<%BANNER%>

The Development of an Energy Monitoring System for the Assessment of Variability in Proctor Compaction Results

Permanent Link: http://ufdc.ufl.edu/UFE0024515/00001

Material Information

Title: The Development of an Energy Monitoring System for the Assessment of Variability in Proctor Compaction Results
Physical Description: 1 online resource (105 p.)
Language: english
Creator: Beriswill, Keith
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2009

Subjects

Subjects / Keywords: base, beriswill, calibration, compaction, compliance, energy, geotechnical, kinetic, proctor, soil, t180, t99
Civil and Coastal Engineering -- Dissertations, Academic -- UF
Genre: Civil Engineering thesis, M.E.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Existing techniques used to calibrate soil compaction equipment do not measure the overall imparted energy into the soil. Instead, various parameters (i.e., drop height, rammer mass, etc.) are measured independently and hence one is forced to assume that the theoretical energy is mobilized. Since the results of the Proctor density tests are critical to field compaction control, a calibration system is needed to ensure consistency in the equipment used by the Florida Department of Transportation (FDOT), consultant, and contractor testing labs. A new portable calibration device has been developed that measures rammer speed and base system forces during impact, and outputs the kinetic of the rammer and base compliance energy of the compaction machine. The calibrator was used to test 30 compactors in the state of Florida. However, there was a trend indicating a lower than acceptable available energy for compaction. Soil density pills were then compacted on several of the machines and the results used to determine a linear regression equation for two target moisture contents in order to describe the effect of the variables on resulting dry density. The variances associated with each of the dependent variables were then used to account for variance observed in the population of dry density results. The results of this study will be used by labs to check and adjust their equipment so that results from disparate labs can be used with increased confidence. FDOT?s Independent Assurance Inspection teams will also be able to provide a performance-based check.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Keith Beriswill.
Thesis: Thesis (M.E.)--University of Florida, 2009.
Local: Adviser: Bloomquist, David G.
Local: Co-adviser: McVay, Michael C.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2009-11-30

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2009
System ID: UFE0024515:00001

Permanent Link: http://ufdc.ufl.edu/UFE0024515/00001

Material Information

Title: The Development of an Energy Monitoring System for the Assessment of Variability in Proctor Compaction Results
Physical Description: 1 online resource (105 p.)
Language: english
Creator: Beriswill, Keith
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2009

Subjects

Subjects / Keywords: base, beriswill, calibration, compaction, compliance, energy, geotechnical, kinetic, proctor, soil, t180, t99
Civil and Coastal Engineering -- Dissertations, Academic -- UF
Genre: Civil Engineering thesis, M.E.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Existing techniques used to calibrate soil compaction equipment do not measure the overall imparted energy into the soil. Instead, various parameters (i.e., drop height, rammer mass, etc.) are measured independently and hence one is forced to assume that the theoretical energy is mobilized. Since the results of the Proctor density tests are critical to field compaction control, a calibration system is needed to ensure consistency in the equipment used by the Florida Department of Transportation (FDOT), consultant, and contractor testing labs. A new portable calibration device has been developed that measures rammer speed and base system forces during impact, and outputs the kinetic of the rammer and base compliance energy of the compaction machine. The calibrator was used to test 30 compactors in the state of Florida. However, there was a trend indicating a lower than acceptable available energy for compaction. Soil density pills were then compacted on several of the machines and the results used to determine a linear regression equation for two target moisture contents in order to describe the effect of the variables on resulting dry density. The variances associated with each of the dependent variables were then used to account for variance observed in the population of dry density results. The results of this study will be used by labs to check and adjust their equipment so that results from disparate labs can be used with increased confidence. FDOT?s Independent Assurance Inspection teams will also be able to provide a performance-based check.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Keith Beriswill.
Thesis: Thesis (M.E.)--University of Florida, 2009.
Local: Adviser: Bloomquist, David G.
Local: Co-adviser: McVay, Michael C.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2009-11-30

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2009
System ID: UFE0024515:00001


This item has the following downloads:


Full Text

PAGE 1

1 THE DEVELOPMENT OF AN ENER GY MONITORING SYSTEM FOR THE ASSESSMENT OF VARIABILITY IN PROCTOR COMPACTION RESULTS By KEITH BERISWILL A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF ENGINEERING UNIVERSITY OF FLORIDA 2009

PAGE 2

2 2009 Keith Beriswill

PAGE 3

3 To Rebecca

PAGE 4

4 ACKNOWLEDGMENTS I thank all of the people that have been there for me throughout the difficult times associated with pursuing higher ed ucation, as well as for all of the encouragement that I have received over the years from those close to me. I would especially like to thank my parents, grandparents and my bride-to-be. I would like to thank Jeff Beriswill for be ing such a wonderful role model and for introducing me to Civil a nd Geotechnical Engineering. Scott Wasman, Dr. McVay and Dr. Bloomquist, thank you so much for all of your time effort and willingness to shar e your knowledge. Chuck Broward and George Lopp, I thank you for your time, skill and tools.

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS...............................................................................................................4 LIST OF TABLES................................................................................................................. ..........7 LIST OF FIGURES................................................................................................................ .........8 ABSTRACT....................................................................................................................... ............10 CHAPTER 1 INTRODUCTION..................................................................................................................12 1.1 Background................................................................................................................ ......12 1.2 Problem Statement......................................................................................................... ..14 1.3 Objective and Scope of Work..........................................................................................15 2 BACKGROUND....................................................................................................................17 2.1 Mechanics of Compaction...............................................................................................17 2.2 Delivered Energy and Soil Compaction..........................................................................17 2.3 Previous Research on Energy Measurement...................................................................19 2.4 Previous Research on the Variability of the Soil Compaction Process...........................19 2.5 Base Compliance........................................................................................................... ..20 2.6 Statistical Background.................................................................................................... .23 3 THE CALIBRATION DEVICE.............................................................................................25 3.1 Operation of the Compaction Machine............................................................................25 3.2 Fundamentals of the Calibrator.......................................................................................25 3.3 Conceptual Designs........................................................................................................ .27 3.3.1 Dynamic Impact Calibrator...................................................................................27 3.3.2 Displacement Based Calibrator.............................................................................29 3.3.3 Acceleration Based Calibrator...............................................................................32 3.3.4 Development of the Photo Gate............................................................................35 3.3.5 The Infrared Photo Gate........................................................................................36 3.3.6 Development of a Photo Electric Gate..................................................................39 3.4 Development of th e Load Cell System............................................................................40 4 VALIDATION OF THE PORTABLE CALIBRATOR........................................................42 4.1 Validation of the Photo Electric Gate..............................................................................42 4.2 Accuracy of the Instrument.............................................................................................46 4.3 Validation of the Testing Procedure................................................................................47 4.3.1 Photo Electrics.......................................................................................................47

PAGE 6

6 4.3.2 Compliance Instrumentation.................................................................................48 5 LABORATORY TESTING AND ANALYsIS......................................................................50 5.1 PDEC Setup Description.................................................................................................50 5.2 Testing Program Overview Using PDEC Instumentation...............................................54 5.3 Testing Results and Analysis of PDEC Data...................................................................56 5.3.1 Kinetic Energy Assessment...................................................................................56 5.3.2 Manual Compaction Rammer................................................................................62 5.3.3 Base Compliance...................................................................................................64 5.3.4 Force Compressive Energy Theory.......................................................................66 5.3.5 Compressive Energy Results.................................................................................66 5.4 Measured and Predicted Dry Density Variance of A2-4 Soil...........................................69 5.4.1 Background............................................................................................................69 5.4.2 Properties of Test Soil............................................................................................70 5.4.3 Regression Analysis of Lab Test Results..............................................................75 5.4.4 Standardized Regression Equation........................................................................77 5.4.5 Variance of Dry Densities, Measured vs. Predicted..............................................79 5.4.6 Summary of Predicted and Measured Variances of A2-4 Tested Soil..................80 6 CONCLUSION/RECOMMENDATIONS.............................................................................81 6.1 Conclusions............................................................................................................... .......81 6.1.1 Background............................................................................................................81 6.1.2 Calibrator Development and Operation.................................................................82 6.1.3 Laboratory Testing a nd Data Analysis..............................................................86 6.1.3.1 Kinetic Energy.............................................................................................86 6.1.3.2 Base Compliance Energy............................................................................87 6.1.3.3 Regression Analysis....................................................................................87 6.2 Recommendations........................................................................................................... .88 APPENDIX A OPERATION OF INSTRUMENTATION............................................................................90 The Photo Electric Gate........................................................................................................ ..90 Saving and Importing Data into Excel....................................................................................91 Kinetic Energy Processing......................................................................................................92 Compliance Measurement Setup and Testing........................................................................92 Compliance Energy Processing..............................................................................................93 B DATA RESULTS...................................................................................................................94 LIST OF REFERENCES.............................................................................................................104 BIOGRAPHICAL SKETCH.......................................................................................................105

PAGE 7

7 LIST OF TABLES Table page 3-1 Summary of Displacement Laser En ergy Measurements (5.5-lb rammer).......................31 3-2 Measured Travel Times from Sens ors and Corresponding Energy Calculations..............39 4-1 Summary Statistics for Valid ation Study Hammer Rotations...........................................45 5-1 Component Summary Table for Machin es used in % Moisture Content Regression Analysis...........................................................................................................77 5-2 Component Summary Table for Machin es used in .5% Moisture Content Regression Analysis...........................................................................................................77 B-1 T-180 Rammer Rotation Summar y Statistics Per Machine...............................................94 B-2 T-180B Base Complianc e Summary Per Machine............................................................95 B-4 Soil Pill Results and Associated Energies 6% Moisture....................................................96 B-5 Soil Pill Results and Associated Energies 7.5% Moisture.................................................98

PAGE 8

8 LIST OF FIGURES Figure page 1-1 Effect of compaction energy on co mpaction of sandy clay (Das 2002)............................13 2-1 Effect of moisture conten t on dry density (Das 2002).......................................................18 2-2 Compaction foundation block w ith wood cushion in place...............................................21 2-3 Effect of wood cushion on soil Proctor curve per soil type...............................................21 3-1 Typical mechanical soil compactor...................................................................................26 3-2 Manual rammer frame .......................................................................................................28 3-3 Compaction laboratory mechanical compactor displacement laser...................................30 3-4 Laser recorded displacement versus time plot...................................................................30 3-5 Detail of guide rods and guide disk...................................................................................32 3-6 Typical acceleration record................................................................................................34 3-7 Standard test setup for infrared photo gate........................................................................37 3-8 Voltage measured from infrared photo detector pairs 1, 2 and 3.......................................38 3-9 Base-mounted setup for fiber optic photo gate..................................................................40 4-1 Linear acceleration of rammer during T-180 fall..............................................................43 4-2 Sensor alignment on compaction machine........................................................................44 4-3 Typical acceleration and time duration plot for base compliance measurement...............49 4-4 Typical force and time duration plot for base compliance measurement..........................49 5-1 Photo electric gate........................................................................................................ ......50 5-2 Illustration of rammer rod at switching point of sensor 2..................................................51 5-3 Data acquisition system setup and sensors........................................................................52 5-4 Base system configuration with mold assembly................................................................53 5-5 Instrumented mold assembly.............................................................................................53 5-6 FDOT district map.......................................................................................................... ...54

PAGE 9

9 5-7 Compactor foundation. A) Steel cushion. B) Plywood cushion........................................55 5-8 Frequency distribution of mean energy data......................................................................57 5-9 Bootstrap frequency distribution of mean energy..............................................................59 5-10 Frequency distributi on of all energy data..........................................................................60 5-11 Scatter plot of energy standard deviation versus mean velocity........................................61 5-12 Scatter plot of energy standard deviation versus mean energy..........................................61 5-13 Frame for maintaining vert ical alignment of rammer........................................................63 5-14 Scatter plot of average maximum force.............................................................................65 5-15 Scatter plot of average maximum acceleration..................................................................65 5-16 Dynamic force and acceleration.........................................................................................67 5-17 Soil compaction results of A-2-4 test material..................................................................72 5-18 Test target moisture content on Proctor curve...................................................................73 5-19 Scatter plot of 6% moisture content results.......................................................................74 5-20 Scatter plot of 7.5% mo isture content results....................................................................75 6-1 Photo electric gate........................................................................................................ ......83 6-2 Photo electric gate on compactor.......................................................................................83 6-3 Compactor base configurat ion and mold assembly..........................................................85 6-4 Influence of cushion on A-2-4 compaction curve..............................................................85 6-5 Instrumented mold assembly.............................................................................................86

PAGE 10

10 Abstract of Thesis Presen ted to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Engineering THE DEVELOPMENT OF AN ENER GY MONITORING SYSTEM FOR THE ASSESSMENT OF VARIABILITY IN PROCTOR COMPACTION RESULTS By Keith Beriswill May 2009 Chair: David Bloomquist Co chair: Michael McVay Major: Civil Engineering Existing techniques used to calibrate so il compaction equipment do not measure the overall imparted energy into the soil. Instead, va rious parameters (i.e., drop height, rammer mass, etc.) are measured independently and hence one is forced to assume that the theoretical energy is mobilized. Since the results of th e Proctor density tests are critical to field compaction control, a calibration system is needed to ensure cons istency in the equipment used by the Florida Department of Transportation (FDOT), cons ultant, and contractor testing labs. A new portable calibration device has been de veloped that measures rammer speed and base system forces during impact, and outputs th e kinetic of the rammer and base compliance energy of the compaction machine. The calibrator was used to test 30 compactors in the state of Florida. However, there was a trend indicati ng a lower than acceptable available energy for compaction. Soil density pills were then compacted on severa l of the machines and the results used to determine a linear regression equation for two target moisture contents in order to describe the effect of the variables on resul ting dry density. The variances associated with each of the

PAGE 11

11 dependent variables were then used to account for variance observed in the population of dry density results. The results of this study will be used by labs to check and adjust their equipment so that results from disparate labs can be used with increased confidence. FDOTs Independent Assurance Inspection teams will also be ab le to provide a performance-based check.

