of material in the cell edges is 0









E*= C -2 +C' (1-: ), (2-13)
E, P C P,

where a, p, a', 3', C1, and Cl'are all constants of proportionality. The above equations

take care of the contribution from the stretching of the cell walls. The next contribution

comes from accounting for the fluid trapped in the cell walls, E*g. From Gent and

Thomas (1963) we have


Eg* p(12v*) (2-14)



where po is the initial gas pressure. Gibson and Ashby (1988) note that if this pressure

happens to be equal to the atmospheric pressure, this contribution is small. Lastly, shear

modulus for the closed-cell foam is shown below

G=C22 p +2(1-) (2-15)
E P, P,

Finally the last contribution to consider comes from the bending on the cell struts, and

that analysis is identical to the open-celled formulation presented earlier.

We note that the above relations are for foams where the cell walls are equiaxed.

Most man-made foams are anisotropic. When this particular foam is sprayed down,

foaming agents cause it to rise and the cells are stretched in the rise direction. Cell shape,

then, can significantly impact the material properties. The shape anisotropy ratio, R, is
R1 -
L2
(2-16)

R13
T3









where L, L2, and, L3 are the lengths of principal cell dimensions and L, is the largest of

the three. Following a similar analysis for the bending stresses within the beams for the

isotropic lattice Gibson and Ashby (1988) go onto define a quantity called the Young's

modulus anisotropy ratio for open-celled foams as

E *3 2R2
3- 2(2-17)
E*1 (1+(1/R)3)

2R
For closed cells an additional term (1- +) appears in (2-17). For the
1+ (1 / R)

anisotropic shear modulus, we have

3 (2-18)
G*1 1+R

The Poisson ratio is once again the ratio of lateral and normal strains and, just like in the

equiaxed case, is independent of the relative density. As such this elastic constant is

dependent solely on the cell geometry. Unfortunately, for the anisotropic foams,

employing the same type of dimensional analysis used to determine the elastic moduli for

the Poisson ratio will not offer any insight into its dependence on cell wall geometry and

Gibson and Ashby (1988) do not discuss this in their text.

So there appear to be two ways to try and determine the properties of the foam

insulation material. The properties can be estimated using the relative density and shape

anisotropy ratio or they can be experimentally evaluated. The BX-265 foam contains

closed cells, so the equations presented earlier for this particular cell wall geometry along

with corrections necessary to account for the anisotropy could be utilized for evaluating

various material properties in terms of the relative density.









Evaluating the all-important relative density, though, may not be as straightforward

as it seems. While it is not difficult to measure the density of the foamed material (one

simply weighs the sample and divide it by the measured volume), getting a handle on the

un-foamed density for this particular kind of polyurethane foam is not so simple unless

the exact same foam is available in the literature somewhere. Also, when NASA was

preparing various foam test panels (from which the test specimens are machined) they

determined the density could vary within the specimen. Also, there are several ways in

which the insulation can be sprayed down. The foam can be applied via a mechanical

process where a machine sprays it down, or it is laid down by hand using special

equipment that delivers the insulation to the testing surface (normally an aluminum

plate).

Thus, since the density could change throughout the test panel, and even vary

depending on the process by which it is sprayed, NASA developed a test program where

a vast database of information was to be established for this particular foam material. It

was decided that standard test procedures for evaluating both elastic properties and

fracture response were to be employed, instead of estimating the properties based on

relations that require the relative density to be known beforehand.

Figures 2-2 and 2-3 display some of the various test specimens used in various

tension and torsion tests. A summary of typical stress-strain data is shown in Figure 2-4.

Within this chart the results of several tests performed with different foam orientations

(denoted as local rise, spray, and axial) at a few temperatures (RT denotes room

temperature and LN2 is the temperature of liquid nitrogen).

























Figure 2-2. Square specimen tensile test used to determine the Poisson ratio

As expected the colder the temperature, the more 'stiff the foam becomes and in general

the foam fractures with little deformation. It should be pointed out that the foam is not a

material that behaves exactly like a generic isotropic steel alloy. Most engineering


Figure 2-3. Foam test specimens







16



Summary of Typical Tensile Stress Strain Curves
BX-265 / Category 4


002 004 006 008 01 012
Strain (inlin)


Figure 2-4. Summary of stress-strain data t

metals, when placed in a uniaxial tension test, have a linear relationship between stress

and strain up until the yield point. Cellular materials generally have a non-linear

relationship between stress and strain and the constitutive relations defined in this chapter

assume that this material has a linear stress strain relationship because a model that

accurately describes the response for foam has not yet been developed.

The aforementioned tension and torsion tests have led NASA to classify the foam

as a transversely isotropic material. These materials are sometimes called hexagonal

materials and five independent elastic constants relate the stresses to the strains in the

constitutive matrix. The 11-22 plane in Figure 2-5 is a plane of isotropy.


Source: personal communication, Doug Wells (MSFC)






















Figure 2-5. Coordinate system for the transversely isotropic material

Using the coordinate axes from Figure 2-4 we can define the constitutive relations for a

transversely isotropic material. The Young's moduli in the 11 and 22 directions are

defined as being equal as are the shear moduli in the 33, or rise, direction. The other

shear modulus, G12, is determined through a relation with Ell and v12: G12= 2(1+ v12)/Ell.

Table 2-1 lists the room temperature values of the elastic constants for the orientation

shown in Figure 2-5.

Since one of the objectives of this study is to analyze the effect of temperature, or

thermal loads, on the K-solution the material properties at various temperatures are

needed for a meaningful analysis. NASA evaluated the foam's elastic moduli, thermal

conductivity, and coefficient of thermal expansion.


E E2 E3
-21 1 0 0 0
Si E E,2 3 011
o22 -31 32 0 0 0 (2-19)

033 El E2 E3 033
723 0 0 0 1 0 0
713 G23 13
712 0 0 0 0 1 0 "12
G13
0 0 0 0 0
G,,

where the five independent elastic constants are










E. = E ;E
E11 E22; E33
G23 G13 (2-20)
V13 V31
,3 12 21
E3 El

Table 2-1. Measured material property values from experiment
En = 950 psi v12 = 0.45 G12 = 328 psi

E22 = 950 psi v31 = 0.3 G23 = 231 psi
E33 = 2400 psi v3 = 0.3 G31 =231 psi

These values are plotted in Figures 2-6 through 2-9. The tests performed by NASA

indicated that the Poisson ratios did not vary substantially with temperature. As of this

writing, only one value for k, the thermal conductivity, is available. It is possible for that

value to vary with direction as well as temperature. It will be assumed, however, that k is

isotropicc,' or has no direction dependency.

-0-Ell
Young's moduli --- E22
-- E33
6000

5000

4000

'. 3000

2000

1000

0
-550 -450 -350 -250 -150 -50 50 150 250
temperature (F)



Figure 2-6. Young's moduli vs. temperature


Source: personal communication, Doug Wells (MSFC)







19



Shear moduli


-500 -400


-300 -200 -100
temperature (F)


0 100 200 300


Figure 2-7. Shear moduli vs. temperature


Secant coefficient of thermal expansion

0.00016

0.00014

0.00012

0.0001

I 0.00008

0.00006

0.00004

0.00002


-500 -400 -300 -200 -100 0
temperature (F)


100 200 300


Figure 2-8. Coefficient of thermal expansion vs. temperature


Source: personal communication, Doug Wells (MSFC)


600

500

400

u' 300

200

100


-- G12
---G13
- G23


---as11
-- as22
--- as33











Thermal conductivity

0.0018
0.0016
0.0014
0.0012
0.001
3 0.0008
I-
0.0006
0.0004
0.0002
0
0 ------------------------------
-500 -400 -300 -200 -100 0 100 200 300
temperature (F)



Figure 2-9. Thermal conductivity vs. temperaturet

Fracture Testing for the BX-265 Foam Material

While many foam materials do not exhibit linear-elastic response before yielding,

brittle foams normally exhibit linear-elastic behavior (while in tension) up until fracture

takes place. Thus, we would like to use LEFM concepts to compute the near-tip stresses.

These analytical expressions assume that we are dealing with a continuum. So if the

crack is very small, cell size could influence these computations. Brittle fracture of

foams has been studied by Fowlkes (1974), McIntyre and Anderton (1978), Morgan et al.

(1981), Maiti et al. (1984), and Huang and Gibson (1991) among others. Perhaps the

most widely accepted theory for the brittle facture of foam is the Maiti et al. (1984)

study. They present a relation to compute the toughness for brittle foams as


KIC = C8f (Tl) 2 (p*/ )3 2 (2-21)









where Cs is a constant of proportionality and cfs is the failure strength of the struts.

Equation (2-21) can also be re-written as

KI, =C, KIJs (p/p* )3/2 (2-22)

where KIcs is the toughness of the solid material. Equation (2-22) assumes that KIcs is

proportional to Cfsl1/2 and once more the desired property is related to the relative density.

The concern, now, is at what length scale is such an equation valid.

Huang and Gibson (1991) analyze the fracture toughness of brittle honeycombs.

The primary aim of their paper is to examine the effect of short cracks and determine

some sort of correction factor for (2-21). They go on to show that cell size does indeed

impact the fracture toughness calculations. Using fractographic analysis, Brezny and

Green (1991) study short cracks in oxidized carbon foams. Their results dovetail with

Huang and Gibson's (1991) in that cell size can influence the fracture response.

However, in both studies short cracks are used in the various experiments. By

short, we mean crack lengths less than an order of magnitude greater than the average cell

size. But in Huang and Gibson's (1991) paper, the well-known relations based on

continuum assumptions seem to be valid as long as the crack length is an order of

magnitude larger than the cell size. Thus, if one is considering cracks much greater than

the size of the cell lattice, it seams reasonable to model the foam as a continuum.

To that end NASA conducted numerous single-edge bending SEN(B), compact

tension C(T), and middle tension M(T) tests to examine how the foam fractures under

load and used the regular, continuum-based relations to determine fracture properties,

such as the plane strain toughness. Pictures of these test specimens are shown in Figure

2-10. Since this material is anisotropic, each time the fracture toughness is calculated,









the material's orientation is supposed to be reported along with it. Figure 2-11

summarizes some of the tested geometries at various temperatures (left and right sides of

the vertical lines), for a few orientations, denoted by the labels along the bottommost

horizontal axis.




