Click for next page ( 252


The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 251
CHAPTER 4 SIGNIFICANT GENERAL FINDINGS INTRODUCTION (chanter 2 examined in detail resilient modulus testing of asphalt concrete, and Chapter 3 examined resilient modulus testing of aggregate base and subgrade materials. This chapter reviews the more significant of these findings, introduces new concepts and integrates this information together to explore the overall utilization of resilient modulus testing in pavement design. The broad aspects of resilient modulus testing are examined in this chapter and put into perspective from a practical viewpoint. OPTIMUM RESILIENT MODULUS TESTING SYSTEM For production resilient modulus testing, a completely automated, modern' electro-hydraulic loading and data acquisition system is a necessity to maximize the number of tests performed am to minimize the poter~fi~fortestiingan~data reduction errors. Another important advantage of using a fully automated testing system is that a laboratory technician can be readily trained to reliably perform the test after the system has been calibrated and made operational. The use of a fully automated testing system is also desirable for research applications. A completely automated electro-hydraulic testing system for repeated load biaxial testing is shown in Figure 123. The loading system has been programmed to automatically perform the complete stress sequence required for a resilient modulus test. Data is also automatically collected and saved using an analog to digital data acquisition system. A testing and data acquisition system similar to this one is being used for routine resilient modulus testing by He Alabama DOT. Excellent success including high production output has been reported using their fully automated testing system. By using an automated testing/data acquisition system, approximately 6 to ~ resilient modulus tests, under ideal conditions, can be performed and the data reduced in one day. One person performs the tests while a two person team prepares specimens. lhe use of a fully autom~edt testing system including Ma acquts*i~n and reductilon is considered a key component ire adapting resilient modulus testing for routine use or extensive research applications. Data Acquisition Important advantages of a date acquisition system compared to a strip chart recorder include: (1) load and displacement are measured more accurately and (2) Me readings are automatically stored in digital form which can be readily manipulated including transferring We data to another computer, if required, and automatic data reduction. If a sufficient number of readings are not taken during the application of He load pulse, however, the true peak response is not obtained since the pulse has a very steep slope with the peak being reached in only 50 ms (0.05 sec.) for a 100 ms (0. ~ sec.) load pulse. A minimum of 200 data points per sec. are required to give reliable results. The number of readings collected on the response 251 ,,

OCR for page 251
- - - - ~ ~e L--3 ~. I, .. , , ~ ~ rear ~ ._ - . ~ ' .%' i., _ 1 1 I_ ~ __.. . ~ ' a,,..-.._. _ ~ . . . Clamps Positioned on the Granular Material Sped men __ ~1 at, ~ . ~ _n =~ ~ -, ~ ~ ~ _ . . Am= ~ _ . 1- ~: ~ .~.~A~ ~ ~ ^ ~... .. . : ' 4 ... :,'''.,.,.,: :. At_ ,''. I'. : i,_ -: .. ' :.. ~ ~. : ~. .:: ,, . ,: ~,, - .,,.- - .- - - _ ,:: , ,., ,:.,.-: I'. i,:' ,',' :~ .. . . ,:.... ~ ,.. .. ~.~ . 1~ ~ ~1~4 I ~ L ~^~ ': -A. ;,. ~ icky F3~ai ~9 Data Acquisition System Figure 123. Fully automated closed-lop, electro-hydraulic testing system 252

OCR for page 251
curve depends upon the (1) capability of the data acquisition software used, (2) time of pulse application, (3) capacity to store and handle the data, and (4) overall speed and operation of the analog to digital converter. The data acquisition system should use as a minimum a high quality 12 bit analog to digital (D/A) board to insure accuracy of the data collected. Also, the data acquisition system should have provisions for outputting the collected data in digital form to facilitate data reduction. In general, a sotcware/data acquisition system should be selected that samples each channel of data at the same time so as to obtain a minimum time lapse between data points collected for each channel. Refer to Appendix E for detailed specifications for the data acquisition system. Overall System Reliability. Some indication of the overall accuracy of the rate of data acquisition, analog to digital conversion and testing system can be determined by examining the coefficient of variation and magnitude of the resilient modulus obtained which is dependent on both the load and displacement measurements. Consider, for example,the average coefficient of variation for Me repeated load biaxial test resilient moduli obtained for five 0. ~ sec. pulses for a granular base. Combined external and clamp mounted EVDT data are used in this comparison. The coefficient of variation (CV) for 102 data point readings/sec./channel was CV = I.4%, for 205 readings/sec./channel CV = 0.6% and for 666 readings/sec./channel CV = 1.0% (refer to Table ill, Chapter 3~. For a given method of measurement (external EVDTs or clamp mounted EVDTs), the resilient moduli values for all three data acquisition rates were within ~ S; of each other. The average resilient modulus determined for 660 readings/sec., which should be closer to He true value of load, was consistently the largest by a small amount (0.42% more than for 205 readings/sec.~. Conclusion -- The most consistent data acquisition rate for the system used is 205 readings/sec. Theoretically, better and more consistent values of resilient modulus should be obtained as the sampling rate is increased. However, as He sampling rate is increased more noise is also obtained in the readings. Cross coupling between channels may also cause noise. The effect of cross coupling on noise can be evaluated by reducing He number of channels of output collected to see if peaks change. Filtering He data to remove noise when measuring peak points is dangerous and should, if possible, be avoided. For the system considered in this example, either 205 or 666 readingsIsec. give good results as shown in the next section. Load and Deformation Variation For tile same aggregate bases discussed in the previous section, the variation in 5 consecutive load pulses (CV = 0.67%) was about 3 times the corresponding variation in EVDT readings for both external EVDTs (CV = 0.21 %) and clamp mounted EVDTs (CV = 0.20%~. Since the coefficient of variation in load increases with decreasing load levels, a 5 kip load cell having a sensitivity of no less than 3mV/V, which was the sensitivity of the cell used in the base study, should be the minimum used for unstabilized 6 in. diameter specimen base testing to reduce the overall variation in the test results. Use of a load cell with a sensitivity greater than 3mV/V would more effectively reduce resilient modulus variability Can increasing the sensitivity of the displacement transducers. A 5 kip load cell was used since the capacity equals the maximum load range used of the testing system. Use of a lower capacity, higher voltage output load cell, although desirable, makes it possible to accidentally overload He cell and hence damage it. At He very low load levels, the maximum practical sensitivity is needed of bow He displacement transducers and He load cell. To obtain the most reliable results, He maximum practical 253