PAGE 12

12 CHAPTER 1 INTRODUCTION 1.1 Background The Florida Department of Transportation (F DOT) and many other entities rely on the compaction of soil for infrastructure construction pr ojects. This requires that a sample of onsite soils be compacted in a laboratory, in order to establish the requirements for field compaction. Explicitly, the characteristics of interest ar e the maximum dry unit weight and the optimum moisture content of the particular soil. Whether the test is being performed by the FDOT or by a private consulting laboratory, the te sting procedure remains unchanged. The Proctor compaction test consists of two pr imary types of tests; the Standard American Association of State Highway Transportation Officials (AAS HTO) Standard T-99 and the Modified AASHTO T-180 test. Fo r each test, a rammer of a specified size, shape and mass is lifted to a specified height and allowed to fa ll until reaching a soil sample. The material is contained in a mold of specified size for a set numb er of impacts per soil lif t. As a result of the rammer impact, compaction or densification of th e soil occurs and should be directly related to the amount of energy that is produced. For a gi ven compaction test, the amount of total energy that is delivered to a volume of soil is: number of numberweightheight of b lows perofofdrop of layerlayershammer hammer E volume of mold (1-1) With an increase in applied energy, an increase in dry density will likely occur if the soil moisture content is maintained (Proctor 1933). Additionally, the optimum moisture content of the soil will vary for differe nt amounts of energy imparted to the soil (Dubose 1952). This dependency of maximum dry unit we ight and moisture content on en ergy is illustrated in Figure

PAGE 13

13 1-1. Due to the effect of compaction energy on th e resulting densities, it is critical that the energy and testing procedure be consistent. In this illustration, mass and drop height remained constant while the number of blows per layer in crease and is representa tive of adding energy imparted to the soil. Figure 1-1. Effect of compaction ener gy on compaction of sandy clay (Das 2002).

PAGE 14

14 In an effort to verify that mechanical co mpaction machines are delivering the prescribed energy, a calibration procedure is periodically performed. On e calibration method, specified by AASHTO, is based on the deformation a lead cylinder upon rammer impact (ASTM D 2168 B). However, the method used by the FDOT utilizes the compaction of a calibration clay (CL) material as prescribed by ASTM D 2168 A. For this method, the CL material is compacted using a manual Proctor compactor and the resu lts compared to a specific machine to be calibrated (AASHTO T-99 or T-180). Utilizing the manual testing procedure, the maximum dry unit weight and optimum moisture content of the soil are determined. It is then performed using the mechanical method (AASHTO T-99 or T-180) and the maximum dry unit weight compared. Attention is focused on the percent difference between the two maximum dry unit weights. When the absolute percent difference between the maximum dry unit weights for the two testing procedures is less than 2%, the mechanical calibrator is considered to be calibrated. 1.2 Problem Statement In previous work performed by the FDOT, larg e quantities of soil were obtained, divided, and sent to state-approved priv ate testing laboratories that use compaction machines calibrated via AASHTO D 2168 A. These samples were then tested in accordance with AASHTO specifications for the AASHTO T-99 Standa rd and AASHTO T-180 Modified testing procedures. The results from this testing re gime displayed differences in the maximum dry densities and optimum moisture content reported from lab to la b. There appear to be two primary possibilities for these differences. Sin ce the same soil was used for all testing, the possibility of soil properties skewing the result s was assumed to be negligible. Thus, it was hypothesized there is some difference in the compac tive energies across labs, or the operator had an effect on the results.

PAGE 15

15 While the energy applied to the soil specime n is known to have a pronounced effect on density, it is not the only factor that needs to be considered. Base support can play a critical effect on both density and optimum moisture co ntent (Chapman and Ray 1954). Knowing that different base conditions do exist between laboratories, this resear ch was launched to investigate both energy delivery and energy diss ipation on resulting densities. 1.3 Objective and Scope of Work As a result of the discrepancies in the ma ximum dry unit weight and optimum moisture content, the primary goal of this research was to take an energy based approach to the calibration of mechanical soil compactors. This involved being able to measure the energy available for compaction and the energy transferre d to the base (base compliance) of the compaction machine. A portable calibration device was developed which measures both the rammer impact velocity and base system energy. It was then tested thr oughout FDOT district co mpaction laboratories. The construction and validation of the device was performed at the Department of Civil and Coastal Engineering Laboratory at the University of Florida (UF) in Gainesville. Tests in several districts were performed on a population of thirty machines mostly as AASHTO D T180 Modified compactors. The results were co mpiled and analyzed to establish a confident variance in energy available for compaction and base system energy. As components of the compaction process, these two quant ifiable values comprise two-thir ds of it, with the variance of the operator involving such things as moisture co ntrol, layer thickness, particle distribution, pretamping, and moisture determination. This i ssue was only identified as a contribu tor in soil density variance, as the time required for a testing was not available. The scope of work found that the variance in compaction energy was half the variance in the base system energy. Although the compacti on variance was small indicating similarity among machines in the test popu lation, the mean energy was le ss than that prescribed by

PAGE 16

16 AASHTO T-180 and ASTM D 1557 Standards. Th e difference is attributed to machine performance due to drop height and impact ve locity, and not on the mass of the rammer. Variance in base system energy is attributed to the many different base foundations discovered in the population. Typically, foundations consisted of some form of concrete, either cast or constructed with concrete blocks, and cemented in place. Soil density tests were perfor med on A-2-4 soil samples on machines tested with the developed energy devices. The results of these density tests in conjunction with the measured energies of the machines were used to develop re gression equations that describe the interactions of the dependant variables on the resulting dry dens ities. These regression equations were then used to quantify the variance projected from the va riability observed in th e dependent variables. This was able to attribute half of the observed va riance in dry density to the variance observed in, moisture content, dry density and base compliance.

PAGE 17

17 CHAPTER 2 BACKGROUND 2.1 Mechanics of Compaction Soil compaction is performed for the purpose of increasing the strength of the soil through densification (Proctor 1933). The process by which this increased density is reached is through the removal of air voids between soil particles (D as 2002). With the removal of the air voids, it is possible to increase the mass of soil in a finite volume. In order to most efficiently remove air, water is added allowing the soil particles to sl ide past one another as mechanical energy is applied. When the same amount of energy is applied and the amount of water is increased, eventually a maximum mass of so il will result. From this point on, the addition of water will serve to displace soil particles resultin g in decreased density (Figure 2-1). 2.2 Delivered Energy and Soil Compaction The energy that an object possesses in free fa ll is the sum of its potential and kinetic energy (Equation 2-1). In this situation it can be assumed that the instant the rammer is at its apogee, its velocity is zero. Thus, the rammer possesses only potential energy and its kinetic energy is zero (Equation 2-2). This case is opposite at the instan t that the hammer has fallen and just prior to impact. Here, the object has no potential energy but possess es only kinetic energy (Equation 2-3) (Halliday 2000). Total energy = ( mass velocity2) + (mass gravity height) (2-1) Energy at drop height = (mass gravity height) (2-2) Energy at impact = ( mass velocity2) (2-3)

PAGE 18

18 Figure 2-1. Effect of moisture content on dry density (Das 2002). Based on conservation of energy, if the loss due to friction is negligible, then the energy at the drop height is equivalent to the energy at the instant of imp act. The drop height and mass are each specified in AASHTO, and once verified to be within tolerance, th e velocities for the T-99 and T-180 tests can be calculat ed through simple algebra. AASHTO T-99 calls for a 5.5-lb rammer to be dropped 12 inches, thus producing 5.5 ft-lb of energy each drop. The T-180 method calls for a 10-lb rammer to be dropped 18 inches, resulting in a theoretical ener gy of 15 ft-lb being delivered (n eglecting friction). AASHTO does however make allowance for differences in the drop height mass of the rammer for both testing procedures. This is 0.06 inches and 0.02 lb, respectively. Assuming the reduction in energy is negligibly affected by friction, for the Standard Proctor compacti on test this resu lts in a range of acceptable energy of 5.45 to 5.55 ft-lb. For the Modified Proctor compaction tests, the range is 14.92 to 15.08 ft-lb.

PAGE 19

19 2.3 Previous Research on Energy Measurement The Transportation Research Board (TRB) ha s published research that took an energy based approach to monitoring the compaction pr ocess. It also used the technique for soil compactor calibration (Sebesta and Liu 2007). Sebesta and Lius research focused on a permanently mounted displacement sensor that utilized a magneto-restrictive rod and sensing ring concentrically mounted above the rammer rod. While this method may be used in a particular lab to measure the impact velocity of a rammer, it does not appear that it lends itself for use on multiple machines in other labs, nor could it be used to establish the energy deliv ered from a manual compaction hammer. The study proved useful in measuring impact velocity but proved to be ineffective as a portable energy calibrator due to the permanent mounting requirements of the displacement sensor. Most critically, they found that successive impacts of a mechanical compaction rammer do not provide equal energy. This means that multipl e successive impacts needed to be monitored. 2.4 Previous Research on the Variability of the Soil Compaction Process Reproducibility of soil compaction has been previously investigated using a 2.5-kg and 4.5-kg (5.51-lb and 9.92-lb) rammer, comparable to the T-99 and T-180 testing procedures by British standards for both manual and mechanical compactors. According to The Roadway Research Laboratory, No significan t differences could be observed between the results achieved by hand compaction and those achieved by machin e compaction (Sherwood 1970). In terms of the observations made in this research, it was al so noted that no faulty hand held rammers were observed in any laboratories and that the masses and drop heights of all handheld rammers were consistent. Thus, they conclude d that differences in testing using the handheld rammer could only be documented by obs erving the actual test.

PAGE 20

20 Of primary interest in their study was the e ffect a single operator had on soil density and optimum moisture content results, as well as the effect of different labo ratories. Their results showed that an operator using th e same machine to run multiple te sts showed little variance, with a maximum dry density COV of 0.13 and 0.84 for optimum moisture cont ent, whereas eight operators on a single machine had a maximu m dry density COV of 2.8 and 7.8 for optimum moisture content for the same soil. The Road Re search Laboratory then looked at results for 36 laboratories and found a COV of 2.1 for maximum dry density and 9.7 for moisture content. These results illustrate that the testing procedur es of a single operator tend to be far more consistent than between individua ls or different laboratories. The research did not seek to identify the source of the differences between the laboratory testing re sults in terms of the mechanical compaction equipment (Sherwood 1970). 2.5 Base Compliance In UF research, an effort to determine the e ffect of a worse-case scenario regarding the base stiffness of a mechanical compactor wa s investigated. Two 0.75-inch thick plywood sheets were bolted to the top of a concrete compac tion foundation block in the Soils Compaction Lab (Figure 2-2). Six manual T-99A tests were then performed, two each on A-3, A-1-b and A-2-4 soils; one with and one without the plyw ood cushion. As expected, th e boundary condition had a profound effect on the shape of the compaction curve as well the maximum dry density (Figure 2-3). The resulting density varied between 0.7% and 3.4% lower with the wood base compared to the T99A standard procedure. For the A-3 soil, the optimum moisture content varied by more than 1%.

PAGE 21

21 Figure 2-2. Compaction foundation block with wood cushion in place. 112.74 116.56 118.10 114.54 122.24 114.19 117.50 113.64 112.00 114.00 116.00 118.00 120.00 122.00 124.00 126.00 128.00 130.00 56789101112131 4 MOISTURE (%)DRY DENSITY (PCF) Wood Cushion in place Foundation Block Only Zero Air Void Figure 2-3. Effect of wood cush ion on soil Proctor curve per soil type. A) Soil A-1-b. B) Soil A-3. C) Soil A-2-4. Foundation Bloc k Plywood Cushion A Dry Density (pcf) Moisture ( % )

PAGE 22

22 103.76 104.59 103.22 104.51 105.86 105.41 105.61 107.07 106.01 108.22 100.00 102.00 104.00 106.00 108.00 110.00 112.00 114.00 116.00 118.00 120.00 9101112131415161718 MOISTURE (%)DRY DENSITY (PCF) Wood Cushion in place Foundation Block Only Zero Air Void 109.17 112.13 114.18 110.06 115.01 110.37 113.29 111.13 108.00 110.00 112.00 114.00 116.00 118.00 120.00 122.00 124.00 126.00 678910111213141 5 MOISTURE ( % ) DRY DENSITY (PCF) Wood Cushion in place Foundation Block Only Zero Air Void Figure 2-3. Continued. C B Dry Density (pcf) Moisture (%) Dry Density (pcf) Moisture ( % )

PAGE 23

23 2.6 Statistical Background Statistical interpretati on and representation of data is cr itical in the analysis of random variables. The variables contained within the to pic of this research have been dealt with as statistically random variables, meaning they vary independently of each other. In most cases these variables have will be sampled a set num ber of times and summarized by their mean and variance (Equations 2-4 and 2-5) n i ix n x11 (2-4) x = Mean of x n = Number of Samples ix= Sample i 2) ( ) ( x E x Var (2-5) ) ( x Var = Variance of x E = Expected Value x = Sample = Mean of x In many cases it is necessary to verify that enough samples have been taken to establish accurately the variance. This calls for the use of a statistic al method of inferring what the variance of the population is from the data obtained. The method for doing this that will be used is a method known as bootstrapping. This method of sampling useful because it does not require that the distribution of the data be know. Bootstrapping is the term used for repetitive re-sampling of data. In order to perform a bootstrap only a sample of data is needed. From th is sample of data a random value is selected a set number of times, the selected values are then summarized by their mean and variance to generate a single boot strap. This is then perf ormed repetitively until the mean of the bootstrap

PAGE 24

24 means and the variance of the bootstrap means reach es a constant value. These bootstraps are then considered representative of the population. Additional summary statistics can be then performed on the bootstraps to indicate if the enough samples were taken from the population. this is done by looking at the variance of the bootstrap variances, if e nough samples from the population were originally acquire d this value will be much sma ller than the variance of the sample population. If too few samples from the population were taken the variance of the variance will be a large number. Another method of determining if enough samp les have been taken is to look at the coefficient of variation of the bootstrap varian ces. If this value ex ceeds a pre determined percentage then more samples should be acquire d in order to accurately determine the variance of the sample population.

PAGE 25

25 CHAPTER 3 THE CALIBRATION DEVICE 3.1 Operation of the Compaction Machine The typical compaction machine used consulting labs in Florida as well as the FDOTs State Materials Office (SMO) is the Rainhardt Model 662 or simila r machine (Figure 3-1). This machine consists of an electric motor that driv es a belt which rotates a flywheel and attached cable to lift a grabber vertically. The lift height is a calibrated, controlled distance specific to the test, either 12 0.06 inches or 18 0.06 inches. This grabber tr avels along the rammer rod and grabs the rod at its lowest point of travel, rotates the rod a fi xed increment during the lift and releases it at the highest point of travel at which time the rammer rod falls vertically. While falling, the rammer rod passes through the jaws of the grabber and is guided by a disk on the rammer assembly along guide rods until impact. Fi gure 3-1 illustrates the main components of the mechanical compactor. 3.2 Fundamentals of the Calibrator Development of an energy based calibrator required that the ener gy delivered to a soil sample be quantified. In order to perform this task, several general issues needed to be addressed. First, the rammer should be allo wed to rotate freely during the compaction procedure. Secondly, the cal ibration device should function without altering the compactor, thereby voiding its calibrati on. This would then require a recalibration per AASHTO specifications prior to being used. Thirdly, no attachments to the rammer or guide rods are possible, since attaching any part of a calibrator to the rammer changes the rammer mass, and thus, the kinetic impact energy.

