15hE'hhEEE7h EhEEmhmhhEmhEI EEEEEmhhmhhhEI · Flow Determination for Plateau Shaped Plots Between...
138
RD-RI53 469 FIELD VALIDATION OF STATISTICALLV-BRSED ACCEPTANCE PLAN 1/2 FGOR BITUMINOUS Al.. (U) CLEMSON UNIV S C DEPT OF CIVIL ENGINEERING J L BURATI ET AL. MAY 84 UNLSSEDOTf A-84/i2-VOL-i DTFAI-i-C- 9857 F/G /5N mhhhRSIFED DOhE RhEEN 15hE'hhEEE7h EhEEmhmhhEmhEI EEEEEmhhmhhhEI
Transcript of 15hE'hhEEE7h EhEEmhmhhEmhEI EEEEEmhhmhhhEI · Flow Determination for Plateau Shaped Plots Between...
RD-RI53 469 FIELD VALIDATION OF STATISTICALLV-BRSED ACCEPTANCE PLAN 1/2FGOR BITUMINOUS Al.. (U) CLEMSON UNIV S C DEPT OF CIVILENGINEERING J L BURATI ET AL. MAY 84
UNLSSEDOTf A-84/i2-VOL-i DTFAI-i-C- 9857 F/G /5N
mhhhRSIFED DOhE RhEEN15hE'hhEEE7h
EhEEmhmhhEmhEIEEEEEmhhmhhhEI
JjS
MICROOPY RSOTO 132 CAR
NATION -BUEU1FSANAD316-
Ilk_._
11111.I*
SII I 111111---
~MICROCOPY RESOLUTION TEST CHART
NATIONAL BUREAU OF STANDARDS-19A3-A
9
',,. .
- . . 0 * .,.
;'1 ". . . . . ." :" ; . - - . -,- . ... . -. •. - - + - ...
- . + . -'. .
. . ° . . . . .- . . .' - . . . . . . .. ..• . . + ' . . • = . . - -. . . , . . . . . . .. 0 -
. o , + ° o " + . . °
w_ _ X_- -A -- -6
DOT/FAA/PM-84/12,1Field Validation of Statistically
Program Engineering Based Acceptance Plan forand Maintenance ServiceWashington, D.C. 20591 Bituminous Airport Pavements
Volume 1 -Correlation Analysis of MarshallProperties of Laboratory-Compacted Specimens
(0
James L. Burati, Jr.Gary D. Brantley
I Frederick W. Morgan
Department of Civil EngineeringClemson UniversityClemson, South Carolina 29631
May 1984
This Document is available to the publicthrough the National Technical Information
C:. OService, Springfield, Virginia 22161.,'.-,
DTICELECT 1 .
__ C MAY 9 198a
U.S. Department of Transportation
Federal Aviation Administrettlon
"45 ' . 1. .* * . - - * - * '-t
NOTICE
This document is disseminated under the sponsorship of theDepartment of Transportation in the interest of informationexchange. The United States Government assumes no liabilityfor its contents or use thereof.
Technical Report Documentation Page
1. Report No. 2. Government Accession No. 3. Recipient's Catalog No.
DOT/FAA/PM-84/12,
4. Ttle and Subttle 5. Report DateFIELD VALIDATION OF STATISTICALLY-BASED ACCEPTANCE MAY 1984PLAN FOR BITUMINOUS AIRPORT PAVEMENTS 6. Performing Organization CodeVOLUME 1 - CORRELATION ANALYSIS OF MARSHALLPROPERTIES OF LABORATORY-COMPACTED SPECIMENS 8. Performing Organization Report No.
7. Auhor, s)
* Burati, J.L., Brantley, G.D. and Morgan, F.W.
9. Performing Organizaton Name and Address 10. Work Unt No. (TRAIS)
- Department of Civil Engineering II. Contract or Grant No.
SClemson University DTFAO1-81-C-10057, Clemson, SC 29631 13. Ty.,e of Report and Period Covered
" 12. Sponsoring Agency Name and Address
U.S. Department of Transportation FINAL REPORTFederal Aviation AdministrationProgram Engineering and Maintenance Service 14. Sponsoring Agency Code
Washington, DC 2059115. Supplementary Notes
16. Abstruct The laboratory phase of a three phase research effort conducted to fieldvalidate a multiple price adjustment system for bituminous airport pavements using
the Marshall properties, stability, flow and air voids, is presented. The purpose ofthe laboratory phase was to identify whether correlations exist among the Marshallproperties within individual tests. To consider'the use of these properties in amultiple price adjustment system, it was first necessary to identify thesecorrelations.
The experimental design consisted of 4 different aggregate gradations and 6different asphalt contents for a total of 24 combinations. A total of 12 replicateswere tested for each combination for a total of 288 Marshall test specimens. Anumber of statistical analyses were conducted on the laboratory test results. Ananalysis of variance was conducted to determine whether time, i.e., order of testing,had an effect on the results. Correlation coefficients among the Mirshallproperties, i.e., stability with flow, stability with air voids and flow with airvoids, were calculated for each of the 24 combinations.
The results of the analysis indicate correlations that are consistentenough to violate an assumption of statistical independence among the properties.The results indicate that the effect of correlations among the properties should betaken into consideration in developing a multiple price adjustment approach based onthe Marshall properties.
17. Key Words 18. Distribution Statement
Acceptance Plans, Price Adjustment, This document is available to the U.S.Marshall Properties, Statistical Quality public through the National TechnicalControl, Correlation Information Service, Springfield,
VA 22161.
19. Security Class$(. (of this report) 20. Security Classif. (of this page) 21. No. of Pages 22. Price
Unclassified Unclassified 133
Form DOT F 1700.7 (8-72) Reproduction of completed page authorized
6~i:- ::-i--:: , .i:. i::: ::;:i: : :_: i: : ;;: i" . : :.i: ; i 2 : i:; .:: : i: .: .:
ACKNOWLEDGMENTS
This research was sponsored by the Federal AviationAdministration. The authors are indebted to personnel of the FederalAviation Administration Eastern Region for information concerningsuitable projects.
The research described in this report was carried out under thesponsorship of the Federal Aviation Administration. However, theanalyses of data and all conclusions and recommendations are theresponsibility of the authors and may not necessarily reflect theofficial views or policies of the Federal Aviation Administration. Thisreport does not constitute a standard, specification, or regulation.
ii
0
.. .. . . . . . . . . . . . .
". .-
TABLE OF CONTENTS
Page
ACKNOLDGMENTS........................... ................ .i
LIST OF TABLES.............................. ........... .V
LIST OF FIGURES ....................................... vi
PREFACE .............................................. x
CHAPTER
I. INTRODUCTION................................................ 1
Basis For Study............................................. 1Research Objective .................................... 2Potential Research Benefits ........................ 2Literature Review.......................................... 2
II. RESEARCH PROCEDURES......................................... 3
-Experimental Design ............ ................... 3Sample Size Determination ...................... 9
-Steps Taken to Reduce Testing Variability ................... 13*Pretesting Preparation.................... ........ 14
Testing Procedures ...................... .. ............. 17
III. DATA ANALYSIS........................................... 22
% Preparation............................................ 22Testing for Time Trend.................................. 22Optimum Asphalt Content Determination andMarshall Property Comparisons............................. 30
Correlation Analysis ................................. 30
*IV. DATA ANALYSIS RESULTS ................ .............. . .. 36
*Marshall Properties-Asphalt Content Relationship ............. 369 -Optimum Asphalt Content Determination ....................... 51
Correlation Results .................................... 51
V. SUMMARY, CONCLUSIONS AND RECOMMENDATIONS.................. 71
Summary .............................................. 71S Conclusions..................................................... 72
Recommendation............................................. 73
BIBLIOGRAPHY ........................................................ 74
AP E DIE .. . . . . . . .. . . . . . . . .. . . . . . . . .7
A. Example Maximum Theoretical Specific GravityDetermination: FAA Lower Gradation 5.0% Asphalt Content ..... 76
B. Marshall Properties and Related Laboratory Test Data ......00.. 77C. Marshall Property Scatter Plots with Outliers from
Replicates Two and Three dentified........................ 90D. Temperature Recorded During Marshall Briquet Mixing
Process in the Laboratory ............................... illE. Summary of Marshall Stability Results ................ *....... 118F. Summary of Marshall Flow Results ........................ 00... 119G. Summary of Air Voids Results. .. .. .. ......... .. . .. . ....... 120
iv
LIST OF TABLES
Table Page
I. Summary of Aggregate Gradation Limits
Considered ......................................... 4
II. Summary of Aggregate Gradations
Chosen for Research Effort ......................... 10
III. Specific Gravity and Absorption Values
for Aggregates Used ................................ 19
IV. Marshall BriquetMaximum Theoretical Specific Gravities ............. 20
V. Correlations Results Using All 12 Replicates (12 Reps)
and Using Only Replicates 2-12 (11 Reps) ........... 23
VI. Correlations Results Using
Replicates 4-12 .................................... 31
VII. Optimum Asphalt Contents ir Accordance
With the ERLPM for Gradations Used ................. 52
catio
IAvailabilitY CodeB--- Avail and/or
Di\st special
k:V
LIST OF FIGURES
Figure Page
1. Aggregate Gradation Limits from the PennsylvaniaSpecifications for loads >60,000 lbs. or tire pressures > 100 psi.(Center line is midpoint of specification limits.) ...... 5
2. Aggregate Gradation Limits from the Virginia
Specifications for loads>60,000 lbs. or tire pressure > 100 psi.(Center line is midpoint of specification limits.) ...... 6
3. Aggregate Gradation Limits from the New YorkSpecifications for loads >
60,000 lbs. or tire pressure > 100 psi.(Center line is midpoint of specification limits.) ...... 7
4. Regional Gradation Limits for Loads >60,000 lbs. or tire pressure > 100 psi.(Center line is midpoint of specification limits.) ...... 8
5. Summary of Aggregate GradationsChosen for Research Effort .............................. 11
6. Relationship Between Power for Testing Ho: p=Ovs. Ha: pO0 and the Number of Replicatesfor a 0.1 Significance Level ............................. 12
7. Bituminous Concrete Laboratory Test Data Sheet ............. 15
• . 8. Flow Determination for
Plateau Shaped Plots Between Stability and Flow ......... 16
- .9. AC-20 Asphalt Cement Viscosity Curve ....................... 21
10. Stability vs. Flow Correlation Coefficient PlotsUsing All 12 Replicates ................................. 24
11. Stability vs. Air Voids Correlation Coefficient PlotsUsing All 12 Replicates ................................. 25
- 12. Flow vs. Air Voids Correlation Coefficient Plots. Using All 12 Replicates ................................. 26
13. Stability vs. Flow Correlation Coefficient Plots" Using Replicates 2-12 ................................... 27
7
vi
** . . . ..
List of Figures (Cont'd.)
Figure Page
14. Stability vs. Air Voids Correlation Coefficient PlotsUsing Replicates 2-12 ................................... 28
15. Flow vs. Air Voids Correlation Coefficient PlotsUsing Replicates 2-12 ................................... 29
16. Stability vs. Flow Correlation Coefficient Plots
Using Replicates 4-12 ................................... 32
17. Stability vs. Air Voids Correlation Coefficient PlotsUsing Replicates 4-12 .................................... 33
18. Flow vs. Air Voids Correlation Coefficient PlotsUsing Replicates 4-12 .................................... 34
19. Marshall Property Plots from the Asphalt Institute's
MS-2 Manual (10) .......................................... 37
* 20. Stability vs. Asphalt Content:FAA Upper Gradation ...................................... 38
21. Stability vs. Asphalt Content:FAA Midpoint Gradation ................................... 39
22. Stability vs. Asphalt Content:
FAA Lower Gradation ...................................... 40
23. Stability vs. Asphalt Content:
JMF Gradation ............................................. 41
24. Flow vs. Asphalt Content:FAA Upper Gradation ....................................... 43
25. Flow vs. Asphalt Content:FAA Midpoint Gradation .................................... 44
26. Flow vs. Asphalt Content:FAA Lower Gradation ....................................... 45
27. Flow vs. Asphalt Content:JMF Gradation ............................................. 46
28. Air Voids vs. Asphalt Content:FAA Upper Gradation ....................................... 47
29. Air Voids vs. Asphalt Content:FAA Midpoint Gradation .................................... 48
vii
0
| . .-
List of Figures (Cont'd.)
*Figure Page
* 30. Air Voids vs. Asphalt Content:FAA Lower Gradation ..................................... 49
31. Air Voids vs. Asphalt Content:JMF Gradation ........................................... 50
32. Stability vs. Air Voids Correlation Coefficient PlotsUsing Replicates 4-12with Gradations Adjusted for Optimum Asphalt Content.... 54
33. Stility vs. Air Voids Correlation Coefficient PlotsUsing Replicates 4-12 ................................... 55
34. Stability vs. Flow Correlation Coefficient PlotsUsing Replicates 4-12with Gradations Adjusted for Optimum Asphalt Content .... 59
35. Stability vs. Flow Correlation Coefficient PlotsUsing Replicates 4-12 ................................... 60
36. Stability vs. Flow Scatter Plot for theFAA Upper Gradation at 5.5% Asphalt Content ............. 61
37. Stability vs. Flow Correlation Coefficient PlotsUsing Replicates 4-12 Less Outlierwith Gradations Adjusted for Optimum Asphalt Content .... 62
38. Stability vs. Flow Correlation Coefficient PlotsUsing Replicates 4-12 Less Outlier ...................... 63
39. Stability vs. Air Voids Correlation Coefficient PlotsUsing Replicates 4-12 Less Outlieru.ith Gradations Adjusted for Optimum Asphalt Content .... 64
40. Stability vs. Air Voids Correlation Coefficient PlotsUsing Replicates 4-12 Less Outlier ...................... 65
41. Flow vs. Air Voids Correlation Coefficient PlotsUsing Replicates 4-12 Less Outlierwith Gradations Adjusted for Optimum Asphalt Content .... 66
42. Flow vs. Air Voids Correlation Coefficient PlotsUsing Replicates 4-12 Less Outlier ...................... 67
43. Flow vs. Air Voids Correlation Coefficient PlotsUsing Replicates 4-12with Gradations Adjusted for Optimum Asphalt Content.... 69
viii
List of Figures (Cont'd.)
Figure Page
44. Flow vs. Air Voids Correlation Coefficient Plots- Using Replicates 4-12 ........................................ 70
ix
7 7
PREFACE
This report presents the findings of a research project entitled"Field Validation of Statistically Based Acceptance Plan for BituminousAirport Pavements", Report No. DOT/FAA/PM-84/12, that was conducted toinvestigate the use of Marshall properties for acceptance purposes. Theresults of the research effort are presented in the series of reportslisted below:
Burati, J.L., Brantley, G.D. and Morgan, F.W., "CorrelationAnalysis of Marshall Properties of Laboratory Compacted Specimens,Final Report, Volume 1, Federal Aviation Administration, May,1984.