OCR for page 251
gain of Me output signal for all of the transducers should be used as well as the maximum possible transducer excitation voltage. To obtain higher transducer sensitivity it is sometimes desirable to apply a transducer voltage in excess of We rated value; check with the manufacturer before doing this. For example, a non-fatigue rated load cell usually has a higher sensitivity than a fatigue rated one having the same rated capacity. One lab has used non-fatigue rated load cells for years without problems as long as the applied load is one-half of the rated capacity. Limited studies, similar to the ones previously described, would be beneficial to conduct for each testing system go establish He optimum data acquisition rates, as well as the accuracy and variation of Be readings obtained from 5 consecutive load pulses. Also, always compare the level of voltage output of the measurement devices used with the [eve! of background noise in the Ma acquisition system. Me output voltage must be sup entry large to insure accurate readings. Adequate transducer sensitivity is often a problem in resilient modulus testing. Data Reduction Large quantities of data are obtained from a full resilient modulus test sequence using a data acquisition system. To minimize the chance for errors, the entire test procedure including load sequencing, data acquisition and data reduction should~ all be automatically integrated together. Ideally, abler a test is set up and started, resilient moduli should be printed out at the completion of the test without the need for operator intervention. In practice, some computer files may have to be saved and moved to another computer and perhaps a few buttons pushed to obtain resilient moduli. A complete, automated testing/da~a acquisition/ data reduction system is necessary to achieve reliable resilient modulus test results, especially when conducted as a routine test by a techn~c~n. TESTING SYSTEM CALIBRATION Accurate, reproducible resilient moduli can not be measured by sewing an inexperienced technician or engineer into the laboratory and having hen or her start running tests ever using a new system. An automated testing system is complicated to set up, and the electronic measurement and data acquisition systems must be thoroughly understood. The entire operation must be verified by quantitative measurements. The system calibration procedure Mat must be followed to measure reliable moduli consists of the following steps: I. Measurement Device Verification. Verify by careful calibration that each individual component of the testing system gives the correct output. For example, check load by statically loading He laboratory reference proving ring and comparing its indicated load with the systems. Individually calibrate displacement transducers by subjecting them to known displacements. 2. Svelter Alignment and Compliance. Measure the alignment and compliance (extraneous deformation) of each part of the loading system and test apparatus. Just because a $100,000 system is new does not necessarily mean parts are aligned and compliance is small. To carefully and thoroughly conduct this step takes several days of concentrated effort. 254

OCR for page 251
Specin~en Proof Tess. To verify the overall accuracy of all components of the testing system, tests must be performed on synthetic reference specimens having known values of resilient moduli. Testing system problems must be identified and corrected until the correct moduli are obtained, without applying empirical adjustments to the data, before routine testing is begun. System c~ib~tion by proof testing synthetic specimens is absolutely essential to insure reliable resilient modulus test rests. If external deformation measurements are made in the repeated load biaxial test, which is not recommended, system compliance must be accounted for in reducing the data. Synthetic specimens are temperature sensitive and hence tests must be performed at the reference temperature or else corrected for temperature difference. Synthetic specimens may also be stress level dependent. Each laboratory should own their own synthetic specimens. These specimens are easy and inexpensive to make or can be purchased ready to test. Synthetic specimens sent from one laboratory to another take a tremendous beating and often become unreliable as a laboratory standard. Wire resistance strain gages about 2 in. In length can be bonded to Me synthetic specimen to obtain accurate, reference strain measurements. Another good approach is to rigidly attach EVDTs to the synthetic specimen. The simplest approach is to measure, using an optical extensometer, deflection over a predetermined gage length on the specimen. Using one of the above measurement methods allows easy periodic verification of the reference resilient modulus of synthetic specimens at any temperature. Me Importance of thorough testing system calibration cannot Be overemphasized. Proof testing using synthetic specimens also serves to train laboratory technicians. Me resilient modulus test should be repeated on the same synthetic specimen at least 5 times and the coefficient of vacation of MR calculated. If the coefficient of variation of the test group is more than about 6% repeat the 5 tests. Calibration procedures are given in Appendix C for asphalt concrete and Appendix D for base and subgrade materials. SET UP AND OPERATION OF NEW TAXIING SYSTEM Laboratories have significant problems in properly setting up and making operational the resilient modulus test including testing equipment, data acquisition apparatus, specimen preparation methodology, specimen set-up and data reduction. Several factors including He sophistication of the testing system electronics and difficulty to visually observe specimen behavior mean a greater level of care is required to obtain meaningful resilient modulus test results Man for most other tests. Before resilient modulus testing is begun, laboratories need to develop a well-planned and carefully supervised program which includes using synthetic specimens for calibration. Some test equipment does not work as advertised and technicians often have too little experience to identify and correct Me source of problems Also, not all laboratories carefully follow calibration and/or testing procedures. A rushed laboratory testing schedule frequently leads to problems. Resilient modulus values should always be calculated at one or more specific stress states and compared to reference values to verify Hey are 255