PAGE 26

26 Figure 3-1. Typical mech anical soil compactor. Observation at FDOT approved compaction labo ratories indicated that in some, height above the compaction device was limited to less than 18 inches from the ceiling. This would have required cutting a hole in the ceiling to utilize Sebesta and Lius device and thus was not pursued further.

PAGE 27

27 3.3 Conceptual Designs Several design strategies were formulated on ways the energy of a rammer compaction blow could be measured or calcu lated. In theory, quantificat ion is a simple physics problem involving potential and kinetic energy and work theory. Using basic assumptions about the operation of the machine, these three aspects ar e related and it was determined that the energy might be measured through displacement, accelerat ion or the work done at impact via a dynamic load cell. 3.3.1 Dynamic Impact Calibrator Initial instrumentation development focused on the principle that the kinetic energy of the rammer fall was equal to the work done. When the rammer impacts the soil, work is done as the soil deforms, the amount based on the soils mo dulus. From the force measured during the deformation, one is able to employ the rela tionship between deformation and work (Equation 3-1). Assuming negligible losses occur, the wo rk calculated would be the same as the energy that was delivered by the rammer. 2 1y yWF(y)dy (3-1) In this equation, W is the work done on the sa mple, F is the force ap plied to the specimen, and y is the deformation (compression) of the soil mass. By inserting a dynamic force sensor at th e location of the ramme r impact point and measuring the displacement of the rammer during impact, Equation 3-1 could be calculated. In order to dampen the blow to protect the dynami c force sensor from excessive force and to provide measurable impact deformations, a rela tively soft impact pad with known material properties was used.

PAGE 28

28 A Micro-epsilon displacement laser with a nd operating frequency of 2,500 samples per second (2.5 kS/s) and precision of 0.001 of an inch was purchased for the project. In addition, a PCB 200C20 quartz piezoelectric analog dynamic fo rce sensor with a range of 50,000 lb was acquired. The testing procedure consisted of co nducting tests utilizing th is instrumentation on a manual compaction hammer with different types of polyurethane and neoprene pads as the surface material. A two-inch diameter metal ring was installed on the compactor hammer to provide a target for the laser to reflect from. A frame was constructed to mount the laser displacement sensor. In additi on, the frame contained two cross members with C-clamps in order to maintain vertical alignment of the hammer during a drop (Figure 3-2). Figure 3-2. Manual rammer frame. After several tests, this c oncept was abandoned since the hammer could not be moved around the mold as it would be during a comp action test run on mechanical compaction equipment. Based on this issue, allowing free ro tation of the rammer became a high priority in the development of the calibration device.

PAGE 29

29 3.3.2 Displacement Based Calibrator Testing then focused on a displacement sensor that could measure the rammers trajectory over its entire fall event. By utilizing an entire free fall set of data, the total displacement of the rammer for each impact could be determined (Equation 2-2, Section 2.2) and compared. However, more importantly, the impact velocity of the rammer also could be calculated and used to quantify the amount of kinetic energy that th e rammer delivers with each impact (Equation 23, Section 2.2). Using this knowledge and the goal of allowing free rotation of the rammer, testing began on the mechanical compactor. The mass of the rammer assembly was determined using a standard digital laboratory scale. A light (0.06-oz) metalized plastic disc target was affixed to the top of the rammer rod to serve as a ta rget for the laser displacement sensor. The displacement transmitter/receiver was then mounted above the lift rod. Fi gure 3-3 illustrates the test setup. The full displacement record was then processed and analyzed to determine the total distance the rammer traveled over the course of the fall. The displacements with respect to time were plotted and a second order trend line fitted through the data using Microsoft Excel. Refer to Figure 3-4 for a typical example of a plot. Using the trend line for each of these displacem ent records, the first derivative was then taken with respect to time. This results in a velocity profile equation for the fall and was then evaluated at the time of impact providing a calculated impact ve locity of the rammer for each impact. The impact velocity was then used to eval uate the kinetic energy. The results of a series of rammer drops, fall distance, impact velocity potential energy, and kinetic energy for each drop are presented in Table 3-1.

PAGE 30

30 Figure 3-3. Compaction la boratory mechanical compactor displacement laser. y = 182.74x2 0.9792x R2 = 1 0 2 4 6 8 10 12 14 00.050.10.150.20.250.3 Time (sec)Displacement (in) Figure 3-4. Laser recorded di splacement versus time plot. Optical Senso r Target

PAGE 31

31 Table 3-1. Summary of Displacement La ser Energy Measurements (5.5-lb rammer) Velocity (ft/sec) Energy (ft-lbs) Displacement (in) Energy (ft-lbs) 7.705.0712.035.51 7.735.1012.085.54 7.745.1212.055.52 7.755.1312.085.54 7.765.1512.205.59 7.815.2112.045.52 7.825.2212.065.53 7.775.1611.935.47 7.765.1412.175.58 7.725.0912.105.55 KineticPotential As expected, this table shows that in all case s, the kinetic energy is less than the potential energy. This is due to frictional losses from the rammer guide rods and disk contacting each other during free fall (Figure 3-5). It is importa nt to note that the m achine used at the UF compaction laboratory is not currently used for soil compaction but rather for prototype development and validation. Hence, it has not been certified as calibrated. However for calculation purposes, the mean drop height measur ed for these 10 impact was 12.07 inches which is very close to the tolerance of 0.06 inches specified in AASHTO standards.

PAGE 32

32 Figure 3-5. Detail of gui de rods and guide disk. These results verify that measuring the full rammer displacement during its fall may prove effective in accurately determini ng potential energy. Additionally this study indicates there are significant differences in the actual versus theore tical kinetic energies, at tributable to frictional losses. However, due to the mounting and cleara nce issues with the laser, its high cost and complexity of aligning the target with the laser, this concept was abandoned. 3.3.3 Acceleration Based Calibrator The investigation then focused on using an accelerometer for dete rmining the kinetic energy of an impact, since a miniature accele rometer could be easily mounted on the rammer Guide Rods Guide Disk

PAGE 33

33 without adding significant weight. The accelerome ter provides information such as the rammer release point, the time of impact and the accelera tion during the fall. A numerical integration of the acceleration data produces a velocity profile and more importantly, the impact velocity. A second integration could then be performed over the same time inte rval and the displacement of the rammer with time determined as well. From the accelerometer information it was thought that one would be able to compare the theoretical potential energy to the actual potentia l energy as well as the kinetic energy just prior to impact. This kinetic energy would be a usef ul check against the energy calculated from the load cell. However in practice when attempting this config uration, significant i ssues arose regarding the processing of the accelerometer data due to the vibration of the compactor during operation and the data having a high-frequenc y noise component in the signal. One of the impact records of a single lift and drop cycle is shown in Figure 3-6. In the graph in Figure 3-6, it is possible to s ee the noise in the signa l. The acceleration of the rammer was expected to be constant or nearly constant at approximately one g, the gravitational constant (32.2 ft/sec2). As can be seen, Figure 3.6 shows that no clear acceleration record is evident.

PAGE 34

34 -1 -0.5 0 0.5 1 1.5 2 2.5 3 00.10.20.30.40.50.60.7 Time ( sec ) Acceleration (g) Lift of Rammer Fall of Rammer Impact Figure 3-6. Typical acceleration record. Advanced numerical signal processing utiliz ing a Fast Fourier Tr ansform procedure was performed on the signal which indicated that dur ing the lift portion of the cycle there was an underlying frequency of 10 Hz. This was attributed to vibration of the motor lifting the rammer. Following the point at which the rammer was releas ed to begin its free fall, the noise frequency increased dramatically. This high frequency is likely caused by the rammers guide disc contacting the guide rods during the fall. Th e high noise amplitudes occurred throughout the Fourier transform, making it nearly impossible to remove and obtain an accurate acceleration record. While it is possible that further advan ced signal processing algorithms could have been employed, the time and effort required to obtain us eful information was considered problematic. Since the goal of this project is to produce a re latively simple and repeatable device that does not

PAGE 35

35 involve advanced data processing and lengthy co mputational effort, it was decided to abandon this approach and look at yet another alternative concept. 3.3.4 Development of the Photo Gate Work began on a more direct way of measuring velocity, in order to compute the kinetic energy of the rammer assembly. A search fo r available velocity sensors was conducted and several transducers were identified. However due to the constraints of the testing environment (i.e., no added mass to the rammer, clearance issu es, etc.), nothing was found that would work for this application. Thus, obtaining an average velocity rather th an an instantaneous velocity concept was pursued. Average velocity is readily computed by accurately measuring an elapsed time over a known distance. Equation 3-2 below provides resu lts sufficiently close to the instantaneous velocity at impact as long as the distance over which the time measurements are taken are sufficiently small. This is illustrated mathema tically by analyzing the li mits of Equation 3-2 as d approaches zero, yielding the in stantaneous velocity of the ramm er. However, physically this is not possible as there is no way to measure change in time as the rammer passes a single point along its path. 21 21dd d Velocity ttt (3-2) In order to apply this method, the rammer inst rumentation needed to be set up to begin a trigger timer as the rammer passed a known point just prior to impact and a second sensor to measure the time as the rammer passed a second point sl ightly closer to the point of impact. The second time measurement is then subtracted from the first and dividing by the distance between sensors, an average velocity is co mputed. By measuring the distance between the sensors with a set of calipers accurate to 1/1000th of an inch, the dist ance between the sensors

PAGE 36

36 can be accurately measured. These measurements can then be inserted into Equation 3-2 and the velocity just prior to impact evaluated. While investigating this concept, it was felt th at the Hall effect or proximity sensors might be a viable choice. However, fu rther investigation showed that it would be virtually impossible to mount such a assemblage of transmitters/rece ivers close to the point of impact without creating measurement errors. Thus, it was decide d that another type of sensor would be investigated. 3.3.5 The Infrared Photo Gate Based on the fact that the da ta acquisition system can read ily read voltages, the idea emerged that a photo gate or optical switch might work. It is based on the principle that when a phototransistor detector senses an emitter diodes IR light, a voltage is produced. In addition, when the phototransistor detector does not sense the emission, the voltage remains zero. Three of these emitter/detector pairs were then planne d to be mounted in sequence and used to obtain the change in time. The rationale for using thr ee sensor pairs rather than two was based on the fact that if time measurements were known at three locations, then three separate velocity calculations were possible. These additional velocity measurements could then be used to determine if the velocity of the ra mmer was within an acceptable profile. Several infrared emitter and detector pairs were purchased with the appropriate resistors in order to create the switch confi gurations. Switch operation was then monitored with a voltmeter for preliminary tests and found to perform properl y. They were then mounted to a compaction mold base plate so that an emitter and its corresponding detector were on opposite sides of the plate (Figure 3-7).

PAGE 37

37 Figure 3-7. Standard test setup for infrared photo gate. In order to prevent rammer bounce, an impact pad was added to dampen the blow. Various materials were tested, since it is important for it to not undergo any permanent deformation. This is because the distance from the surface to the last detector would then change slightly and alter the velocity calculations. Having a proper impact pad also allowed for the lower emitter and detector pair to be aligned such that the vo ltage drop will occur the instant the rammer comes into contact with the pad. After numerous tests, several issues became evident. First, there were occasional voltage spikes prior to switch detection as well as random volta ge irregularities. Effort was spent trying to eliminate interference from out side infrared sources, as these were suspected of causing the problem. The increase in voltage prior to impact was determined to come from the reflection of infrared light reflecting off the bottom of th e rammers face prior to the rammer actually breaking the line of sight of the detector. These issues made it impossible to accurately Emitter 2 Emitter 1 Emitter 3 Detector 2 Detector 3 Detector 1 Lines between emitter-detector pairs are for illustration purposes only as infrared is not in the visible spectrum.

PAGE 38

38 determine the time of travel. An example of the data obtained from the infrared photo gate is presented in Figure 3-8. 0 0.05 0.1 0.15 0.2 0.25 00.0050.010.015 Time ( sec ) Voltage (V) Detector 1 Detector 2 Detector 3 t (sec) Sensor 1 to Sensor 2 t (sec) Sensor 2 to Sensor 3 Figure 3-8. Voltage measured from infr ared photo detector pairs 1, 2 and 3. Several rotations of the hammer with ten imp acts each were performed and velocities were calculated. These measured velocities were then compared to impact velocities obtained from differentiation of the lasers displacement reco rd. The results are presented in Table 3-2. Table 3-2 shows the importance in accurate time determination, since small differences in the measured time have a large adverse effect on the velocity calculations. It was determined that for accurate kinetic energy measurement, th e infrared emitter detector sensors were not adequate unless significantly improved. Attemp ts were made to obtain better quality voltage records by replacing the infrared emitted diodes with laser diodes, however, this also proved unsuccessful. Attention was then directed to an existing type of through-beam photo electric sensor which was available on the market.

PAGE 39

39 Table 3-2. Measured Travel Times from Sensors and Corresponding Energy Calculations E1 (from V1) (ft-lb) E2 (from V1 ) (ft-lb) 10.004860.0046887.4590.818.258.90 20.003140.00439135.3596.8119.7710.11 30.008560.0043549.6597.702.6610.30 40.007610.0055455.8576.713.376.35 50.006800.0056962.5074.694.226.02 60.004120.00417103.16101.9211.4811.21 70.004620.0041191.99103.419.1311.54 80.006560.0074864.7956.824.533.48 90.006700.0044263.4396.154.349.98 100.002540.00542167.3278.4130.216.64 Kinetic Energy V2 Avg Velocity Sensor 2 to Sensor 3 (in/sec) V1 Avg Velocity Sensor 1 to Sensor 2 (in/sec) Impact Sensor 1 to Sensor 2 t (sec) Sensor 2 to Sensor 3 t (sec) 3.3.6 Development of a Photo Electric Gate Traditionally, photo electric sensors have been utilized in manufacturing for product detection, but it was hypothesized and later proven that these sensors were able to effectively detect the presence of the rammer as it passed by a sensor in the same manner that the infrared sensors operated. The primary i ssue with these devices was thei r switching response times since all of the available models were digital (compared to analog). After significant searching, Keye nce Corporation had the precis e instrument to resolve the issues encountered in the testing of the infrared photo gate. After reviewing the specifications, two Keyence FS-M1H fiber optic amplifiers were purchased. These sensors operate on the same principle as the infrared photo ga tes in detecting the rammers presence. However, they boast more advanced electronic featur es for velocity measurements. The Keyence M1H fiber optic sensors operate dig itally. They are essentially a switch in the traditional sense of the word, with the output from the sensor either a fixed portion of the excita-

PAGE 40

40 tion voltage or zero. This function allows for simple determination of when the rammer passes the line of sight of the detector. While detecti on remained an issue for the infrared sensor, it was not the only issue solved by using this instrume nt. The fiber optic sens ors also offer a fixed sampling period of 20 microseconds which by calcula tion is more than sufficient for a spacing of 0.950 inches, the anticipated sensor spacing. Th ese features alone are reason enough to utilize these sensors. In addition, they feature pair sp ecific light modulation to prevent cross talking and false detections of the rammer which ensure accurate reportin g of rammer detection times. The sensor heads were mounted in the same configuration as the infrared sensors on the base plate of the compaction mold as a pair for testing (Figure 3-9). Figure 3-9. Base-mounted set up for fiber optic photo gate. 3.4 Development of the Load Cell System During the development stage of measuring th e impact kinetic energy, several compaction units were observed in various laboratories. It became apparent during these laboratory visits that while all of the machines were calibra ted per AASHTO standards, there was one major Sensor Spacing Sensor 1 Sensor 2

PAGE 41

41 difference that appeared fairly regularly. As per soil compaction standards, the compaction machine must be mounted on a rigid concrete ba se with a mass greater than 200 lb (AASHTO T99 and T-180 Note 7). All the machines observe d did satisfy this requirement, however, some had a plywood cushion beneath their steel base plates while others used aluminum spacers. In fact, several had nothing supporting their bases. As previously shown in Figure 2-3, an obser vable shift in the maximum dry density and the optimum moisture content occurs that is not consistent with AASHTO specifications. This is likely due to the varying amount of energy that is transferred into the soil during compaction. With the understanding of dampening and its energy e ffects, the possibility that the base stiffness itself may have a significant effect on soil compaction was surmised. In order to quantify what effect the base stiffness has on energy transfer standard compaction mold base plates, 4 and 6 inch respectively, were instrumented with both a PCB 200C20 load cell and a PCB M352A60 accelerometer. A 0.25-inch thick piece of neoprene pad was then cut to f it the impact surface of the load cell to protect it during im pact. These base plates were then fastened to the base of the machine and clamped in place. The compactors rammer could then be set up at the standard drop height angled slightly from the impact pa d to account for the rotatio n of the rammer. The machine could then be switched on for a sing le impact on the face of the load cell. Due to the accuracy and ease of repeatability of this test, it could then be used to measure the base stiffness of a sample population of comp action machines. From this information, the energy losses due to variables in the mounting configura tion of the machines could be quantified.