Burati, J.L., Seward, J.D. and Busching, H.W., "StatisticalAnalysis of Marshall Properties of Plant Produced BituminousMaterials," Final Report, Volume 2, Federal AviationAdministration, May, 1984.
* Burati, J.L. and Seward, J.D., "Statistical Analysis of Three
Methods for Determining Maximum Specific Gravity of BituminousConcrete Mixtures," Final Report, Volume 3, Federal AviationAdministration, May, 1984.
Nnaji, S., Burati, J.L. and Tarakji, M.G., "Computer Simulation ofMultiple Acceptance Criteria," Final Report, Volume 4, FederalAviation Administration, August, 1984.
Burati, J.L., Busching, H.W. and Nnaji, S., "Field Validation ofStatistically Based Acceptance Plan for Bituminous AirportPavements - Summary of Validation Studies," Final Report, Volume5, Federal Aviation Administration, September, 1984.
The application of multiple price adjustments is significantly moreinvolved than the case when only one property, e.g., density, isconsidered. Since the Marshall properties (i.e., stability, flow andair voids) are physically related, they can be expected to bestatistically correlated. If this is truly the case, then it may not besufficient to treat each of the three properties individually. It isnecessary to determine whether correlations exist among theseproperties, and whether such correlations should be considered whendeveloping acceptance plans.
0 The objectives of the research described in the reports listedabove include:
1. Review current methods of determining maximum specific gravityfor use in air voids calculations for possible incorporation intothe FAA Eastern Region P-401 specification,
SX
x
[.6
2. Investigate the use of price adjustments when more than onecharacteristic is being used for acceptance purposes and recommendto the FAA potential procedures for dealing with multiple priceadjustments,
3. Develop the procedures necessary to evaluate the performance ofmultiple properties acceptance plans,
4. Implement proposed Marshall properties acceptance plans ondemonstration projects under actual field conditions, and
5. Attempt to correlate values of asphalt content and aggregategradation with those from Marshall tests to determine whether ornot correlations exist among these properties.
This report, Volume 1, presents the findings of a laboratoryanalysis to determine whether correlations exist among the Marshallproperties. How correlations can be considered in the development ofprice adjustment systems is presented in the subsequent volumes.
xi
0
* . . . . . . ** * * *
CHAPTER I
INTRODUCTION
The Federal Aviation Administration (FAA) incorporated statisticalconcepts into its bituminous surface course (Item P-401) acceptance planin 1978 by using the mean and the range of mat density tests todetermine acceptability. In 1980, as a result of an FAA-sponsoredresearch effort (1), the mean and standard deviation for mat densitytests were incorporated into a statistically based price adjustmentsystem. The final report of the research effort indicated that therewere not sufficient data available to warrant the implementation of amultiple characteristic acceptance plan that included price adjustmentsfor Marshall properties as well as mat density.
The final research report (1) recommended further study todetermine the feasibility of applying multiple price adjustments usingthe Marshall properties. These properties are physically related (i.e.,determined from a single test) and therefore can be expected to bestatistically correlated. Because of this, it is necessary to identify
0any correlatons existing among these properties before attempting toapply multiple price adjustment factors. If a high correlation doesexist among 2 or more of the properties, then it can be argued that thecorrelated properties are, in fact, measuring the same characteristic ofthe mix and the price adjustment system should incorporate only 1 ofthese properties to avoid penalizing the contractor twice for the samedeficiency.
Basis For Study
The current research is a direct result of the recommendations made
in the final report (1) of the initial research effort. Therecommendations proposed a 3-phase research project. The ultimate goalof the recommendations was to obtain a multiple price adjustment systemusing the Marshall properties for the FAA Eastern Region to apply onbituminous paving projects. The 3 phases of research proposed were thelaboratory phase, the field phase, and the computer simulation phase.This report presents the results of the laboratory phase of the researchthat developed as a result of that proposal
To establish multiple price adjustment factors, it is necessary todetermine whether any correlations exist among the Marshall propertiesas a result of their physical relationship. The purpose of thelaboratory phase of the research is to identify whether suchcorrelations exist and to estimate their magnitude.
0
Research Objective
The objective of the research is to determine whether significantcorrelations exist among the Marshall properties of stability, flow, andair voids. The specific correlations to be considered are stabilitywith flow, stability with air voids, and flow with air voids.
Potential Research Benefits
The results of this research will be used in the investigation ofthe feasibility of establishing a multiple price adjustment system usingMarshall stability, flow, and air voids for the Federal AviationAdministration's Eastern Region. If correlations are found to existamong the properties, multiple price adjustment factors can beestablished to account for the properties that measure the samecharacteristic of the mixture. This will prevent the contractor frombeing penalized twice for the same deficiency. If correlations exist,either some of the tests required by the FAA Eastern Region couldpossibly be eliminated, or it may be necessary to develop a method ofincluding the correlatons in the acceptance decision.
Literature Review
Although the Marshall test has been used for more than 40 years,with many research projects on the Marshall properties, no previousresearch relating to within-test correlations among the properties wasfound. The correlations existing between each property and ashpaltcontent are well established, but it appears little is known of thecorrelations existing between each Marshall property for a given asphaltcontent and gradation.
02
S7
- 4
CHAPTER II
RESEARCH PROCEDURES
The research effort was divided into 3 principal areas. Theinitial step was the experimental design. This included defining therange of asphalt cement contents and aggregate gradations to use as wellas determining the number of replicates to provide the desired level ofconfidence. The next step was to procure and perform related tests onthe required materials. Once this was accomplished, the mixing andtesting of Marshall briquets could begin. The Marshall test method wasused rather than other methods of testing because the Marshallparameters (stability, flow, air voids) are used for acceptance purposesby the Federal Aviation Administration.
Experimental Design
The research was designed to identify correlations among theMarshall properties. The goal of the experiment was to determinewhether the levels of correlation among the Marshall properties vary
6with asphalt content and aggregate gradation. Other variables were held
as constant as possible to prevent them from influencing the Marshalltest results. Such variables included aggregate quality, asphalt type,temperature (mixing, compaction, and testing), equipment, operators,etc. To determine the effects of asphalt content and aggregategradation on the correlation analysis, a range of asphalt contents andaggregate gradations was required. It was desired to cover as broad arange of asphalt contents and aggregate gradations as possible.
To determine the appropriate number of asphalt contents andaggregate gradations to use, the standard FAA Eastern Regionspecifications and the various state versions of the specifications foraggregate gradation and asphalt content for aircraft loads greater than60,000 pounds and a maximum 3/4-inch stone were compared. Each of thespecificatons required the bitumen content to be between 5.0% and 7.5%of the total mix weight. As defined in the Eastern Region LaboratoryProcedures Manual (ERLPM) (9), the laboratory procedures for developingthe job mix formula for a paving project require the aggregategradations to be tested at 0.5% asphalt content increments. Since itwas desired to span as broad a range of asphalt contents as possible,based on these specifications and the ERLPM, 6 asphalt contents (5.0%,5.5%, 6.0%, 6.5%, 7.0%, 7.5%) were selected for testing.
The aggregate gradation limits specified by Pennsylvania, Virginia,New York, and the Eastern Region standard specification are presented inTable I and Figures 1-4, respectively. The center lines in the figuresare the midpoints of each specification band. After comparing thespecifications, several groupings could be made. The specificationlimits for the Virginia, New York, and Eastern Region were similar and,therefore, considered the same. It was also concluded that the midpoint
. o- • . .. o° . .. , - • . ° . • ° . , • . . - ° ,
Table I. Summary of Aggregate Gradation Limits Considered
Percent Passing
SieveSize Pennsylviania Virginia New York Regional 2A
3/4 in. 100 100 100 1001/2 in. 77-96 82-96 82-96 82-963/8 in. 68-89 75-89 - 75-891/4 in. - - 65-79 -
No.4 48-73 59-73 - 59-731/8 in. - - 51-65 -
No.8 34-60 46-60 - 46-60No.16 23-48 34-48 - 34-48No.20 -- 29-43 -
No.30 16-38 24-38 - 24-38No.40 -- 20-33 -
*No.50 10-27 15-27 - 15-27No.80 -- 10-20 -
No.100 6-18 8-18 - 8-18No.200 3-6 3-6 3-6 3-6
4
B90
-~70E-4
30
°N
10 3/4I/2O38.. ...6.0.S 10 0
" 90 I I
I t
K40
80 1/ 3/------_,--0550 100.-.
tion fo lod ? 6000 's ortr rsue 10 s. (etrlnis midoi t ofseifcto-lmt.
100
90
80
70
SO.-
3 0-
02
.0.-
p °40
P .0.
3/4 1/2 3/8 4 8 16 30 5 100 200
SIEVE SIZE
Figure 2. Aggregate Gradation Limits from the Virginia Specifications.* for loads > 60,000 lbs. or tire pressures 2: 100 psi. (Center line is
midpoint of specification limits.)
6
' .. t , I. .
100-
90.
a0
703/4 1/ 14 /820408020
9~ ~ SIN SZ
60
Fiue35grgt0rdainLmt-rmteNe okSeiia~n* fo lods 60,00 bs. r tre ressres 10 ps. (ente lie i
midpint f spcifiatio limtsQ
7t
30.....................................-f.
07.SO
60
0
I r\3O0 1
I~~ N ::
20-
0
10
3/4 1/2 3/8 4 8 16 30 50 100 200
* SIEVE SIZE
Figure 4. Regional Gradation Limits for loads k 60,000 lbs. or tire0pressures 100 psi. (Center line is midpoint of specification limits.)
8
I % .. .... . . . . . .
60 .- - - - - - - - -. '
and upper specification limit for the Pennsylvania specifications andthe lower and upper specificaton limits for the Regional, respectively,were close enough to be considered the same. After making thesegroupings, 4 distinct aggregate gradations were established that spannedthe entire allowable limits established by the Virginia, Pennsylvania,New York, and Eastern Region lower, midpoint, and upper specificationlimits. It was concluded that the lower Pennsylvania specification wasnot very realistic because of its extremely low gradation band incomparison with the other specifications and it was therefore deleted.
Based on experience obtained from mixing and testing severalpractice Marshall briquets, it was decided that approximately 20 to 25briquets could be mixed on 1 day and tested the next without rushing theprocess. By mixing and testing all combinations of asphalt content andaggregate gradation at the same time, the effects due to time should berestricted to variatons between each replicate and not within eachreplicate. With this in mind, and the fact that 6 asphalt contents weredesired, 4 gradations were selected. This would provide a total of 24combinatiuns for each replicate. As noted earlier, the aggregatespecification limits could be spanned with only 3 gradations with thedeletion of the Pennsylvania lower specification limits. The job mix
4formula (JMF) gradation used for a Rochester-Monroe County airportpaving project in New York was chosen as the fourth gradation. TheRochester-Monroe project was 1 of 5 airport paving projects studied inthe field phase of the 3-phase FAA research project. The 4 gradationsare sumnarized in Table II and plotted in Figure 5. The Upper,Midpoint, Lower, and JMF bands in the plot represent the FAA Upper, FAAMidpoint, FAA Lower, and the Rochester job mix formula gradation bands,respectively.
After the aggregate gradations and asphalt contents wereestablished, the number of replicates for each combination of gradationand asphalt content required to produce the desired level of confidencein the results was determined.
Sample Size Determination
Since it is not possible to determine exactly the correlationcoefficient for each asphalt content/aggregate combination, it wasdesired to determine whether any s:ignificant correlations existed amongthe properties. To determine whether a correlation was present, powercurves depicting the probability of detecting a certain magnitude ofcorrelation for a given cimple size, or number of replicates, wereobtained (11). The power of the test is the probability of rejectingthe null hypothesis. If the null hypothesis is not true, then thechance of rejecting the hypothesis should be as large as possible, i.e.,a large value is desired for the power. A plot showing the relatonshipbetween power of the test at the 0.1 significance level and the numberof replicates is presented in Figure 6.
To determine the relationship between sample size and the power of
4 the test, the null hypothesis was that Ihe true correlation was 0 (Ho:0=0), and the alternate hypothesis was for a correlation riot equal to 0(Ha: P 0). The power was determined for the cases when the true
1 2 .
p,... --.
Table II. Summary of Aggregate Gradations Chosen for Research Effort
Percent Passing
Sieve FAA Regional FAA Regional FAA Regional Rochester-- Size Lower Midpoint Upper Monroe
Job-Mix
" 3/4 in. 100 100 100 100
1/2 in. 82 89 96 98.6
3/8 in. 75 82 89 84.6
No.4 59 66 73 66.5
No.8 46 53 60 55
No.16 34 41 48 42
No.30 24 31 38 31
No.50 15 21 27 20
No.100 8 13 18 8.5
No.200 3 4.5 6 3.8
[.
10
: , i . i i- . . . *... . . . *. .*
100
90-
S 70 N
50
Fiur 5 S.rm~y f gc~cat Gadizon Cosn orReeachXfr
20
.00.d
0.0 -
LEGN/ SAFEto / /2
Fiur 6 RlaioshpBewen owr orTetngth NllHyotessHo p=O, vesu /h lent yohss H:p n h ubroRelcae fo a. 0. SinfcneLvl
LI ,/~' 12
correlaton coefttcte,, , was equal to 0.1, 0.2 ........ s, 0.9. Thetrue correlation cU, fir mit is represented on the horizontal axis inFigure 6, and the probahilit.v of detecting a correlation, i.e.,rejecting the null hypothes.is that there is not a correlation, is alongthe vertical axis. Ea:h curve represents a different number ofreplicates. As would be expected, a i the number of replicatesincreases, the probahilitiv of detr -ting a correlation also increases.
Consi deri no t hi s, a min inimum of 9 r-r< I icates were req i red toprovide_, the levl i'.;!sificarnr icc - : :l. The desirable ],v I ofsignific;tnce was; ie det ection of a vet y high correlation wi-ih aprobability near 1 .0 and detection of a moderate correlation of 0.5 to0.6 with a prohabi ity about 0.5. If there im; a true correlationcoefficient of 0.8 or greater amoci the Marana Hr iperties, then, for 9replicates, the o is aiprxinatelv t 904 chanre of rejecting the nullhypothesis that no 7-,r elatii)t exists and savyrin n here is sooecorrelation among the properties. Ior a true ci relation coefficient of0.6, there is approximatlv a 55% chance of rejecting the nullhypothesis. After practice samples were mixed and tested, a maximum of12 replicates were possible, given the time and resources available.With 12 replicates, there is approximately a 96% chance of rejecting thehypothesis of no correlation if a true correlation coefficient of 0.8 orgreater exists among the properties, and approximatm!y a 71% chance of
I rejecting the hypothesis of no correlaion if a true correlationcoefficient of O.n exists among the properties.