OCR for page 251
reasonable. The laboratory validation studies showed the test of reasonableness to be all too frequently absent from laboratory test procedures. Typical ranges of resilient moduli, for specific stress conditions and materials, should be tabulated so that technicians and engineers can quickly and easily verify the laboratory test results. R - ommendation. Consideration should be given to obtaining outside help in setting up, calibrating and establishing the resilient modulus test as a routine laboratory procedure. RESILIENT MODULUS TESTING METHODS AND PROCEDURES Resilient modulus testing equipment and procedures are specified to help insure accurate results. A number of different resilient modulus test procedures are available for both the diametral and repeated load biaxial tests. Many of the differences in these procedures involve small changes in stress conditions used and other testing details. The effects on resilient modulus of poor or lack of system calibration and choice of instrumentation far outweigh the influence of most testing details. Asphalt Concrete Based on findings from this study a new protocol for resilient modulus testing of hot mix asphalt concrete was developed and presented in Appendix C. The Protocol has been written by incorporating the- findings of this study into the final version of SHRP P07 Protocol (November I, 1992~. It was decided to rewrite SHRP P07 instead of the existing ASTM D4123 procedure, as the SHRP protocol had already made significant improvements to He ASTM standard. Conclusions. The following general conclusions are made concerning resilient modulus testing of asphalt concrete specimens: I. Resilient modulus decreases when testing is repeated on an axis mutually perpendicular to the axis initially tested. 2. The resilient modulus decreases significantly with increase in temperature. Thus, it is important to run the resilient modulus test at the desired test temperatures. 3. Poisson's ratio is one of the most important parameters influencing the resilient modulus. The variation in MR values due to the testing axis dependency and different lengths of rest periods are almost negligible compared to the magnitude of difference in He MR values from assumed and calculated Poisson's ratios. Poisson ratio should be evaluated using the EXSUM deformation measurement system. A mountable extensometer device, compared to the stand-alone EVDT measurement device, provides less variance and hence better repeatability within He five consecutive cycles used for resilient modulus determination. However, using the SHRP EG device EVDTs gave comparable performance to the mountable extensometer. Mountable deformation measurement devices are recommended for resilient modulus testing because of He smaller variability. 256

OCR for page 251
6 7. 8. 9. The SHRP EG device minimizes rocking of the specimen. The main features of the SHRP EG device are the use of two guide columns, a counterbalance system, an innovative semi-rigid connection between the upper plate and the load actuator, and its sturdiness. The disadvantages are its bulkiness, complication of use, possible inertia from He counter-balance system, friction in He guide columns, and limitation of the size of He sample Hat can be tested. The concept behind the use of EVDTs mounted along a small gage length (l in.) on the surface of the specimens as in the Gaze-Point-Mounted setup. is sound. The main drawbacks for its use . . . . . _ in repetitive testing are its heavy dependence on the alignment and homogeneity of specimens. The gage length of ~ in. seems to be too small for reasonable results with the asphalt concrete specimens used in this study. The proposed measurement system, the EXSUM setup, provides a promising measurement memos for determination of consistent and reasonable Poisson's ratios. At 41F, however, increase in variability occurs due to misalignment and rocking. Use of the SHRP EG device, or its modification, together with He EXSUM setup could ensure reasonable values even at low temperatures. The use of the EXSUM setup requires an increase in testing time compared to conventional measurement systems because of the significant time required for mounting the EVDT on the specimen. For research applications, improved reliability can be obtained by mounting an EVDT on both He front and back surfaces of He specimens. Corrections for bulging and non-uniform stress distribution using finite element analyses could make the analysis more relevant. A square load pulse produces significant specimen damage and smaller resilient moduli compared to a haversine pulse. The haversine pulse also better simulates the field loading condition than a square pulse. As a result the haversine load pulse is recommended for resilient modulus testing. The loading time significantly affects the MR values. A loading time of 0.2 sec. considerably reduces MR, and produces more damage as compared to a shorter loading time of 0.05 sec. A shorter loading time of 0.05 sec. is representative of high vehicle speeds, but is hard to accurately apply and monitor. Also, accurate load control at higher temperatures is difficult using very short loading times. The usually used loading time of 0. ~ sec., represents slow traffic conditions Hat cause significant damage to He pavement and should be continued to be used for resilient modulus testing. 10. Rest period to loading period ratios of 4, 9, 19, 24, and 29 used in the study did not make a significant difference in the resilient moduli. Also a rest period to loading period ratio greater ~- ~an ~ has been shown to generate no significant beneficial effect by past research. A rest period to loading time ratio of 9 gives a rest period of 0.9 second and a loading frequency of ~ Hz. This is the loading condition specified by SHRP P07 and a change in it is not justified. Three levels of preconditioning were studied. There was no significant difference in the variation of resilient moduli and Poisson's ratio between five cycles for the selected preconditioning levels, 2 and 3. However, MR values did decrease with increasing number of preconditioning cycles. 100 preconditioning cycles are recommended at 41 and 77F and 50 cycles at 104F. 257