PAGE 42

42 CHAPTER 4 VALIDATION OF THE PORTABLE CALIBRATOR 4.1 Validation of the Photo Electric Gate In order to ensure accurate velocity measur ements using the fiber optic photo gate, it was critical to compare the veloci ties measure with a known velocity Thus, the laser displacement sensor was again mounted above the mechanical compaction machine and used to continuously measure the displacement of the rammer. The ph oto gates were mounted in the configuration pictured in Figure 3-9 as shown in Section 3.3. 6. The mechanical compaction machine in the T-180 configuration, was then switched on and a llowed to complete five full rotations of the impact rammer, in this case several of the impacts were missed by the photo gate when the rammer was near perpendicular to the sensor as discussed previously. For the 50 impacts, 29 were captured by the photo gate. However, this is not a problem, since the operator would simply wait until a sufficient number of data points are collected. The displacement record from the laser was then parsed such that the data for the fall could be analyzed. This data was then processed in a very similar way to that presented in Figure 3-4. However, through a study of the laser displacem ent data, the acceleration was not constant. Using a central difference scheme on the displacement versus time records, accelerations for points along the time record were calculated dire ctly. These results are presented in Figure 4-1. The results clearly show the acceleration cha nges with respect to time linearly throughout the fall event of the rammer. This prevents use of a second-order equation for derivation of impact velocity of the rammer as a valid method. Rather, it dictates th e use of a third-order polynomial equation for the description of the ramme rs fall with displacement in order to allow the linear change in acceleration with time.

PAGE 43

43 27 28 29 30 31 32 33 0.000.100.200.300.40 Time (sec)Acceleartion (in/sec2) Figure 4-1. Linear accelera tion of rammer during T-180 fall. Thus, a third-order equation was used to determ ine the impact velocity of the rammer at the time of impact. Impact velocities were t hus calculated from the laser displacement data and compared to the impact velocities measured using the photo gate m ounted on the compaction mold base plate. The results of this T-180 te st configuration showed a mean impact of 14.19 ft-lb with standard deviation of 0.24 as measur ed by the displacement laser. The photo gate mean kinetic energy was 14.30 ft-lb with a standard deviation of 0.99. Th e results from this test show fairly poor agreement between the two measurem ents. These discrepancies are due to the distance between the emitter and de tector optical fibers as well as the difficulty in precisely aligning the sensors. Since the mold and appurtena nces limit the installation height to 6 inches, and to bypass the issue with the rammer being perpendicular to the phot o gate and its signal being missed, the sensors were then relocated to the top of the compactor. Now, the time that the rammer breaks the line of sight of the detector an d the time that the line of sight is restored is used. The only line of sight of each other, for bo th sensors, is when the rammer is in contact with the impact pad (Figure 4-2).

PAGE 44

44 Figure 4-2. Sensor alignm ent on compaction machine. This setup was verified to operate correctly 100% of the time regardle ss of the orientation of the rammer. The verification study was perf ormed again using the laser displacement record and the velocity of impact measured by the phot o gate. The test operated for five complete rotations, and all 48 impacts were recorded by both sensors. A third-order polynomial derivative to describe the impact velocity of the rammer was compared to the velocity measured with the photo electric gate. This testing configuration resulted in much better agreement than the base mounted configuration. These results have b een summarized in terms of the rotation of the hammer as well as in terms of the entire population of rotations. However, for the summary of all rotations, it is important to note that the firs t rotation has been removed from the data set due to the improper function of the photo gate during this initial te st run (Table 4-1). Detector1 Emitter 1 Emitter2 Detector 2

PAGE 45

45 Table 4-1. Summary Statistics fo r Validation Study Hammer Rotations This table illustrates the precision in the meas urement of mean veloci ty for each rotation of the rammer. With the removal of the data acquired from the first rotation of the rammer, the summary statistics of the population are based on 39 impacts. The mean impact velocity measured using the photo elec tric gate was within 0.21 in./sec of the actual impact velocity of the rammer as measured by the displacement laser. This is less than a 0.2% difference. This maximum difference in measured velocities thus re sulted in a maximum percent difference in the mean energy calculations of 0.37 ft-lb while the other three useable rotations resulted in an absolute percent difference of less than 0.16 ft-lb. As an additional step to verify the accurate measurement of the impact energy, attention focused on the standard deviation of the ramme rs impact energy for each rotation. Analysis shows there to be good agreement between the standard deviation of the energy of each individual rotation as well as for the populati on as a whole for each ro tation of the hammer. Velocity Measured Using Photo Electric (in/sec) Vimp Lase r (in/sec) Kinetic Energy Measured (ft*lb) Kinetic Energy Calculated (ft*lb) Potential Energy Measured (ft*lb) Percent Difference between Kinetic Measured and Calculated (%) Percent Difference Between Kinetic Measured and Potential Measured (%) A verage 114.24115.8414.0914. 4815.272.737.77 Standard Deviation 0.700.810.170. 200.070.550.97 A verage 115.85115.7614.4814.4615.28-0.165.19 Standard Deviation 0.660.600.170. 150.040.681.12 A verage 115.73115.6514.4614.4315.29-0.155.44 Standard Deviation 0.630.630.160. 160.040.630.90 A verage 115.82115.6114.4814.4215.30-0.375.35 Standard Deviation 0.790.870.200. 220.030.671.32 A verage 115.91115.9514.5014. 5115.300.075.27 Standard Deviation 0.860.780.220. 200.060.731.26 A verage 115.83115.7414.4814.4615.29-0.155.31 Standard Deviation 0.720.720.180. 180.040.671.12 All Impacts in Rotation 2-5 Rotation 1 Rotation 2 Rotation 3 Rotation 4 Rotation 5

PAGE 46

46 From this study it was verified that the ve locity measured using the photo electrics mounted on the top of the compaction machine at a spacing of 0.950 inches is sufficiently accurate to establish the velocity at impact of the compaction ra mmer, as well as the variance of the energy during operation. 4.2 Accuracy of the Instrument In an effort to verify that the Keyence M1 H photo electric sensors accurately measure the velocity of a passing object, a time study was perfo rmed that utilized the operational frequency of the sensors and the distance between them. The design distance of the sensors was set at 0.950 inches. For an AASHTO T-180 compaction test, Newtonian physics shows that for a free fall of exactly 18 inches, an impact velocity of 117.89 inches per second should result. This velocity was then used to determine the time re quired for the rammer to travel a distance of 0.950 inches or approximately 0.00806 seconds. The operational frequency of the M1H photo electric sensors is 50 kHz (50,000 samples per second). This means that every 1/50,000th of a second, the sensor outputs a voltage corresponding to its line of sight. This results in an accuracy of 0.00002 seconds in the detection of the rammer at either sensor. Thus, for a two-se nsor system, the precisi on in the time of travel measurement could be off by a maximum of 0.00004 seconds. In order to determine the effect a time of tr avel error of 0.00004 seconds might have, this tolerance was applied to the 0.00806 time determ ined previously. A maximum error was found to result in an impact velocity range of 117.28 to 118.45 inches per second. This translates into a range in energies from 14.84 to 15.14 ft-lb, resulting in a tolerance of 1% of the actual energy for a single impact.

PAGE 47

47 It should be noted that this is the maximum error for any given impact. This error in time determination is considered random error and could thus occur for either sensor, translating into 0.00002 seconds and the mean centered about 0 s econds. As such, as long as multiple impacts are being measured, any error is offset when taking the average of the values. 4.3 Validation of the Testing Procedure 4.3.1 Photo Electrics The next step in the validation of the photo el ectric sensor was to validate the testing procedure. The same data that was used in the sensor validation study was used. However, now that it has been verified that the photo electric velocity measured was valid for use as the impact velocity, only these values were utilized. A bootstrap analysis was performed for each hammer rotation in an effort to verify that the mean energy for a single rotation of the hammer was representative of the mean of the multiple rotation s of the machine. In addition, the variance of a single rotation of the rammer was representative of the variance of the machine. For this analysis, the mean energy for each rotation was found to be the mean of the bootstrap for all cases. In comparison to the mean the variance of the bootstrap was found to be small and in all cases less than 0.0024. Since this value was small and the mean of the sample was equal to the mean of the bootstrap, statis tically, the mean for any single rotation is representative of the mean of the machine. In an effort to validate the variance of the mean energy for a single hammer rotation, the bootstraps for each rotation were again utiliz ed. The bootstrap mean variance for a single rotation was compared to the variance of the energy for each rotation a nd found to be within 0.01 of the variance of the rotation.

PAGE 48

48 Another method used to determine if a single rotation is enough to qua ntify the variance of the machine is to take the coefficient of variati on of the bootstrap standard deviation. This is done by Bootstrapping the kinetic energy measurem ents of a single rotation 4000 times. The standard deviation of the boot st rap standard deviation is then calculated and divided by the mean of the bootstrap standard deviation. A sm all value of the COV of the bootstrap standard deviations indicates that there are enough samp les to accurately quantif y the variance of the machines kinetic energy. If this value is small, that indicates the variance of the kinetic energy is representative of the machine. 4.3.2 Compliance Instrumentation In an effort to ensure the repeatability of th e results of the load cell and accelerometer base compliance device, several tests were run at UF and FDOTs SMO. These tests consisted of placing the compaction mold base plate in the compactor, using th e standard base plate vise. Several impacts were then created on the face of the load cell and recorded. The data generated by the impacts were then plotted as load and acceleration with respect to time. The results of these tests showed the ease of accurate repeatability for a given machine (Figures 4-3 and 4-4). This is due primarily to the accuracy under which these instruments have been calibrated.

PAGE 49

49 0 5 10 15 20 25 30 35 40 00.0050.010.0150.020.0250.03 Time ( sec ) Accelerations (g) Series1 Series2 Series3 Series4 Series5 Figure 4-3. Typical accelera tion and time duration plot for base compliance measurement. 0 200 400 600 800 1000 1200 1400 1600 00.0050.010.0150.020.0250.03 Time (sec)Force (lbs) Impact 1 Impact 2 Impact 3 Impact 4 Impact 5 Figure 4-4. Typical force and time durati on plot for base compliance measurement.

PAGE 50

50 CHAPTER 5 LABORATORY TESTING AND ANALYSIS 5.1 PDEC Setup Description Use of the portable dynamic energy calibrator (PDEC) relies on two main components, the photo electric gate and a base system complian ce mold. The photo electric gate pictured in Figures 5-1 easily mounts on the top of the compactor using two C-clamps. Adjustment to the photo electric gate then needs to be performed to ensure that the velocity is measured across the last 0.950 inches of travel or ju st prior to the rammers impact on the impact pad. This can be done by loosening the adjustment thumb bolts on the back of the mounting post and sliding the C channel section vertically until sensor pair 2 is at its switching point (see Figure 5-2 as well as Figure 4-2 in Section 4.1 and Appendix A). Figure 5-1. Photo electric gate A) Front view. B) Rear view. A

PAGE 51

51 Figure 5-1. Continued. Figure 5-2. Illustration of rammer rod at switching point of sensor 2. Adjustment thumb bolts B

PAGE 52

52 With the plywood and sorbothane impact pa d in place, the compaction machine can be started. The digital output from the Keyence M1H photo electric sensors are simultaneously sampled by a Measurement Computing 1608H data acquisition system at a rate of 50,000 samples per second (50 kS/s) and the laptop conver ts it to the impact ve locity of the rammer (Figure 5-3). Since the rammer ma ss (including the rod) was determ ined at the time of testing, by measuring the impact velocity, the kineti c energy of the rammer for each impact was calculated. Figure 5-3. Data acquisition system setup and sensors. Following the velocity measurements, forces and accelerations from a single rammer impact are measured at the base of the compactor (Figure 5-4). The device is stationary and there is a single impact location, so the rammer must be positioned correc tly to impact the force sensor. Figure 5-5 shows an instrumented 4-inch mold assembly with a force impact sensor and 500-g accelerometer affixed to the base. The respective mold assembly, 4-inch or 6-inch (not shown), is placed into the compactor and used to measure single impact for ces and accelerations. Data Ac q uisition S y stem Keyence M1H Photoelectric Amplifiers (2) Power Supply for Photo Electrics

PAGE 53

53 Figure 5-4. Base system configuration with mold assembly. Figure 5-5. Instrumented mold assembly. Accelerometer Housing and Sensor Impact Force Sensor

PAGE 54

54 The force sensor is positioned such that cont act is made with it at the centroid of the rammer. The accelerometer is located two inches aw ay from the force center, center-to-center. The sensor instruments are connected to the po rtable data acquisition system/notebook PC for data sampling (40,000 samples per second (40 kHz)) and storage. 5.2 Testing Program Overview Using PDEC Instumentation The PDEC was taken to 16 state and independe nt compaction laboratories in Florida for testing on T-99 and T-180 mechanical compactors. Laboratories were identified per FDOT districts as ones certified to pe rform T-99 and T-180 tests. Distri cts 2, 5, and 7 (see Figure 5-6) were visited. Figure 5-6. FDOT district map The data obtained provided a sample population with summary statistics, such as the mean, median, and standard deviation in impact velo city (function of drop he ight), kinetic energy