.'t eps Taken to Reduce Testin , Variability
Factors inf[uenring the variability ot the test results were eithereliminated or reduced as much as possible. In addition to the factorsdescribed below, others were listed in the experimental design sectionof this chapter. To reduce the learning curve effect on the laboratorytechnicians, 96 briquets, or 4 replicates, were mixed and tested astrial runs. The same 3 technicians were used thicughout the mixing andtesting operations to reduce variations among ,perators. One wasresponsible for sieving and weighing the aggr,-_gate. The secondlaboratory te,_hnician was responsible for assisting the third during themixing and compaction processes and was responsibie for clean-up. Thethird techuirian supervised all laboratory efforts and performed allmixing, testinig and briquet weighing.
* - A random select i on[ pre:, 'as used to elimrinate bias in the orderof mixing and testing the briquets. The testing order was the same asthe mixing order for er-ch 1- 1 i te. This order wa- di,termi ned by
drawing a sl ' ,f pa- ' -"' .roita , er rontain- 1 0 '4 ; i-nt1ial slips;I for ea: 0 cornbi n.it i.or of t asphalt contents and 4 ,r;tda 1 tions. Thisprocess wis repe,t td for eac.h replicate.
hiue to t!l- 1;.-, rumlber of briquiets in ech rep! irat-e, eachreplicate wa mix'', t One day and tested the next. By mixing andtesti.n'4 iii thi 1- . variations in the results due to time should beconf ned to , between t112 repl iCates, it is very unlikelythat tiniu woO I I.''e' v.iri-t ionus within the replicate from mixing on one
• day and t n; g he n,.Yt. Even if time did have an effect, all of the
0..;
-A
* replicates should be affected in the same f a.-hion.
The same equipment was used throughout the research effort. TheMarshall testing machine used (Model 850 manufactured by the PineInstrument Company) automatically records the stability and flow. Thetesting machine was checked for proper calibration at the start of eachweek. A Pine Instrument Company Model PMC4 automatic compactor was aLsoused. A dial thermometer was checked for calibration before each mixingday.
Due to the large number of briquets mixed in a single day (24) and-. the limited supply of Marshall molds (15), 12 briquets were mixed in the
morning and 12 in the afternoon. This was required to enable 9 of themolds to cool and be extruded, cleaned, and reheated for the second setof 12 briquets in the afternoon.
A data sheet was developed to record pertinent information. Thisdata sheet is presented in Figure 7. The order for aggregate gradationand percent asphalt content was determined by a random process as
* described above. The asphalt weight was shown to aid the laboratorytechnicians when weighing the asphalt cement into the aggregate mixture.To analyze the effect of temperature on the test results and ensure that
4| the various temperatures were within the specification limits, theasphalt, mixing, and compaction temperatures were recorded. The testingtemperature was not recorded due to the small variation allowed by thespecification limits (1400+/- 1.80F). The asphalt temperature was thetemperature of the asphalt cement before adding it to the aggregate.The mixing temperature was taken as the temperature of the aggregate andasphalt cement combination immediately prior to mixing, and thecompaction temperature was the temperature of the mixture immediatelybefore compaction. The thickness of the Marshall briquets was measuredto the nearest 1/32-inch and then converted to decimal equivalent. Thebriquets were weighed the day of testing. The measured stability andflow were recorded from the Marshall test plots. On some plots, theflow formed a plateau at the maximum staility. For these plots, theflow value was determined as illustrated in Figure 8.
Pretesting Preparation
After the experimental design was finalized, the necessarymaterials were acquired and tests were performed to determine specificaggregate properties. The aggregate and asphalt cement used in theresearch were materials commonly used in the FAA Eastern Region. Theasphalt cement was obtained from West Bank Oil, Inc., Pennsauken, NewJersey. The asphalt cement was AC-20 grade with a penetration value of79 at 770F. The absolute viscosity was 1,971 poises at 140'F and thekinematic viscosity was 376 centistokes at 2750F. The aggregateconsisted of limestone obtained from the General Crushed Stone Companyquarry at Honeoye Falls, New York, and a natural sand from BaughamMaterials in West Bloomfield, New York.
A gradation analysis was conducted in accordance with ASTM C-136 onthe limestone and natural sand. It was found that adequate quantitiesof limestone were available to meet the gradations to be used in the
14
"4 .' ? - i . ' ' '. - " . . V , , : - - . . . . - .--
FAA BITUMINOUS CONCRETE TEST DATA
REPLICATE
DATE OF MIXDATE OF TEST:
-JAspnaic TEMPERATURE 'F Thick- WGT --- GRAMkS Meas. FlowSP. Grada- Asphalt Contentpa MxnCo. ness In fIn Sat.Su Stab. Units of
it tion IWgt. PercentAspatMxn Co. inches k4ir W4ater Drv lbs 10oojn
I
2
3
4
5
6
10
11
1 ~ ~~12__ ____
13
14
15
16
17
18 ___ ___
19
* 20
21
23
0 Figure 7. Bituminous Concrete Laboratory Test Data Sheet
15
Flow at Right-Hand Sideof Plateau
0110
Flow
Figure 8. Flow Determination for Plateau Shaped Plots Between Stabilityand Flow
16
experiment. It was determined from the gradation analysis that therewas sufficient natural sand to supply approximately 15% of the aggregateweight. Due to limitations on the the availability of natural sand in
icertain sieve ranges, the actual amount of natural sand used was 14.8%by weight of the total aggregate in the mix.
Natural sand was used because the FAA Eastern Region permits thecontractor to use natural sand to facilitate field compaction of thedense-graded mix, and this is typical of current practices within theEastern Region. The primary purpose of the natural sand is to increasethe workability of the mix as it is being placed.
Specific gravity and absorption tests were performed after theaggregate was sieved. The tests were run in accordance with ASTM C-128for fine aggregate and ASTM C-127 for coarse aggregate. The results areshown in Table III along with the specific gravity and absorption valuesprovided by the material suppliers' laboratories. The values for bothtests were slightly different. Confidence in the results of theresearch technicians led to the use of these values rather than thesupplier's. These values were used in determining the maximumtheoretical specific gravity for each of the 24 Mar~hall briquets which
* are presented in Table IV. The procedure for determining the maximumtheoretical specific gravity was in accordance with the FAA EasternRegion Laboratory Procedures Manual (ERLPM). An example calculation ispresented in Appendix A.
Testing Procedures
The laboratory procedures followed in preparing, mixing, weighing,and testing of the Marshall briquets were in accordance with theprocedures outlined in Section 2 of the ERLPM with only one exception.A piece of filter paper was placed on top of the asphaltic concretemixture immediately prior to compacting the first side of the briquet toprevent material from adhering to the compaction hammer. The laboratoryprocedures for the FAA Eastern Region are the same as those outlined indeveloping a job mix formula in The Asphalt Institute's Manual SeriesNo. 2 (MS-2) Mix Design Methods for Asphalt Concrete (10) publicationwith the following exceptions:
1. The FAA Eastern Region specifies the compaction temperatures as2500+/- 50F whereas MS-2 specifies the compaction temperature asthe temperature that produces a kinematic viscosity of 280 +/- 30centistokes.
2. The FAA Eastern Region permits reheating the aggregate andasphalt mixture to 2500+/- 50F for compaction if the container iscovered to prevent oxidation and the temperature is not below200*F. The MS-2 manual does not allow reheating.
I
.. 17
I
*,*. , * . - .- *I. : . . .. . . - . .. . -, .. • ... .. .- " .,. - . . . ,.. "" . . ' -, i.
0T . . . .. . .
3. The FAA Eastern Region specifies the temperature of thebreaking head to be maintained from 1000F to 140°F whereas MS-2specifies 70°F to 1O0°F.
4. The method for determining the optimum asphalt content differs,and is explained in a later section of this report.
Before the mixing and testing of briquets could begin, severalpreliminary decisions had to be made. A weight of 1,155.7 grams wasneeded, based on practice replicates, to obtain the required 2.5-inchthickness for the Marshall briquets. From the asphalt viscosity curvein Figure 9, the mixing temperature at 170 +/- 20 centistokes wasdetermined to be 297°F to 307°F. After these preliminary decisions wereresolved, mixing and testing began and continued for approximately 8weeks.
18
Table III. Specific Gravity and Absorption Values for Aggregates Used
Aggregate Limestone Limestone Natural(coarse) (fine) Sand
Supplier's Lab
Apparent SpecificGravity 2.715 2.646 2.660
Percent
Absorption 0.69%
Research Lab
Apparent SpecificGravity 2.700 2.684 2.660
PercentAbsorption 0.73% 1.40% 1.30%
41.
. 19
0
0W 6
Table IV. Marshall Briquet Maximum Theoretical Specific Gravities
Gradation Asphalt Max. TheoreticalContent % Specific Gravity
JMF 5.0 2.4835.5 2.4646.0 2.4466.5 2.4287.0 2.4107.5 2.393
*FAAL 5.0 2.4845.5 2.4656.0 2.4476.5 2.4297.0 2.4117.5 2.394
FAAM 5.0 2.4835.5 2.4646.0 2.4466.5 2.4287.0 2.4107.5 2.393
FAAU 5.0 2.4825.5 2.4646.0 2.4456.5 2.427
*7.0 2.4097.5 2.392
20
100000-
C
E- 1000-
00-
120 170 220 2"70 320
TEMPERATURE (,F)
Figure 9. AC-20 Asphalt Cement Viscosity Curve
:: 21
.. ,.
II
..- .. -: -:.. .- v '. .- ' 0.. __._-_,, _____' ____. ._. ._..'-_-_'.---_- -_"____-',-. ______ ._-_______... ..._"", "I ' ," . ' : , ',I-, - - . - . I -, - : . ' . - - . . . . : . - : , : - - " : , : . " : ," . . ...- . . " " .
CHAPTER III
DATA ANALYSIS
After the 12 replicates of Marshall briquets had been mixed andtested, and the results stored in the computer, the analysis of the datacould begin. Various programs within the Statistical Analysis System(SAS), a commercially available package of statistical programs, wereused. An analysis of variance (ANOVA) and Duncan's Multiple Range Testwere performed first to determine whether time, i.e., order of testing,had an effect on the results. Then, the Marshall properties wereplotted against asphalt content, and a correlation analysis among theMarshall properties was performed.
Preparation
The computer was used for all data analysis, but before anyanalysis could be performed, the test results had to be stored andconverted to usable data. All test data, including the asphalt, mixing,
4and compaction temperatures, were originally recorded on preprinted-forms and then transferred to computer storage. A simple computer
program was written to perform the necessary calculations in accordancewith the ERLPM for determining the apparent specific gravity, percentvoids in the total mix, percent voids filled, unit weight, and stabilitycorrections due to briquet volume fluctuations. The refined data werearranged by gradation and asphalt content and are presented in AppendixB. SAS was utilized to manage the data and assist in the analysis.
Testing for Time Trend
An analysis of variance (ANOVA) and Duncan's Multiple Range Testwere performed on the data. The primary reason for this analysis was to
- determine whether time had an effect on the data. Blocking for time,i.e., testing for differences in stability, flow, and air voids betweenreplicates, using the ANOVA and Duncan's Multiple Range procedures inSAS revealed differences existed in the stability and flow results atthe 0.05 significance level. In particular, Duncan's test indicated thestability and flow results were statistically significantly differentfrom the other replicates for replicate 1, and the flow results weredifferent for replicate 2. This was not conclusive enough to warrantthe omission of any replicates from the data base. However, acorrelation analysis of the Marshall properties with and without thefirst replicate was conducted. This analysis indicated a notabledifference in the correlation coefficients determined with and withoutreplicate 1. It was therefore decided that the first replicate shouldbe deleted. The correlation coefficiets among the properties with andwithout the first replicate are shown in Table V and in Figures 10-12and 13-15, respectively.
22
_ , ', i , .. ...,. ,,,, . . t . . . . . - ." .. . .
"* Table V. Correlations Results Using All 12 Replicates (12 Reps) andUsing Only Replicates 2-12 (11 Reps)
Gradation Asphalt Stability Stability FlowContent vs. vs. vs.
Flow Air Voids Air Voids
12 Reps 11 Reps 12 Reps 11 Reps 12 Reps 11 Reps
FAA 5.0 0.511 0.582 -0.547 -0.594 0.126 -0.035LOWER 5.5 0.316 0.360 -0.446 -0.296 -0.152 -0.264
6.0 0.676 0.781 -0.268 -0.290 -0.440 -0.4386.5 0.665 0.713 0.328 0.121 -0.289 -0.3907.0 0.368 0.373 -0.274 -0.276 0.076 0.0567.5 -0.274 -0.273 -0.489 -0.542 0.359 0.417
FAA 5.0 0.386 0.506 -0.519 -0.538 0.294 0.052MIDPOINT 5.5 0.582 0.593 -0.276 -0.416 0.007 -0.239
6.0 -0.527 -0.591 0.261 0.275 -0.454 -0.4566.5 0.080 0.040 0.489 0.370 -0.502 -0.7367.0 -0.123 0.195 0.233 -0.007 -0.552 -0.4587.5 0.077 0.040 -0.281 -0.402 -0.104 -0.222
FAA 5.0 0.118 0.112 -0.413 -0.400 -0.511 -0.510UPPER 5.5 0.341 -0z031 -0.666 -0.525 -0.589 -0.465
6.0 -0.416 -0.391 0.383 0.357 -0.865 -0.9316.5 -0.421 -0.278 0.506 0.011 -0.812 -0.8147.0 -0.684 -0.690 0.485 0.483 -0.481 -0.4817.5 0.015 -0.062 -0.095 0.019 -0.669 -0.661
JMF 5.0 0.634 0.281 -0.855 -0.576 -0.692 -0.3625.5 0.198 -0.090 -0.771 -0.702 -0.088 +0.5776.0 0.181 -0.097 -0.701 -0.762 -0.509 0.1386.5 0.312 0.003 -0.646 -0.722 -0.907 -0.6117.0 -0.327 -0.391 -0.230 -0.439 -0.504 0.1497.5 -0.108 0.282 -0.116 -0.354 -0.882 -0.794
23
1.0 +
0.9+
- .0.8 +
0.7 + L ------*0.6 +
C I0 -0.5i + LR I
- 0.6 +ME U L
-0.8 +L
- 0.9 +
0.0 -1.0
0 -0. +- m -+
1 -0. MIPON U UPE
th cor0laio cofiin+~=0
008
-0. +. . . . . . . .* * . .
-w. ---. - --- . ,. -- - - J - - - ,- -. . -'. = . , . .- --.. -.- -. S- . -- .- . .
1.04
0.9+.1
0.8 +
0.7 +----------------------------------------------------------------------------------------------
0.6 *C I0 0.5 + M U
R 0.4+ UE I LL 0.3 +A I H MT 0.2 +I I0 0.1 +N
0.0 +C - I0 -0.1 + JE IF -0.2 +F I J
I -0.3 + N UC II -O.s + UE I LN -0.5 + m LT IL
-0.6 +I ------------------ --------------------------------- J --------------------------------------
-0.7 + J
-0.8 +
-0.9 +
-1.0 +
---------------------------------------------------- --------
5.0 5.5 6.0 6.5 7.0 7.5
ASPHALT CONTENT
LEGEND: GRADATION J = JMF L = LOWER
M = MIDPOINT U = UPPER
NOTE: Dashed lines are 95% confidence limits for the hypothesis that
the correlation coefficient, p = 0.