OCR for page 251
A significant difference exists between resilient moduli and Poisson's ratio values computed using the SHRP P07 analysis and the elastic analysis which is similar to the ASTM analysis. The SHRP ~ ~ , e ~ ~ ~ , , ~ ~ , analysls gives nigher values when an assumed Poisson s ratio is used as compared lo tne elastic analysis with an assumed Poisson's ratio. 13. 14. 15. 16. The 4 in. diameter specimen is acceptable for testing medium gradation mixes but a 6 in. diameter specimen should be used to test coarse gradation mixes (mixes with more proportion of coarse aggregate or mixes wig large aggregate such as base courses or large-stone mixes). Grain size distributions for medium and coarse gradations are given in Appendix B. Table Bet. SHRP protocol P07 recommended load amplitudes are suitable for testing at 41F and 77F, but at 104F, a smaller load should be used. Load levels corresponding to 30, 15, and 4 percent of the indirect tensile strength at 77F are recommended for testing at 41F, 77F, and 104F, respectively. The relatively large seating loads recommended by the SHRP P07 protocol may not be necessary as high seating loads seem to damage He specimen at higher temperatures. Instead, 5, 4, and 4 percent of He total load are recommended at 41 F, 77F, and 104F, respectively. However, at 104F, a minimum load of 5 Ibs. must be maintained to avoid possibility of separation of the loading strip from the sample surface. The maximum seating load should not exceed 20 Ibs to ensure minimum damage to the specimen. The following configuration of test apparatus is recommended for use in resilient modulus testing: Load Device: A device comparable to the SHRP EG device, possibly with the following modifications: I. Reduction of the upper plate weight using high strength, light weight materials and thus elimination of the counterbalance weights, 2. Reduction of the size of the device so that it can be easily used in commonly available environmental chambers, and 3. Capability for the testing of 6 inch diameter specimens. The MTS deformation measurement device was used for the final phase of the testing program mainly due to time and budget constraints. Although the control of rocking would be a little inferior to He recommended device, the testing device gives comparable results especially as extensometers are to be used for measurement of horizontal deformation. Also, the testing device can be used in a typical environmental chamber. Measurement System: The EXSUM setup is recommended for use. However, a faster curing glue with non-sagging properties is required to reduce the time required for testing. Also, in- depth finite-element analyses might be required to make corrections for bulging and non-uniform stress distributions. The capability to use two mounted EVDTs, one each at the front and back face of the specimen, might make results more trustworthy. Although accurate and convenient, 258

OCR for page 251
extensometers are expensive, and a cheaper mountable measurement system wig comparable accuracy should be developed. Base and Subgrade Materials The repeated load biaxial test is recommended to evaluate the resilient modulus of base, subbase and granular subgrade materials. A repeated load test performed on an unconfined specimen is recommended for cohesive subgrade soils. Test procedures are given in Appendix E. The round-robin tests (Appendix H) show for base and subgrade materials that very large variations in MR values were observed between labs when axial deformation is measured outside of the biaxial cell. Therefore, the recommendation is made to make area deformation measurements inside the ceil. An optical extensometer, non-contact proximity gages and EVDTs mounted on clamps can all be used as previously discussed in Chapter 3. The use of an inside defo~ation measurement system neither eliminates or reduces the new for minding system compliance (i.e., extraneous deformation in the loading and testing system) or measuring and correcting test apparatus alignment (Appendix D). Comparison of Test Procedures. A general comparison between the proposed base and subgrade resilient modulus test procedures and those of AASHTO [! 12] and the AASHTO version of SHRP [! 13] are given in Table 58 and 59, respectively. The complete proposed test procedures are given in Appendix lo. Major Issues - Dee foRowzag major msdient modulus test issues completely overshadow other test details which usually have relay rely minor influence on the measured resilient modulus: (~) fully automated loading and Ma acquisition system, (2) accurate measurement of arid deformation, (3) cohesive specunen aging, (47 envuo1unen~ly induced changes in MR and {5) soil structure of compacted cohesive specimens. Failure to properly account for any of the above major factors can easily led to errors of 30 to 100% or more in the measured resilient modulus. Test System - An automated, closM loop electro-hydraulic testing system including automated data acquisition and data reduction is specified in the proposed procedure to make He test practical. This requirement leads to higher productivity while at the same time minimizing the chances for testing and data reduction human error. Although the SHRP test memos given in AASHTO TP46 [! 13] requires a closed loop, electro-hydraulic system, a fully automated test is not specified. The AASHTO T292-911 procedure does not specify either loading or data acquisition systems. Also, He proposed procedure requires and describes a detailed equipment calibration procedure which is not true for ache other two procedures which to varying degrees just mention calibration. Axial Deformation of Granular Material ~ During actual testing, the most critical feature of the resilient modulus test itself is the accurate measurement of specimen axial deformation. The proposed test method requires for granular materials Hat axial deformation be measured directly on the specimen using any one of Tree methods given in Tables 58 and 59. Displacement measurement on the specimen minimizes the very serious problem of test system compliance (i.e., extraneous deformations). The SHRP TP46 procedure [~13] specifies measurement of deformation outside the biaxial cell. Although AASHTO T292-911 [~121 requires deformation measurement using clamp mounted EVDTs on the specimen, no provision is given for also allowing He use of either an optical extensometer or non-contact proximity type gages which are bow permitted in He proposed procedure. 259