PAGE 55

55 (function of mass and impact velocity squared) and compliant characters (peak force, peak acceleration, peak time). The Modified (T-180) Proctor configuration was most frequently encountered in the laboratories. Thirty T-180 co mpactors and four T-99 machines were tested. The compactors were set on a foundation of cast-in-p lace concrete or block with aluminum, steel or plywood cushions between the machine base and the foundation (Figures 5-7). The results and analysis presented are from the thirty T-180 compactors. Figure 5-7. Compactor f oundation. A) Steel cushi on. B) Plywood cushion. SteelNuts Base of Mechanical Com p acto r A

PAGE 56

56 Figure 5-7. Continued 5.3 Testing Results and Analysis of PDEC Data 5.3.1 Kinetic Energy Assessment Impact velocity was measured for each mach ine. The kinetic energy of each impact is calculated from the impact velocity and rammer ma ss of each machine as shown in Equation 1-2. The kinetic energy data was summarized for th e variance per machine and variance among all the machines by considering the mean per machine. Figure 5-8 presents the frequency distribution of the mean energy of the sample population. The mean energy is taken as the sum of all kinetic energies in a single round of rammer impacts (8-10 impacts) divided by the number of impacts and represents a single machine. B

PAGE 57

57 0.0% 10.0% 20.0% 30.0% 40.0% 50.0% 60.0% 70.0% 80.0% 10111213141516 Mean Energy (Ft-lbs)Frequency0.0% 10.0% 20.0% 30.0% 40.0% 50.0% 60.0% 70.0% 80.0% 90.0% 100.0%Cumulative Frequency Population Frequency Cumulative Population Frequency Mean 13.66 Median 13.86 Standard Deviation 0.67 Bin Size 0.6 Ft*lbs Allowable Energy Range according to AASHTO Figure 5-8. Frequency distri bution of mean energy data. A bootstrapping procedure was then performed to ensure that the sample was representative of the entire population of machines. The functionality of bootstra pping is a numerical simulation by re-sampling real data. The process of bootstrapping takes the values of the sample population and randomly selects a value a specifie d number of times to generate a single bootstrap. From those values randomly selected, su mmary statistics can be used to describe the characteristics of that bootstrap, namely the m ean and variance. In an effort to ensure enough repetition in the bootstrap proce dure, 4,000 bootstraps were generated. The mean of all 4,000 was then calculated as well as the variance and mean of the bootstrap variances. These summary statistics for the bootstraps are then used to make statistical inferences for the entire population. In order to verify that the sample is sufficien t to represent the population, two key values are analyzed: the variance of the bootstrap variances; and the mean variance of the bootstrap. As long as the variance of the bootst rap variances is small, it indi cates that enough samples were

PAGE 58

58 taken from the population to establish the variance of the bootstrap population accurately. In this case, the mean variance of the boot straps can then be compared to the variance of the sample population. Good agreement of these values i ndicates the sample is representative of the population. From the sample of thirty machines, a bootstra p analysis was performed on thirty values (with the possibility of repetition) randomly c hosen to generate each bootstrap. From the summary statistics, the mean of the variance for all of the bootstraps is 0.43. When compared with the variance of the sample populati on (which had a variance of 0.45 (0.672)), it is apparent they are in excellent agreement. This shows th at the population sample of thirty machines is representative of the entire popu lation of proctor compaction machines. Next, the variance of the bootstrapped variances was calcul ated to be 0.03. Since it is significantly sm aller than the mean variance of the bootstraps, it shows ther e is high accuracy in the determination of the variance of the population usi ng those thirty machines. Also using the bootstrap method, one is able to gain more confidence in the distribution of the data collected. The bootstrap procedure qua ntifies the uncertainty of the mean through statistical inference. Shown in Figure 5-9 is the frequenc y distribution of the bootstrap procedure for the 30 mean energy data values. Th e distribution type sugge sted by Figure 5-9 is normal with a mean and median very close to that of the sample population (13.66 ft-lb and 13.86 ft-lb, respectively). The distribution of the variance of the data mean is small (0.32 = 0.09). A comparison of the data and bootst rap distributions show good agreem ent. It is important when comparing the bootstrap distribution and the da ta distribution for one to note the summary statistics, to see if the mean en ergies coincide and the median values are close to one another.

PAGE 59

59 For example, the median value for the bootstrap is within 1.5% of the median value for the laboratory data set. 0.0% 10.0% 20.0% 30.0% 40.0% 50.0% 60.0% 70.0% 80.0% 10111213141516 Mean Energy (Ft-lbs)Frequency0.0% 10.0% 20.0% 30.0% 40.0% 50.0% 60.0% 70.0% 80.0% 90.0% 100.0%Cumulative Frequency Bootstrap Frequency Cumulative Bootstrap Frequency Mean 13.66 Median 13.70 Standard Deviation 0.30 Bin Size 0.6 Ft*lbs Allowable Energy Range according to AASHTO Figure 5-9. Bootstrap frequency distribution of mean energy. Figure 5-10 shows the frequency di stribution of all the calculate d kinetic energies from the data; these are not a representati on of the mean values. These ar e all energies for a single round of impacts on each machine. Compared to Figure 5-8, Figure 5-10 provides better insight into the percent of the population be low the allowable range of en ergy based on AASHTO standards of rammer mass and drop height. Through close examination and interpre tation of Figure 5-10, it is possible to see that the cumulative freque ncy portion of the plot showed no impacts above the AASHTO specified energy. Rammer masses were measured for each machine and it was observed that all rammer masses were within the specified tolerances. This indicates that low energy available for compaction stem s from low impact velocities.

PAGE 60

60 0% 10% 20% 30% 40% 50% 60% 70% 80% 10111213141516 Energy (ft*lb)Frequency0.0% 10.0% 20.0% 30.0% 40.0% 50.0% 60.0% 70.0% 80.0% 90.0% 100.0% Frequency Cumulative Frequency Mean 13.62 Median 13.85 Standard Deviation 0.73 Bin size 0.6 ft*lbs A llowable Energy Range according to AASHTO Figure 5-10. Frequency distri bution of all energy data. With the summary statistics and the mean velo city and energy, Figures 5-11 and 5-12 were generated as scatter plots. In this case, a scatter pl ot provides a good visu al identification of machines which may be outliers of the population a nd where to look for the sources of error. For example, in Figure 5-11, there is an extreme outli er with a standard deviation of approximately 4.5 and mean velocity of 103 inches/second. Si nce the velocity, V, is a function of the drop height and free fall acceleration, this indica tes there is a problem with the machine not consistently dropping the mass from the same he ight and/or large in consistent frictional impedance during the free fall acceleration. A point near the low end of the standard deviation, for example 0.5, indicates a cons istent deviation about the mean for this machine, essentially showing that the drop height and or the frictional forces on the rod are consistent for each fall of the rammer. According to drop heights in AASHTO and ASTM standards, the range of allowable velocities based on free fall is shown in red in Figure 5-11.

PAGE 61

61 R2 = 0.5172 0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00 4.50 5.00 101.00103.00105.00107.00109.00111.00113.00115.00117.00119.00 Mean Velocity of Rammer Blows (ft-lbs)Standard Deviation of Machine Blows for a Single Rammer Rotation A llowable Velocity Range according to A SHTTO Figure 5-11. Scatter plot of energy sta ndard deviation versus mean velocity. R2 = 0.4747 0.00 0.20 0.40 0.60 0.80 1.00 1.20 10.0011.0012.0013.0014.0015.0016.00 Mean Energy of Rammer Blows (ft-lbs)Standard Deviation of Machine Blows for a Single Rammer Rotation A llowable Energy Range according to A SHTTO Figure 5-12. Scatter plot of energy st andard deviation versus mean energy.

PAGE 62

62 Figure 5-12 shows the scatter plot for the mean kinetic energy of th e sample population. The results show a definitive trend in the data. That is to say, machines that display a low standard deviation in the energy delivered by each blow are more likely to have a higher mean energy, whereas machines that have a large en ergy standard deviation typically have a lower mean energy associated with the machine. In general, the majority of the population has small standard deviations ( 0.10 to 0.20) and all are below the range of allowable energy according the AASHTO standards T-180. 5.3.2 Manual Compaction Rammer In order to establish a baseline for understand ing the kinetic energy measurements from the mechanical compaction machine, testing was pe rformed to quantify the typical kinetic energy available from a T-180 manual rammer. Six manual T-180 compaction rammers were te sted. Three of the six were different commercial models. The rammers were attached to a temporary frame using C-clamps to maintain vertical alignments (Figure 5-13). Th e photo electric gate was then attached to an adjustable height table and the plywood and sor bothane impact pad placed beneath the impact point of the rammer.

PAGE 63

63 Pipe coupler clamps Sorbothane and wood impact pad T-180 manual rammer Figure 5-13. Frame for maintaini ng vertical alignment of rammer. The adjustment described for the mechanical compactor was performed for this setup as well, i.e., the sensors were set su ch that the rammer was located just beyond the switching point. The mean of at least 25 impacts were then recorded for each of the six manual T-180 rammers. These six means were then summarized by thei r mean and variance, 14.25 ft-lb and 0.005 ft2-lb2, respectively, and then used to perform a bootst rap procedure to ensure sufficient tests were conducted. The bootstrap mean was calculated as 14.25 ft-lb as well, with the mean variance of the bootstraps determined to be 0.004 ft2-lb2. The most probable reason why the theoretical energy of 15 ft-lb (10 lb x 1.5 ft drop height) was not achieved is due to friction between the rammer and hammer housing. It is virtually impossible to provide a friction free fall since even a

PAGE 64

64 slight inclination of the hammer will create concomitant friction between the moving mass and its housing. Of significant importance with the bootstrap pr ocedure is the variance determined to be on the order of 6 10-6, which suggests that the calculation of the bootstrap means is highly precise. This high precision indicates th at six values for the energy associated with the manual compaction rammer is a sufficiently large populat ion to accurately determine the mean and variance to describe the populati on of all T-180 manual rammers. The small variance suggests that for any manual rammer, the energy is like ly to be extremely cl ose to 14.25 ft-lb. 5.3.3 Base Compliance Following the measurement of the kinetic ener gy for each of the machines tested at the sixteen compaction labs, force and acceleration at the base plate were measured for multiple impacts of the rammer (three to eight) on ea ch machine in the population using the base compliance instrumentation (Figure 5-5). Th e mean force and acceleration for each machine were plotted and reflect similarly to each othe r for any given machine (Figures 5-14 and 5-15), machines with a high mean force display high m ean accelerations. The standard deviations are relatively low compared to the magnitudes, altho ugh there are four types of base systems in the population and measurements are also a function of velocity.

PAGE 65

65 0.00 200.00 400.00 600.00 800.00 1000.00 1200.00 1400.00 1600.00 1800.00 2000.0005101520253035MachineForce (lbs) Mean 1501.72 Median 1468.91 Standard Deviation 171.29 Figure 5-14. Scatter plot of average maximum force. 0 5 10 15 20 25 30 35 40 45 50 05101520253035 MachineAcceleration (g) Mean 36.70 Median 36.10 Standard Deviation 4.00 Figure 5-15. Scatter plot of average maximum acceleration.

PAGE 66

66 The primary reason for measuring the force an d acceleration at the base of the compaction equipment was to quantify the energy or th e purpose of identifying a possible source of difference in compaction density re sults. As previously discussed, the stiffness of the material supporting material beneath the compaction mold has an influence on the resulting maximum dry density and optimum moisture content. 5.3.4 Force Compressive Energy Theory In an effort to quantify the effect that diffe rent base types have on the compaction process, the energy at the base of the compaction device n eeded to be quantified. Compressive energy at a point along a Standard Penetrati on Test (SPT) rod can be expresse d as an integral of the force squared with respect to time multiplied by the dynamic impedance of the SPT rod (Palacios 1977) (Equations 5-1 and 5-2). In the case of the load cell being the same cylindrical shape and displaying 1-D wave transmission down the length of the cylinder (the load cell housing in this case), the same assumptions hold tr ue and validate the use of SPT compressive energy equations in this study. c Dynamic Impedence Ma (5-1) t 2 impacti 0c EnergyF Ma (5-2) C = wave velocity (ft./sec) M = Youngs Modulus (psf) a = cross sectional area (ft.2). 5.3.5 Compressive Energy Results For each machine, a numeric integral of the force squared was calculated for the time over which it occurred. The mean of the integrals we re then recorded for each machine (Table B-2).

PAGE 67

67 Force and acceleration were plotted for severa l machines and used to verify that the dynamic impedance of the base compliance system was the same for every machine (Figure 516). The relationship of measured force to acceleration of the syst em was observed to be linear. This relationship verifies that the dynamic impe dance for the system is a constant value. 0 200 400 600 800 1000 1200 1400 1600 05101520253035 Acceleration (g)Force (lbs) Machine 1 Machine 2 Machine 5 Figure 5-16. Dynamic force and acceleration. Since this relationship is linear the dynamic im pedance of the system can be determined by dividing the particle velocity occurring at a point in time by the dynamic force measure at the same point in time (Equation 5-3). Since the accelerometer was mounted on the same surface and close to the load cell, the accel eration record is useful in dete rmining the particle velocity at times corresponding to force measurements.