Figure 11. Stability vs. Air Voids Correlation Coefficient Plots Using* All 12 Replicates
25
_, .. .0, i ,. _ - . . " : _ - i: i - ,. -> > " ,, T ,
0.9 +
*0.8 +
*0. 7 +I --- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
0.6 +*C I
0 0.5 +R IR 0.4 +E I LL 0 3 +MA IT 0.2 +
*I I L0 0.1 + LN I7
0.0 +
C I0 -0.1 + JiE LF -0.2F1 -0.3 4 LC I1 -0.'4 +
N -0.5 +U J 4T I
I ------------------------------------------------------------------------------- U---0.7 + J
-0.8 *UJU
-0.9.
5.0 5.5 6.0) 6.5 7.0 7.5
A'3PIIALT CONIFNT
LEA (,1J I ,. 'iN J Jl 0m LOWER
M MID~t)INT U = UPPER
4 NOTE D~ nfidj.t er, ,- Ii ts for the hypothesis thate3 r I. , n 0 .
Figure 12. Flow vs. Air Voids Correlation Coefficient Plots Using All12 Replicates
2ob
10
1.
0.9* +L
0.7k +
0.6 +LC I ":R
R -0.4 +E ILL
-0.6 +J
-10.1 +
-0.0 +
FEND GRADT.O + M OEF MIION UL PE
E) I
-08
-0. +
1.0 +
0.9 +
0.8 +
0.7 +0.6 +
C
R IR -0.3 +LE I JL -0.3 + 14A IJT -0.25+
- 0.6 + L
-0.07+ __ _ __ _
0-0.8 +
IF -0.9 +
-0.0 +LLL
1 -0. + U - + M
N 050 5. 6.U. . .ASTL IONTEL
LE0END GRD+O J jM OE
Repicte 2-1
I2------------------------ -----------------------
° . - -
1.0 +
0.9+
0.8 +
0.7 +---------------------------------------------------------------------------------------- ------
0.6+ JC I0 0.5+R IR 0.4+ LE IL 0.3+
*A IT 0.2+ I J J0 0.1+N INM
0.0 + "
qC I L0 -0.1 +E IF -0.2 + -F I L1 -0.3+C I
I 0.4 + LE I U L MN -0.5 + U UT I
-0.6 +GJ-------------------------- --------------------------------------------
-0.7 +
" M IPIT PE
-0.8 + uj
-0.9 +U
-1.0
-------- +------------------------------------------- ------ 4---
5.0 5.5 6.0 6.5 7.0 7.5
ASPHALT CONTENT
LEGEND: GRADATION J = JMF L = LOWERM - MIDPOINT U = UPPER
NOTE: Dashed lines are 95% confidence limits for the hypothesis thatthe correlation coefficient. p = 0.
Figure 15. Flow vs. Air Voids Correlation Coefficient Plots Using Rep-licates 2-12
29
6I
At this point, a more detailed investigation of the scatter plotsof the properties for each gradation at each asphalt content wasconducted. This investigation revealed a significant number of outliersfrom replicates 2 and 3 influencing the correlation coefficients. Thescatter plots containing the outliers, with the outliers identified, arepresented in Appendix C. The ANOVA results were reviewed and replicates2 and 3 failed to show significant differences from the other replicatesfor stability, flow, or air voids. A third correlation analysis wasperformed deleting the first 3 replicates. The results of this analysisare shown in Table VI, and Figures 16-18. The analysis showedcorrelation coefficients markedly different from the coefficients in theprevious analysis with replicates 2-12. Because of these differences,the first 3 replicates were eliminated from the correlation analysis.
Optimum Asphalt Content Determinationand
Marshall Property Comparisons
The Marshall properties, stability, flow, and air voids, were eachplotted against asphalt content. This was performed primarily todetermine the optimum asphalt content for each gradation and to comparethe research data with the generally accepted plots of stability, flow,and air voids versus asphalt content. The first replicate was deletedfrom these plots for the reasons previously discussed.
Correlation Analysis
The objective of the research was to determine whether or notcorrelations exist among the Marshall properties. To be more specific,the research was intended to determine how well Marshall stabilitycorrelated with Marshall flow, Marshall stability correlated with airvoids, and Marshall flow correlated with air voids for each gradationand asphalt content combination. Since each property appeared to berelated to the optimum asphalt content, the correlation coefficientswere plotted with each gradation adjusted for its respective optimumasphalt content.
A correlation coefficient is a measure of the amount of associationbetween 2 variables. The correlations in this research effort were
* evaluated based on a linear relationship between 2 variables, and weredefined by:
30
--. . . . . . . . . . . .
Table VI. Correlations Results Using Replicates 4-12
Gradation Asphalt Stability Stability FlowContent VS. VS. VS.
Flow Air Voids Air Voids
FAA 5.0 0.449 -0.286 0.368LOWER 5.5 0.522 -0.329 -0.010
6.0 0.860 -0.415 -0.4956.5 0.644 0.378 -0.1377.0 0.770 0.133 -0.2567.5 '-0.263 -0.619 -0.023
1 0FAA 5.0 0.645 -0.532 -0.220MIDPOINT 5.5 0.422 0.444 0.084
6.0 0.116 -0.200 -0.6826.5 0.674 -0.010 -0.1847.0 0.692 -0.401 -0.1487.5 0.350 -0.532 -0.252
FAA 5.0 0.298 -0.416 -0.529UPPER 5.5 -0.223 -0.396 -0.308
6.0 0.278 -0.473 -0.7536.5 -0.005 -0.113 -0.7677.0 -0.547 -0.407 0.0177.5 0.540 -0.631 -0.345
JMF 5.0 0.006 -0.156 -0.1475.5 0.048 -0.601 0.5376.0 0.282 -0.643 -0.4936.5 0.032 -0.635 -0.7317.0 -0.260 -0.515 0.3067.5 0.383 -0.457 -0.688
31
6L
L0.7 + M
I -M -- - - - - - - - - - - - - - - - - - - - - - L -- -- -- -- ----
C I L- -- U
R ILR 0-0.1+
-03+C
I 0 . +
N 0.5 +T IU
0 0.6 +
II -
1 -0.3 +
N-0.8 +
-10+
-0. + +-
5.0 5.5 6.0 6.5 7.0 7.5
ASPHIALT CONTENT
LEGEND: GRADATION J =JMF L LOWERM = MIDPOINT U = UPPER
NOTE: Dashed lines are 95% confidence limits for the hypothesis thatthe correlation coefficient, 0 0.
* Figure 16. Stability vs. Flow Correlation Coefficient Plots Using Rep-licates 4-12
3L
0.9+
0.8 +
0.7 +
it I M-0.6 +
-0.2 +
-1.0 +
0.1 + PON U = UPENOE Dahdlnsae9jcniec lmt o h yohssta
th co0ela5o +ofiinp0
I- -- ~-.-------- - -- ----- - -.. . . . . . ------------ U--
-0. +
0.9 +
--------------------------------------------------------------------------------------------1.6 +
0 1)., +R IL 0. 1 +
L (J.,I +J
A I
(3. +
o 1).1 + MN I>
). U
C I0 -o1. +
E IJi L Mr -1).2 + M
IL M
I -0. I U
E
- I U-u. 6 +
I U
-(1.a +
-11.9 4
-1.01 +
------------------------4------------------------------- ---------------------------
5.U 5.5 6. u 6.5 7.0 7.5
ASPIALT CONTENT
LEGEND: GRADATION J = JMF L =.LOWER
M = MIDPOINT U - UPPER
NOTE: Dashed lines are 95% confidence limits for the hypothesis that
the correlation coefficient, p = 0.
-
'.3"
Figure 18. Flow vs. Air Voids Correlation Coefficient Plots Using Rep-
*g licates 4-12
34
N ZVjul
rs
:1 XiR2~ (y_Y)2
where:
r sample correlation coefficientN - number of samplesX - one variable
mean of X. variablesY- the other variable
- mean of Yj variables
The sample correlation coefficient (r) can range from -1.0 (exactnegative correlation) to +1.0 (exact positive correlation). When thesample data are scattered or form either a horizontal or vertical line,the (X-X)(Y-Y) cross products will be approximately half positive andhalf negative, the sum of which will be close to zero, resulting in a
*Q low correlation coefficient. But, for variables showing a nearone-to-one relationship, the sum of the cross products, (X-X)(Y-Y), is
* either positive for a direct correlation or negative for an inversecorrelation, and approaches the magnitude of the denominator in the
-' equation, thereby resulting in a high correlation coefficient, i.e., oneclose to +1.0 or -1.0.
Only 9 replicates were used for the correlation analysis, because,as discussed previously, outliers in the first 3 replicates caused thecorrelations among the properties based on 12 replicates to varymarkedly from the correlations based on 9 replicates. Therefore,referring to Figure 6 presenting the power curves, there is a 0.4probability of detecting a correlation if the true populationcorrelation coefficient between 2 parameters is 0.5. If the truepopulation correlation coefficient is 0.7, then there is a 0.7probability of detecting a correlation.
35
o -- ~ .. . .-
CHAPTER IV
DATA ANALYSIS RESULTS
In this chapter, the results of the data analysis are presented.Some of the findings were as expected, while other results wereunexpected. The results of the optimum asphalt content determinationare presented first, followed by the results of the correlationanalysis.
Marshall Properties-Asphalt Content Relationships
The primary purpose in plotting the Marshall stability, flow, andair voids against the percent asphalt content was to determine the
optimum asphalt content for each gradation and to compare the researchdata with the accepted plots of the Marshall properties published inMS-2 (10).
The comparison of the Marshall p-operty plots was conducted firstsince several of these plots were required to determine the optimumasphalt content for each gradation. Typical plots of stability, flow,and air voids versus asphalt content as presented in MS-2 are shown inFigure 19. The patterns exhibited by the properties in Figure 19 arealso present in the research data.
Stability versus Asphalt Content
In the stability versus asphalt content plots (Figures 20-23),stability increases to a maximum and then decreases as the percentasphalt content is increased. The horizontal reference line at 1,800pounds corresponds to the minimum specification limit for the FAAEastern Region. In Figure 20, the maximum stability would be somewherebelow 5.0% asphalt content. Another interesting point to be made is therange, i.e., the difference in the minimum and maximum values, in thestability values among gradations for a given asphalt content.
All stability means for each gradation and asphalt content arewithin the spcificaton limits exceut the mean for the Midpoint gradationat 7.5% asphalt content. However, the stability may have a range aslarge as 700 to 800 pounds for a given gradation and asphalt content.For instance, the stability results for the Lowei gradation at 5.0%asphalt content vary from 2,130 to 2,890 pounds; a range of 720 pounds.The smallest range in stability for a given gradation and asphalt
content is 165 pounds, which is from the results of the JMF gradation at7.5% asphalt content. Although the stability results appear to varysignificantly, the standard deviation for each asphalt content variesbetween 53 and 224 pounds with a mean standard deviation of 137.2 pounds
for all asphalt contents and gradations. The standard deviationgenerally decreases as percent asphalt content increases. A summary ofthe statistics for all gradations is presented in Appendix E.
36
4
Ism 5Z00--- -_ (N)
!1600 -- 000
1700 7500.\
41600 - - - 000
4.045 5.0 U. 6.0
% AC BY WGT. OF MIX
25 25
20----------------20 E0 1
0 C01I5 -- 5 0
5. 5 .J
Ba.
4.0 4.5 5.0 5.5 60
% AC BY WGT OF MIX
a5 -
04
E 3
, - "- -
4.0 4.5 5.0 53 6.0
% AC BY WOT OF MIX
Figure 19. Marshall Property Plots from The Asphalt Institute's MS-2Manual (10)
37
3500
3100
* 2100
2700
T 500AA
360
121600 La300
Y
T 1000
i 3S00
T~ML 2500CH
3300
3300
300
21000£
200
2400
L*
3239
atG
07
3300-
3200-
3100-
2000-
2400-
2800- A
3A 0A A
LIA
2100.
1700 A
1000
15001
3.0 5.5 9.0 3.5 7.0 7.3
RS~l4ALT CONTENT
04
3300-
3000
9100-
200
2700-
2600 A
300
3200- AA
1700
040
.05.3 6.0 6.5 7.0 7.5
* ASPMALT CONTENT
q Ficure 23. Stability vs. Asphalt Content: JMF Gradation
41
Flow versus Asphalt Content
In the flow versus asphalt content plots (Figures 24-27), the flowincreases with increasing asphalt content, as indicated in MS-2. Thehorizontal reference lines at flows of 8 and 16 correspond to theminimum and maximum specification limits for the FAA Eastern Region.All mean flows for each gradation and asphalt content are above theminimum specification limit. However, the mean flow for the Upper andMidpoint gradations above 6.5% asphalt content and mean flow for theLower and JMF gradations above 7.0% asphalt content are also above themaximum speci limit.
As the mean flow for each asphalt content increases with increasingpercent asphalt content, the range in the flow values tends to increase.The maximum range for flow, which occurs in the Upper gradation at 7.5%asphalt content, is 5.5 (1/100-inch). The minimum range for flow is 1.3(1/100-inch) for the JMF gradation at 5.0% asphalt content. Thestandard deviation, like the range, generally increases with increasingpercent asphalt content. A summary of the statistics for all gradationsis presented in Appendix F.
Air Voids versus Asphalt Content
In the air voids versus asphalt content plots (Figures 28-31), theair voids decrease as the asphalt content increases and then graduallylevel off at approximately 1 to 2 percent depending on the gradation.The horizontal reference lines at air voids of 2 and 5 percent are theminimum and maximum specification limits for the Eastern Region. Theair voids means for all gradations at 5.0% asphalt content are above themaximum specification limit. At 6.5% asphalt content, the air voidsmeans are below the minimum specification limit for the Upper, Midpoint,and Lower gradations, while the JMF gradation air voids mean does notfall below the minimum specification limit until 7.5% asphalt content.
The range in air voids at each asphalt content, like the range instability, tends to decrease with increasing percent asphalt content.The maximum range in air voids is 2.75%, which occurs in the Lowergradation at 5.0% asphalt content. The minimum range in air voids is0.26%, which is for the Lower gradation at 7.0% asphalt content. Therange at 7.5% asphalt content for the same gradaton is only slightlyhigher (0.29%). The standard deviation for each asphalt content varies
*O from 0.08% to 0.71%, with a mean standard deviation of 0.23%. Thestandard deviation for air voids generally decreases with increasingpercent asphalt content. A summary of the statistics for all gradations
. is presented in Appendix G.
42
0 .
e ., .