*0













Figure 2-10. From left to right: M(T), C(T), and SNE(B) fracture test specimens

While NASA utilized many geometries to test the fracture toughness, the C(T) test

was the most extensively tested specimen used to estimate a toughness. The plane strain

fracture toughness is usually obtained with slow loading rates (Barsom and Rolfe, 1999).

By plane strain, we mean thick plates, or test specimens, with deep cracks. The foam is

brittle and fracture is sudden with little or no stable crack growth. Wells (1966)

suggested that the fracture behavior near the crack tip can be characterized by crack tip

opening displacement (CTOD). He showed that CTOD is analogous to the crack

extension force (sometimes referred to as Go in the literature) and CTOD could be related

to the plain strain fracture toughness-KIc.

If the material is brittle and the subsequent load vs crack mouth opening

displacement (CMOD) curve (Figure 2-12) is linear-elastic and the C(T) specimen meets










the standards mandated by the ASTM, KI, can be calculated from this test method and

this approach is that NASA used to calculate the fracture toughness for the foam. In

order to determine KIc, a preliminary fracture toughness, KQ, is determined by the

following relationship (Anderson, 1991).

Where PQ (Figure 2-13) is maximum load applied to the C(T) specimen before

failure, B is the specimen thickness, W is the width of the specimen measured from the

centerline of the pin holes to the rear of the specimen, and f(a/W) is a dimensionless

polynomial function. The ASTM E399 standard denotes which function to use

depending on what specimen one is using. If the a/W ratio and B (thickness) are correct

per ASTM procedure and as long as Pmax is < 1.10PQ then KQ = KIc.


BX265 / Geometry and Constraint Overview I CAT 4
30.0 320F RT 320F R RT -320F 20F RT



25.0 -



20.0 ---------- ------ ------------------------------


I I

a <* i 0.
$ g
15.0 ------------ -------- --------------------------------------------










0.0
SEN(B)(A-S) M(T)(A-) M(T)(33-S) C(T)(A.S) C(T)(33-S)


Figure 2-11. Summary K-values for various fracture specimens









P
K B f (a/W) (2-5)




Clip




A CMOD







Figure 2-12. Clip gauge on C(T) specimen used to measure CMOD

Additional Testing: Divot Test Specimens

While NASA was conducting tests to evaluate the material properties at various

temperatures, additional experiments were devised to examine how and why the foam

becomes detached from the tank. Gibson and Ashby (1988) discuss the design of

sandwich panels with foam cores. These types of structural members are normally

comprised of two, stiff, skins separated by a lightweight core. These parts are utilized

heavily by the aircraft industry, particularly in applications where resistance to bending

and buckling loads are important, such as helicopter rotor blades. One mode of failure

for these types of structural members is decohesion of the adhesive bond between the

skin and the core.









Load, P



Pmax = PQ

/







A CMOD


Figure 2-13. Typical load vs CMOD curve for a brittle material

Often the epoxy bond is stronger than the core material itself. So if the interface

between the skin and core is defect-free, delamination is usually not a concern. The

situation changes if there are defects within the interface, however. This type of analysis

is complicated by the difference in elastic constants between skin, adhesives and core.

The delamination described by Gibson and Ashby can be likened to 'weakening' the

bond between the core and the outer skin. When the strength of the bond becomes too

weak, the foam core and skin can peal apart. The failure mode that NASA is

encountering, however, is more of an explosive and sudden 'blowout' of foam from the

ET.

Since the foam loss seems to be the greatest near areas where the surface of the

tank is somewhat uneven (from bolts and fittings), NASA believes that voids created

when the foam is first being sprayed down are the primary reason why the foam is able to

break off. These are indeed defects between the ET, or skin, and the foam, or core

material. To model this phenomenon, NASA devised an experiment to examine how









various voids within foam test panels could contribute to large pieces of foam being

blown out under certain conditions. Rectangular test panels like the one shown in Figure

2-14 are used for the 'divot testing.'


















Figure 2-14. Divot test panel

In this experiment a cylindrical bore is machined into the panel. At the top of that

bore, a sharp notch is created to simulate a crack, or defect, near this void. Liquid

nitrogen is then pumped into the bored hole and a heat flux is applied across the other

side of the foam panel. Two-dimensional schematics of this process are shown in Figures

2-15 and 2-16.

The applied flux warms the liquid nitrogen and eventually a phase change takes

place. The experiment takes place in a thermo-vacuum chamber to simulate the

environment the foam is exposed to during liftoff. As the Shuttle ascends to space, the

atmospheric pressure is dropping, but the gaseous N2 applies pressure on the walls of the

void and the crack faces. With enough force, the flaw turns and propagates toward the

surface.






















Figure 2-15. 2-D view of divot test

In Figure 2-17, we see the aftermath of one of these experiments. A frustum

shaped piece of foam has been blown out and the crater left behind resembles a 'divot,'

similar to the removal of a small piece of turf after a golf shot has been played. It is here

that predicting the crack turning angle could have a practical application.

Heat Flux

O\\\ \,\\


X Crack









Figure 2-16. 2-D view of test panel after heat flux has been applied

For a given void and flaw size, along with the applied tractions, it might be possible to

calculate this angle, denoted as coc in Figure 2-16, by calculating near-tip stresses and

applying a turning theory such as the maximum tangential stress criterion. The concept

of the crack turning angle and methods to compute it will be covered in a later chapter.


Sharp Notch


Void
Liquid N,























Figure 2-17. Foam blow out from divot test

Summary

The pertinent theory (Gibson and Ashby, 1988) regarding cellular solids is

presented and various formulas are available depending on what type (open or close-

celled) of foam one is analyzing. Formulae for anisotropic foams are presented as well.

The elastic constants, for both isotropic and anisotropic constants can be estimated

through the relations presented in the early sections of this chapter or via experiment.

While there is extensive literature available on foams, NASA decided to perform tests to

evaluate the elastic constants and fracture properties at various temperatures. This

enabled NASA to have a large collection of experimental data for several material

orientations, over a wide range of geometries.

Once the tests were conducted, determining the fracture properties can be done

using the well-known empirical relations for standard test specimens. One of the main

assumptions behind these equations is that they are to be applied to a continuum. The

foam is a porous material and it has been shown that cell size can impact fracture

properties if one is conducting analyses with short cracks. However, continuum









assumptions are acceptable as long as the crack is an order of magnitude longer than the

cell length.

Lastly, additional and highly specialized tests were performed to investigate how

the foam is able to free itself from the tank. NASA believes that voids near fittings and

bolts allow liquefied air to collect in these cavities prior to liftoff. During liftoff, the air

rushing over the outside surface of the insulation warms the liquid nitrogen which

converts it into a gas. The pressure, while small, appears to be sufficient to drive a crack

toward the exterior and this seems to account for the foam loss outlined in chapter one.

An idealized case of this process is examined through the divot test. These

experiments entail machining oval or cylindrical-shaped voids into a foam test panel and

pumping in liquid nitrogen. The cryogen is heated when a flux is applied to the top part

of the foam panel. At the top of the void, a sharp notch is inserted and when the nitrogen

is turned into a gas, the pressures exerted on the void walls and crack faces are sufficient

to drive a crack toward the surface. It is here that a determination of the crack turning

angle could be of some use to NASA and there a few theories prevalent in the literature

on this topic. Most of them require the evaluation of near-tip stresses and/or strains.

Since this an anisotropic material, the general equations for isotropic materials are not

applicable in most cases. In the next chapter a detailed discussion covering linear,

elastic, anisotropic fracture mechanics is covered and these concepts are applied in

chapter four when crack turning theory is presented.














CHAPTER 3
AN INTRODUCTION TO FRACTURE MECHANICS

Material Definitions

For local crack tip calculations there are material definitions, or classifications, that

need to be defined before moving on with discussions on how to compute near-tip

stresses for materials that are anisotropic, or have direction-dependent properties. In

general, most texts on fracture mechanics focus on materials that are isotropic. Here the

constitutive matrix contains three elastic constants, two of which are independent, E and

v. For isotropic materials, there are an infinite number of planes of material symmetry

and the strains and stresses are related to each other via equation (3-1).

E 1/E -v/E -v/E 0 0 0o ~ x
Eyy 1E -v/E 0 0 0 a
S1/E 0 0 0 C- (3-1)
y" 2(1+v)/E 0 0 a
,, 2(1+v)/E 0
7, sym 2(1+v)/E_ c

Many other types of materials can be characterized by the number of planes of

internal symmetry (Dieter 1976). Cubic materials, for example, have three independent

material parameters and nine planes of symmetry. Materials with three internal planes of

symmetry are known as orthotropic materials and many engineering metals fall into this

category, such as rolled aluminum. A special class of orthotropic materials has a single

plane of symmetry that is also isotropic. These are known as transversely isotropic

materials and the constitutive relations for this material were defined in the previous









chapter. Monoclinic materials possess a single plane of material symmetry, but the

behavior in that plane is not isotropic.

Being cognizant of the constitutive relations for the particular material in question

(be it isotropic or fully anisotropic) is important because many of the initial derivations of

near-tip stress fields for anisotropic materials tried to take advantage of material

symmetry to decouple 'in-plane,' and 'out-of-plane' displacements, which, in turn, makes

the mathematical formulation a little less complicated.

Isotropic Fracture Mechanics

For isotropic materials, the near-tip stress fields have been analyzed by

Westergaard (1939), Sneddon (1946), Williams (1957), and Irwin (1960) among others.

These solutions pertain to specific cracked configurations with applied far-field loads and

are governed by a single parameter, K, the stress intensity factor (Anderson, 1991).

K
0, (r, o) = f,j (o) + higher order terms (3-1)

y



P(x,y)

Crack faces





Figure 3-1. Coordinate system used in equation (3-1)

The coordinate system for equation (3-1) is shown in Figure 3-1, where the fij term is a

trigonometric function. We note the Greek letter, theta, is normally used to define the

angle with the horizontal in the polar coordinate system presented above, but theta will be








used later on to define the material orientation. A different variable name is selected to

avoid confusion. The key points about (3-1) are that the near-tip stresses have no

dependence on the elastic constants and the first term is proportional to 1//r and,

therefore, the stresses approach infinity as r is made smaller and smaller. Also, as r

approaches zero the 'higher order terms' in (3-1) get smaller or remain at some finite

value. Thus, in many texts and papers the analysis is restricted to small values of r such

that the higher order terms are neglected.