PAGE 68

68 Dynamic Impedance F V (5-3) F = Force (lbs) V = Velocity (t g A ) (ft/sec) g = Gravitational constant (32.2 ft/sec2) t = Time (sec) Values were pulled from figures 5-14 and 5-15 for typical maximum forces and accelerations from T-180B machines and used to evaluate Equation 5-3. A typical dynamic force of 1500 lbs and an accelerati on of 37 g were used to evaluate the equation using a time of 0.00375 sec. Calculation of the pa rticle velocity yields 4.5 ft/s ec, resulting in a calculated dynamic impedance of 0.003. This illustrates that a value on the order of 1/300 ft/(sec-lb) accurately assesses the dynamic im pedance of the base compliance device and the base system for T-180B machines. It has been illustrated that the force and accelerati on relationship for the machines tested was constant and the dynamic impedance of the base system evaluated. Therefore, the use of the accelerometer is not required for determinati on of the compressive energy on an individual machine basis. Due to the difference in the wave transm ission through the different sized compaction mold base plates required for the T-180A and T180B tests (4 inch and 6 inch diameter molds), the sample population of base compliance data wa s subdivided by base type for further analysis. This resulted in a subpopulati on of nine T-180A machines and twenty-one T-180B machines. The subpopulation of the T-180A machines was too small to be used in further calculations. Of the twenty-one T-180B machines, several diffe rent base types were represented. These consisted of two machines with plywood m ounted between the machine and the foundation block, three machines mounted di rectly to concrete, and the re maining sixteen machines were

PAGE 69

69 mounted with aluminum spacers or steel nuts be tween the machine and their foundation blocks. The integral of the force squared as well as the maximum force and maximum accelerations for machines mounted with plywood displayed unchara cteristically high compliance measurements. Due to this unrepresentative behavior and uncommon occurrence in the population (two machines), they were removed from the final an alysis so as to not misrepresent the sample population of T-180B compaction machines. Summary statistics were perfor med on the compressive energies of the nineteen machines as well as a bootstrap procedure to ensure that the population of ni neteen machines was adequate for quantifying the variance of the population. A summary table contai ning the measurements from the nineteen machines is presented in Table B-2 in Appendix B. The mean compressive energy was 11.66 ft-lb with a variance 1.39 (ft-lb)2 and the mean of the bootstrap means was 11.65 ft-lb and variance mean of th e bootstraps equal to 1.32 (ft-lb)2. The variance of the bootstrap variances was then calculated as 0.082, more than two orders of magnitude smaller than the variance of the sample, thus indicati ng that nineteen machin es is large enough to establish the mean and variance of the population of T-180B base compliances. The coefficient of variation (COV or CV defined in Equation 54) was then calculated for both the sample and bootstrap populations, and were found to be in go od agreement, i.e., 0.12 and 0.11, respectively. COV or CV = / (5-4) = standard deviation = mean. 5.4 Measured and Predicted Dry Density Variance of A2-4 Soil 5.4.1 Background As identified earlier, soil compaction is affect ed by the amount of energy imparted to the soil sample, the stiffness of the base on which th e sample is compacted as well as the moisture

PAGE 70

70 content of the sample being compacted and ot her operator influences. Examples of operator influences are mixing of soil, placement and thic kness of lifts as well as placement of mold in frame (i.e. tightening of bolts). Of interest, is the relative influence of each variable on the resulting dry densities as measured after compacti on. The latter effects may readily be assessed by establishing a linear relationship between dry density, moisture, kinetic energy, and compliance and then assessing the variance of the relationship. Subsequently, if the measured kinetic or compliance variance were substituted into the linear variance relationship, a variance of the dry density may be predicted. Reasons why the predicted variance may not agree with the measured variance is that the relationship are; one the relationship may not be linear and two, operator effects. To ensure the relationship is linear, a small moisture content range should be considered on both the dry and wet side of optim um. In the case of operator influence, its variance is unknown. However, by comparing the expected that variance between the measured dry densities and predicted variance due to kineti c energy, compliance and moisture content, the difference may be assumed to be ope rator effects (i.e. lumping all othe r influences). In addition, the linear variance relationshi p will assist in identifying wh ere improvements in compaction testing should be undertaken. The latter is ve ry important, since curr ent ASTM specifications require comparisons of dry densities between manual and automatic compaction equipment, and not necessarily the dependent variables (e.g. comp liance stiffness). A discussion of soil selected, testing requirements, statistical analysis and results follows. 5.4.2 Properties of Test Soil An A-2-4 embankment fill soil was selected for the soil density compaction study due to its prevalence in Florida and e xpected large range (vs. A-3 so il). A large sample of A-2-4 material was obtained from a Florida DOT borrow pit (Lake C ity) and tested at the SMO laboratory. The compaction results for T-180 are given in Figure 5-17, content from the FDOT

PAGE 71

71 State Materials Office evident from Figure 5-17, the moisture vs. dry density curve is highly nonlinear. To assume it to be linear over a smal l range, the soil testing had to be broken up into two moisture ranges, if the study was to be perf ormed on the whole compaction curve. Since there are different processes occu rring on either side of optimum mo isture, both should be tested. For instance on the wet side of op timum as more energy is imparted to the soil, the zero air void line is approached, this results in a maximum a ttainable dry density for any moisture content, causing lower variance in the overall density re sults. On the dry side, small energy changes usually result in larg er density changes. Consequently, moisture contents at 6 and 7. 5 were selected for te sting. Subsequently, samples of the test material were then distributed to several state and private labs and asked to assess dry densities at two moistures, one wet of optimum and the other dry of optimum.

PAGE 72

72 Figure 5-17. Soil compaction resu lts of A-2-4 test material.

PAGE 73

73 Each participating laboratory was then asked produced 20 soil density pills on specified T180 machines without reconstituting the soil. In each test performed, the technician running the test was to use a portion of the sample large enou gh to prepare a pill, hydrate it to the specified moisture content and compact the pill as they would ordinarily do in accordance with AASHTO T-180 procedures. The use of a new sample of material for each pill was requested so as to preserve the structure of the soil and remove any error possibly associated with repeatedly drying and hydrating the soil. Ten pills were compacted on the dry side of opt imum at 6% moisture content and ten were compacted on the wet side of optimum at 7.5% moisture content (Figure 5-18). 129.0 130.0 131.0 132.0 133.0 134.0 135.0 2.03.04.05.06.07.08.09.010.0Moisture Content (%)Dry Unit Mass (lb/ft3) Figure 5-18. Test target mois ture content on Proctor curve In accordance with the AASHTO T-180 standard for soil compaction, the moisture content of each pill was determined by the machine operato r utilizing an oven dried sample of the pill material. A total of six machine test results were recei ved back from the labs. Shown in Figures 519 and 5-20 are scatter plots the data obtained. Since each machines results are identified, the variability of both machine and by labs are evident in Figures 5-19 and 5-20. Also, the plots Optimum Moisture Content 6.0% Moisture Content 7.5% Moisture Content Dry Unit Mass (lb/ft3) Moisture Content ( % )

PAGE 74

74 clearly depict the inability for testing to be performed at a single given moisture content as evident from the spread of data across a moisture content range. The latter spread occurred over different machines as well as from any single machin es test. In addition, so me of the results, the moisture content of the tests were far from the specified testing moisture content and so these values were removed from the analysis so as to not invalidate the assumption of linearity. In addition, it should also be noted that the moistures reported are average values and do not consider any variabil ity within the pill. 130.5 131.5 132.5 133.5 134.5 135.5 5.05.56.06.57.0 Moisture Content (%)Dry Density (lb/ft3) Machine 1 Machine 11 Machine 13 Machine 14 Machine 20 Machine 27 Figure 5-19. Scatter plot of 6% moisture content results.

PAGE 75

75 130.5 131.5 132.5 133.5 134.5 135.5 6.06.57.07.58.0 Moisture Content ( % ) Dry Density (lb/ft) Machine 1 Machine 11 Machine 13 Machine 14 Machine 26 Machine 27 Figure 5-20. Scatter plot of 7.5% moisture content results. 5.4.3 Regression Analysis of Lab Test Results As identified earlier, a linear relationship between dry density with moisture, kinetic energy and base compliance Equation 5-5 was assumed: D ance BaseCompli C rgy KineticEne B Ad ) ( ) ( ) ( (5-5) In order to prepare the data for analysis th e results from the dens ity testing performed on T-180B machines were subsequently paired by m ean kinetic energy and base compliance (Tables B.4 and B.5) in order to establish the constant coe fficients in Equation 5.5. The Summary data tables (Tables B.4 and B.5) were then used to calculate a regression equation by utilizing the least squares method of analysis from the MiniTab software.

PAGE 76

76 The following linear equations describe the dry density as a function of moisture content, kinetic energy and compliance for the 6% and 7.5% moisture content por tions of the compaction curve, respectively (Equations 5-6 and 5-7). 122 489 0 007 0 846 0 CE KEd (5-6) 144 187 0 202 0 05 2 CE KEd (5-7) = moisture content (%) KE = Kinetic Energy (ft-lbs) CE = Base Compliance Energy (ft-lbs) The evaluation of each of these equations using the mean of each vari able illustrates the ability of the equations in determining the expe cted values of dry density. For example, Equation 5-6 was evaluated using the mean moisture content of the % moisture content data, mean kinetic energy and base compliance (Tables 5-1) for the six machines tested, and the predicted dry density was 132.7 pcf. When comp ared to the mean value of the measured dry densities used to generate this regression equation, a mean of 132.2 pcf was found. Evident, the regression equation does quite well in estimatin g the mean. In the case of 7.5% moisture content, and the measured kinetic, compliance (T able 5-2) and moisture content, the regression equation (Equation 5-7) predicted a dry density of 133.6 pcf. The actual measured mean dry density of the data is 133.2, or a difference of 0.4 pcf. Because of the good agreement between the expected dry density and the mean dry dens ity, the variance of the densities on the wet and dry side may now be explored.

PAGE 77

77 Table 5-1. Component Summar y Table for Machines Used in % Moisture Content Regression Analysis Machine Number Mean Kinetic Energy (ft-lb) Compressive Energy (ft-lbs) 113.2012.29 1113.4310.87 1314.1512.58 1413.9211.31 2014.1510.35 2713.7511.55 Variance0.150.71 Mean13.7711.49 Table 5-2. Component Summar y Table for Machines Used in .5% Moisture Content Regression Analysis Machine Number Mean Kinetic Energy (ft-lb) Compressive Energy (ft-lbs) 113.2012.29 1113.4310.87 1314.1512.58 1413.9211.31 2612.228.56 2713.7511.55 Variance0.482.07 Mean13.4411.19 5.4.4 Standardized Regression Equation Since the units of each variable are not the sa me, the magnitude of the coefficients of each variable cannot be directly compared to determ ine which parameter has the most effect on the outcome. Because this is an issue of interest a process of standardiz ing the data for each parameter was carried out. Sta ndardization removes the effect of the mean and the variance on the determination of coefficients in th e regression analysis (Equation 5-8).

PAGE 78

78 i iX z (5-8) Xi = Original Variable Value = Mean of Variable = Standard Deviation of Variable zi = Standardized Variable This process forces the mean of each of the vari ables to be zero and the variance to be one. The entirety of tables B.3 and B.4 were then st andardized and used to generate a standardized regression equation for the each of the moisture contents at 6% and 7.5% respectively (Equations 5-9 and 5-10). 071 0 572 0 004 0 235 0 Std Std Std dStdCE KE (5-9) 0012 0 293 0 152 0 679 0 Std Std Std dStdCE KE (5-10) From this standardization analysis, it can be noted that in terms of direct effect on the resulting dry density without the effect of units, base compliance energy has the greatest influence on results at 6% mois ture content (Equation 5-9). It is shown that kinetic energy has almost no discernable effect on the resulting dry density at 6% moisture. The latter may be attributed to the small coefficient in Equation 5-6 for the Kinetic energy which was an outcome of the machines that data was recorded, i.e. ve ry small variance of KE and constant, mean. Generally, the kinetic energy has a large effect on density, i.e. T-99 vs. T-180. If more machines were used to determine the regression equation it is likely that a more discernable trend would occur, resulting in a larger coefficient. For the analysis of the 7.5% mo isture content, Equation 5-7, it is evident that the standardized base compliance had the largest effect on resulting densities. In the non-standardiz ed form, the kinetic energy has th e largest effect on the resulting

PAGE 79

79 dry density. This is directly related to the fact that the measured values of kinetic energy are larger than those of base compliance and the st andard deviation is smaller on the sub population of kinetic energies. While the st andardized regression e quation is of value, the more useful of the regression equations is the non-standardized form, which allows an analysis of the variance of each of the parameters and the vari ance of the resulting dry densities. 5.4.5 Variance of Dry Densities, Measured vs. Predicted Application of the second central moment to the coefficients of the mean equations (Equations 5-6 & 5-7) allows the variances of eac h parameter to be summed to find the expected variance of the resulting dry dens ities (Equation 5-11). Note, the va riance of the constant D in the regression Equation (5-5), is zero, which l eaves the sum of the component variances (Tables 5-1 & 5-2) times the square of each of their corresponding coefficients. ) ( ) ( ) ( ) (2 2 2ance BaseCompli Var C rgy KineticEne Var B Var A Vard (5-11) The Variance form of the regression analysis equation, Equation 5-11 is particularly useful for determining the effect that a parameters variance has on the expected variance of the outcome. For instance, when the variance of the moisture content, kinetic energies (Tables 5-1 &5-2) and the base compliance (T able 5-1 &5-2) energies are us ed to evaluate Equation 5-11 this will result in a value for the expected vari ance in the resulting dry densities. For instance, the predicted variance of dry de nsity for the tests performed at 6% moisture is 0.20 (pcf)2 Where as the actual measured variance of the dry de nsities recorded from the six laboratory tests was 0.52 (pcf)2. This shows the ability to account for th e cause and magnitude of nearly half of the variance observed in density re sults. Carrying out the same approach with data from the 7.5% moisture results, the predicted va riance in the dry density is 0.48 (pcf)2. The measured variance of laboratory data at 7.5% moisture was 0.84 (pcf)2.

PAGE 80

80 The observed differences in the measured and calculated variances of the dry densities are likely due to the operator of the machines effect on the dry density results. This effect cannot currently be completely quantified due to the complexities in the evaluation human ability. Some of the sources of possible variance that can be attributed to the operator are: -Variances in the process of mixing the soil and water -Lift Thickness -Amount of tamping applied to the sp ecimen prior to compacting each lift -How tightly the compaction mold is assembled -Variance in sampling to determine moisture content of the pill -Accuracy in trimming final lift from the top of the compaction mold 5.4.6 Summary of Predicted and Measured Variances of A2-4 Tested Soil The developed regression equations for mean and variance of dry density vs. moisture content, kinetic energy and compliance can only be interpreted to be representative of the machines used to generate the equations. If more data is us ed to generate the regression equations, a change in the coefficients of the te rms in the equation may occur. Similar changes would occur if different soils were tested. Howeve r, for the sake of interest in what type of results might be acquired from th e possibility that the regression equations ar e representative of the entire population of machines, the variance of the population of thirty ki netic energies (0.45) and nineteen T-180 base compliance measurements (1.39) were used in the variance Equation 510, an expect variances of 0.36 (pcf)2 at 6% moisture content and 0.45 (pcf)2 would be computed for 7.5% moisture. Interestingly, 7.5% moisture content variance agrees quite closely with the variance recorded for the 6 machine results in this study. The latt er suggests that more machine tests need to be performed on th e dry side and especially with different kinetic energies. Also from a comparison of the measured and predicted variances of the machines tested significant variance (i.e. 50%) may be attributed to human operator differences.