24-
23 &
21
&a
to0 &
IIt
L 17-
12.
&I.
5.03.56.06.57.0 7.3
MSrMNLT comrgmr
Fig-ure 24. Flow vs. Asphalt Content: FAA Upper Gradation
43
2-
23
3a31 *21.
to A
is
L 17 £a &
162L-A A
LI-.to
5.0 5.5 6.0 ,., 7..0 7.
0srNmLr cONTeNT
0
Figure 25. Flow vs. Asphalt Content: FAA Midpoint Gradation
44
0
".' ', -. ,- -; -' -" -' , -i. .. . . . . .• . . . . . . . .. . . .. . -- '." '- '. -- -" " - '-
"4
22
21
L &7 &a
14
13
12
5.0 5.5 6.0 6.5 7.:0 7.5
ASPPMALT CONTENT
Figure 26. Flow vs. Asphalt Content: FAA Lower Gradation
45
at
L17
an
12
1.5 a.
ALI'TCMTN
Fiue2. Fo1s shltCnet M rdto
46*r£
--- --- --- --- -- - -- - -- ---. 7 --- -. . . . . *- 7
7.5
6.5
6.0
V
0 .0
2.0
5.0 5.5 6.0 6.5 7.0 7.5
* NRSPIIALT CONTENT
Figure 28. Air Voids vs. Asphalt Content: FAA Upper Gradation
47
7.5
7.0
U.S-
1.0
5.5
2.0.
V .5 2 .aa.5 7.0 7.5
ASPHiAL T CaNTE[NT
, Figure 29. Air Voids vs. Asphalt Content: FAA Midpoint Gradation4
I 48
7.5-
7.0a
0.5
5.5
2.0 --------- --- ---- -- ------------------- ---
9 49
7.0
7.5
7.0
3.5
vA
3.0
5.5A
5.0. - - - ------- -
0: 5: - - : .ISHL ONE~T
Figur 31. ir Vods vs AspatCnet rdto
".so
Standard Deviation versus Coefficient of Variation
It should be noted that when the coefficients of variation are*l compared, a similar increasing or decreasing trend with asphalt content
is not apparent. This indicates that the trends in standard deviationare associated with corresponding increases or decreases in the meanvalues of the results as asphalt content increases. The coefficients of
*- variation for stability, flow, and air voids appear in Appendices E, F,and G, respectively.
Optimum Asphalt Content Determination
The ERLPM and MS-2 determine the optimum percent asphalt contentdifferently. The MS-2 manual specifies the optimum asphalt content asthe average of the asphalt contents at the maximum stability, maximumunit weight, and the midpoint of the air voids specification limits.The percent air voids specification limits are 3-5 in MS-2 and 2-5 inthe ERLPM. The ERLPM determines the optimum asphalt content using onlythe asphalt content at the midpoint of the air voids specificationlimits. The optimum asphalt contents were determined for each gradationin accordance with the ERLPM procedures and are presented in Table VII.
* The Marshall properties are related to the optimum asphalt contentfor each gradation. The stability values in Figures 20-23, have atendency to reach maximum values within approximately 1/2 percentasphalt content of the ERLPM optimum. This maximum stability andsubsequent decrease is due to the aggregate particles becoming coatedwith a thicker film of asphalt cement as the asphalt content isincreased. The asphalt cement is used in bituminous mixtures primarilyto provide durability and act as a binder between aggregate particles.The asphalt cement alone cannot provide stability. As the asphalt filmgets thicker, the aggregate particles tend to slip. If the asphaltcontent is increased to percentages well above the optimum, the airvoids decrease. As a result, the aggregate becomes suspended in theasphalt cement and the ability to sustain applied loads is reduced.Similarly, at asphalt contents above optimum, the flow (Figures 24-27)rises sharply and air voids (Figures 28-31) decrease to a minimum.
Correlation Results
The results from the correlation analysis were investigated fromboth a generalized and detailed perspective that included each gradationand asphalt content. The correlations between stability and air voids,stability and flow, and flow and air voids were analyzed. The resultsfrom the analysis, which are based on 9 replicates (4 through 12), areplotted in Figures 32-35 and 44-45. The Upper, Midpoint, Lower, and JMFplots represent the correlation coefficients for the FAA Upper, FAAMidpoint, FAA Lower, and Rochester job mix formula gradations,respectively. The correlation coefficients can range from a perfectnegative correlation of -1.0 to a perfect positive correlation of +1.0.The horizontal reference lines at +/- 0.67 for each correlation plotcorrespond to the 95% confidence limits for the null hypothesis that the
41 true correlation coefficient is zero.
51
Table VII. Optimum Asphalt Contents in Accordance with the ERLPM forGradations Used
OptimumGradation Asphalt Content
(percent)
FAAU 5.6
* FAAM 5.5
FAAL 5.7
JMF 6.3
n
- 52
0 .: : -. .. :,, - --"" " " ... - ," ' • ,.... "' ,- ,. -. - : . .. -. -.--- ---:-..-. . .,,---. , ,.-". •. .. • . "
As mentioned in the previous section, the Marshall properties arerelated to the optimum asphalt content for each gradation. Furthermore,it can be expected that the correlation coefficients among theproperties may also be related to the optimum asphalt content for eachgradation. There is a large difference between the gradation with thelowest optimum asphalt content (FAAM with 5.5%), and the gradation withthe highest optimum asphalt content (JMF with 6.3%). Due to thedifferences in the optimum asphalt contents and the effect the optimumasphalt content for a particular gradation has on the Marshallproperties, each gradation was adjusted for its respective optimumasphalt content to compare the correlations from each gradation. It wasfelt that the true relationship among the properties would not berevealed from correlation plots with the gradations not adjusted fortheir respective optimum asphalt contents. Therefore, the correlationcoefficient plots without the gradations adjusted for their respectiveoptimum asphalt content are not discussed, but are included for thereader's reference, (Figures 33, 35, 38, 40, 42, and 44).
Although the correlation coefficients are not impressively large,consistent patterns among the results lead to speculation about the truerelationships among the properties. Coefficients in this researcheffort ranging from approximately 0.3 to 0.4 were considered to be
* slightly or mildly correlated, coefficients ranging from 0.4 to 0.7 wereconsidered moderately correlated, and significant correlations between 2properties were considered to exist for coefficients of approximately0.7 or higher.
Stability and Air Voids Correlation Results
A general analysis of the stability and air voids correlationsreveals a low to moderately low negative correlation for all gradationsat asphalt contents around optimum and below (Figure 32). At asphaltcontents between 0.5% and 1.5% above optimum, the correlationcoefficients are dependent upon the gradation. For greater than 1.5%above optimum asphalt content there is a low to moderate negativecorrelation. However, the asphalt cement is speculated to becontrolling the properties at asphalt contents more than 1.5% aboveoptimum.
To obtain these general conclusions, each gradation was analyzedindividually. Referring to Figure 32, the Lower and Upper gradationsare both mildly negatively correlated up to approximately 0.5% above theoptimum asphalt content. At this point, the Lower gradation correlationbecomes positively correlated and then reverts back to a negativecorrelation at 1.5% above optimum asphalt content. Although the Uppergradation follows the same pattern as the Lower gradation, thecorrelation coefficient at approximately 1.0% above optimum asphaltcontent does not become positive.
The Midpoint gradation, with the exception of the correlationcoefficient at optimum asphalt content, also follows the same generalpattern established by the Lower and Upper gradation correlationcoefficients. The exact reason for a positive correlation at theoptimum asphalt content is unknown. But, due to natural variaton in the
53
. ..
0.9
(0A. +
0..7 +
I --- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -11.6 +
C I0 0.5 +
R + L
L -U.3 +1
A I
N 0(.2 +JI I L
-0.6 + J..,
0 -U----+
-0.28+
2**. --t-+--- ----- - -- + -- +- --
-1L -1. -0.5 0. 0.J. ..
OTSE FIO OPIU ASHLMO N
LEGEND GRDT J -M L - LOWER-I~~~~ ~ ~ ~ MIDPOINT----------- U --------------------- UPPERNOTE: Dahe +ie r ofdnelmt o h yohssta
* ~~~~~~~-. th+orlto ofiin,~=0
-U9
ReOiatsE:1 Dithe Gradsaion Adjustfiee forit Foptih Asphlts Cotet
54
I . , - '' ,, . - - . . . j ; . , , ". - .- ,, . .-- -' . - ,. . . 7 ; ,, ,, ., . .- - r - -, " - - :
1.0 *
0.9 +
0.8 +
0.7 +----------------------------------------------------------------------------------------------
0.6+C I0 0.5+R Im
L 0.3 +A I
T 0.2+
! L
I I
0 -0.1+
0 -0.6+ L
F -0.7 +F I
-0.8 +
-0.9 +
-1.0 +---- +
5.0 5.5 6.0 6.5 7.0 7.5
ASPHALT CONTENT
LEGEND: GRADAT!CN J = JMF L = LOWER
M = MIDPOINT U = UPPER
NOTE: Dashed lines are 95? confidence limits for the hypothesis that
the correlation coefficient,. = 0.
Figure 33. Stability vs. Air Voids Correlation Coefficient Plots Using* Replicates 4-12
55
,I : --.--- - - --... -V.... ._: ...........--
data, there is a chance of the data showing a positive correlationcoefficient even if there is really a negative correlaton betweenstability and flow. Since the other three gradations show a consistentnegative correlation, in addition to the negative coefficients at 0.5%below and 0.5% above optimum asphalt contents for the Midpointgradation, it is believed that there is a negative correlation betweenstability and flow around optimum asphalt content and below, and thatthe positive coefficient at optimum asphalt content is simply due tonatural variation in the data.
For the JMF gradation correlations in Figure 32, from approximately0.5% asphalt content below optimum to 0.5% asphalt content aboveoptimum, the same pattern established by the Lower and Upper gradationsis apparent. The negative correlation in this range increases in
*magnitude slightly, and at asphalt contents higher than 0.5% aboveoptimum, the correlation coefficient decreases.
It should be noted that due to the high optimum asphalt content ofthe JMF gradation, the lowest asphalt content tested (5.0%) for the JMFcorrelation coefficient stands alone when plotted with the othergradations adjusted for their respective optimum asphalt contents. Whatthe Lower, Midpoint, and Upper gradation coefficients would be at
*Q asphalt contents 1.0% below the optimum cannot be extrapolated from theavailable data. In Figure 32, the JMF gradation could indicate there isnot a correlation between stability and air voids at asphalt contentsextremely below the optimum, which is what would be expected if therewas an insufficient amount of asphalt cement in the mix to bind theaggregate particles together. However, it is difficult to make apositive statement on the correlation based on a single correlationcoefficient from only 9 replicates.
In practice, the production of asphalt concrete at the plant should* not vary as much as 1.0% above or below the optimum asphalt content.
*Since the main concern of the research is the relationship among theproperties at asphalt contents encountered in asphalt concrete mixturesin the field, the correlation coefficient for the JMF more than 1.0%below optimum is not significant to the overall results of thecorrelation analysis.
The moderately low correlation coefficient for each gradation would*Q not have significance if viewed individually, but, with all of the
gradations taken as a whole, the final conclusion is that a moderatelylow negative correlation between stability and air voids exists frombelow to slightly above optimum asphalt contents. If there was not acorrelation between stability and air voids, ideally, each of thegradations should show erratic correlation patterns averaging about the
* zero correlation coefficient.
Stability and Flow Correlation Results
An initial analysis of the 4 gradations for Marshall stability andflow correlations revealed a slight positive correlation at and belowapproximately 0.5% above the optimum asphalt content. From 0.5% to 1.5%above the optimum, the correlations appear to be dependent upon the
560
gradation (Figure 34). At more than 1.5% above optimum, thecoefficients, on the average, form a slight positive correlation.However, as noted in the stability and air voids correlation results,the asphalt cement is probably controlling the results of the mixturedue to the high asphalt content.
Referring to Figure 34 that presents the stability and flowcorrelation coefficients for each gradation adjusted for its respectiveoptimum asphalt content, the JMF and Upper gradations correlationcoefficients, although not consistent in any pattern, tend to averageabout a zero correlation coefficient. At 1.0% below optimum asphaltcontent the JMF gradation correlation coefficient is essentially zero.It increases to a mild positive correlation slightly below the optimum
*asphalt content before becoming a small negative correlation atapproximately 0.5% above optimum. The correlation is then positive atmore than 1.0% above the optimum asphalt content.
The Upper gradation is mildly positively correlated at about 0.5%below optimum and, as asphalt content increases, becomes a negativecorrelation before starting a cycle in which the correlation coefficientbecomes negative and then reverts back to positive. Because thesecorrelations tend to average about the zero correlation coefficient, the
* correlation between stability and flow for these 2 gradations was firstthought to be zero. But, as will be shown later in this section, thecorrelations are now believed to be slightly positive.
The Lower and Midpoint gradations in Figure 34, with the exceptionof the Midpoint correlation coefficient at 0.5% above optimum asphaltcontent, form a pattern. From approximately 0.5% below optimum toslightly more than 1.0% above optimum asphalt content, the correlationcoefficients are moderately positive with the exception noted above. Asdiscussed previously, at 1.5% above optimum asphalt content and greater,the asphalt cement is speculated to control the Marshall properties.
The correlations appear to be dependent upon the gradation, and ifall of the correlation coefficients are considered as a whole, theresult is a weak positive correlation. But, a detailed analysis of thestability and flow correlations revealed an outlier in the scatter plotsof stability versus flow that significantly influences the results ofthe correlation coefficients. The outlier, as shown in Figure 36, inthe 5.5% asphalt content Upper gradation scatter plot, causes thecorrelation coefficient to be -0.223. If the outlier is deleted, thecoefficient becomes +0.382.
If the outlier is deleted and the new correlation coefficientsplotted (Figure 37), a moderately low positive correlation betweenstability and flow, on the average, becomes apparent around 0.5% aboveoptimum asphalt content and below. From around 0.5% above optimumasphalt content to aproximately 1.5% above optimum asphalt content, thecorrelations appear to be dependent upon the gradation. The Lower andMidpoint correlation coefficients are significantly positive, whereasthe JMF correlation goes from a slight positive correlation coefficientto a moderate negative correlation coefficient.
57
[:
The other correlations were also influenced by the deletion of theoutlier. The stability and air voids correlation coefficient for theUpper gradation at 5.5% asphalt content changed from -0.396 to -0.722.There was a slight decrease, from -0.308 to -0.151, in the Uppergradation at 5.5% asphalt content for the flow and air voidscorrelations. The stability and air voids and flow and air voidscorrelation plots without the outlier are shown in Figures 39-42,respectively. The purpose of deleting the observation was to show howmuch influence one outlier can have on a correlation coefficient and toshow that a moderately low positive correlation probably does existbetween stability and flow.