There is typically a roman numeral written next to the stress intensity factor to

denote the mode of loading and Figure 3-2 displays the three modes of deformation.













Figure 3-2. From left to right: mode I (opening), mode II (shearing), mode III (tearing)

While the stresses are proportional to 1/Nr, the displacements are related to
(3-2).

0) 2 0) 2 0
fu2 2 2 2 KI
x 2, rcos K -1+2sin2 0) sin K+1+2c05 C o3
uy =-s sin fK+l-2cos2 -cos-2 K-l-2sin2j 0 KII (3-2)
u 2, 2 2s 2J 2l 2( 0
Zu KIII
0 0 sin-
2









where K = 3-4v for plane strain and K = (3-v)/(1+v) for plane stress and ps is the shear

modulus.

Linear Elastic Fracture Mechanics for Anisotropic Materials

In general, the equations in the previous section cannot be applied to this problem

directly because the material properties are direction-dependent. The analytical work

geared at developing the near-tip stress fields for elastic anisotropic materials has been

examined by Sih et al. (1965), Bowie and Freese (1972), Barnett and Asaro (1972), Bogy

(1972), Tupholme (1974), Kuo and Bogy (1974), Rathod (1979), Hoenig (1982), and

Dhondt (2002) among other researchers.

The important work of Sih et al. is one of the most cited references for determining

the stress fields near the crack tip for anisotropic materials. Following Westergaard's use

of stress functions and complex variables to determine near-tip stresses for isotropic

materials, Sih et al. use a similar approach for materials with direction-dependent

properties. Whenever stress functions are utilized to solve this kind of problem, there is

normally a corresponding characteristic equation (a detailed explanation of the

characteristic equation can be found in Appendix A). For the most general case, the

order of the characteristic equation is six. In Sih et al.'s paper the material in question is

monoclinic and the crack is resting in the symmetry plane. The constitutive relations for

monoclinic materials are shown in equation (3-3).

EX S1 S,, S3 0 0 S16 x
EY S22 S23 0 0 S26 (y
e, S33 0 0 S3 (33)
Yyz S44 S45 0 O
7Y55 0 O
7, sym S66-









This configuration is advantageous because the in-plane and out-of-plane

displacements become decoupled, which in turn reduces the order of the characteristic


equation to four.


Re _L 2 LRe[2 1 Re

Re 17r2 Rer i r1 1

Re I 1 _1) Re [_ 1 1 P
A 0 /2 0 Q1 Q A -lP2)Q1 Q2

0 0


C. = (S31 + S32, + S36- )/S33

where the elastic constants in compliance form are defined as

, = Sc j


Re[ r (/ipP2Q P2 )] Re( 2 ('2p, P1)]

Re 1 '2 2q q -qQ)] Re [ (q2Q2 -ql)

0 0


0

Re Q3
C45 + P3C44


KI (3-7)
KII
[Kill]


The roots l1i, [l2, and [l3 for this particular formulation come from the special case for

monoclinic materials listed in Appendix A.

P, =S',11 u -S',16 u +S ', (3-8)

q, = S',12 / S'26 + S'22/ (3-9)

Sih et al. enforce a plane strain condition and the reduced compliance matrix becomes


1
: 2-- r


IyyI

y0([7
(7,
CyY


0

0

0

-Re ii

Re[1
.a3


KI
KII (3-4)
KIJJ


(3-5)


U
/dzI


(3-6)










S' = S, (3-10)
S Y S33


Q, = Jcos ) +/ sin (3-11)

The C45 and C44 terms in (3-7) are the elastic constants which are in stiffness form

cr = Cec (3-12)

where

C, = S, (3-13)

The aforementioned decouplingg' of the anti-plane displacement is now evident in

equation (3-7). The Uz displacement does not depend on KI and KII. This is similar to

the isotropic case in equation (3-2); the z component of displacement only depends on

Kill.

When analyzing the foam problem for NASA, a crack, or defect, can be at any

orientation relative to the insulation. In that sense, it does not seem very practical to

employ the Sih et al. solution unless we know the crack always lies in the symmetry

plane. More general solutions, ones that do not require a decoupling of the anti-plane

displacements, will be used instead.

The asymptotic solution derived by Hoenig is the one we will adopt to get the near-

tip stress fields. Materials with virtually any orientation, irrespective of the location of

the crack front, can be modeled with this method. Hoenig uses the pioneering work of

Lekhnitksii (1963) to derive the stresses and displacements given below by using

complex variables and stress functions to derive a coupled set of differential equations

whose solution entails determining the roots of the ensuing sixth order characteristic









equation. We notice that K once again arises as a scalar multiplier just as it did for the

isotropic solution.

1 A3 2NJ 1K-
-x = Re /cos + u sin Ro
1 3 'K
0- = ReZ
,!2,r ,=1 /cos o) +/ sin a
1 3 / A,uN, Z K
Se= uNRe J'K (3-14)
v 2;r ,=1 /cos cw + sin o
-1 3 N
" Re- AN, 1Kj
/2,r ,=1 /cos o0 + ,u sin oa
-1 e N, 1KK
r Re
x 2 r ,=1 ,/cos 0 + /, sin o

S= -(31 + S32yy + S34yz +S350, + S36 x )/S33 (3-15)


u = Re m NJlK, N (cos0 + j sin)) (3-16)


Here an important distinction is made between the general case and Sih et al's

formulation. Notice in (3-16) how the mode I-III SIFs are to be summed over all three

modes of displacement. Thus, all three components of displacement are dependent on

KI-KIII.

n 1, = S1 ', 2 S'16+S' +A, (S'15 S'14)
=' -S\ 24q (3-17)
m,, = S', -S'26- S2 +A, S S,24(3-17)


m3, = S 41 /, -S'46 42 S45 S'44
/A /A









The u, are the roots that occur in conjugate pairs and they arise from the sixth order

polynomial, equation (3-18). The coordinate system used in equations (3-14) and (3-16)

is the same as the one used in (3-1).

14 (/) ()32 (/)=0 (3-18)

Where

1 ) = SLp-2 2S'4p + S44
13 ()= S 3 (S'4 + S6)2 + (S5 + S6)- S4
4 (p) = S[4 2S'6p3 + (2S2 + 66)2 2S6p + S

The N matrix in equation (3-14) is defined as


N, = -p -2 -J3 (3-20)


Where j is defined as 13 (02)/l(u,).

There are two situations, also called degenerate cases, where this solution is

invalid, however. If anti-plane shear and plane strain displacements becomes decoupled-

for monoclinic materials, say-Hoenig's formulation falls apart. He does present an

alternate solution for monoclinic materials which turns out to be consistent with the Sih et

al solution mentioned earlier. His solution also encounters difficulties if the crack is

lying in a plane that is isotropic. The N-matrix, equation (3-20), becomes singular

because l1 and 02 = i. However, we have near-tip solutions for the isotropic cases.

These are shown in equations (3-1) and (3-2).

Determining the Stress Intensity Factor-K

In the above sections, it is necessary to develop solutions for determining the stress

intensity factor, K. Over the past 50 years, numerous methods have been developed to









determine this parameter. For simple geometries, a handbook (Tada et al, 1973) of K-

solutions could be consulted. Alaibadi and Rooke (1991) present an excellent summary

of the literature on the subject of numerical evaluations of SIFs. They divide many of the

available methods into three main categories, denoted as stages in Table 3-1.

Table 3-1. Various methods for determining the SIF
Stage 1 Stage 2 Stage 3
Handbooks Superposition Collocating, or mapping
Stress concentration Integral transform
Stress distribution Body force method
Green's function Method of lines
Approximate weight function Finite element method
___Compounding Boundary element
Alternating technique

As one might have guessed, the procedures listed in the Stage 3 category are the

most accurate. Since computers have become less expensive and more powerful in recent

years, many of the current studies geared toward K-solutions incorporate the use of the

finite element (FE) method or boundary element (BE) techniques. Boundary element

methods are particularly attractive because the dimensionality of the numerical problem

is reduced, i.e. 2-D problems involve discretizing the line-boundary of the domain and

for 3-D problems, just the surface of the domain is discretized.

This analysis will call on the FE method to help compute K, and there are a few

ways of determining it with this method, such as: extrapolation of near-tip stress and

displacement fields, Rice's J-integral (1968), strain energy, and the virtual crack

extension method.

FRANC3D Next Generation Software

The Cornell Fracture Group has developed software capable of creating crack

meshes and computing an anisotropic K-solution. This particular method of solution

would fall under Stage 3 in Table 3-1, and in this case Rice's J-contour integral along









with displacements output from FE software will be used to determine K along the crack

front. This software is very versatile in that it is not restricted to analyzing any one kind

of material; K-solutions can be generated for isotropic and fully anisotropic materials

alike.

Since many FE models were built and the subsequent K-solutions computed with

this software, it would be prudent to discuss the pertinent theory (Banks-Sills et al, 2005)

that governs this particular way of calculating K. Rice's J integral, a conservation

integral that measures the energy flux into the crack tip, is equivalent to the strain energy

release rate (denoted as G) for small scale yielding. Furthermore, it can be shown for a

plane problem that only has KI and KII that

G = J = a(K2 + KII2) (3-21)

Where a = (-v2)/E for plane strain and a = 1/E for plane stress. One can see that the J-

integral equation, as is, cannot give us the SIFs on an individual basis.


J = Wdy T ds (3-22)


with W = (1/2)oij8ij, also known as the strain energy density, T = ijnj and ui is the

displacement vector.

The path, F, is defined in Figure 3-3

y
n


~~=== ---V-x


Figure 3-3. Path of the J-integral









The J-integral was initially formulated for 2-D problems, however since this

material has properties that are direction dependent, our analysis requires the use of 3-D

FE models. The J-integral, then, needs to be modified so it can be used in integration

that take place over volumes. Following Li et al (1985), the J-integral can be rewritten as


iJ= cO, W8 q ds (3-23)


In this formulation, the crack is undergoing a 'virtual extension,' and only a small portion

of the crack front is being advanced. When using the formulation from the Li et al

paper, the J-integral is now evaluated in a cylindrical domain centered over the crack

front.