PAGE 81

81 CHAPTER 6 CONCLUSION/RECOMMENDATIONS 6.1 Conclusions 6.1.1 Background Laboratory Proctor compaction is performed by state agencies and private laboratories for determination of unique dry density-moisture cont ent relationship of soil s used for embankment construction. This type of compaction is pref ormed through imparting the kinetic energy, to a confined soil mass, of a sta ndard rammer mass falling from a st andard height (AASHTO T-99, T-180 and ASTM D689, D1557). For quality control of Proctor compaction thr ough out the state of Florida, the Florida Department of Transportations State Mate rials Office (FDOT-SMO) routinely compares compaction results, i.e., maximum dry density, fr om mechanical compactors. This, indirectly, assess calibration, or compliance, of the compact or that undergoes a standard calibration (ASTM D2168) based on obtaining similar ma ximum dry densities, within tolerance, between the manual and mechanical Proctor compaction process. The dry densities obtained using the compactors are influenced by the kinetic energy available, the base that the m achine rests (foundation and cushion), and the operator. Variance of these pa rts will lead to, perhap s, unsatisfactory variance in the densities and without quantifying the va riance of each part compactors may be out of calibration accordi ng to standards. The objective of this research was to devel op a portable, electronic calibration device for use on mechanical Proctor compaction machines and determine sources of variance in soil density results. The Portable Dynamic Energy Calibrator (PDEC) was de veloped and used to quantify the kinetic energy and the compressive impact energy on T-180 mechanical compaction machines throughout central and nor th Florida. Variance in soil densities was quantified for six

PAGE 82

82 of T-180 machines so as to quantify the operato r influence and variance. The results provided valuable insight into the variability of co mpaction machines, both i ndividually and as a population, and the effect of their parameters on soil density. It is expect ed that the FDOT will continue to use this equipmen t in the future to verify th at compaction machines are in compliance with appropriate standards (AAS HTO T-180 and ASTM D1557). A summary of major findings in each category follows. 6.1.2 Calibrator Development and Operation The calibrator was developed to capture the rammers kinetic energy at impact and the compressive force from rammer impact at the comp actor base. Portability and rapid employment was a necessity throughout the development as co mpactors throughout the stat e will be observed. Data collected and stored on a notebook comput er can then be processed through a macro command with in an Excel spreadsheet allowing the user to immediatel y observe results. A photo electric gate (Figure 6-1) was developed to be at tached to the top of the compaction machine and capture the passing ra mmer rod (Figure 6-2). Validation and determination of accuracy consisted of a laser displacement sensor companion observation of the rammer and rod during op eration. It was determined that velocities measured using the photo gate are representative of the rammers impact velocity and used in ki netic energy calculation with an accuracy of 1%. After the photo electri c gate captures the rammer impact velocity, the user determines the rammer mass and enters into the spreadsheet for calc ulation of the kinetic energy. Since the operation of the machine cons ists of the rammer rotating about a point, a complete rotation is observed allowing for multip le kinetic energy values to be calculated.

PAGE 83

83 Figure 6-1. Photo electric gate. Figure 6-2. Photo elec tric gate on compactor. Detector 1 Emitter 1 Emitter 2 Detector 2

PAGE 84

84 It was shown that the base system, i.e., f oundation and cushion (Figure 6-3), influence the compressive energies and thereby the maximum dry densities and optimum moisture content (Figure 6-4). This is due to the differences in the modulus and density of the two base systems. The velocity, c, of a compressive wave is a function of materials m odulus and density. A dynamic load cell rigidly to a compaction mold plat e (Figure 6-5) and placed at the base of the compaction machine is able to capture the impact compressive force. By integrating the impact force over the event duration (t2 t1) and multiplying by material dynamic impedance (Equation 6-1), the compressive energy from an imp act can be determined (Equation 6-2). c Dynamic Impedence Ma (6-1) t 2 impacti 0c EnergyF Ma (6-2) c = wave velocity (M/ ) (in/sec) M = Youngs Modulus (psi) A = cross sectional area (in2) The use of the dynamic load cell on machines w ithin the population is valuable in that it allows for comparison between machines and its influence on the density results. The compressive force measurement device is fixed into position at the compactor base and a series of impacts (5-10) are captured. Si nce the device does not rotate as does the rammer, the machine must be operated on single blows. This requires placement of the rammer off center to the load cell such that upon lif t and rotation the rammer strikes the center of the load cell. Next, the user enters the impedance into the sp readsheet and keystrokes the macro for calculation of the compressive energy.

PAGE 85

85 Figure 6-3. Compactor base conf iguration and mold assembly. 109.17 112.13 114.18 110.06 115.01 110.37 113.29 111.13 108.00 110.00 112.00 114.00 116.00 118.00 120.00 122.00 124.00 126.00 6789101112131415 MOISTURE (%)DRY DENSITY (PCF) Wood Cushion in place Foundation Block Only Zero Air Void Figure 6-4. Influence of cush ion on A-2-4 compaction curve.

PAGE 86

86 Impact Force Sensor Figure 6-5. Instrumented mold assembly. The Portable Dynamic Energy Cali brator (PDEC) is easily tran sportable and can be readily employed in a compaction laboratory. The photo electric gate is ad justable and affixes to the top of the compaction machine with no alterations needed to the mach ine. The dynamic force device is placed in the compactor and captures single rammer blows. A data acquisition system, which accompanies the devices, samples the sensors and st ores the data for immediate analysis. The user can run a macro on the data collected and, with a few variables entered, can observe the rammer kinetic energy and base compressive energy. 6.1.3 Laboratory Testing and Data Analysis 6.1.3.1 Kinetic Energy Sixteen state and privat e laboratories were tested with the PDEC for the quantification of energy their machines display. Thirty T-180 machines were tested for kinetic energy by monitoring a single rotati on of the compaction hammer. Each machines tested displayed an energy less than the theoretical energy pres cribed by AASHTO for T-180 soil testing. The

PAGE 87

87 laboratory testing results also in dicated that there is a well defi ned proportionality to the mean and standard deviation. Machin es that display high variances in kinetic energy also display lower than normal mean kinetic energies. The population of machines tested displayed a mean kinetic energy of 13.66 ft-lbs with a variance of 0.45 ft2-lbs2. Kinetic energy was measured for six ma nual T-180 compaction hammers, each hammer displayed nearly identical mean kinetic en ergies (14.25 ft-lbs) and variance (0.005 ft2-lbs2). This low variance of the mean kinetic energy indi cates that all manual compaction hammers are essentially the same. 6.1.3.2 Base Compliance Energy Due to the small population size, 21 T-180B m achines with many base types, it was found that the mean and variance of base compliance energy could not accurately be assessed for any specific base type. However, it was determined that the population of machine base compliances was large enough to be analyzed as a whole fo r T-180B machines. The population mean of T180B base compliance energies was found to be 11.66 ft-lbs and the variance 1.39 ft2-lbs2. Using this analysis it was found that the C OV of the compressive energy (0.12) of the compaction machine was much larger than COV the kinetic energy (0.05). 6.1.3.3 Regression Analysis Soil density results in conjuncti on with energy measurements from six machines were used to determine regression m odels that describe the effects of dependent variables on dry density at moisture contents wet and dry of optimum (Equations 5-6 & 5-7). The models were validated by calculating a predicted mean dry density that compared closely to the mean density of the laboratory tests. The regression equations developed were used to determine the effect of each parameters variance on the variance of the repor ted soil densities. In the cases of tests run both wet and dry

PAGE 88

88 of optimum the sum of the parame ters variances were able to qua ntify approximately half of the total variance observed in density results. Base compliance contributes more variance to the resulting densities than kinetic en ergy. This was largely due to the larger variance observed in the base compliance measurements used to de velop the regression equations. The remaining variance that cannot be quantif ied through summing measured variances of the regression parameters are likely attributed to the operator and any issues associated with the fit of the regression equation. 6.2 Recommendations Current Proctor compaction calibration proce dures do not quantify th e parts of system (kinetic energy, base compressive energy, operator) which directly influence the density results. The regression analysis strongly supports identifying the sources of variability in resulting densities. It is highl y recommended that a large scale testi ng procedure be carried out. This should involve the following: All machines be tested with the base compliance and kinetic energy devices Other soil types, i.e., A3 and A-1-b, be tested for influence of soil type Soil compacted at moisture contents wet, dry and possibly at optimum Technicians practice greater quality cont rol over the determination of moisture A low variance in moisture content duri ng the testing proced ure would allow the regression analysis to better determine the effect s of the kinetic and base compliance energies on resulting dry densities. The findings also support that control be ex ercised over the base system. A universal cement block and mounting configuration woul d help to minimize the variance of base compressive forces in the population and potentially negate the need for base compliance energy measurement in the future. As has been observed in this study the variance of base compliance energies is a large contributor to the dry density variance.

PAGE 89

89 Following future regression testing, it is r ecommended, a threshold value for the mean kinetic energy and variance of the kinetic energy for individual m achines be established. If a machine falls outside of this boundary, it may be n ecessary to adjust, service or replace parts on the machine.

PAGE 90

90 APPENDIX A OPERATION OF INSTRUMENTATION The following steps are critical to the functionalit y of the Portable Electronic Calibration Device. The Photo Electric Gate Assemble the guide post and base plate of the photo electric gate. The opening in the guide post is perpendicular to the lead ing edge of the base plate. Remove the thumb bolts from the C cha nnel containing the photo electric sensors. Place the C channel on the guide post such th at the line of sight of the photo electric sensors is parallel to the ba se plates leading edge. The orientation of the C channel should be such that the photo electrics are on the bottom side of the C channel. Refer to the C channel for determining top/botto m of channel for proper orientation. Thread the thumb screws through the guide post and into the back of the C channel and lightly tighten. Mount the photo electric assembly on the comp action machine such that the rammer guide rod and guide rod bushing fall within the radius cut in the base plate and the base plate is clear of all moving parts. Use quick connection clamps to fasten the ba se plate to the top plate of the compaction machine. Making certain the clamps used fo r mounting will not interfere with the moving parts of the compaction machine. Check the rotational alignm ent of the photo electric sensors to ensure that the guide post is in proper orientation such that the lenses of th e photo electric sensors will not be struck by the rammer rod. Turn on laptop and insert the USB cable for the measurement computing Data Acquisition system into any USB port on the laptop. Turn on power strip in instrumentation box. Allow sufficient time for data acquisition sy stem and instrumentation to warm up and stabilize approximately 20 minutes. In the mean time select the a ppropriate size impact pad for th e machine configuration to be tested. Place the impact pad on the compaction machines base and clamp in place using the mold clamp for the machine. o Note: Two impact pads have been pr ovided and have been adequate for the testing performed by the University of Florida. The purpose of the impact pad is to provide a safe surface for impact of the rammer and dampen the blow of the

PAGE 91

91 rammer such that the rammer is in at re st before being lift ed by the grabber. Should a particular machine display inst ances in which the rammer is picked up by the grabber during a bounce, following the previous impact changes to the impact pad must be made to eliminate this effect. However the impact pad should have sufficient rigidity such that there is little deflection or deformation when the rammer is at rest on the surface of the impact pad. Place the rammer face on the impact pad. From the top of the machine adjust the vertical alignment of the C channel. The line of sight of the bottom pair of sensors should be just slightly above th e top of the rammer rod so that the time of impact can be recorded The switching threshold can be observed by watching the lights on the photo electric amplifie rs change from green to red. Lift the rammer rod slightly to ensure that the amplif ier switches in all orientations of the rammer this will ensure that the impact energy can be measured for all impacts. From the desktop of the provided laptop open TracerDAQ Pro. With Strip Chart selected from the list of options click Run. On the file menu select load configurati on and choose named configuration select Kinetic Energy. Make certain the fiber optic cables of the in strumentation stay clea r of the rammer and grabber assembly during monitoring. Select the Play button on the top of the scr een. This will begin the acquisition of data from the photo electrics. Turn the compaction machine to the on position and record one complete revolution of the rammer. Once the rotation is complete sw itch the compaction machine off. Select the Stop to end the acquisition of data. Saving and Importing Data into Excel From the file menu of TracerDAQ Pro select Save As. Change the file type to .txt and save th e record of impacts in the directory of your choice. From the desktop open Microsoft Excel 2003.

PAGE 92

92 From the file menu, select Open. On th e File type drop down select All Files. Navigate to the file that was just recorded, select it and choose Open. The Import Text Wizard will be opened within excel, select delimitated and choose Next. Select Comma as a delimiter and click Next. Click Finished. Now The Data is ready for automated processing. Kinetic Energy Processing On the keyboard of the laptop press the c ontrol key and the k at the same time. Fill in the cell requiring the mass of the rammer after verifying its mass. Table generated now displays the kinetic energy of each imp act of the rammer. If a Boostrap of the kine tic energy is desired, press the control key and m The photo electric sensors can now be removed from the top of the compaction machinery. Compliance Measurement Setup and Testing Connect the coaxial cable from the load cell on the compaction base plate to the PCB ICP. Turn on the PCB ICP. Allow the load cell and ICP to warm up and stabilize. Place the compaction mold into the compaction machinery and clamp in place using the base plate clamp for the machine. Align the rammer such that when the machine is switched on it will impact the load cell squarely on the impact surface. The resting he ight of the rammer face should be on the same horizontal plane as the face of the load cell a nd the grabber for the machine in its lowest point of travel and gripping the rammer rod. When the machine is switched on the rammer wi ll begin to be lifted, rotate approximately 35-40 degrees and freefall. Attention to the lo cation to the impact s hould be checked. Upon verifying the rammer will impact squarely testing is ready to be performed.

PAGE 93

93 On the File menu of TracerDAQ Pro select Load Configuration, choose named configuration, select Base Compliance to load. Once the configuration loads, sele ct scan rate/ trigger settings and set the trigger ot begin acquisition at 0.003V. Press the play button to begin recording. Switch the machine to the on position moment arily and return it to the off position immediately after the rammer has been dropped, successive impacts of the rammer could damage the load cell or the rammer face. It is imperative that only one impact occur at a time and that it occur squarely on the load cell face. Realign the rammer for the next impact again this alignment will be approximately 35-45 degrees from the load cell face. Following the alignment steps outlined above. Switch the machine on for a single impact. Continue repeating rammer alignment and turning the machine on for a single impact on th e load cell face until at least 3 impacts have been recorded. o Note the macro to handle data proce ssing can handle a maximum of 5 impacts. Follow the same steps presented in the segm ent about saving and importing the data into excel. Compliance Energy Processing On the keyboard of the laptop press the control ke y and the r at the same time. This begins the automated data processing of the compressive energy. The table generated from this data displays the compressive energy from each of the impacts recorded. The variance of these values is di rectly related to accu racy of the rammer alignment by the user and thus not reported. The mean is reported a nd representative of the machine as long as proper care was taken by the operator to align the rammer properly before switching the device on. On the plot of the forces note the im pact number of any erroneous impacts and remove them from the summary for the machine.