6
-V..4
S58
x7
1.0 +
4).9 +
0.8. I +
0.7 + L P
0.6 +C I '0 u.5 + L
E IL u.3 +U UA I
J -
10.2 +
N I -J ,_C
o-0.1E IF -0.2 +
F L
I -0.4 +
EIN -U.5 +
-4o.6 +I ------------------------------------------------------------------------------------------
-42.7 +
-08+
-0.9 +
-------------------------------------------------- ----------
-1.5 -1.0 -0.5 0.44 0.5 1.0 1.5 2.0
OFFSET FROM OPTIMUM ASPHALT CONTENT
LEGEND: GRADATION J -JMF L = LOWERM =MIDPOINT U = UPPER
NOTE: Dashed lines are 95% confidence limits for the hypothesis thatthe correlation coefficient, p =0.
Figure 34. Stability vs. Flow Correlation Coefficient Plots Using Rep-licates 4-12 with Gradations Adjusted for Optimum Asphalt Content
59
1.0 +
0.9.+I L
0.8 +
0.7 + H~
----------------- N -------- L0.6 +
C I LU
C IL0 -0.4 +
E IL -0.3 +UC I
0 -0.1 NI I\N -05-0.6 + U-0.71
F 0.82+
1 -0.49+
N -0.0 +
ASTL CONEN
1.0.Figure---35. --Stability--vs.---lo---Correlation---Coefficie---t--Plots--Using---ep-
5. 5icte 4-125 .07.
60HLTCNTN
LEED RDAIN J-JM OE
M - MDPOIT U -UPPE
3400 +I
3200 + I
IA AI I A
3000 + IAAI
I I Rep. 92800 + II
T IA 2600 +
T 2400 +I
2200 +II
2000 +
1800 ----------------------------------------- I--------------------------------------- I------
1600 +
0 1 2 3 4 *5 6 7 8 9 10 11 12 13 14 15 16 17
FLOW
Figure 36. Stability vs. Flow Scatter Plot for the FAA Upper Gradation* at 5.5% Asphalt Content
61
1.0 +
0.9 +I L
0.8 +
o.6 +C I U0 0.5+ L
R 0.11 + U.M J
L 0.3* U i U
T 0.2 +I I
0 0.1+ M
N 0 .- J0.0 +-U
C I0 -0.1 +E IF -0.2 +F LI -0.3 +CI -0.4 +E IN -0.5 +T I U
-0.6 +
-0.7 +
-0.8 +
-0.9 +
-1.0 +
------------------- ----------------------------------------
-1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0
OFFSET FROM OPTIMUM ASPHALT CONTENT
LEGEND: GRADATION J = JMF L - LOWERM = MIDPOINT U = UPPER
NOTE: Dashed lines are 95% confidence limits for the hypothesis thatn the correlation coefficient, p = 0.
0
* Figure 37. Stability vs. Flow Correlation Coefficient Plots Using Rep-licates 4-12 Less Outlier with Gradations Adjusted for Optimum AsphaltContent
A. 62
"0
.0+
0.9 +
I LI __ L
0.7 + MI-M--------- --- ------- --- - ---------- ---------
0.6 +
C I U0 0.5 + L
R 0.4 + P4 J
L 0.3 +U jA
IT 0.2 +
0 0.1 MN I
0.0 +.-UC I0 -0.1 +E IIF -0.2 +IF I L
I - 0.3 +C II - 0.4 +E IN -0.5 +T I UJ
-0.6 +I-------------------------------------------------------------------------------------------
-0.7 +
-0 I-0.9 +
-1.0 +
----------- + --------------------------- ------- +----------
5.0 5.5 6.0 6.5 7.0 7.5
ASPHALT CONTENT
LEGEND: GRADATION J -JMF L = LOWERM = MIDPOINT U = UPPER
NOTE: Dashed lines are 95% confidence limits for the hypothesis thatthe correlation coefficient, p =0.
Figure 38. Stability vs. Flow Correlation Coefficient Plots Using Rep-licates 4-12 Less Outlier
63
62
1.0 +
0.9 +.
.0.8 +
0.6 +C0 0.5 +R I M
E I
L -0.3 +C IT -0.2 + U~
I* I L
-0.8 +
F -0.9 + M
-1.0 +L
-1. -0.0 -0+ 0-0--5- 10-1.-2.
OTSE FRO OPIUMSHL OT
010
* ~ ~ -. epiae4-1.0 Less 0ut.0 with Graaton Adjste fo2OtiuAsphalt ContentU SPALCNTN
LEED GRDTO J = JM L - OE
M =. m-P N U = UPPER
NOTE Dahe lie rC 5%cniecelmt ortehpthssta
1.0 +
0.9 +
0.8-+
0.7 +
0.6
0 -0.5 +J
- 0.7 +L
-0.8 +
- 0.9 +
-1.0 + M
0. 5.51 6. 6.U. .
1 -0. MPON U UPE
th corela5o cofi+nt 0
-066
-i-e-0-Sailt-v.Ai-oisCorlain-offcen-losUsnReplicates -4- Less-------------- Ou l e
-0.7 65
I: .:
1.0 +
o.9 +
0. 8 +
0.1 +
0.6 +C J
0 0.5 +
L 0.3 + J
1 0.2+
o 0.1 +
0.0
0 -0.1 +E I J LMF -0.2 + M
*L MI -0.3 +C I U
I 0.1 +
N - 0.5 + LT I U
-0.6 +
-u. 7 + - - - - - - -
IJ UU-0.8 +
* -0.9+
* .- 1.0 +
* .- 1.5 -1.0 -U.5 0.0 0.5 1.0 1.5 2.0
OFFSET FROM OPTIMUM ASPHALT CONTENT
LEGEND: GRADATION J = JMF L = LOWERM = MIDPOINT U = UPPER
NOTE: Dashed lines are 95'1 Confidence limits for the hypothesis thatthe correlation coefficient, p = 0.
Figure 41. Flow vs. Air Voids Correlation Coefficient Plots Using Rep-licates 4-12 Less outlier with Gradations Adjusted for Optimum Asphalt
S Content
66
'1.0 +
0.9 +
0.8 +
0.7 +
0.6 +C IJ0 0.5 +R I
*R 0.4 +E I L
L 0.3 +JA IT 0.2 +
0 0.1 + 14N I
C I0 - 0.1 +E IJi U L 1IF -0.2 +1 MF 1 L '41 -0.3 +C I U
E 1-0.5 +J
I U-0.6 +
-0.7 + ;i <jI V j
-0.8 +
-0.9 +
-1.0 +
------------------------------------------------------- --------
5.0 5.5 6.0 6.5 7.0 7.5
ASPHALT C0ONTENT
LEGEND: GRADATION J = JMF L - LOWERM =MIDPOINT U = UPPER
NOTE: Dashed lines are 95S. confidence limits for the hypothesis thatthe correlation Loefficient, o 0.
Figure 42. Flow vs. Air Voids Correlation Coefficient Plots Using Rep-licates 4-12 Less Outlier
67
Flow and Air Voids Correlation Results
The last correlation analysis considered was between flow and thepercent air voids in the total mix. From a general analysis of the 4gradation correlation coefficients, there appears to be, on the average,a mild negative correlation between flow and air voids at optimumasphalt content and above. As shown in Figure 43, from approximately1.0% below optimum asphalt content to around 0.5% above optimum, thegeneral trend is an increasing negative correlation. At asphaltcontents greater than 0.5% above optimum asphalt content, the trend isfor the correlation coefficients, on the average, to decrease.
The 4 gradations follow a similar correlation pattern throughoutthe asphalt content range. The JMF correlation more than 1.0% belowoptimum asphalt content is negative (Figure 43). The correlationbetween 2 properties based on only 1 correlation coefficient isdifficult to determine. But, at asphalt contents far below the optimum,the correlation in probably zero due to the small amount of asphaltcement in the mixture. From approximately 1.0% below optimum asphaltcontent to slightly above optimum asphalt content, the JMF correlationswitches from a moderate positive coefficient to a significant negativecoefficiet. At approximately 0.5% above optimum asphalt content, the
O correlation coefficient becomes mildly positive, but around 1.0% aboveoptimum asphalt content, the correlation reverts back to a moderatenegative correlation coefficient. Since the correlation coefficientsabove and below the coefficient at approximately 0.5% above optimum areeither moderate or significantly negative, and the other 3 gradations inthe same range also show a negative correlation, it is believed thepositive correlation coefficient for the JIMF is due to the naturalvariability in the data. The true correlation is probably negative,despite the positive value for this sample.
The Lower, Midpoint, and Upper correlations form a very consistentpattern. Although the Lower coefficient is mildly positive, and theMidpoint and Upper coefficients are slightly and moderately negativelycorrelated around 0.5% below optimum asphalt content, respectively(Figure 44), all 3 correlation coefficients decrease in magnitude atoptimum asphalt content. The correlation coefficients increase tomoderate to significant levels at 0.5% above optimum before againdecreasing in magnitude to slight, but negative, correlationcoefficients from 1.0% to 1.5% above optimum asphalt content. As noted
--. above, at asphalt contents around 1.5% above optimum asphalt content andabove, the asphalt cement is probably controlling the results of theMarshall properties.
The erratic nature of the correlation patterns for each gradationprovides little information on the true relationship between flow and
*" air voids if viewed individually. However, with all of the gradationstaken as a whole, the final conclusion is that there is a mild negativecorrelation between the properties at optimum asphalt content and above.
6["
k
6o °-
1.11
(1.9
1). 1
I L
+A
I .? +
o L
0 .1 + L M
-, I
o - .2 + M
F IL1 -(J.3 + UCU
FN -,),5 +T I U
6 +.
I U U
%. . -1 i
-I5 -1 -5i ' LII 0.3 1.3 1. .
t F It l H M OPI IMUM AP'iA[ CONI[NI
"N . I m, " :,w' seits for the h-, ,e,
Figure 43. Flow vs. A r Vo-ds Corte13t:cn Coefficient Plots Uszng Re-• licates 4-12 with Gradat.oris Ai:uste for Apt'n Asphalt Ccnent
. .. .- 0
1.0 +
0.9 +
0.8+
0.7 +
0.6 +
0 0.5 +R IR 0.4 +E I LL 0.3~ + .A INT 0.2+
0 0.1 + M
0.0kC0 -. 1 ISE LIJF -0.2 + M MF I LM
I -0.3 + UC IU
E IN -0.5 +T IU
-0.6 +
-0.7 + j
-0.9 +
-1.0 +
-------------------------------------------- ------------------
5.0 5.5 6.0 6.5 7.0 7.5
ASPHALT CONTENT
LEGEND: GRADATION J = jmF L = LOWERM = MIDPOINT UJ UPPER
NOTE: Dashed 1 ine3 are 95', confidence l imits for the hypothesis that* the correlation coefficient, p =0.
Figure 44. Flow vs. Air Voids Correlation Coefficient Plots Using Rep-licates 4-12
70
CHAPTER V
SUMM1ARY, CONCLUSIONS AND RECOTENDATIONS
Swrmrar
This research effort is the laboratory phase of a 3-phase research
project for the Federal Aviation Administration. The research wasconducted to evaluate the possibility of implementing a multiple price
adjustment system for bituminous airport pavements using the Marshall
properties: stability, flow, and air voids. The Marshall propertiesare physically related, i.e., determined from a single test, and,therefore, can be expected to be statistically correlated. To use these
properties in a multiple price adjustment system, the correlations amongthe properties needed to be identified. This research effort was
designed to identify such correlations.
The experimental design consisted of 4 aggregate gradations
spanning the FAA Eastern Region specification limits for loads greater
than 60,000 pounds and maximum 3/4-inch stone, and 6 asphalt contents
from 5.0% to 7.5% at 0.5% increments, for a total of 2A differentcombinations for 1 replicate. Twelve replicates were made for a total
of 288 Marshall briquets. Each replicate was mixed on one day andtested on the next. Other than asphalt content and aggregate gradation,
*factors influencing the Marshall test results were held to a minimum.
The aggregate was crushed limestone and natural sand, and the asphalt
cement was AC-20. The laboratory procedures followed were in accordancewith those specified in the FAA Eastern Region Laboratory Procedures
Manual (ERLPM) (9).
Several analyses were conducted on the data. An analysis ofvariance (ANOVA) and Duncan's Multiple Range test were performed on the
data to determine whether time, i.e., order, had an effect on the test
results. The Marshall properties were plotted against percent asphalt
content, and the optimum asphalt content was determined for each
gradation in accordance with the ERLPM. The Marshall property plotswere also compared to accepted plots in the Asphalt Institute's Manual
Series No. 2 (MS-2), Mix Design Methods for Asphalt Concrete.Ccrrelation coefficients among the Marshall properties for each asphalt
content/aggregate gradation combination were clculated using the
Statistical Analysis System (SAS) computer program. The correlationswere compared by plotting the correlation coefficients against thepercent asphalt content with each gradation adjusted for its respective
optimum asphalt content.
Conclusions
From the results of the Marshall property correlation analyses, thefollowing conclusions were reached.
1. The data conform to the accepted correlation patterns amongMarshall stability, flow, and air voids versus asphalt content.
*i 2. The standard deviation for each Marshall property varies withregard to aggregate gradation and asphalt content. Generally, thestandard deviations for stability and air voids decrease as percentasphalt content increases, and increase as percent asphalt contentincreases for flow.
3. There were no significant correlations among the Marshallproperties. However, a moderately low negative correlationcoefficient exists between stability and flow from below to
approximately 0.5% above the optimum asphalt content. And, a mildnegative correlation exists between flow and air voids at optimum
asphalt content and above.
4. The correlations for stability and air voids, and stability and*flow appear to be dependent upon the aggregate gradation from
approximately 0.5% above optimum to approximately 1.5% above-- optimum asphalt contents.
I-" .5. There appears to be no correlation among the Marshall*& properties at more than 1.0% below the optimum asphalt content.
But, due to the low optimum asphalt contents of the gradations
chosen, and the range of the asphalt contents tested, only onegradation had a correlation coefficient more than 1.0% below the
*optimum asphalt content. This makes it difficult to make a- positive statement on a correlation between two properties based on-" a single correlation coefficient.
6. The results of the correlation analysis are not significant
enough to justify eliminating one or more of the Marshallproperties from use in a multiple price adjustment system, but they
appear to be significant enough to violate an assumption ofstatistical independence among the properties.
These conclusions are based on the asphalt cement, asphaltcontents, aggregates, and aggregate gradations, considered in thisinvestigation.
7
* . . * •.. *.
Recommendation
The results of the laboratory analysis indicate that it is not. sufficent to consider the Marshall results to be statistically
independent. This means that it is probably not appropriate to considerthe properties separately and then to multiply the individual results toarrive at an acceptance decision for the Marshall properties. It isrecommended that computer simulation analyses be conducted toinvestigate methods for treating the case of correlated multipleacceptance properties. The results of such analyses are presented insubsequent volumes of this report series.