Crack front
X2

Virtual crack advance










X3

Figure 3-4. Cylindrical domain and virtual crack advance for a 3-D crack front

Following Figure 3-4 we can define incremental area of virtual crack advance as

A = Aama q (s) ds (3-24)
AL

The definition of J in equation (3-22) requires F to be very small. It is difficult to

compute stresses and strains about a vanishingly small path about the crack tip. Instead,









area integrals (Moran and Shih, 1987) are used (for 2-D problems) because they are

easier to implement numerically. So, the q term in (3-23) is simply a mathematical tool

that enables us to recast the J-integral into a slightly more usable form.

This virtual extension method is the modem-day approach to solve computational

fracture mechanics problems. One of the main advantages of such a method is the ability

to compute the energy release rates across the crack front for 3-D problems. Since the

energy release rates can vary over the front that implies K can also change with respect to

crack front position as well

JJ(s)q,(s)ds j
J = (3-25)
Sq,(s)ds Aq

where qt(s) is the crack tip perturbation.

Rice's J-integral does not allow us to look at the SIFs on an individual basis for a

mixed mode crack problem. This hurdle was first surpassed by Ishikawa (1980) where

he decomposes the stress, strain, and displacement fields into symmetric and anti-

symmetric parts. This allows him to use the J-integral to compute the Ks individually.

Ishikawa's method, though, has not been extended for anisotropic materials. Yau et al.

(1980) uses the M-integral developed by Freund (1978) to determine KI and KII,

individually, by using the idea that two separate and independent equilibrium states

(denoted by superscripts (1) and (2)) for a linear elastic, isotropic material can be related

to a third equilibrium state, denoted with no superscript. That is to say the stresses,

strains, displacements, and SIFs from these separate equilibrium states can be superposed

()= 1 + (2 KI= KI( + Kij(
S (1) + E(2) KI = KII(1) +KIY(2) (3-26)
S= (1) u (2) Kill =KIII) + KIII2









The J-integral can now take the form

J= J1 + (2 +M1,2) (3-27)

The M(1,2) term is known as the M-integral initially derived by Freund and it can be recast

in terms of the aforementioned equilibrium states

-F (2) FUn(1) O1 q
M/(1,2) = 0-(1,2) +j(2)t -(2)(I'2)IJ (3-28)
F a 9x, a9 ] }9x

This integral describes the interaction between the two equilibrium states, and sometimes

the M-integral is called the 'interaction' integral in the literature.

The first step in the procedure to determine the SIFs is to define the strain energy

release rate using Irwin's (1957) crack closure integral. Consider the crack in Figure 3-5.

Now a compressive, or closure stress, is applied in such a fashion to clamp the crack

faces down along the length 6-r. The work it takes to perform this closure can be related

to the energy release rate, G. Irwin goes on to use the force displacement curve and

equate that to the work required to close the crack. Substitutions are made for near-tip

displacements and stresses and eventually an integral expression is obtained that relates

the energy release rate to the stress intensity factor.

In a similar manner, Hoenig (via Equation 3-30) provides us an analytical

expression for the energy release for general anisotropy in terms of the SIFs and material

constants. Irwin's work was initially developed for brittle, isotropic materials; his

expression for the energy release rate includes the isotropic Poisson ratio and the Young

and shear moduli.

1
G lim-a (3 r, 0)-.u (r, r)dr (3-29)
0


















IX





Compressive stress








Figure 3-5. Definition of Irwin's crack closure integral

Now, if the stress and displacement fields are substituted in (3-29), we have

G= = [KI m(m,,N, 1K,)+KIIIm(ml,N, 1K,)+KIIIm(m,,N, 1K)] (3-30)

When we use the relations for SIFs in (3-26), we get

1 (KI( + KI(2) 1M(K2 N ) + (KII() + KII(2))m(m2 N, -1K,)
G 2 +(KIII(1) + KIII2))m(mN,1K) (3-31)


The terms in (3-27) have bars over them because a certain definition of J is used in (3-

23). Let us define an alternate form of (3-27) where J is not normalized with respect to

the extended area as

J = (1) + (2) + M(1,2) (3-32)








Now we substitute (3-31) into (3-32) and for clarity the individual terms of that equation

are listed below


2 (3-33)
(2) = [K(2) m(2,N -(2))+ KI(2) Im(m-,N1K(2)) + () I(m3,N 1K(2))]

M(1,2) = [KI Im( 1K(2)) +KI(2) Im 1K(1))

+KII) Im (mN, 1K 2)) + KI(2) Im (m1,NJ K(1)) (3-34)
+KIIm ( K 2)) + KI (2) Im (m3N 1K 2)

It is now evident how the decomposition of displacements, strains, and stresses into

separate equilibrium states is helpful. One can see that we have two definitions of the M-

integral, equations (3-28) and (3-34). If they are equated, the individual SIFs (KI(1),

KII'), KIII1)) can be computed. But before that step, we must define 'auxiliary'

solutions for the second equilibrium state, denoted by superscripts (2a, 2b, 2c). The

terms with the (1) superscript are going be computed numerically via FE software.

The three auxiliary solutions are defined as: 2a, 2b, and 2c. For case 2a, KI(2a)= 1

and KII(2a) = KIII(2a) = 0. For case 2b, KII(2b) = 1 and KI(2b) = KIII(2b) = 0, and for case 2c,

KIII (2c) = 1 and KI(2c) = KII(2c) = 0. Since the SIFs are being prescribed beforehand, the

stresses, for those equilibrium states, can be computed from (3-14). Now the two

relations for the M-integral are equated

KI(1) Im(m2,Nj-1K (2))+
"(2) d(1)(2) zN1K))+

x 01 0 1 KII() m(mlN 'K (2))
(3-35)
A, 2 KII(2) Im(1,N- 1K (l))+
KIII() im(m3, N 1Kj (2)+
KIIIl(2) m(m3, N-1K (2)









From (3-35) we can assemble a system of equations that contain the three unknown SIFs

we are after

22m (m2A1) I (mN1+ m, +AN) I \(M +11 m2N31 )KI (1 H 'M1/A
Im (M A +m 1 2 Im (mA,) Im ( i3 KIJ(1) ={ (1,2b)/Ad (3-36)
Im m2^N 1 +ml^N1) =Imhn^N^ + N )
Im m2A31 +M3I1) Im MilA31 +M3 21 2Im (M3 331) K-I(1) (1,2) A


Finally, we note that (3-36) involved the extraction of stresses near the crack tip. These

stresses are taken from the Gauss point locations within the elements that encompass the

crack front. Also, the integral in (3-36) is evaluated numerically using Gauss quadrature.

Computing SIFs Using Displacement Correlation

Displacement correlation (DC) is perhaps the oldest and simplest method to

compute SIFs via FE displacements. Chan et al (1970) were among the first group of

researchers to employ such a method to obtain SIFs this way. In general, the correlation

point is selected where the crack tip displacements are the largest so the relative error in

displacements is small. Another advantage in using such a method is the SIFs are

separated into the three modes defined in Figure 3-2. In its most basic form, crack tip

opening displacement (CTOD) can be used to determine the SIF by equating the

displacements at the points in question to some analytical solution.

From equation (3-2) with co = t7 and Figure 3-6 for a plane strain problem dealing

with an isotropic material, we have

K b- (uyb ,) 2u (3-37)
K+1 V r

(u(uxb-)2l/ 2z (3-38)
K+1 V r









where [s, again, is the shear modulus determined by E/2(1+v) and K = 3-4v for plane

strain We note that Uy and Ux are the opening and sliding modes of displacement. The

above method is sometimes called a 'one-point' matching technique since it involves

extracting displacements from one point near the crack tip. The downside of such a






a
y








Figure 3-6. Correlation points typically used for displacement correlation

method is that very refined meshes are needed for accurate results. Improved accuracy is

obtained through the use of a special element specifically designed to capture the

singularity present at the crack tip.

From equation (3-1), we see how the stresses have a 1/'r singularity at the crack

tip. For accurate computations, we would like our FE models to also have this

singularity. Quarter point elements are one way to do this. The first attempts at

modeling cracks were done so using quadrilateral elements with the mid-side nodes

moved to quarter point locations (Barsoum, 1976) When this is done the desired strain

singularity is achieved. However, one downside of using this type of element is that the

singularity is only present on the edges that contain the quarter point. If triangular-

shaped elements are used instead, the singularity exists both along the edges and within

the element (Anderson, 1991).

In Figure 3-7, we see how an eight node quadrilateral element can be transformed

to a triangle and then certain nodes are moved to the quarter point locations. Let us









derive how the required strain singularity is obtained along the edge of the quadrilateral

element with the mid-side nodes moved to the quarter point locations. Consider the

bottom edge of the element on the right in Figure 3-8. Nodes one, five, and two lie along

this boundary. In order to show how the strain singularity comes about, we will need to

list the shape functions for this particular element. The shape functions (Logan, 2002) for

the isoparametric element at the corner nodes are

N, = -(1 + ,)(1 + 1,)(1 +r, -1) (3-39)

where i is the number of the shape functions at the corresponding node (i.e. N1 is the

shape function for node 1) and

S=1,1,1,-1 (i = 1,2,3,4)
77, =-1,-1,1,1 (i = 1,2,3,4)

We also have shape functions for the mid-side nodes

N, = 1 ( 2)(1+ ) t=-1, (i= 5,7)
2 (3-41)

N, = I(I- )(l+7 72) =-1,1 (i= 6,8)

The global origin is placed at the corer of the element. We will need the shape functions

at nodes one, two, and five. Setting r = -1 we have

N1 =- (1 ~)
2
N, = (1+) (3-42)
=2 )
N5 =(I 2)









Isoparametric
quadrilateral element


Degenerated isoparametric
quadrilateral element /


SNodes moved to
quarter point locations
1/4 pt elements placed circumferentially
about the crack tip

Figure 3-7. Triangular quarter point elements


7 3 4



L 6


1 x 5 2


7 3


8


-/-- 3L/4
L/4


Figure 3-8. Isoparametric element and degenerated element with mid-side nodes moved
to quarter point locations

The global x coordinate is related to the isoparametric coordinate system via


(3-43)


where the repeated index is an implied summation. Setting xl


0, x2 = L, and x5 = L/4


we have

1 )L
2 4


x=Nx,


(3-44)









We can now solve for


= 2J -1 (3-45)

In the same way that we can relate the global x coordinate to the shape functions, we can

also determine the displacements using the shape functions

u = Nu, (3-46)

Again determining the displacements along the edge containing nodes one, two, and five

we write


2 2
u= 1-( l + (l++)u,+(1_- (3-47)


Now we can substitute (3-45) into (3-47) to obtain


u-- 1+2 2-2 u+ -1+2 2 u2+4 5 (3-48)


We note that the displacement shown in (3-48) has a /lx term present. This is consistent

with equation (3-2) where the displacements vary with /r. We need to differentiate

equation (3-48) to obtain the strain in the x direction

u 1 3 4 1 4 U 2 4
SCJ7 q. 4 I+- /_+- u2+ U5 (3-49)
a x 2[xL L 2 f Zj 2L fxL- L

Finally, it is shown in (3-49) how the strain displays the desired singularity (x-1/2).