OCR for page 251
Table 58 . Comparison of selected AASHTO, SHRP and proposed resilient modulus test requirements for aggregate base TEST DETAIL AASHT0 St ~PIX)POS~ T292-911 TR46 (ED.1) GENERAL TESTING SYSTEM not "pecitied electro-hydraulic fully-automated, electro-hydraulic DATA ACQUISITION strip chart or not specif fed AID Data computer Acquisition Sy$. . TYPE TEST repeated load biaxial repeated load repeated load (R.L.) GRAN ULAR triaxia I trioxia I COHESIVE same (see note 2) same (see note 2) R.L. unconfined LOAD PULSE SHAPE haversine, rect., haver$ine haversine triangular TIME ~variable (see note 1) 0.1 ~0.1 FREQUENCY (~1 ~0.33-1 LOAD CELL LOCATION insideloutside - Extemal (implied) inside AGGREGATE BASE . _ COMPACTION METHOD vib.; impact kneading, static vib ratory vib ratory CONDITIONING (psi) cs3=20, od=1~; a3=15; ad=1 3.5 03=15; <5d=15 N=1000 reps. N=500-1000 reps. N=200-1000 reps. AXIAL DEF. inte m al clamps; 2 external LVDTs intern al clamps, non MEASUREMENT 2 LYDTs contact, or optical STRESS STATUS (pai) c,3=20,15,10,5,2; 03=3,5,10,15,20 <73=3,4.5,6,1 4 18 oh's ~ N=50 reps. 15 oh's ~ N=100 15 oh's ~ N=50 PERMANENT rapid shear rapid shear; or DEFORMATION not considered as approx. repeated load perrn. t6Gt: ~=5O,COO raps. COMMENTS PROPOSED METHOD Fully automated testy ate acquisition/ reduction greatly speeds test, reduces trances for errors. Unconfined test for cohesive soils is much simpler, saves time; easy to measure optically asocial strain. Need to use one pulse approx. puree shape for consistency. Use one valve for ease of test, reduce error; empirically correct for pulse time, it desired. Inside location eliminates friction on piston. Use one method for consistency; vib. slightly better than impact. N=200 adequate for good base Critical to measure def. on specimen; optical best. Stress states not critical for granular mat. Perm. def. measure is critical; repeated load test best. NOTES: 1. PULSE TIME DEPENDS UPON VEHICLE SPEED AND DEPTH BELOW SURFACE. 2. SESAMES INDICATES THE SAME AS GIVEN ABOVE IN THE SAME COLUMN. 260

OCR for page 251
Table 59. Comparison of selected ~SHTO, SHRP and proposed resilient modulus test requirements for subgrade soib: ~ ~ PSI TEST DETAIL "SHTO S H R P PIX)~ED _ T292-911 TP" (ED.1) _ COHESIVE AND GR4M)LAR SUElGRADE __ _ coMPACnOY hIETHOD ~HESNE |veriable (~e note 2) | static | v~r~bl~ (am note 2) | C~ANW ~vib; impact vibratory vibratory knasding;static CON DIT ONING (PSq COHESIVE 1000 N ~ 0,-3, 501~ N ~ m=6, 2 - N. a' =0, ~s6 (A ~s3.6 off ~ . C;RANUL4R 1000 N ~ o3 = 1S "me (~e noto 5) soo-1000 N ~ all-6, ~-1 2 . c';~8 AXIAL DEF. IdEASURIAENT _~_ ~ _ _ ~, . ~Intcrnal optical, non. COHESIVE internal c'.mp.(~), 2 xtcrnal LVOTe conted, cbmpe; top to 2 LVDTe bottom LVDTI non contact. GRANUUR "~ (~e note 5) "n. (~e note 5) Internal optical, non~ contact, clempc _ _ . STRESS STATES (PSI) Repr - entatives CIj; ~-6,~,2; t5 ~'. Unconfined (~0); Ca~VE ~3,5,~,10,15 ~ t4= 100 "ch ts. = 2,4~6,E, 10 ~N=SO each ~ t. = 50 GRANlJLAR c~=15,10,5,2; ~me (~ note S) o~2,3,~,6 . 1J' ~'s ~ Na 50 12 o,.s ~ N = 50 ~ ~_ _ ~not consider" Rap~d Sh~r T - t Rep~d Sh~r Tod . CO~TS PROPOSED KT~ hlothod u~d has brge aneet on coil tructure nd hIR vib. alightly teen" than impact; be consi~ent. - , Yery simpic t~t it u" optical or top~ottom ~. wiem; relhLb. Critkal to measure dd. on soocimen: optical b - t NOTES: 1. S~E "~ TA~E F" AG - ~AW 8~ F" C - P"~ ~ GENE~ UST 2. COMPAC~ - ~ - ""N" U~ "PEC~D ~ ~RE C~ - AT n" ~ C - PAC~ ANC LONG~; "~O "LO - STA=, IMPACt, - ~G; ~D ~ - ~ STA~ - D IMPACT. 3. USE <5~1= 3 pei If SNEAR STRENGTI1 S 1000 pef. 4. US EXTERtlAL LYDTs ON P~TON FOR SOfT SOILS; CALIBRAtE FOR SYSTE.I CONIPLIANCE. 5. S"IE- INDK:ArES THE S - E AS Gn,EN "OVE IN THE SA~E COLU~N. 261

