National Academies Press: OpenBook
« Previous: Chapter 1 - Introduction and Research Approach
Page 14
Suggested Citation:"Chapter 2 - State of Practice." National Academies of Sciences, Engineering, and Medicine. 2005. Aggregate Properties and the Performance of Superpave-Designed Hot-Mix Asphalt. Washington, DC: The National Academies Press. doi: 10.17226/13844.
×
Page 14
Page 15
Suggested Citation:"Chapter 2 - State of Practice." National Academies of Sciences, Engineering, and Medicine. 2005. Aggregate Properties and the Performance of Superpave-Designed Hot-Mix Asphalt. Washington, DC: The National Academies Press. doi: 10.17226/13844.
×
Page 15
Page 16
Suggested Citation:"Chapter 2 - State of Practice." National Academies of Sciences, Engineering, and Medicine. 2005. Aggregate Properties and the Performance of Superpave-Designed Hot-Mix Asphalt. Washington, DC: The National Academies Press. doi: 10.17226/13844.
×
Page 16
Page 17
Suggested Citation:"Chapter 2 - State of Practice." National Academies of Sciences, Engineering, and Medicine. 2005. Aggregate Properties and the Performance of Superpave-Designed Hot-Mix Asphalt. Washington, DC: The National Academies Press. doi: 10.17226/13844.
×
Page 17
Page 18
Suggested Citation:"Chapter 2 - State of Practice." National Academies of Sciences, Engineering, and Medicine. 2005. Aggregate Properties and the Performance of Superpave-Designed Hot-Mix Asphalt. Washington, DC: The National Academies Press. doi: 10.17226/13844.
×
Page 18
Page 19
Suggested Citation:"Chapter 2 - State of Practice." National Academies of Sciences, Engineering, and Medicine. 2005. Aggregate Properties and the Performance of Superpave-Designed Hot-Mix Asphalt. Washington, DC: The National Academies Press. doi: 10.17226/13844.
×
Page 19
Page 20
Suggested Citation:"Chapter 2 - State of Practice." National Academies of Sciences, Engineering, and Medicine. 2005. Aggregate Properties and the Performance of Superpave-Designed Hot-Mix Asphalt. Washington, DC: The National Academies Press. doi: 10.17226/13844.
×
Page 20
Page 21
Suggested Citation:"Chapter 2 - State of Practice." National Academies of Sciences, Engineering, and Medicine. 2005. Aggregate Properties and the Performance of Superpave-Designed Hot-Mix Asphalt. Washington, DC: The National Academies Press. doi: 10.17226/13844.
×
Page 21
Page 22
Suggested Citation:"Chapter 2 - State of Practice." National Academies of Sciences, Engineering, and Medicine. 2005. Aggregate Properties and the Performance of Superpave-Designed Hot-Mix Asphalt. Washington, DC: The National Academies Press. doi: 10.17226/13844.
×
Page 22
Page 23
Suggested Citation:"Chapter 2 - State of Practice." National Academies of Sciences, Engineering, and Medicine. 2005. Aggregate Properties and the Performance of Superpave-Designed Hot-Mix Asphalt. Washington, DC: The National Academies Press. doi: 10.17226/13844.
×
Page 23
Page 24
Suggested Citation:"Chapter 2 - State of Practice." National Academies of Sciences, Engineering, and Medicine. 2005. Aggregate Properties and the Performance of Superpave-Designed Hot-Mix Asphalt. Washington, DC: The National Academies Press. doi: 10.17226/13844.
×
Page 24
Page 25
Suggested Citation:"Chapter 2 - State of Practice." National Academies of Sciences, Engineering, and Medicine. 2005. Aggregate Properties and the Performance of Superpave-Designed Hot-Mix Asphalt. Washington, DC: The National Academies Press. doi: 10.17226/13844.
×
Page 25
Page 26
Suggested Citation:"Chapter 2 - State of Practice." National Academies of Sciences, Engineering, and Medicine. 2005. Aggregate Properties and the Performance of Superpave-Designed Hot-Mix Asphalt. Washington, DC: The National Academies Press. doi: 10.17226/13844.
×
Page 26
Page 27
Suggested Citation:"Chapter 2 - State of Practice." National Academies of Sciences, Engineering, and Medicine. 2005. Aggregate Properties and the Performance of Superpave-Designed Hot-Mix Asphalt. Washington, DC: The National Academies Press. doi: 10.17226/13844.