One can see why the quarter point element has been the standard for modeling cracks for

the past 30 years. No special programming, or internal tampering, with the FE code is

required. Any FE program that carries quadratic, or higher order, elements can support

these special ones with the nodes moved to the quarter points.










Let us now return to our discussion of using displacement correlation to extract

SIFs from FE models. If quarter point elements are used, equations (3-37) and (3-38) are

modified because r = L/4


KI -(ub )4p 2 (3-50)
K+1 L


KI= (U -Ux,)4p 27+l (3-51)
K+1 L

The SIFs extraction via the DC method need not be restricted to just one set of nodes, or

correlation points. One can also consider the displacements over the whole element. The

form of these equations depends on the shape functions of the elements. Ingraffea and

Manu (1980) derive a DC extraction method using 3-D elements. This formulation is

particularly useful to the present study since 3-D simulations are used to model the

fracture response of the foam material. Ingraffea and Manu use a special labeling

convention shown in Figure 3-10 for a collapsed 20 node brick element (Figure 3-9), also

known as a 15 node wedge.



3 20 7 3 20 7
515
4 10 17 1 14 *10 14

12" 2 19 16 11 2 19 15
9 ___13 9 y
1 A 18 5 1,4,12 17.18 5,8, 16


Figure 3-9. 20 node brick and 15 node wedge elements

The shape functions for a 20 node brick element are listed below. The local (S,r, )

coordinate system is placed at the center of the brick element. For the corner nodes at

positions i = 1,2... 8 and i, ri, (i = +1









N, = )( )(l (a, + +C, 2) (3-52)
8

For the nodes at positions i = 17, 18, 19, 20 and ji = 0, ri = 1, (i = 1


N, = (3-53)
4

At i = 10, 12, 14, 16 and i = +1, ri = 0, (i = +1


N, = (3-54)
4

Finally, for the nodes at positions i = 9, 11, 13, 15 and ji = 1, ri = 1, i = 0

(1+ C, ) (1 + ~ ) )(1+ ;2)
N, = (3-55)
4

For anisotropic materials, Saouma and Sikiotis (1986) extend Ingraffea's work for

anisotropic materials. They use the displacement relations developed by Sih et al. A

similar procedure is followed in this study, but instead of using displacement equations

from Sih et al's paper, the relations in (3-16) are to be employed. The equations to

determine the anisotropic Ks, using Hoenig's displacement equations, are as follows

KI
KII =[B] [A] (3-58)
KIII 2L

where [B] is

mIN1, (cos a + p, sin C)
[B]= Re m2 Nl (cos m+ ,u sin ) (3-59)
m3ljN (cosu +, sin o)














C'
x, u A BC'


r I


2C C





















+12 (vF +v> Dv -v u, -
C E B '
.e

'' l




Figure 3-10. Collapsed 20-node brick element


2v, v, +2vE +v, 2v, +vcv, 2vE +Fv, VD + (-4v, +v, +4v, v, + 4v, ve, 4v,, + v,,)

+- (V2 +VC -2VD -F -v, 2vD)
2 f T o t F (3-57)
HC -+ 2" +u, -1 .+uc, -- .+uC,-uD,+-UD (-4u,+uc+4u- UF +4u,-u,,-4uE,+U,')

+2 (2F +UC ".-UF'-C -C )

11 -wc+ +w, -1 2.+w,,-1 +wp, -wD, +- (-4w, +wc +4w, +4w, w, -4wE, +FW,

+-^ ( -+W F-:. '+WC' -2.. F CF


Once again, the repeated indices are an implied summation. The terms in the B matrix

only depend on the roots of the sixth order characteristic equation.

Effects of Temperature on the SIF Solution

In chapter two it is pointed out that temperature plays no small role in causing the

foam to break off from the ET. For example, when Shuttle is resting on the launching










pad, one side of the foam is exposed to the temperature of liquid hydrogen and oxygen;

roughly -3000F. The other side of the foam is exposed to air, perhaps at 75 oF prior to

liftoff. Over typical acreage sprays, the foam is approximately three inches thick. This

small thickness coupled with a large thermal gradient suggests that thermal strains could

be significant. As shown with the divot testing in chapter two, even though there are no

applied far-field mechanical loads, the thermal gradient seems to be sufficient enough to

impart stresses capable of driving a crack toward the surface of the ET resulting in a loss

of material.

To account for thermal strains, we simply add this term to our constitutive matrix.

Following Hyer (1998), the stress-strain relations can be written as

E -T, (T, T7Y) SS,1 S1 3 0 0 0 a0-

E2 Th (T, 7f) 21 S2 S 23 0 0 0 a22

E33 E33 (T,ef) 31 32 S 33 0 0 0 33 (3-60)

723 44 0 0 023

713 S5 0 13
712 sym S66 '12


where the total strain is {fll 822 833 723 713 712 Th11(T, Tref) is the free thermal strain in

the 11-direction, and Tref is the reference temperature. The mechanical strains are

1nmech 11 Thll(T,Tref)
mech Th
Ell -11 \ IJTre/)
.22mech E22 -E Th22 (T, Tf )

E33ech 33 Th33(T Tf ) (3-61)
mech
723 Y23
mech
713 Y13
mech
Y12 Y712

or, stated more concisely the total strain is

S mech thermal (3-62)
Total= me + (3-62)









If the thermal expansion is linear with the temperature change, we can rewrite (3-60) as

11 -a AT S11 S12 S13 0 0 0 o011
E22 a2AT S2 S22 S3 0 0 0 022
33 -a3AT S31 S32 S33 0 0 0 033 a
723 S44 0 0 023
713 55 0 13
12sym S66 012

The ai term is the coefficient of thermal expansion and AT is the difference of

temperature between the state of interest and the reference state.

The above equations allow us to compute thermal strains in the numerical models.

Incorporating thermal and mechanical loads will involve two separate steps because a

thermal gradient implies that a temperature distribution must be obtained before solving

the model that contains both the thermal and mechanical boundary conditions. The first

step is to solve the conduction problem using the thermal conductivity coefficient listed

in chapter two. The nodal temperature distribution is saved for later use. In a subsequent

run, now with mechanical loads, the nodal temperatures are carried over as body forces

along with any applied mechanical loads.

Summary

Summing up, the analytical expressions from Hoenig's near-tip solution is

presented. His equations depend solely on the roots of the characteristic equation and the

stress intensity factor, K. Hoenig's solution is not as well known as Sih et al's paper and

as such the vast majority of studies dealing with near-tip solutions for cracked anisotropic

bodies use the Sih et al formulae to compute the stresses. This is acceptable as long as

one realizes those particular equations are valid for certain configurations where a

decoupling of in-plane and out-of-plane displacements takes place. Hoenig's solution is









more general and his formulation of the energy release rate is the one chosen to compute

the SIFs within the FRANC3D software. The DC technique presented also uses Hoenig's

equations, but only the ones that pertain to displacements.

These two methods, interaction integral and DC, are presented to show how the

mode I-III SIFs can be computed. These formulations are different in a few ways.

Computing K with the J-integral is inherently more accurate because only a moderate

level of mesh refinement is needed for the solution. However, the implementation of

such a method is very involved. Alternatively, the DC approach is much easier to apply

but the accuracy of the solution does depend on the level of mesh refinement.

Finally, the relations dealing with handling thermal strains are presented. These

equations assume an expansion that is linear with temperature change. Handling thermal

loads is straightforward by first solving the conduction problem to obtain the nodal

temperature distribution. With that, those values are forwarded as body forces that are

applied alongside the mechanical loads.














CHAPTER 4
CRACK TURNING THEORY AND FINITE ELEMENT MODELING

Crack Turning Theory

There are three theories prevalent in the literature to predict incipient crack turning

angles for isotropic materials: maximum energy release rate, minimum strain energy

density, and maximum hoop stress. All of these theories are based on LEFM concepts

and were initially developed from experiments involving brittle plates.

The turning theory proposed by Hussain et al. (1974) seeks the direction where the

energy release rate, G, will be a maximum. The minimum strain energy density theory

(Sih, 1974) postulates crack turning in the direction where the strain energy density is at a

minimum. The maximum hoop stress theory proposed by Erdogan and Sih (1963)

predicts crack propagation in the direction whereoo, defined in equation (4-1), is a

maximum relative to the crack tip. Crack turning, in this study, denotes the incipient

turning angle. We define the crack turning angle, wc, below in Figure 4-1.

For isotropic materials, the near-tip stresses (in the r, co polar coordinate system)

have been derived by Williams (1957). If the derivative with respect to theta of equation

(4-1) is taken and set to zero, the critical angle, oc, can be determined; equation (4-4).

1 C 3 a) 3
) = COS- KI COS- KIIsin (4-1)
2irr 2 2 2


S= cos- KI 2+sin- +-KI sino-2KIItan- (4-2)
2 r 2 2 2 2
























Figure 4-1. Definition of crack turning angle

o- = 2 r cos [KIsinco+KII(3cos- 1)] (4-3)



) = 2 tan 1l 1 + J (4-4)
4(KII/KI)

The maximum hoop stress theory is the easiest to implement and that is perhaps the

main reason why it is widely used for turning angle predictions. The minimum strain

energy density criterion is also very popular and there is some debate (Gdoutos, 1984) as

to which one is superior. Some might argue that it is not correct to use just a single

component of stress to predict the incipient turning angle, whereas a quantity such as

strain energy density seems to be more comprehensive as all components of stress and

strain are included. Maccagno and Knott (1989, 1992) study mixed-mode facture

behavior of PMMA and lightly tempered steel alloys and in both cases the crack turning

angles are well approximated by the maximum tangential stress theory.