OCR for page 251
40,000 psi. For He design parameters given on the figure, experimental error in measuring the resilient moduli of each layer causes about a 12 to 14% increase of the total equivalent aggregate pavement Sickness. Experimental error In measuring the resilient modulus of the subgrade, assuming no other MR testing error Is present, accounts for only about 20% of the tote] possible effect of resilient modulus testing error. In determining the effect of error in measuring the resilient modulus of specific layers, the measurement Of MR for the other layers is assumed to be perfect. Now consider the behavior of a pavement having a relatively den 6 in. A.C. surfacing with a weak base. For a weak base modelled by a resilient modulus of 20,000 psi, He pavements exhibits an ~ to ~ ~ % increase in He equivalent base Sickness compared to Hat required in the absence Of MR testing errors. Experimental error In measuring the value Of MR for a weak base accounts for 77 to SS% of the possible combined effect of all MR testing errors. In contrast, variation in the resilient modulus of the strong base due to experimental error accounts for only 25 to 60% of the combined effect as the subgrade resilient modulus decreases from 10,000 psi to 2000 psi, respectively. AASHTO Reliability Analyses Figure 127 shows the effect on total equivalent base thickness of experimental error in measuring resilient moduTi using He AASHTO type reliability analysis. Reliability levels of 50, 85 and 98% are shown for pavements having a moderate strength base (MR = 30,000 psi). This pavement is loaded with 6x106 ESALs. The general trends shown for the AASHTO reliability analyses are similar to those obtained for the Monte CarIo method. The same coefficients of variation of MR for each layer were used in all analyses. Change in thickness determined by the AASHTO type reliability analysis due to resilient modulus testing errors, however, are about one-half of the ~ to 14% increase in total equivalent base thickness (i.e., no A.C.) typically predicted using the Monte CarIo analysis. The AASHTO reliability method, which is practical to perform, was developed for a single set of conditions as Indicated by taking the partial derivative of pavement behavior at a point. In contrast, the Monte CarIo method uses the statistically developed AASHTO equation which should be reasonably valid over a wider range of variables, and should show trends better than the AASHTO method. Relationships Between MR and Structural Coefficients The relationships used in this analysis to express the structural coefficient as a function of resilient modulus for the A.C. surfacing and base are shown in Figures 128 and 129. These relationships are almost the same as given in the 1986 AASHTO Guide [1141. Using the Monte CarIo method results in the use of a reasonably wide variation in resilient moduli. Therefore, He AASHTO curves were slightly modified to give more realistic relationships between He structural coefficient and resilient moduli at its extreme values. Reliability Design Implications Table 60 gives the approximate relative importance of obtaining a reliable resilient modulus measurement for the asphalt concrete surfacing, aggregate base and subgrade. This table was developed from the results of the Monte CarIo reliability study using the AASHTO 1986 Guide for thickness design. Generalizations of these results to design are summarized as follows: 267

OCR for page 251
23 ~ me c, ~2 Z Z ~ Y I C' :: , ~ - , ..! - u cn m 21 ~9 17 15 13 11 7 - _ DESIGN VARIABLES 6, - 0,000 ESALe ~ IN. A.C. SURFACE A.C. OR ~ ~'~ BASE FIR - APS1 ~ 2 R R~81L~ - EFFECT OF LAB SIR ,48~BLm -' . ~ V ` -A 7 , r R ~ 98% R 8S% 1 . ~ 6 8 t0 12 14 16 R 50Ye 0 2 4 SUBGRADE RESILIENT MODULUS (KSI) Figure 127. Influence of lab variability on base thickness for AASHTO reliability analysis: Base MR = 3D,000 psi 268

OCR for page 251
:c hi: J :' _ 0.5 ~,~ 0.4 a:~ Hi: o C) 0.6 0.3 0.2 0.1 it. - _ _= - z / r 0.0 ~ 200 400 600 800 1000 ASPHALT CONCRETE RESILIENT MODULUS AT 68F, MR (KSI) Figure 128. Variation of asphalt concrete structural coefficient with resilient modulus used in reliability analysis 269 l

OCR for page 251
In ~ - :E 0 ~ to - J oO 20000 ce In m / 42 ~ 0-~49 LOGlo(Ess). o.9n 7 ./ J ~,-~- ~ -"I; 42 ~ 4.~361~MR n ~R "~ PSI O- ' 1 - 1 0.0 0.1 0.2 BASE STRUCTURE COEFFIClE~, A2 Figure 129. Variation of base structural coefficient with resilient modulus used in reliability analysis Table 60. Approximate relative effect on pavement thickness of resilient modulus test variabiliny of each layer (l,2) 0.3 LAYEF Condition Comment AC Base . ~. Full Depth AC 6 0 6 in. AC Surface Strong Subgrade 5.5 2.5 and Strong Deep Base Weak Subgrade 2.5 6 6 in. AC Surface Strong and or 0.3 8.3 Weak Deep Base Weak Subgrade Subgrade 4 2 1.5 2 Notes. 1. NOTATION: Maximum Importance = 10 No Importance = 0 2. AASHTO 1986 Design Guide Used for Pavement Thickness Design 270