PAGE 94

94 APPENDIX B DATA RESULTS Table B-1. T-180 Ramm er Rotation Summary St atistics Per Machine Machine Number Rammer Mass (lbs)MeanMedian Standard DeviationMeanMedian Standard Deviation 110.00110.59110.982.02 13.2013.290.48 210.00110.92110.980.82 13.2813.300.20 310.00112.96113.101.14 13.7813.810.28 49.98114.61114.460.4414.1514.110.11 510.02114.36114.460.4514.1414.170.11 610.00113.12113.370.7013.8113.870.17 79.98111.22111.240.4613.3213.330.11 89.99114.24114.180.6914.0714.060.17 910.00102.57100.424.6111.3710.881.03 1010.00114.61114.730.8014.1714.200.20 1110.01111.51111.760.6713.4313.490.16 1210.00114.79115.010.3714.2214.280.09 1310.00114.54114.730.8214.1514.200.20 1410.00113.54113.370.9013.9213.870.22 1510.00113.91114.180.5314.0114.070.13 169.99113.61113.640.3513.9213.920.09 179.99113.70113.640.4313.9313.920.10 1810.00109.82109.950.8113.0213.050.19 1910.02113.40113.500.8513.9013.920.21 2010.00114.50115.010.7414.1514.280.18 219.98112.95112.830.7013.7413.710.17 229.99109.44109.701.4412.9112.970.34 2310.01112.47112.560.5813.6613.680.14 2410.00115.54115.570.5914.4114.410.15 2510.00109.10108.941.5612.8512.810.37 2610.00106.40106.501.5512.2212.240.36 2710.01112.78113.643.2513.7513.950.77 2810.00114.88115.150.7814.2414.310.19 299.92116.06115.850.3914.4114.360.10 309.92112.41112.560.6413.5213.560.15 Velocity (in/sec)Kinetic Energy (ft-lb)

PAGE 95

95 Table B-2. T-180B Base Co mpliance Summary Per Machine Machine Number Average Integral of Force2 (lbs2-sec) Base Mounting Compressive energy (ft-lbs) 13686.23Direct to Concrete12.29 53533.44Direct to Concrete11.78 63672.25Direct to Concrete12.24 73176.89Aluminium Spacer10.59 93260.51Aluminium Spacer10.87 103381.72Aluminium Spacer11.27 113104.20Aluminium Spacer10.35 123691.42Aluminium Spacer12.30 154213.78Aluminium Spacer14.05 163393.28Steel Spacer11.31 183491.33Aluminium Spacer11.64 193774.38Steel Spacer12.58 203863.64Steel Spacer12.88 213433.66Aluminium Spacer11.45 233934.11Aluminium Spacer13.11 242566.72Aluminium Spacer8.56 263492.05Aluminium Spacer11.64 273334.38Aluminium Spacer11.11 303465.89Steel Spacer11.55 Note: Does not include plywood base mounting.

PAGE 96

96 Table B-4. Soil Pill Results and As sociated Energies 6% Moisture Test Number Percent Water Content (%) Dry Unit Weight (pcf) Kinetic Energy (ft-lbs) Compliance Energy (ft-lbs) KECE 15.7134.013.2012.29 25.7134.513.2012.29 35.9133.113.2012.29 45.6133.813.2012.29 55.8133.013.2012.29 65.5133.513.2012.29 75.7133.813.2012.29 85.7134.713.2012.29 95.8133.813.2012.29 105.6134.313.2012.29 15.8132.713.9211.31 25.9132.713.9211.31 35.8133.213.9211.31 45.7133.313.9211.31 55.9133.113.9211.31 66133.213.9211.31 76133.213.9211.31 85.9133.213.9211.31 95.8133.113.9211.31 106132.713.9211.31 16.2134.514.1512.58 25.6133.614.1512.58 35.9133.614.1512.58 46.3134.214.1512.58 56.1133.514.1512.58 66.1133.714.1512.58 75.8133.614.1512.58 86.1134.014.1512.58 95.8133.614.1512.58 106.3133.814.1512.58 6% Machine 1 Machine 14 Machine 13

PAGE 97

97 Table B-4. Continued. Test Number Percent Water Content (%) Dry Unit Weight (pcf) Kinetic Energy (ft-lbs) Compliance Energy (ft-lbs) KECE 15.4131.913.4310.87 25.6132.513.4310.87 35.6132.113.4310.87 45.7132.413.4310.87 55.8132.313.4310.87 65.5132.513.4310.87 75.7132.213.4310.87 85.6132.313.4310.87 95.7132.613.4310.87 105.7133.013.4310.87 16.2133.114.1510.35 25.713314.1510.35 36133.314.1510.35 46.1135.114.1510.35 56133.314.1510.35 6 7 8 9 10 15.9132.6713.7511.55 25.8133.313.7511.55 35.8132.0413.7511.55 45.8132.5413.7511.55 55.9132.513.7511.55 65.9132.313.7511.55 75.8133.313.7511.55 86132.413.7511.55 96132.913.7511.55 105.8133.513.7511.55 6% continued Machine 11 Machine 27 Machine 20 Note: Only five test results reported on Machine 20

PAGE 98

98 Table B-5. Soil Pill Results and Associated Energies 7.5% Moisture Test Number Percent Water Content (%) Dry Unit Weight (pcf) Kinetic Energy (ft-lbs) Compliance Energy (ft-lbs) KECE 17.0134.413.2012.29 26.6134.413.2012.29 36.4135.113.2012.29 46.8134.813.2012.29 56.7134.713.2012.29 67.3133.913.2012.29 77.4133.713.2012.29 87.4133.913.2012.29 97.2134.213.2012.29 107.2134.113.2012.29 17.5133.313.9211.31 27.5133.213.9211.31 37.7132.813.9211.31 47.6133.113.9211.31 57.7133.313.9211.31 67.6133.213.9211.31 77.6132.713.9211.31 87.6133.113.9211.31 97.5133.013.9211.31 107.8132.513.9211.31 17.4133.114.1512.58 27.5133.214.1512.58 37.5133.014.1512.58 47.4133.314.1512.58 57.5133.314.1512.58 67.5133.314.1512.58 77.5133.214.1512.58 87.5133.014.1512.58 97.5133.414.1512.58 107.4133.414.1512.58 7.5% Machine 1 Machine 14 Machine 13 Table B-5. Continued.

PAGE 99

99 Test Number Percent Water Content (%) Dry Unit Weight (pcf) Kinetic Energy (ft-lbs) Compliance Energy (ft-lbs) KECE 17.5133.613.4310.87 27.7133.413.4310.87 37.2134.113.4310.87 47.2134.013.4310.87 57.2134.113.4310.87 67.1134.513.4310.87 77.2134.013.4310.87 87.1134.513.4310.87 97.3133.913.4310.87 107.3134.313.4310.87 17.3133.4213.7511.55 27.4133.4213.7511.55 37.5133.1813.7511.55 47.5133.0513.7511.55 57.513313.7511.55 67.9132.413.7511.55 77.5132.613.7511.55 87.5132.613.7511.55 97.3133.113.7511.55 107.3133.513.7511.55 17.7131.712.228.56 27.6131.812.228.56 37.7131.812.228.56 47.7131.712.228.56 57.613212.228.56 68.2131.212.228.56 77.7131.812.228.56 87.4131.212.228.56 97.8131.712.228.56 107.7131.712.228.56 7.5% continued Machine 11 Machine 27 Machine 26

PAGE 100

100 Regression Analysis 6%: versus KE, CE The regression equation is = 122 + 0.846 + 0.007 KE + 0.489 CE 55 cases used, 5 cases contain missing values Predictor Coef SE Coef T P Constant 122.496 3.501 34.99 0.000 0.8462 0.5711 1.48 0.145 KE 0.0069 0.3192 0.02 0.983 CE 0.4892 0.1167 4.19 0.000 S = 0.602469 R-Sq = 33.5% R-Sq(adj) = 29.6% Analysis of Variance Source DF SS MS F P Regression 3 9.3455 3.1152 8.58 0.000 Residual Error 51 18.5114 0.3630 Total 54 27.8569 Source DF Seq SS 1 2.7298 KE 1 0.2351 CE 1 6.3806 Unusual Observations Obs Fit SE Fit Residual St Resid 8 5.70 134.700 133.422 0.166 1.278 2.21R 22 5.60 133.600 133.486 0.307 0.114 0.22 X 44 6.10 135.100 132.818 0.209 2.282 4.04R R denotes an observation with a large standardized residual. X denotes an observation whose X value gives it large leverage.

PAGE 101

101 Regression Analysis 6%: standardized versus, standardized, KE standardized, CE standardized The regression equation is stdzd = 0.071 + 0.235 stdzd + 0.004 KE stdzd + 0.572 CE stdzd 55 cases used, 5 cases contain missing values Predictor Coef SE Coef T P Constant -0.0715 0.1152 -0.62 0.538 stdzd 0.2349 0.1586 1.48 0.145 KE stdzd 0.0037 0.1733 0.02 0.983 CE stdzd 0.5721 0.1364 4.19 0.000 S = 0.838813 R-Sq = 33.5% R-Sq(adj) = 29.6% Analysis of Variance Source DF SS MS F P Regression 3 18.1160 6.0387 8.58 0.000 Residual Error 51 35.8840 0.7036 Total 54 54.0000 Source DF Seq SS stdzd 1 5.2917 KE stdzd 1 0.4557 CE stdzd 1 12.3686 Unusual Observations Obs stdzd stdzd Fit SE Fit Residual St Resid 8 -0.68 2.086 0.307 0.232 1.779 2.21R 22 -1.19 0.555 0.396 0.427 0.159 0.22 X 44 1.32 2.643 -0.534 0.291 3.177 4.04R R denotes an observation with a large standardized residual. X denotes an observation whose X value gives it large leverage.

PAGE 102

102 Regression Analysis7.5%: versus KE, CE The regression equation is = 144 2.05 + 0.202 KE + 0.187 CE Predictor Coef SE Coef T P Constant 143.652 1.875 76.63 0.000 -2.0540 0.2450 -8.38 0.000 KE 0.2019 0.1868 1.08 0.284 CE 0.1871 0.1001 1.87 0.067 S = 0.422038 R-Sq = 79.9% R-Sq(adj) = 78.8% Analysis of Variance Source DF SS MS F P Regression 3 39.704 13.235 74.30 0.000 Residual Error 56 9.975 0.178 Total 59 49.679 Source DF Seq SS 1 32.937 KE 1 6.145 CE 1 0.622 Unusual Observations Obs Fit SE Fit Residual St Resid 3 6.40 135.100 135.471 0.198 -0.371 -0.99 X 40 7.30 134.300 133.403 0.077 0.897 2.16R 56 8.20 131.200 130.878 0.189 0.322 0.85 X 58 7.40 131.200 132.521 0.137 -1.321 -3.31R R denotes an observation with a large standardized residual. X denotes an observation whose X value gives it large leverage.

PAGE 103

103 Regression Analysis 7.5%: standardized, versus standardized, KE standardized, CE standardized The regression equation is stdzd = 0.0012 0.679 stdzd + 0.152 KE stdzd + 0.293 CE stdzd Predictor Coef SE Coef T P Constant -0.00121 0.05938 -0.02 0.984 stdzd -0.67891 0.08097 -8.38 0.000 KE stdzd 0.1518 0.1405 1.08 0.284 CE stdzd 0.2931 0.1568 1.87 0.067 S = 0.459930 R-Sq = 79.9% R-Sq(adj) = 78.8% Analysis of Variance Source DF SS MS F P Regression 3 47.154 15.718 74.30 0.000 Residual Error 56 11.846 0.212 Total 59 59.000 Source DF Seq SS stdzd 1 39.117 KE stdzd 1 7.298 CE stdzd 1 0.739 Unusual Observations Obs stdzd stdzd Fit SE Fit Residual St Resid 3 -3.37 2.0564 2.4603 0.2157 -0.4039 -0.99 X 40 -0.41 1.1846 0.2068 0.0844 0.9777 2.16R 56 2.56 -2.1937 -2.5450 0.2062 0.3513 0.85 X 58 -0.08 -2.1937 -0.7542 0.1488 -1.4395 -3.31R R denotes an observation with a large standardized residual. X denotes an observation whose X value gives it large leverage.

PAGE 104

LIST OF REFERENCES American Association of State Highwa y Transportation Officials (AASHTO). AASHTO standard test methods T 87-86 Dry Preparation of Disturbed Soil and Soil Aggregate Samples for Test; T 99-01 Moisture Density Relations of Soils Using a 2.5 kg (5.5 lb) Rammer and a 305 mm (12 in.) Drop; T 180-01 Moisture Density Relations of Soils Using a 4.54 kg (10 lb) Rammer and a 457 mm (18 in.) Drop; T 88 Particle Size Analysis of Soils. American Society for Testi ng and Materials (ASTM). ASTM standard test methods D 216802a Calibration of Laboratory Mech anical Rammer Soil Compactors. Clinch, J. R. (2006). Soil compaction: a comp arison of standard Proctor and modified Proctor tests. Honors Thesis. Depa rtment of Civil and Coastal Engineering, University of Florida, Gainesville. Das, B. M. (2002). Principles of geotechnical engineering 5th ed. Pacific Grove: Brooks/Cole, 100. Dubose, L. A. (1952). Evaluating taylor marl clay for improved use in subgrades. Texas Engineering Experiment St ation Research Report 35 Texas A&M College, Texas. Halliday, D., Resnick, R., and Walker, J. ( 2000). Fundamentals of physics. 6th ed. New York: Wiley. Palacios, A. (1977). The theory and meas urement of energy tran sfer during standard penetration test sampling. Ph.D. Disserta tion. Department of Civil Engineering, University of Florida, Gainesville, Fl. Proctor, R.R. (1933). Description of field and labor atory methods. Engineering News Record September 7, 286. Ray, P. N., and Chapman, T. G. (1954). Bri tish standard compaction test for soils: study of some factors affecting test results. Geotechnique Vol. 4 (4), December, 169. Sebesta, S., and Liu, W. ( 2007). Improving calibration and qua lity control of Proctor-style laboratory compaction Transportation Research Board, July. Sherwood, P. T. (1970). The reproducibility of the results of soil classification and compaction tests. RRL LR 339 Road Research Laboratory, 27. Texas Department of Transpor tation (TexDOT). Laboratory Compaction Characteristics and Moisture Density Relationship of Base Materials. TexDOT-113-E

PAGE 105

BIOGRAPHICAL SKETCH Keith Beriswill was born in Lakeland, Florida. He lives with his parents and two siblings. Keith has been involved in his lo cal church and youth ministry and was active in other outside activities that have model his life. He was an active member of th e Boy Scouts of America achieving the highest rank of Eagle Scout in 2001. Keith graduated from George Jenkins High School in Lakeland, Florida in 2002 and was accepted to enter the University of Central Fl orida the following fall where he studied and graduated with his BSCE, majoring in civil engineering in the spri ng 2007. His desire for further education led him to the University of Florida to pursue a Master of E ngineering degree with a focus in geotechnical engineering. Upon graduation, Keith will be entering into industry in the field of geotechnical engineering, where he will be working in the area of dams and water resources.