There have been some topical investigations for crack turning within materials that

have direction dependent properties. Buzcek and Herakovich (1985) predict the crack









extension angle for orthotropic materials and some of the recent work in this area mirrors

their ideas. They assume the tensile strength, T, of the orthotropic material varies with

direction, r. Since it is difficult to measure the strength of a material in all possible

orientations, they assume that T = f(r, 0, XT, YT) where r is the angle in a polar

coordinate system, 0 is the material orientation and XT and YT are the strengths of the

material in the axial and fiber directions respectively (they are analyzing orthotropic

composites). The equation for T, also denoted as a fracture resistance parameter, must be

independent of r if the material is isotropic. The lowest order function that satisfies these

requirements is an ellipse. This function is plotted in Figure 4-2 in (T,,, r) polar

coordinates. Now T, can be expressed as

T7 = X, sin2 7 + cos2 77 (4-5)


YT





XT XT





YT


Figure 4-2. Elliptical relationship for T,

The crack will turn, according to Buzek and Herakovich, in the direction where the

ratio of the hoop stress to T is a maximum. The hoop stress, sometimes called the

tangential stress (or oy), is defined in Figure 4-3. Consider a point P and a vector









connecting that point to the crack tip. The hoop stress is normal to this vector that

connects P to the crack tip.

Y






P





Figure 4-3. Definition of hoop stress

Mixed mode crack propagation in anisotropic materials is also investigated by

Saouma et al. (1987). The near-tip stresses are obtained via Sih et al. (1965) and in their

analysis the maximum tangential stress theory is used to predict the angle of propagation.

Saouma et al. use a slightly different criterion for crack turning, but it is of similar flavor

to what Buzek and Herakovich propose. Instead of using strength as a fracture resistance

parameter, the plane strain fracture toughness is utilized.

Boone et al. 1987) use FE models to study crack propagation in orthotropic

materials. The idea of maximizing the ratio of the hoop stress to some strength parameter

is once again utilized to predict the crack turning angle. Chen (1999) analyzes crack

propagation in aircraft fuselages. He, too, makes use of a ratio of the hoop stress to the

plane strain fracture toughness. In this case, the material in question has isotropic elastic

moduli, but the fracture properties are direction dependent. A similar analysis is

conducted by Pettit (2000). He examines crack turning in rolled aluminum, which again,

has isotropic stiffness properties but anisotropic fracture characteristics.









Carloni et al. (2003) and Nobile and Carloni (2005) consider incipient crack

turning for an orthotropic plate under biaxial loading. The former researchers utilize

essentially the same fracture resistance parameter as Buzek and Herakovich, whereas the

latter study uses a slightly modified version of the Saouma et al. turning criterion.

Carloni et al. and Nobile and Carloni also call on complex variables and stress functions

in their formulations for near-tip stresses. Both studies, however, make assumptions that

reduce the order of the characteristic equation that arises in such derivations (see chapter

three), down to order four.

While Pettit's analysis is for a material with isotropic stiffness properties, he

generalizes the fracture resistance parameter to three dimensions. His analysis accounts

for an arbitrary location of the crack with respect to the material. Thus, we would like to

use the method to obtain stress intensity factors given by Banks-Sills et al. (2005) in

conjunction with Pettit's formulation for a fracture resistance parameter. This approach

is a bit more generalized than the aforementioned studies because their work seems to be

confined to specific configurations that usually decouple the in-plane and anti-plane

displacements.

Pettit's method for determining the fracture resistance is formulated for orthotropic

materials which possess cubic symmetry, i.e. three orthogonal planes of symmetry and

within each plane lay two principal toughness values resulting in six principal toughness

values: K12, K21, K23, K32, K13, and K31. The first number in that nomenclature is the

vector normal to the crack plane and the second number corresponds to the direction of

propagation. It is difficult to obtain a toughness for all possible orientations for materials

that exhibit some form direction dependence. Essentially we are interpolating between









the six principal toughness values for the orthotropic material. This gives us an estimate

of the toughness in any direction we desire. Consider Figure 4-4 and the material

orientation as shown. What we see in that figure is block of foam with various M(T)

specimens oriented within it. If the M(T) specimens are aligned with the principal axes

of the large block of foam, there are six possible orientations and with each orientation

there is a corresponding toughness.

Pettit's interpolation equation, then, requires six unique toughness values for the

orientations shown in Figure 4-4. In our case, only three toughness values are needed,

however. In Figure 4-4, consider the M(T) specimens that have the K31 and K32 labels.

Here the load is applied in the 3-direction and the crack is propagating in the 1 and 2-

directions, respectively. Since the 3-direction is the rise direction and the 1-2 plane is a

plane of isotropy, it is reasonable to assume that K31 = K32. That is to say, the crack is

resting in the plane of the knit line for those two cases. For the other four cases, the

fracture plane is normal to the knit line plane. Thus, K21 = K12 and K13 = K23. So the

total number of required toughness values has been reduced to three and NASA did

conduct tests, at room temperature, to obtain them: K32 = 22.4 psi/in, K12 = 17.4 psi/in,

and K23 = 19.5 psi/in.

Pettit's equation that will determine the effective K needs several variables besides

the six principal toughness values. Remember, the goal here is to try and predict the

direction of propagation. We can conveniently describe this direction with a vector,

denoted as a. That vector lies in a plane tangent to the fracture surface, whose normal

vector will be defined as n. These two vectors are shown in Figure 4-5.
































Figure 4-4. Orthotropic toughness values


Crack faces


Figure 4-5. Definition of the a and n vectors

To develop the equation for the effective K, we first define the trace angles of the

vector a makes with the principal planes. These are


tan (,) = a3
a2


tan (o) = a
a3


tan(o93) = a
a1


The subscript in the theta term denotes the axis normal to the principal plane. Using

equation (4-5) we can write a 2D interpolation function of the form


(4-6)









Kk (,k) = K, cos2 (ok) + K, sin2 (ok) (4-7)

We also observe the following trigonometric identities


cos2 tan b -- C-
c b 2 + (4-8)

sin2 tan- b- b22
c b +c

and the components of unit vector a must satisfy the relation

a1 +a2 +a2 =1 (4-9)

Pettit goes on to define equations that can be interpreted as the fracture resistance

components of a in the principal planes. Consider Figure 4-4 once more where the 1-axis

is the loading direction. To estimate an effective K on this plane, we will interpolate

between two toughness values, K12 and K13. Using (4-7) through (4-9) we can write K1

as


K, (a) (K2a22 + K,3a32) (4-10)
1- a2

In a similar fashion we can determine K2 and K3

K2(a)_1 1 2 (K23a32 +K21a12)
K, (a) = -- Ka + Ka
a2 (4-11)
K3 (a)- 1 2 (K31a12 + K32a22)
1 a3
I-a

Finally, Pettit sums the relations in (4-10) through (4-11). The effective K, denoted as

Kp, can now be written as


K,(a,n) = K1i2 +K2n22 +K3n32 = (K 2 2 +K,13a2)
1 a- 2
S2 ( (4-12)
+ I22 (K23a32 +K2a2)+ -n2(K3,2 + K32a22)
1-a 2 1-a3









FE Modeling of the M(T) Specimen

While the importance of crack turning was made evident by the divot test, this

experiment is highly specialized and is not recognized as a standard fracture test

specimen. A more conventional way of analyzing crack turning can be performed with

the middle tension, M(T), test specimen. Material orientation, mode mixity, and

boundary conditions can all be carefully controlled with the M(T) specimen.

The idea behind creating FE models of M(T) specimens is two-fold. One of the

objectives of this study is to examine the effect of mode mixity on the K-solution, or

stated differently, as the material orientation is changed how does this impact KI-KIII

along the crack front? Also, several foam M(T) specimens were fabricated and the

resulting crack turning angles measured. Numerical predictions, based on Ks extracted

from the FE models are used to make predictions via the maximum hoop stress theory

discussed earlier.

M(T) FE models are built using ANSYS FE software. Crack meshing is handled

by FRANC3D, a FE software package developed by the Cornell Fracture Group. Figure

4-6 shows a schematic of the M(T) specimen and Figure 4-7 displays one of the FE

models used in this analysis. The FE model has a unit applied pressure at the top and

bottom faces and is simply supported. The model is comprised of entirely of SOLID92

and SOLID95 elements (ANSYSTM, 1999) which are used to make 3D models and have

the capability to handle anisotropic material properties. The model is twelve inches high,

five-and-a-half inches wide, and an inch-and-half thick. The boxed in area in Figure 4-7

highlights the densely meshed region containing the crack.










i i




2a-







Figure 4-6. Schematic of M(T) specimen


Figure 4-7. M(T) model created in ANSYS
Since the foam has direction dependent properties, this needs to be accounted for in

the FE models. This is best done by using direction cosines. Using Figure 4-8, let us

look at simple transformation of axes to define a new material orientation. Consider two

sets of axes: x, y, z and x', y', z'. Initially, the both sets of axes are aligned and then a

rotation, anti-clockwise (of angle zeta) about the x axis, takes place. Table 4-1 lists the









direction cosines, i.e. the cosines of the angles between the primed and unprimed axes.

Here aq is the cosine of the angle between the x' and x axis, U2 is the cosine of the angle

between the y' and x axis, and so on. We will examine several material orientations with

different initial crack inclinations. In this way, mode mixity is introduced into the

problem and we also examine the effect of material orientation on the predicted

propagation path.


K Z
Z' Y








X, X'

Figure 4-8. Sample transformation

Table 4-1. Direction cosines
x y z
x' ai Pi Yi
Y' Ga2 32 Y2
z' X3 P3 73

As mentioned in an earlier section, for typical acreage sprays there is little offset

between the material and substrate coordinate systems except when the foam is applied

near fittings and/or bolts where the foam can to rise at angles as high as 30 degrees

relative to the ET. Twelve M(T) fracture test samples with varying material orientation

and crack inclination were fabricated and tested. A broken M(T) specimen is shown in

Figure 4-9 and a comparison between the numerical prediction and the measured turning






67


angle, as well as how the foam M(T) specimens are fabricated, will be made in chapter

five.