* 73
BIBLIOGRAPHY
1. Burati, J. L., Jr., and J. H. Willenbrook, Acceptance Criteria forBituminous Surface Course on Civil Airport Pavements. Report No.FAA-RD-79-89. National Technical Information Service, 1979.
2. "Characteristics of Bituminous Mixtures". Bibliography Number 35and Supplement. Highway Research Board, Washington, DC, 1963.
3. "Construction Practices - Flexible Pavements". Bibliography Number42.- Highway Research Board, Washington, DC, 1966.
4. Proceedings. First International Conference on the StructuralDesign of Asphalt Pavements. University of Michigan, Ann Arbor,Michigan, 1962.
5. Proceedings. Second International Conference on the StructuralDesign of Asphalt Pavements and Supplement. University ofMichigan, Ann Arbor, Michigan, 1967.
o 6. Monismith, C. L., R. G. Hicks, and Y. M. Salam, Basic Properties ofPavement Components. Report No. FHWA-RD-72-19. Federal HighwayAdministration, 1971.
7. Proceedings: Association of Asphalt Paving Technologists TechnicalSessions. Vols. 16-51, Years 1947-1982.
8. "Relation of Physical Characteristics of Bituminous Mixes toPerformance of Bituminous Pavements". Bibliography Number 43.Highway Research Board, Washington, DC, 1967.
9. Laboratory Procedures Manual. De!partment of Transportation, FederalAviation Administration: Eastern Region, 1981.
10. Mix Design Methods for Asphalt Concrete and Other Hot-Mix Types.Manual Series No. 2 (MS-2). The Asphalt Institute, College Park,Maryland, 1979.
11. David, F. N., "Tables of the Ordinates and Probability Integral ofDistribution of the Correlation Coefficient in Small Samples".Cambridge University Press, Cambridge, 1938.
. . , - . - . "
U- T - - ----
APPENDI CES
75
Appendix A
Example Maximum Theoretical Specific GravityDetermination: FAA Lc.;er Gradation 5.0%
Asphalt Content
Aggregate % by Tctal Wg: Sp Gr.
Coarse (-3/4, +No. 4) 33.95 2.700
Fine - Limestone (-No. 4, +Pan) 42.01 2.684
Fine - Nat. Sand (-No. 16, +Pan) 14.04 2.660
Asphalt Cement 5.0 1.020
Max. Theoretical Specific Gravity =
lO0%
38.95 42.01 ld.O4 5.002.700 2.684 2.66C 1.020
*Max. Theoretical Specific Gr avit= 2v.484 (FAA ..... - 5.0% Ac)
0
76
Appendix B
Marshall Properties and Related LaboratoryTest Data
-. 77
c~~~z ~ z~2 C\~Z :c~ U2
r..-N
. . . . . . . . . . . . ..... . . .
Nz -: -1C
u<-x
. ..- . .. . .
C : Z
•~ ,- z
*~~~ 'N 'N N- ' -~.
-. ,", ', C , "I
Z:Z
~~22 C C\N N. NNNN N C.N N N N
* < <:< < < < <<< << << . -
.. l-';'-- - -< << -< << -<<<-<:<<- -- - -- ---" " --"..
-. ',z Z
Il22
. .-. . . . ..- . .
C' A. %I •
- I I
"
-~N Z,- -
T -
N %' N~ N N
- - - - -- - - - - - - -- - -- -- -,-
IZ > -- .2 '-. . .. . . . .... - - -C\ - 71~ :c--
\0 Iz
I~ C-. N ~ :1, NN N
. . .N . . . . . . . . . . . .. . . . .
C., --
I <---- <<<<< <<<- <<<~ <<<<<<Z'-
cc -- C \.
1 2 _
-- - - - - -
Z . .\ .. . . . . . .
-7 R*~~~~~- - -~,.' ' -- ~ -..- -r'- -
",'2 :~~~~~~~~~ ~~~ ~ .- '- .2 -' ... ... , . " "-'- .'-' ,- .-- -, .- .- .'- .- . .- . ., . -. . , -.'.', . -. . "-. .. ... " . "" "."
' ... N
-- Z ^ a z'
--- - - - -- - - - -----
'I Z N
.. . . .. . . . . . . . . . .
_ ~- -' -~ - - -~ ~ . ... - - -'
z Z
-o C_ Z :
-' QQ -.
. .- -- --- - . . . .------ -... .-- -- . .
-~
.C . . . . . . . .
-- - -- - - ----- ------
- -- z- -- --- - ~- -- - - C- - - - -- --
""<< < , < < ,< < < < < < .,< < < < ,, < . .
Cr.. . . .
I Z •
- U.- - - - - - - - -
* 8O
>770. .f . . . . . . .7- 'N 7 . . . - . - . . -
. . 7. .7. . . . . . . . .-
--- - - --- -- - -
I~~~; C,.- - - - .- - . - . - -, z :r- --
-- - - -.. ~ - - - -
. . . . . .
- - .--. - --- - - - - - - . - -
. . . . . . . . . . . . . . . . . .
zo -
I . -<
= =- - :" ---.----- = = - -
z. ~- -2 - - - - - .- - - - -
<- <
-L. ::-% -
-2> -- --- - - - - - - - ----
-- - - --- - - - - -- --
-<~ z-..x
- - - - --
< < < < < < < <.
z6
-% z
. .- .-
-7Z-
7- 7
> - -' ---- .
0
-p -:-7
Iz z
. . . 2. 2. .X.. 2. . '. 2. - .
z z
-~-Ri53 469 FIELD VALIDATION OF STATISTICALLY BASED ACCEPTANCE PLAN 2/2FOR BITUMINOUS Al-.(U) CLEMSON UNIV S C DEPT OF CIVILENGINEERING J L BURATI ET AL. MAY 84
UNCLASSIFIED DOT/FAA/PM-84/i2-YOL-i DTFASi-8i-C-18857 F/G 1/5 M
EE/l/IEI/I/IlIEIIIIIIIIIIIu
2.2-
ail. L36 _1
1111"-- I .1 I lui
MICROCOPY RESOLUTION TEST CHART
NATIONAL BUREAU OF STANDARDS 1963-A
,ma"oo*.,
. J CC 7 Z. O -
-..
- -
" ZC
- Cl \c2CE~
C - 1
cc 1Z -- r'
I
[:-: '- -.~~~~~~~~~~~~ .-..'2 .-.'''Y .: 2 .- , " "- -. . ..-" "- . .-. .'-. .- . ..-. .-- .-. . -. .• ... . .
C. % N, N
7- '.Z 7- Z:c
=. NA CO
.. . .
-Z I N C.ZV N* N,.%, N*
< L;
-zz-
Ir-
' , .~~ ~~~~~~~~~. .. _• . .o I . . . ._ ; . . .-, . . . . . . o . - ,
---- ---
- - --- - - - - -
In '
-'
< <
*, - 0
CI - .
-- .- < < - << -< << - -.....-- - -- --
In 7- 11-i
-I'
.- n.....,.
\. O
- 5 .. .. .. .. .. .. .. .. . . .................... . . .
,:.'I- =-
=-C-
-I --I
c Nz
ZZC
"I i's " " = - = z -
C i • . . . . . . . . . . . . . . .
= .- " -
- = 5 = - "= z C "N c :7 - - =
N N c nN% N N.N' N N 'Jn N C\, NJN
m 0%1 ON co-
. . . . . . . ... . . . . ., , . . , . . . . .
.Z. Z Z7 Z Z o ,
• 8C,
- --
00 Z0 04' -1 CNN .3 - I=0 _NC c N
a. Z - T,-. - == -t
<In
rI % ~ ~ -- %* - -
L61 0 uc% , - =N - % c -N z n- = -\ % = N
0 8>
z
C . . . .
ezz Cz~ N N '--
C ~ CN.,. C
>~~ ~ ~ . . . . . . . . ~.. \ . .~ .' c .
I--Z
C: I1 ne n - 1I I n m 1 nrl-
- Z - - - - --- - - - - - - - -
, - I\ X - : - 5
.l° .. . . . . . . . . .
0-
l].N -Z '="N
m Z
. ." .." . . . .
I0
-:--J
zoo -
-o\.O
". .. ... . .. ... . . ... . .. .
V- Ni . .. . . . . . . . . . ..N N N C f. N.. . . . . . . . ..N
- M = -
-m-I
I-- j . . N, C\ . . . N, N.. . . . .
I Z- ~ '. . . . , .. . ..3 ~ z'~
-.'----
,+ .,
w C% .r
Tj NNJC R, M K .^ R N
< --- -- - -- -- - - - -
<1z
... . .." ..: . .2. .i ; .. .. .' .i .. .... .. .. ....--.
"- - ,* .. . .. . .4w ;.. .'-..'--
I -
I'-z-",. ---- - - , ".- ._ , -,. , ,- , --. " - - -- - -
. . . . . . . . ..- = Z . .
. .. . . . . .,.. . . .......
-: - - - - -' - - . - - - - , - --
>1 I
==
I~~~C ZZ Z' .
Z- z :c---- z z-z-Z
I --- -- -- ~-- - - - - - - - -
CZZ
0
-o. .o .o' .. °- .. .. °o. ., .o. .- = •' ",.-. .' - - -.. .. -.. .. . .. .. . . . . . ,. . - . . -. ' .-. . . ... ...-." -.3.c. - -
Appendix C
Marshall Property Scatter Plotswith outliers from Replicates Two and
Three Identified
09
S i 4 .- L . , ,- l -. , .. \ .. . . •
S
IE:..-o.0
-' - - -'. - .--- : -1 -
+N
a +
+
O aa4+I
0<<
N 4-<
a aN
I a
Z 0 a
00 I6 a-
a I
aa II IO
+
I Of
a ,
I
+ 0,0
I I
£ I- IIN
- - - - - - - - - - - - - - - - - - - - - - - - - - - - --
a I
IN
- - IN- - a
0W
+
aaN
+
III
+
+0
+ ON
.'+' +- ~---- -- - - - - - - - - - - ------- ------. . .. "-. .. @ . .. @ . . @ . . -- Im
00.. I I-
6n= I un
> . .f" e m
"-. " "" "
4 4
3:Iz
IN
C.3.
I I
IN
L.CL
IN
a0
-9
I
0. 0
"" L + "'
-~I-
'C I-
- II
I
I-
I I
- 4. ----. -- -4 - --..-- 4 .. . ----.-.- - - 4. ----. - I
o
o 00
-lil
Ioo IN
I
I I
o o~
£--I
+ LI
I " -
alI
IN
ul I aa
< =
.+
I I
a a0 0, c6 I IA IIn C1 6a r C
en 04I
INI I
0 ,I
a I
N I
4 a IN
II I I
a a
z *
Lr- cf9a
:a . :.--:...- - a.:". :.2. .:.i. . - , - - ,. , . . - ' - '. a--.• :. ._ -_ .Z . . . . " " ,: . .0- =, , n, \.m 4-.- -- ,,, -,,a a, - - . - - i
I I
! +
U I
I I
* I
I I m
C- I
z I-
Z 0 I
+
I ," I
T p_ rCd , C
>"l~.< I'- I I
- I I I
*0 I " <
I-0 I
I I
4 --
C I
I I
I
+ '
I I
o . C- .\ . . . + -
95
"0
t I
I *(
I I
I I
< a.
I J
I -
II I
"" II
- I.I t - I
' '
- < ..-
_ I
- I
I I
.0 + .I
- 4-
I
P ! 0
I -I
I o -
,. + ,
4'n
C~C>
IfI
In +z an <w 4.
* Coll
< a
a( a' a aren a~
97a
I a
oo
a+
- - a-- a
4
I- I a
I-I
a -a
I
"II 9-
I , % . • . . . ., • . % . ° o' .°. , . '. -. " -. " ,°° ' O " . '. o - . " °o .o . • o. ' , "O .o+o. . .-. , - , . -.. . . ,. . - . . . - . ., . ,. -. . . : .- -. , . -- . - ,- . ," . -, '. . _.I. % % '..a a
L 4.
a
a II
a a
.- a
4-
0 S
a.- .
I a
a
+1 In
I +
I
dC I
II--
i J
III
I I
0 I +
-- -
L..
U,--
I .~
I *
VIN Oj0 f.- ,\ N-
-" -*-i . . " " ' ' " ""
z\ C~ I I
00 I100
I,,,..
a I
a I
=..-
4 -
I° .
+
o I
if a+ .. . + ... + . . . +-... .+ . .. -
[: : - I
- U "
I I
I I
II ,"
I °
,4-
, I
0 I
-< I
IL I
-- <-
, ,
z.( ,o ,<
I 1I
.U L IIC..( 127 2' ' . - • .. • - . "I. ' . - - , 4-.. -i " . . .
.° 4.
+,
- - - - -- - - - - - - - - - -
ItI
6 I* I
C.) I
a-6- I
0 +
%n 6Y 0
10
I-".{ I 4"
- .( - -
_ L._
I IIf U4,,- -- 6*
<
I6 ,,
a--~ I
4-t- . . . . . 4-....
40103
a'" -..-''" - . , -- 6.,- . - --'''' , '' .,- , • "" '. - - ' " . . " " " ,- " .
. 4..
III
d ' -
"- -' 0
-wz I
« 0 )1 >
'°n " • , . t, + . -
•I , <
• I
II-
,O,
'I -
. .ii~ ~i :. i.:i~i .''- -.',' ' . .. " ." .'" - - -'-- . i~ i . . . " " -,i ." - 2 ., L. -;-, ,'3 i -' -... _ : : - : • . . ..'._ ,' , _''. : i "_:',i , " _ __' " , . _ .. =. .,.. ,.=.. ;,= , . .,. ,. __ - I, .
0.I . °
/ I
I I.
I I
* I
I IQ
0 C 11
-< I
If.~ I4-.
cflI- I•
1- -I
I I"
105 ~I10
I
"-,, • - . .. . . . -. .. . .. . .. ." '-. -. .. .... - .. . -, ..".' .. -'. ., .. ... . ..i:"' : : . :' : ." - " " ' --
Cd~ i
: 4
*~q - t
1/ 0 *
cJ II ,I -"
< < I I 0
.,. t-<Cl C.
i I
o
I-I
00 d 0-
-°-• i".- -- " ."- --'-'" . •. .. .--£ ':'." .. .1- - r , " - " " " - " -" " - -"
* I
a a
I IO
a I
•-a IL+
"III,-a a
- .-
rJ~ Io a . . . +Ia . . + a-,
::.-'a
- - a: . . . .. . . . . ' . . . "a . .z _:: "" '"" ' - " "" -a- "-" "- "I"-: -- "- ' I ' " " "" -
- . -C ~ - ----
S aa
a a +, a a
a* a a
h £
a . I
a aIIC'.* a< +
a ..
a I I-- 4 .. ,- . . . . . . . . a-. . . . , . . a-. . -
108
I 4
aL"i'a
I I
I Iga a
a I aI IC'
I a
.0 +
I
I 4C
a a-
a I a* a aI a ai~~* a 4..* a ii-a a aa S a* a I* a aa S II a aO~
4..S a II a a* I Ia a aI S Ia a at.,a I 4..a a a.oI I Ia I a
-. 0 a II a Iw I a aa a ai~-I a 4..U, I Ia a
0 I * I aN a
I a ISI I I'C I I 4..