OCR for page 251
Subgrade. The reliability of subgrade resilient modulus evaluation has less effect on pavement Sickness than the reliability of the MR measurement for the asphalt concrete and for the M R of thick aggregate bases. This finding suggests that as a practical alternative the subgrade resilient modulus could be evaluated from empirical relations. Such relationships give resilient modulus as a function of easy to measure variables such as density, water content, Atterberg limits, etc. The practical concept of use of empirical relationships for resilient modulus is discussed in the next section. Base. The reliability of the aggregate base resilient modulus evaluation is more important than for Be subgrade when the base is thick. Base evaluation is also more important than for the subgrade for thin to moderately thick A.C. surfacing and (~) strong, deep bases on a weak subgrade or (2) weak, deep bases on either strong or weak subgrades. These results indicate in areas where marginal base materials of varying characteristics are common, resilient modulus evaluation of base materials should be carried out on a routine basis. Asphalt Surfacing. The reliability of the asphalt concrete resilient modulus evaluation is most important for full depth or deep strength pavements with strong bases and subgrades. EMPIRICAL RESILIENT MODULUS RELATIONSHIPS Resilient modulus testing at the level of sophistication needed to obtain satisfactory results is probably, for most laboratories, more suitable for a research project Can for routine production type testing. A very attractive approach for obtaining resilient moduli for use in design, for at least most agencies, is to calculate values using generalized empirical relationships. Such relationships give resilient modulus as a fimction of statistically relevant, easy to measure physical properties of He material such as percent compaction, moisture content, etc. These relationships for resilient modulus can be established Trough carefully designed and conducted research projects for each class of material: asphalt concrete, aggregate base and subgrade soils. Statistically based equations, graphs or charts would then be develops for each class of materials for the range of properties routinely used in design within He region of interest. Empirical Equation Justification The environmental pretty cycle causes variations in resilient moduli with time as large as one order of magnitude as observed in this study. Also, back calculated resilient moduli of the base from FWD field tests have been found to have a coefficient of variation (CV) of 23% for a carefully controlled test section [! 151. Depending upon the time of measurement, CVs have been observed varying from 26 to 45% for 3300 ft. sections of roadway [! 161. Also, both Monte CarIo and AASHTO type reliability analyses, using He 1986 AASHTO Design Guide, indicate modest errors in evaluating resilient moduli as large as 10 to 15% of He mean MR value have in general a relatively small effect on He overall required pavement thickness. Additional variability is introduced due to varying materials properties during construction, traffic variability and variability in the design equations. Because of all these factors, empirical correlations between resilient modulus and easy to measure properties can be used in pavement design considering the large expected overall variation in resilient modulus as material properties and environmental conditions change with both time and location. This finding is in agreement with the conclusion of Thompson [851. 271

OCR for page 251
Asphalt Concrete Quite useful generalized resilient modulus relationships for asphalt concrete, for example, were developed by Miller, et al. [151 as discussed in of Chapter 2. The resilient modulus prediction equations presented by Miller, et al. were determined from cyclic biaxial compression testing. Resilient moduli of asphalt concrete determined from bending [! 17] and also from dimetrial tests (perhaps to a lessor degrees are usually smaller Han obtained from biaxial compression testing. The tensile stress caused by the bending and diametral tests is more severe of a condition than biaxial compression which helps to explain the difference in resilient moduli observed between tests. The upper portion of the asphalt concrete pavement layer beneath a wheel loading is subjected to compression and the lower portion to tension. As a result, probably the correct value of resilient modulus for use in design falls somewhere between the two extreme types of tests. The Miller et al. empirical correlation approach, which has been programmed for the P.C., offers an excellent siting point for developing design relationships for asphalt concrete for a particular region. Resilient modulus testing of the specific types of mixes (including aggregate typed used in the region is needed to modify the approach as necessary for local conditions. Aggregate Base The resilient modulus of an aggregate base is strongly dependent upon the state of stress to which an element of material is subjected. The resilient modulus is also affected to a much less degree by Me following additional factors given in approximately decreasing order of importance: (~) degree of saturation, (2~' aggregate size, (3) angularity, (4) density, and (5) surface roughness. For best performance a minimum of 100% of AASHTO TI8O density should always be specified. Agency specifications also control, within reasonably tight limits, aggregate size and grading. Variation in moisture content from the optimum value can be considered using moisture sensitivity factors analogous to those discussed in the next section. Therefore, for practical design application only a limited number of independent variables influencing resilient modulus need to be considered by most agencies. As a result, a table of design resilient moduli can be prepared based on laboratory testing of the base and subbase materials used by a particular agency. In areas where marginal materials of variable quality commonly are encountered, caution must be exercised including performing periodically resilient modulus tests. Marginal materials include materials with more than ~ to 9% fines, materials that contain plastic fines or Hose Hat degrade significantly upon compaction. Cohesive Subgrades The resilient moduli of cohesive subgrades are well suited for developing empirical correlations with basic physical parameters. The resilient moduli of cohesive subgrades vary from 2000 psi to 30,000 psi or more. Hence, as found in the study of the effect of variability on pavement Slickness, a modest variation of up to 10 or 15% of the true resilient modulus has, for most conditions, only a small effect on the overall design thickness. 272

OCR for page 251
I Em pirical correlations for cohesive subgrades have been developed for use in design by Thompson [85] In Illinois' Woolsturrn [118] in Nebraska, Li and Selig [57, 110] in Massachusetts, Pezo and Hudson [119] In Texas, Sandla [120] in Georgia, Drumm et al. [121] in Tennessee, Elliott et al. [122] in Arkansas and Burczyk et al. [123] in Wyoming. The most general approaches for estimating resilient moduli are by Li and Selig [57, 110] and the approach developed during this study. Both of these very useful approaches are summarized In Chapter 3. The other approaches referenced above generally use statistics to relate the resilient modulus measured in the repeated load biaxial test to simple to measure physical parameters. Figure 130 shows the reasonably good correlation found between actual and predicted resilient moduli(MR) using Me universal type equation of Uzan Table IS, Chapter 3) for selected cohesive soils found in Georgia [1201. FiDy percent of the estimate values of resilient modulus (MR) were within i7% of He measured values while 90% of the estimated values were within + 18%. Moisture Sensitivity Evaluating the moisture sensitivity of base, subbase and subgrade soils is an important aspect of design. Moisture sensitivity is the change in resilient modulus (or permanent deformation) caused by a change in moisture content of the material. Repeated load biaxial tests to evaluate He effects of the environmental moisture cycle, as previously discussed, can be performed to evaluate moisture sensitivity. For cohesive subgrade soils, Figure 131 and Table 61 illustrate two practical approaches used for estimating the relative effects of moisture sensitivity on resilient modulus. The empirical correction, such as given in Figure 131 and Table 6l, of a reference resilient modulus for varying moisture conditions offers a practical way of handling moisture sensitivity. PERMANENT DEFORMATION The resilient modulus of pavement materials has received considerable publicity in recent years because of its introduction in the 1986 AASHTO Design Guide. Me evaluation of permanent deformation charactenstics of the asphalt concrete, base and s~ilgrade materials is just as unporta~ as the KS ~= am hence saw Bier be fo~offen nor neglected. Tnd~, the evaluation of the permanent deformation characteristics of He base is usually even more important than the evaluation of resilient modulus. As previously discussed in of Chapter 3, the permanent deformation characteristics of the base and subgrade can be readily determined as an extension of the proposed resilient modulus test. This test is conducted using a repeated load test apparatus. The permanent deformation behavior of asphalt concrete can be evaluated using a loaded wheel tester or else as an extension of the repeated load diametral test. A loaded wheel tester could perhaps also be used to evaluate the permanent deformation behavior of aggregate base and subgrade materials although this has apparently not been done in He past. RESILIENT MODULI FOR DESIGN Introduction . . . . . . . The AASHTO design approach. as well as presently used methods based on linear elastic theory, employ a single resilient modulus value for each layer. I-he values of resilient moduli selected for each layer of a pavement system can have a strong influence on the required pavement thickness. The single 273