Figure 4-9. Fractured M(T) specimen: the dotted line is added to show the location of the
knit line














CHAPTER 5
RESULTS AND DISCUSSION

Effect of Material Orientation on the SIF Solution

On most sections of the ET, the foam is sprayed on in such a fashion that it rises

normal to the surface. In that sense, there is very little offset between the tank (substrate)

coordinate system and the foam (material) coordinate system. Also, when modeling

these specimens, we should note that the curvature of the ET is rather large; some 28 ft in

diameter and our specimens tend to include small defects (usually two inches). Thus, it

seems reasonable to assume that curvature effects are minimal. But when the spraying

process takes place near a bolt or fitting point, the foam must be applied around, and over

top of, these parts. In this situation it is possible to have the foam rising at different

angles relative to the surface.

The anisotropic nature of the foam requires 3-D FE models to be employed so the

SIFs can be evaluated along the crack front. To that end, the effect of material

orientation on the K-solution will be examined within models that have both straight and

diagonal cracks. Before moving on with this discussion, let us define a few important

terms that we will use extensively throughout this chapter. The crack inclination angle,

(p, is the angle the crack makes with the horizontal as shown in Figure 5-1. M(T) models

use though-cracks, or flaws that are placed through the thickness of the specimen, or

model, in question. The crack front distance defined as the length of the crack in the

thickness direction, shown in Figure 5-2.

























a O aCT a

Figure 5-1. Definition of crack inclination angle

To analyze the effect of material orientation of the SIF solution, 17 material

orientation cases are defined; see Figures 5-4 and 5-5. M(T) models are constructed

using ANSYSTM FE software and the flaw is inserted using FRANC3D Next Generation.


Figure 5-2. Definition of crack front distance


""YAZO--










A unit pressure is applied to the simply supported M(T) models and in all cases, the

crack length is two inches. A picture of the ANSYS model (with dimensions) is shown

below. The material orientations are also input into the FE models along with the loads


Figure 5-3. M(T) model built in ANSYS

z


x

Figure 5-4. Definition of material orientation




























Figure 5-5. Top view of the cones shown in Figure 5-4

and boundary conditions. There are many different orientations at which the foam can

rise relative to the tank. But when the foam is sprayed down, at some locations, the rise

direction relative to the tank can be as high as thirty degrees. An example of this

happening is shown in Figure 5-5.


Portion of External Tank


Figure 5-6. Hypothetical material orientation relative to the ET

One can see how the foam can rise at angle when it sprayed out the bolt shown in the

above in Figure 5-6. As such we would like to have a reasonable 'tolerance' that the rise









direction can fall into. This would give rise to a range, or set, of possible orientations that

could occur when the foam is being applied. Since the foam can rise at thirty degrees, it

might be interesting to examine the possibility of the foam rising at another, larger, angle;

45 degrees, for example. Thus we now have two possible cases to examine; a rise angle

that is typical near fitting points and bolts, and one that is slightly larger.

The material orientations used in this study can be visualized by setting up two

concentric cones. The points on the 'rims' of the cones denote where the z'-axis (local

rise direction) will be for each material case. Let us consider an example. Point one in

Figure 5-4 lies on the inner, 30 degree, cone. To define the material axes for this case, a

single transformation of 30 degrees, denoted by zeta, in Figure 5-7 is performed. This

can be likened to the knit lines being oriented within the M(T) specimen shown in Figure

5-7. Table 5-1 lists the direction cosines, i.e. the cosines of the angles, between the

primed and unprimed axes. Here ua1 is the cosine of the angle between the x' and x axis,

c2 is the cosine of the angle between the y' and x axis, and so on.




z



Case 1 ..


Y 30-



X, X

Figure 5-7. Definition of case one material orientation for the M(T) FE model









Table 5-1. Direction cosines for the case one material orientation
x y z
x' i = 1 pi = 0 Y = 0
y' C2 = 0 P2 = 0.866 Y2 = 0.5
z' 3 = 0 P3 = -0.5 y3 = 0.866

Figure 5-8 is a plot of the mode I SIF against normalized crack front distance for

cases 1- 4. The mode I SIF for the case zero orientation, the case where the material and

substrate axes are coincident, is also plotted to make a comparison. Figure 5-8 pertains to

a straight crack that is parallel to the horizontal (no inclination). Even though this

material is anisotropic, for this geometry, KII and KIII remain small compared to KI.

This is not always the case, however, particularly when the crack is inclined. To examine

the effect of mode mixity, the crack in the M(T) model is inclined by 30 degrees. All

cases are rerun and in Figures 5-9 through 5-11, the mode I-III SIF is plotted versus the

normalized crack front distance for cases one through four. As expected, when the crack

is inclined KI decreases, and KII-KIII increase. This is consistent with isotropic

assumptions. For the configuration shown in Figure 6-1, the KI and KII are related to p,

or the crack inclination angle (Anderson, 1991)

KI= o cos2 ((P) Q.a
(5-1)
KII = sin (P) cos (P) /-a

We can see that KII is zero when p = 0 and that KI will decrease for increasing values of

(p.

















KI vs normalized crack front distance
(p =0 degrees


---Case 0
--Case 1
- C-ase 2
-e Case 3
-- Case 4


normalized crack front distance


Figure 5-8. Mode I SIF for cases 1-4; 0 degree crack inclination


KI vs normalized crack front distance


--- Case 0
-- Case 1
Case 2
----Case 3
- Case 4


(P =30 degrees


normalized crack front distance


Figure 5-9. Mode I SIF for cases 1-4; 30 degree crack inclination
















KII vs normalized crack front distance

(p =30 degrees


---Case 0
-A Case 1
- C-Case 2
c-- Case 3
- Case 4


normalized crack front distance


Figure 5-10. Mode II SIF for cases 1-4; 30 degree crack inclination


Kill vs normalized crack front distance


---Case 0
-A-Case 1
C--Case 2
-- Case 3
- Case 4


(P =30 degrees


normalized crack front distance


Figure 5-11. Mode III SIF for cases 1-4; 30 degree crack inclination









The mode I SIF for (p = 0 is listed in tables 5-2 and 5-3. Here we can see the variation of

the mode I SIF as the material orientation is changed. For these cases, the ones that

resulted in the largest KI are: 11 and 15 (2.27 psi /in). When you compare that number

to the KI for case zero, the orientation for typical acreage sprays, the largest KI is just

14% higher. This is an interesting result in that even if we 'lean' the z-axis by quite a bit,

the resulting mode I SIF is not much greater than the mode I SIF for the case 0

orientation. However, it is possible to orient this material in an infinite number of ways

and it could be possible to obtain a KI value for an arbitrary orientation that is greater

than the largest value determined in this section (2.27 psi /in).

Table 5-2. KI for cases 0-8, 0 degree inclination
Case 0 1 2 3 4 5 6 7 8
KI (psi Vin) 1.99 2.01 2.14 2.25 2.14 2.01 2.14 2.25 2.14

Table 5-3. KI for cases 9-16, 0 degree crack inclination
Case 9 10 11 12 13 14 15 16
KI (si in) 2.07 2.19 2.27 2.19 2.07 2.19 2.27 2.19

Crack Turning Predictions

We can numerically predict the crack turning angles using the maximum tangential

stress criteria since we have SIFs from FE models and the near-tip stress solutions from

Hoenig (1982). As discussed in chapter four, we will use a ratio of the tangential stress,

CGo, to Kp for turning angle estimations. Let us denote the ratio of a>m/Kp as R and

where R is a maximum, that is the direction of predicted propagation. To make a

comparison, NASA has donated several broken M(T) specimens with various material

orientations. The turning angles within these specimens will be measured and compared

to their numerical counterparts.









M(T) Specimen Fabrication and Determination of Local Knit Line Axes

Fabrication of the M(T) specimens involves spraying the foam on a metal plate, it

rises and eventually a rind forms. When the foam has settled, the next layer is applied

until the desired thickness is achieved. From here, the foam is cut, or machined, from

this parent block so the knit lines are running in the desired directions. One of the foam

samples is shown in Figures 5-12.

Three M(T) foam test specimens are analyzed in this section. These samples have

various orientations and crack inclinations ((p values). The first step in this analysis is to

determine what the material orientation is for each specimen and we do this by measuring















Figure 5-12. From left to right: front, left, right, and rear sides of the M(T) specimen

the angles of the knit lines on the specimen's surface. Consider Figure 5-13. We need to

determine the orientation of the x', y' and z' vectors in order to define the material

orientation. This information is used to define the element coordinate system in ANSYS.

The intersection of two planes is determined by taking the cross-product of their

normal vectors. If one wants the 'trace' angle on a particular face of the M(T) specimen,

the dot product can be used. In our case, the normal to the knit line plane is not known

beforehand. Instead we will measure two angles, 0 and F, and use them to define the knit









line plane shown in Figure 5-13. What we are interested in, is the trace, or intersection,

the knit line plane makes on the x and y = 0 faces. Along the x = 0 face, we need the

angle of the knit line measured from the horizontal, or y-axis. In this case the knit lines

on the x = 0 face are relatively straight; 0 is approximately zero. On the y = 0 face the

angle, F, is also measured relative to the horizontal, or the x-axis. Thus, the traces on the

y and x = 0 faces define the x' and y' material axes respectively. Using the cross product,

the axis normal to knit line plane is determined, i.e. z' = x' x y'.









Knit line plane








r X
Ie I / Iz



Y Side view
X

Figure 5-13. Determination of knit line plane

Local Crack Tip Computations

For crack tip computations involving stresses, there are a few subtleties that should

be addressed in this section because the material classifications discussed in chapter three

are in a 'global' sense. When looking at a foam M(T) specimen, one would say it is

indeed a transversely isotropic material. However, the post-processing of near-tip









stresses requires material properties rotated into a local crack front coordinate system.

Let us consider an example. First, assume we have the case one orientation; a material

rotation (0 = 300) takes place about the global x axis shown in Figure 5-13. If the

inclined crack is oriented in such a fashion so that it rests in the plane of symmetry, the

out-of-plane and in-plane displacements become decoupled and this is a so-called

degenerate case described by Hoenig. His formulation cannot be applied for this special

case. As described in chapter three, when this situation arises (crack lying in a symmetry

plane) the usual isotropic near-tip formulas are utilized.