IS ~----------------------s-.---- -. . ,- a IA5- I I az a a a* UJU~ I a- ~ . I S Iz 0 I .( S I(no I I IIA
- I * \. ( 4..I C.'I ~ * 1* U,
OS- ii a ~J ( a a 0a ~< a a -
.~ .( I ( a I 0a S a >a S.(cn 0 a a so ~
.' 2 I .( ** w a a It., .(Sn o a a> w a a
* . ~.I I aa I
0-~ a a I..JIl I S
* - a~..O I S 4..* S* I.-. a aa a a
-. 0 I a aS aI a0 S a aa-S I 4..* a INa I Ia I aS I S* I a* a aa 5 5-I I 4..
a nj* I I* a I I - -* a a* a aa Ia a ISAa a 4..
a-I a aa I Ia I Ia a aI I Ia a IC~* a 4..0 I a aoa a a
-4. ---- 4. ---- 4. -.-- 4. ---- 4. ---- 4. ---- 4. ---- 4.----e - a0 sA 0 IA 0 IA 0 IA 0IA N 0 IA N 0 IAN N N - - - -
S
110
0
Appendix D
Temperatures Recorded During Marshall BriquetMixing Process in the Laboratory
TemperatureGradation Replicate Asphalt
Content Asphalt Mixing Compaction
FAAL 1 5.0 304 307 250
5.5 305 307 250
6.0 305 307 2506.5 304 307 2507.0 307 298 2507.5 307 301 250
FAAL 2 5.0 307 307 2505.5 307 292 2506.0 301 305 2506.5 301 300 2507.0 304 301 2507.5 307 300 250
FAAL 3 5.0 303 302 2505.5 297 307 2506.0 306 307 2506.5 300 307 2507.0 300 306 2507.5 302 297 250
FAAL 4 5.0 302 307 2505.5 300 307 250
6.0 305 307 2506.5 300 300 2507.0 307 307 2507.5 307 307 250
FAAL 5 5.0 305 301 2505.5 301 302 2506.0 299 307 2506.5 300 307 2507.0 297 305 2507.5 305 301 250
FAAL 6 5.0 298 307 2505.5 302 301 250
111
> . .. . . . -. .. .. .. • - . . ... .. •
6.0 300 305 2506.5 307 307 2507.0 300 300 2507.5 298 307 250
FAAL 7 5.0 302 304 2505.5 305 300 2506.0 298 302 2506.5 297 307 2507.0 297 307 2507.5 300 306 250
-- FAAL 8 5.0 307 299 2505.5 297 307 2506.0 300 303 2506.5 297 301 2507.0 307 305 2507.5 302 304 250
FAAL 9 5.0 303 306 2505.5 307 304 2506.0 302 304 250
0 6.5 305 301 2507.0 300 300 2507.5 303 302 250
FAAL 10 5.0 303 302 2505.5 301 304 2506.0 307 307 2506.5 305 306 2507.0 297 307 2507.5 299 300 250
FAAL 11 5.0 297 305 2505.5 300 298 2506.0 297 305 2506.5 300 307 2507.0 300 301 2507.5 297 302 250
u FAAL 12 5.0 306 307 2505.5 305 305 2506.0 307 304 2506.5 307 305 2507.0 301 299 2457.5 307 306 250
FAAM 1 5.0 306 306 2505.5 307 307 2506.0 304 305 2506.5 303 307 2507.0 297 303 2507.5 307 301 250
112
o0-
FAA 2 5.0 302 299 2505.5 306 297 2506.0 300 300 2506.5 298 307 2507.0 301 305 2507.5 304 306 250
FAA? 3 5.0 302 302 2505.5 303 302 2506.0 297 305 2506.5 300 307 2507.0 307 307 2507.5 301 305 250
FAAM 4 5.0 307 307 2505.5 307 303 2506.0 307 307 2506.5.- 299 305 2507.9 305 305 2507.5 298 306 250
FAAM 5 5.0 300 305 2505.5 303 299 2506.0 300 307 2506.5 307 307 2507.0 299 307 2507.5 307 300 245
FAA1 6 5.0 301 301 2505.5 301 307 2506.0 305 307 2506.5 299 307 2507.0 297 307 2507.5 307 307 245
FAA? 7 5.0 303 300 2505.5 307 302 2506.0 295 307 2506.5 307 305 2507.0 307 306 2507.5 302 307 250
FAA? 8 5.0 307 307 2505.5 300 305 2506.0 302 307 2506.5 297 306 250
7.0 297 302 2507.5 300 305 250
FAA? 9 5.0 302 301 2505.5 307 303 2506.0 299 305 2506.5 305 301 2507.0 299 307 250
113
7.5 300 307 250
FAAM 10 5.0 297 301 250-. 5.5 299 300 250
6.0 307 301 2506.5 297 307 2507.0 307 307 250
7.5 297 305 250
FAAM 11 5.0 307 305 2505.5 307 302 2506.0 303 304 2506.5 304 301 2507.0 297 300 2507.5 307 299 250
FAAM 12 5.0 297 301 2505.5 297 298 2506.0 301 298 2506.5 299 297 2507.0 297 307 2507.5 301 300 250
FAAU 1 5.0 304 305 2505.5 297 307 2506.0 300 303 2506.5 307 301 2507.0 300 298 2507.5 300 307 250
FAAU 2 5.0 307 307 2505.5 307 292 2456.0 305 299 2506.5 305 301 2507.0 303 304 2507.5 300 299 250
FAAU 3 5.0 300 305 2505.5 307 307 2506.0 305 297 250* 6.5 300 307 2457.0 305 300 2507.5 304 304 250
FAAU 4 5.0 307 30S 245
5.5 307 307 250* 6.0 299 305 2506.5 307 304 2507.0 307 305 2507.5 298 306 250
FAAU 5 5.0 307 307 2505.5 307 302 2506.0 307 304 250
114
6.5 305 307 2507.0 306 307 2507.5 305 304 250
7AAU 5.0 305 302 2505.5 306 307 2506.0 297 307 2506.5 300 307 2507.0 300 301 2507.5 300 301 250
FAAU 8 5.0 305 307 2505.5 297 302 2506.0 307 305 2506.5 297 307 2507.0 307 306 2507.5 297 300 250
FAAU 8 5.0 300 304 2505.5 297 307 2506.0 300 305 2506.5 301 307 2507.0 299 301 2507.5 307 305 250
FAAU 1 5.0 305 307 2505.5 307 303 2506.0 305 306 2506.5 301 307 2507.0 305 304 2507.5 301 307 250
FAAU 10 5.0 307 302 2505.5 304 303 2536.0 298 305 2506.5 298 307 2507.0 299 307 2507.5 300 307 250
FAAU 11 5.0 307 303 2505.5 307 307 2506.0 300 301 2506.5 301 300 2507.0 300 307 2507.5 28 304 250
FAAU 12 5.0 305 307 2505.5 298 304 250
'".6.0 297 307 250i 6.5 297 299 250S7.0 305 305 250
I7.5 307 301 250
.. JMTF 1 5.0 300 307 250
iO 11s
5.5 ' 300 307 2506.0 301 297 2506.5 302 307 2507.0 307 301 2507.5 305 302 250
JMF 2 5.0 303 304 2505.5 304 305 2506.0 299 301 2506.5 306 307 2507.0 305 300 2507.5 304 307 250
JMF 3 5.0 307 304 2505.5 307 307 2456.0 304 307 2506.5 305 305 2507.0 297 301 2507.5 307 306 250
JMF 4 5.0 300 305 2505.5 299 307 2506.0 304 307 2506.5 298 307 2507.0 307 302 2507.5 303 305 250
JMF 5 5.0 307 295 2455.5 297 307 2506.0 300 307 2506.5 307 307 2507.0 307 304 2507.5 300 307 250
JMF 6 5.0 297 307 2505.5 300 307 2506.0 302 303 2506.5 305 306 2507.0 298 306 2507.5 298 305 2500
JM"Y 7 5.0 303 301 2505.5 305 307 250
- .6.0 301 307 2506.5 306 302 2507.0 298 301 2507.5 307 304 250
-' JF 8 5.0 307 304 2505.5 306 307 2506.0 307 306 2506.5 305 307 2507.0 307 307 2507.5 307 305 250
116
* - *'.'
JMF 9 5.0 307 307 2505.5 305 307 2506.0 297 307 2506.5 301 307 2507.0 307 302 2507.5 298 307 250
JMF 10 5.0 307 300 2505.5 305 304 250
6.0 307 307 2506.5 306 307 2507.0 300 307 2507.5 307 307 250
3MF 11 5.0 300 301 2505.5 302 305 2506.0 307 307 2506.5 307 307 2507.0 300 302 2507.5 298 307 250
JMF 12 5.0 302 299 2505.5 299 307 2506.0 307 305 2506.5 307 304 2507.0 306 303 2507.5 300 304 250
117
..
Appendix E
Sumnary of Marshall Stability Results
Asphalt Standard Coef. of Min. Max.Gradation Content Mean Deviation Vari. * Value Value Range
(%) (lbs) (lbs) (%) (ibs) (ibs) (ibs)
FAA 5.0 3036 160 5.3 2800 3360 560Upper 5.5 2987 130 4.3 2770 3160 390
6.0 2836 128 4.5 2600 3010 4106.5 2517 90 3.6 2360 2670 3107.0 2165 115 5.3 1880 2280 4007.5 1848 88 4.7 1699 2045 346
FAA 5.0 2805 78 2.8 2720 3000 280Midpoint 5.5 2835 168 5.9 2540 3090 550
6.0 2620 199 7.6 2140 2860 7206.5 2406 132 5.5 2230 2700 4707.0 2136 97 4.5 1990 2240 2507.5 1773 132 7.4 1500 1940 440
FAA 5.0 2405 224 9.3 2130 2890 760Lower 5.5 2523 137 5.4 2270 2680 410
6.0 2505 160 6.4 2290 2810 5206.5 2424 213 8.8 2100 2700 6007.0 2201 139 6.3 2040 2430 3907.5 1905 133 7.0 1730 2120 390
JMF 5.0 2345 140 6.0 2150 2670 5205.5 2420 173 7.1 2141 2780 6396.0 2450 167 6.8 2170 2670 5006.5 2513 132 5.2 2340 2730 3907.0 2360 108 4.6 2179 2550 3717.5 2119 53 2.5 2035 2200 165
• Values for Coefficient of Variation are based on Mean andStandard Deviation Values before rounding for inclusion in Appendix.
118
. - .°
. . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . .
Appendix F
C Summary of Marshall Flow Results
Asphalt Standard Coef. of Min. Max.Gradation Content Mean Deviation Vari. * Value Value Range
M% (1/100"1) (11100") M (1/11o)(1/11oo,)(1/lOO,,
FAA 5.0 9.7 0.42 4.3 8.9 10.5 1.6Upper 5.5 10.4 0.98 9.4 9.2 12.3 3.1
6.0 13.0 1.22 9.3 11.8 15.8 4.06.5 15,7 1.05 6.7 14.5 18.2 3.77.0 18.8 0.89 4.7 17.6 20.5 2.97.5 21.9 1.73 7.9 20.0 25.5 5.5
FAA 5.0 9.5 0.56 5.9 8.7 10.4 1.7* Midpoint 5.5 11.1 0.54 4.8 10.5 12.1 1.6
6.0 12.9 1.10 8.5 12.0 15.6 3.66.5 15.8 1.24 7.8 14.1 18.5 4.47.0 18.5 1.02 5.5 16.8 19.8 3.07.5 21.9 1.39 6.3 20.0 24.1 4.1
FAA S.0 9.4 0.55 5.9 8.6 10.3 1.7Lower 5.5 10.1 0.75 7.5 9.2 11.9 2.7
6.0 11.4 0.71 6.2 10.4 12.5 2.16.5 13.6 1.16 8.6 12.1 15.9 3.87.0 16.8 0.98 6.2 14.3 17.5 3.27.5 18.9 1.18 6.3 17.0 21.4 4.4
JMF 5.0 8.8 Q.43 4.9 8.3 9.6 1.35.5 9.1 0.54 6.0 8.0 9.7 1.76.0 9.6 0.52 5.4 8.8 10.5 1.76.5 10.9 0.50 4.6 10.0 11.7 1.77.0 13.4 0.60 4.5 12.3 14.1 1.87.5 16.5 0.79 4.8 15.1 17.9 2.8
* Values for Coefficient of Variation are based on Mean andStandard Deviation Values before rounding for inclusion in Appendix.
119
0--- -.': . ., . -. - ;- . ,_ . - i -i. , i .: . i : ::: . i .
Appendix G
Summary of Air Voids Results
Asphalt Standard Coef. of Min. Max.Gradation Content Mean Deviation Vari. * Value Value Range
(%) (%) (%) (%) (%) (%) (%)
FAA 5.0 5.3 0.34 6.4 4.81 5.85 1.04Upper 5.5 3.7 0.29 7.9 3.12 4.23 1.11
6.0 2.5 0.21 8.3 1.98 2.69 0.716.5 1.7 0.13 7.3 1.52 1.98 0.467.0 1.3 0.15 11.7 0.97 1.50 0.537.5 1.2 0.12 9.6 0.97 1.37 0.40
O FAA 5.0 5.0 0.31 6.3 4.53 5.42 0.89Midpoint 5.5 3.4 0.28 8.3 3.16 4.12 0.96
6.0 2.3 0.27 11.6 1.97 2.91 0.946.5 1.5 0.20 12.9 1.08 1.85 0.777.0 1.3 0.12 9.5 1.12 1.48 0.367.5 1.1 0.09 8.3 0.95 1.34 0.39
FAA 5.0 5.8 0.71 12.4 4.89 7.64 2.75Lower 5.5 4.1 0.26 6.3 3.60 4.57 0.97
6.0 2.8 0.31 11.4 2.23 3.33 1.106.5 1.9 0.20 10.5 1.63 2.19 0.567.0 1.4 0.08 6.2 1.26 1.52 0.267.5 1.2 0.11 9.0 1.10 1.39 0.29
JMF 5.0 6.9 0.22 3.2 6.48 7.29 0.815.5 5.6 0.34 6.0 5.07 6.16 1.096.0 4.3 0.27 6.1 3.91 4.85 0.94
6.5 3.1 0.25 8.1 2.55 3.37 0.827.0 2.0 0.17 8.4 1.83 2.31 0.487.5 1.6 0.10 6.6 1.42 1.76 0.34
* Values for Coefficient of Variation are based on Mean andStandard Deviation Values before rounding for inclusion in Appendix.
120
FILMED
6-85
* DTIC7.-. 7.* 1. -"A a**P