OCR for page 251
- - - I IEE~ F~ lilE P~ ~ ~L ~vs. E~DI~ ~ A ~ 1 ~S, B ~ 2 O3S, El<:. 1 S=:L tfiED 1N ~: ~ E~IY IS ~1 ~L 1 1 ~1 1 25400 + '* + 1 / 1 srn + ,' i 1 ~1 2~0 ~A' I A ,~MA I 17500 + AA'' AB Ak + I A A ~A I 15000 + B AA CYY ~A I A B Aj3 OCR for page 251
4 ~ 3 A (3 2 ~ 1 a , , , . - . 1 ., ~ F ~ . \ 1 a\ . ~ . , . 1 , ~ . ~ . 10 15 20 Moisture Content, w (%) 25 30 Figure 131. Moisture sensitivity corrections for resilient modulus for cohesive Texas soils (after reference 119) Table 61 . Moisture sensitivity correction factors for selected Illinois soils (after reference 85 and 111) USDA Textural MR Decrease / 1% Classification Moisture Increase Nisi / %) clay, silty clay, silty clay learn 0.7 silt loam I.5 10am 2.! 275

OCR for page 251
resilient modulus required for design has to be evaluated from laboratory test results at one reresentative stress state. In laboratory testing, however, the use of a sequence of stress states is quite desirable to insure reliable results. The resilient modulus for all pavement layers is a sensitive function of stress state to which the material is subjected. Little, if any, guidance, however, is usually given in design methods concerning He stress state to use in determining the resilient modulus for a particular pavement geometry and wheel loading. This very important aspect needs to be given more consideration in the future. For empirical approaches, such as He present AASHTO memos, the stress states to use~should be clearly defined by the memos if correct results are to be obtained; this is not done in the AASHTO memos. Use of He correct stress state in selecting the resilient modulus, even if it is representative of the material, does not insure this resilient modulus is He appropriate one to use in empirical approaches such as the AASHTO memos. Vertical stress and confining pressure acting on a representative element of material, as discussed previously in Chapter I, are the sum of the effects of the (~) wheel loading, (2) weight of material above the element and (3) locked in residual stresses. . The representative stress state in the subgrade, base and subbase varies wig several factors including the thickness of He layers and magnitude of wheel loadings. Layered theory should, therefore, be used to develop generalized tables or charts for estimating the stress state to use in selecting design resilient moduli for each layer. Development of these tables or charts must be carried out as an integral part of He development of He design procedure. Base and Subgrade Proposed stress states at which the resilient modulus of an aggregate base can be evaluated for are Riven in Chapter 3. The stress states given take into relative comparisons of different materials consideration nonlinear material behavior and, at least approximately, the effects of residual horizontal stresses. The total vertical stress on He substrate in general decreases with increasing depth until material ~. ~' -- -A weight becomes more important than the applied wheel loading. The vertical stress on He top of He subgrade typically varies from about 3 or 4 psi for thick flexible pavement sections to about ~ to 10 psi for Win sections. The confining pressure generally increases wig depth due to the weight of material above. A representative confining pressure typically varies from about 2 to 4 psi for He subgrade. Asphalt Concrete An asphalt concrete surfacing or base mix suitable for use in a pavement should have a resilient modules of at least 450,000 psi when measured in the diametral test at 68F. Using He relationship between resilient modulus and structural coefficient given in He 1986 AASHTO Guide (refer to Figure 128), a mix wig a 450,000 psi resilient modules has a spectral coefficient of 0.45. Frequently, however, agencies use a maximum structural coefficient of about 0.45 in design. T/u~refore, if a limiting value of 0.45 is used in design, hide justification emus for performing a diametral resilient modulus test on high qualify minces following the 1986 AASHTO design procedure. These findings indicate the design procedure needs to be carefully reexamined. This problem hopefully will be corrected by the year 2002 lo. .. __ r A ~ ~ ~ ~ ~ ~ after the tHWA provides a mechanistic based pavement design procedure. 276

OCR for page 251
AASHTO Design Guide Incorporation of reliability considerations and resilient modules concepts into the 1986 AASHTO Design Guide [l 14] is certainly an important step forward. One of He weakest links in He design process, however, is how He AASHTO approach utilizes the resilient modulus in flexible pavement design. The resilient modules concept was added during He 1986 revision, and it does not appear to mesh very well with the overall design process. Caution should be exercised when it is used. Also permanent deformation considerations need to be incorporated into the method. . 277