Knit Line














Figure 5-14. M(T) model with inclined crack

Now set


us denote this as a 'general' case and a rotation is needed to set the properties in the local

crack front system. The way the constitutive matrices are transformed depends on how

they are defined. If they are in compliance form








{} = [S]{a-} (5-2)

then the transformation formulas (Ting, 1996) for the elastic constants are

[S,']= [Q][S][QS] (5-2)

where [S'] is the matrix of elastic constants in the rotated system and

a,2 AP2 '112 12y a,/71 a/,P,
a2 2 /22 72 22/2 a2 2 2)2
Q a,2 A32 732 A3'3 a3 3 3) (5-3
[es]= (5-3)
2a2a3 2/623 2 273 /7273 + 2/73 y2 a3 +a23 a282 + -2a3
2a3a, 2/,3/31 27371 ,37"1 + 73y1 a/zy + y3a, a,31 + 3,a,
2a1a2 2/1/32 22172 712 + 2 2 a y2 + y1a2 a12 + P1,a1

The order of the terms in and (5-2) and (5-3) is based on the convention commonly used

by the composites and elasticity community where the stresses and strains are ordered as

{ ox, o, % z, T Yz, y, T, xT. If the constitutive matrix is in stiffness form

{) = [C]{}) (5-4)

there is an alternate definition of the [Qs] matrix

a'12 2 "12 2/7121 2a,/1 2a,/81
a2 2 /22 "22 2/2212 2a/222 2a232
[C a3 / 3 2 7232 27323 20a323 2a33 (5-5)
a2,a3 /2/33 2,223 /3213 +2/233 22a3 + a2213 a2/62+ 2a3P,
a3a1 /33/1 7371 A/31 + y731 3a,2 + 2"y,3 a/31 + PAa
a1a2 /01/2 7172 A/322+7/372 a'12 + 2y1a a/1A2+ A11a2

The transformed elastic constants in stiffness form are therefore determined by

C'= [Qc [C]e [ ] (5-6)

The two [Q] matrices are related

[Q] = [ l] (5-7)









Equations (5-2) or (5-6) allow us to transform the material properties from one Cartesian

coordinate system to another. The only information we need for such a transformation is

the direction cosines of the angles between the coordinate axes where we start, or the

system where the material properties are initially defined, and the system where we want

to set the properties.

The incipient turning angles analyzed in this program use the maximum tangential

stress theory. In order to apply it, the first thing we need is the hoop, or Go,, component

shown in Figure 5-15. The near-tip stresses derived by Hoenig are in a (x, y, z)

Cartesian coordinate system. To obtain the hoop stress, we transform axes, in this case a

rotation anti-clockwise about the z-axis by co degrees. This is done using standard

transformation equations (Boresi et al 1993).

Y





j ic









Figure 5-15. Cartesian and cylindrical stresses

Calculating co is done via

o)o = aO22 0x 22 + 22 + 2 2/ + 22yz + 2/202"zx + 2a2i2oxy (5-4)

Here the (X2, 32, Y2 terms are the direction cosines of the angles between the old and new

axes









Now we can calculate the hoop stress at any r and co of our choosing. The next step

is to divide the hoop stress by the effective K, Kp. We do this because not only are the

elastic constants direction dependent, but the fracture properties are as well. The crack is

predicted to turn in the direction where R is a maximum. The crack turning angle,

defined as co, is shown in Figure 5-16. Here we have a close-up view of a M(T)

specimen containing an inclined crack. The inclined crack offset by angle, yp, and the

direction of incipient turning is denoted by the dashed line.




























Figure 5-16. Crack turning angle

Let us pause briefly and quickly summarize the procedure needed to compute near-

tip hoop stresses and the predicted turning angle

* Enter material orientation (denoted by case number in Figure 5-2) into a FE model
coordinate system









* Run the model and extract the mode I-III SIFs via FRANC3D

* Rotate material properties into a local crack front coordinate system

* Compute near-tip stresses using Hoenig's formulas

* Transform yy to cm via the transformation equations found in Boresi et al

* Divide yy by Kp and determine where this ratio is a maximum to predict the
turning angle, oc



Table 5-4. Measured and predicted turning angles (in degrees)
co, (measured) oc (predicted) o, isotropicc prediction)
specimen A 27 32 43
specimen B 9.7 21 27
specimen C 12 19 27

For the three test specimens, one of them had a crack inclination of 30 degrees

(specimen A). The other two specimens (B and C) had crack inclinations of 10 degrees.

All three specimens have different material orientations. Table 5-4 lists the measured and

predicted incipient turning angles. The SIFs can vary along the crack front and since the

free surface can influence the numerical K-solution, the measured and predicted angles

are computed at the mid-plane. For comparison's sake, isotropic predictions for the same

M(T) geometries are also tabulated. The isotropic angle is the critical angle, 0o, from

equation (4-4). The experimental tests were carried out at 750F and as such, the elastic

properties at that temperature are used to predict the turning angle.

The turning angle was measured using a Brown and Sharpe coordinate measuring

machine. These devices are used to inspect various geometries such as cylinder heads or

gear boxes. This instrument uses a probe that physically touches the work piece and the

points are recorded digitally. In this case, we wish to use the points to define the fracture

planes. The angle between these planes is the crack turning angle. Within each specimen









are four fracture surfaces. The fracture angle was measured at a location near the middle

and an average of the four values is taken and listed in Table 5-4.

Specimen A had the best agreement with experiment whereas the other two

predictions fall somewhere between the isotropic prediction and actual values. Even

though both the direction dependence of both the fracture toughness and near-tip stresses

are accounted for in these predictions, the results presented here still vary by large

amount. One possible source for the difference is how the material orientation is defined

and modeled. When tasked to predict the turning angle, NASA furnished us with the

three test specimens but the orientations were not specified beforehand. These particular

specimens were initially fabricated to analyze mode mixity, not the effects of material

orientation. As such, the orientations of the knit lines vary throughout the specimens.

When defining the knit line plane, the knit lines (denoted with black dots in Figure 5-12)

closest to the crack were used to define the orientation, even though the material

orientation can vary over the specimen.

Based on the results seen here, it seems that perhaps the crack is seeking a pure

mode I condition, or that the crack is seeking a direction that is normal to the highest

principal stress. In such cases, if a crack, for example, is orientated at 30 degrees relative

to the horizontal, the crack will turn -30 degrees to it is running straight across or in a

direction normal to the maximum principal stress. However, only three specimens are

analyzed in this section and to some extent, the material orientation can be more carefully

controlled. Obviously more samples would have to be analyzed to determine if the

cracks favor a direction that is normal to the highest principal stress.









Influence of Thermal Loads on the SIF Solution

FRANC3D is very versatile in that it can generate different types of crack meshes

and also compute a K-solution for both isotropic and fully anisotropic materials but at the

time of this writing, it is unable to handle problems with a thermal gradient. This is main

reason why the DC technique outlined in chapter three is utilized. To examine the effect

of a thermal gradient throughout the M(T) specimen we will examine the case two

orientation with a straight and inclined crack. This varies the amount of mode mixity and

two different sets of temperatures are applied to the top and bottom faces of the M(T)

specimen shown in Figure 5-17.

In the figures that follow the primary aim is to see how the SIFs change when

temperatures are applied to the M(T) specimen. Since the Ks are evaluated using a

different method, it is useful to compare the results of these two approaches when the

same set of boundary conditions is applied. We have results tabulated for several cases at

room temperature. Selecting one of them, case 1 for example, we can compare the SIF

results for a straight crack. The mid-plane KI at room temperature is 2.01 psi /in. Using

the DC method we obtain a value of 1.8 psi /lin, a difference of approximately 10%. The

quality and density of the mesh does impact the alternate solution presented here. In

Banks-Sills et al. (2005) study a similar routine is used to check the M-integral

computations. Recall that the interaction, or M, integral is used to compute the

anisotropic K-solution in FRANC3D. Excellent agreement is shown between their

various schemes, but much denser meshes are utilized within their FE models.

In this study it is possible to refine the mesh, perhaps by changing the location of

the quarter point elements by some degree, but this may not be feasible because there is a









node and element limitation within this FE software. Banks-Sills et al. use 2-D models

and that allows one to change the mesh density extensively without adding a lot of

computational cost in terms of physical model size or run time.

Consider Figure 5-17 with (p = 0 degrees (straight crack). Here T1 = OoF and T2 =

1000F. The same unit pressure is also applied. All SIF data from FRANC3D is at room

temperature (750F). The DC method used to compute the SIFs is able to determine these

parameters at any temperature. To examine the influence of thermal loads on the K-

solution, the mode I-III SIF for case two at both room temperature and with a thermal

gradient is plotted in Figure 5-18. From Table 5-2, the room temperature mid-plane

value for KI is approximately 2.14 psi /in. With the prescribed thermal boundary

conditions, the mid-plane KI drops to 1.44 psi vin; a 33% reduction.

Also, in a few of the figures, some odd activity takes place near the free surface.

When looking at Figure 5-18, and the line that pertains the to KI, for the DC solution, the

third point drops down to approximately 0.946 psi /in. This happens again in the rest of

the figures that pertain to the DC solution. One possible explanation is that the free

surface along with the boundary conditions is influencing the displacement results at that

point.

Sometimes quadratic elements are prone to spurious oscillations. Within Dhondt's

(2002) paper on computing mixed-mode anisotropic SIFs, he performs a smoothing

operation via an averaging technique to remove the unwanted oscillations. Likewise Man

et al (1993) encounter a similar problem when using boundary elements to model a

contact problem. Their solution involves removing the higher order elements in the









problem area, and replacing them with linear elements instead. No such averaging

technique or element modification is employed here.

There is a marked change in KI, and the other SIFs (KII and KIII) also had fairly

substantial changes for this particular orientation and geometry. In this case KIII remains

small across the crack front. But KII increases by quite a bit, just like KI does. KII at the

mid-plane is roughly -0.04 psi /in at room temperature. With the addition of temperature

to this model, KII rises to 0.315 psi /Yin. Granted for this geometry and set of loads, KII

and KIII still remain small when compared to KI. Let us now consider a case where we

have a larger gradient applied to the M(T) model in Figure 5-17. The next set of results

for this material orientation are for T1 = -200oF and T2 = 1000F.

In Figure 5-19 we see how the larger thermal gradient impacts the mode I SIF.

With these boundary conditions, KI, at the mid-plane, is 1.78 psi /in. But this time, KII

and KIII both show a fairly large increase after this gradient is applied. Here KII at the

mid-plane reaches -0.122 psi /in and KIII= 0.355 psi /in.

















///////Figure 5-17. Application of thermal loads
( T

Figure 5-17. Application of thermal loads