×
Page 27
Page 28
Suggested Citation:"Chapter 2 - State of Practice." National Academies of Sciences, Engineering, and Medicine. 2005. Aggregate Properties and the Performance of Superpave-Designed Hot-Mix Asphalt. Washington, DC: The National Academies Press. doi: 10.17226/13844.
×
Page 28
Page 29
Suggested Citation:"Chapter 2 - State of Practice." National Academies of Sciences, Engineering, and Medicine. 2005. Aggregate Properties and the Performance of Superpave-Designed Hot-Mix Asphalt. Washington, DC: The National Academies Press. doi: 10.17226/13844.
×
Page 29
Page 30
Suggested Citation:"Chapter 2 - State of Practice." National Academies of Sciences, Engineering, and Medicine. 2005. Aggregate Properties and the Performance of Superpave-Designed Hot-Mix Asphalt. Washington, DC: The National Academies Press. doi: 10.17226/13844.
×
Page 30
Page 31
Suggested Citation:"Chapter 2 - State of Practice." National Academies of Sciences, Engineering, and Medicine. 2005. Aggregate Properties and the Performance of Superpave-Designed Hot-Mix Asphalt. Washington, DC: The National Academies Press. doi: 10.17226/13844.
×
Page 31
Page 32
Suggested Citation:"Chapter 2 - State of Practice." National Academies of Sciences, Engineering, and Medicine. 2005. Aggregate Properties and the Performance of Superpave-Designed Hot-Mix Asphalt. Washington, DC: The National Academies Press. doi: 10.17226/13844.
×
Page 32
Page 33
Suggested Citation:"Chapter 2 - State of Practice." National Academies of Sciences, Engineering, and Medicine. 2005. Aggregate Properties and the Performance of Superpave-Designed Hot-Mix Asphalt. Washington, DC: The National Academies Press. doi: 10.17226/13844.
×
Page 33
Page 34
Suggested Citation:"Chapter 2 - State of Practice." National Academies of Sciences, Engineering, and Medicine. 2005. Aggregate Properties and the Performance of Superpave-Designed Hot-Mix Asphalt. Washington, DC: The National Academies Press. doi: 10.17226/13844.
×
Page 34
Page 35
Suggested Citation:"Chapter 2 - State of Practice." National Academies of Sciences, Engineering, and Medicine. 2005. Aggregate Properties and the Performance of Superpave-Designed Hot-Mix Asphalt. Washington, DC: The National Academies Press. doi: 10.17226/13844.
×
Page 35
Page 36
Suggested Citation:"Chapter 2 - State of Practice." National Academies of Sciences, Engineering, and Medicine. 2005. Aggregate Properties and the Performance of Superpave-Designed Hot-Mix Asphalt. Washington, DC: The National Academies Press. doi: 10.17226/13844.
×
Page 36
Page 37
Suggested Citation:"Chapter 2 - State of Practice." National Academies of Sciences, Engineering, and Medicine. 2005. Aggregate Properties and the Performance of Superpave-Designed Hot-Mix Asphalt. Washington, DC: The National Academies Press. doi: 10.17226/13844.
×
Page 37
Page 38
Suggested Citation:"Chapter 2 - State of Practice." National Academies of Sciences, Engineering, and Medicine. 2005. Aggregate Properties and the Performance of Superpave-Designed Hot-Mix Asphalt. Washington, DC: The National Academies Press. doi: 10.17226/13844.
×
Page 38
Page 39
Suggested Citation:"Chapter 2 - State of Practice." National Academies of Sciences, Engineering, and Medicine. 2005. Aggregate Properties and the Performance of Superpave-Designed Hot-Mix Asphalt. Washington, DC: The National Academies Press. doi: 10.17226/13844.
×
Page 39
Page 40
Suggested Citation:"Chapter 2 - State of Practice." National Academies of Sciences, Engineering, and Medicine. 2005. Aggregate Properties and the Performance of Superpave-Designed Hot-Mix Asphalt. Washington, DC: The National Academies Press. doi: 10.17226/13844.
×
Page 40
Page 41
Suggested Citation:"Chapter 2 - State of Practice." National Academies of Sciences, Engineering, and Medicine. 2005. Aggregate Properties and the Performance of Superpave-Designed Hot-Mix Asphalt. Washington, DC: The National Academies Press. doi: 10.17226/13844.
×
Page 41
Page 42
Suggested Citation:"Chapter 2 - State of Practice." National Academies of Sciences, Engineering, and Medicine. 2005. Aggregate Properties and the Performance of Superpave-Designed Hot-Mix Asphalt. Washington, DC: The National Academies Press. doi: 10.17226/13844.
×
Page 42
Page 43
Suggested Citation:"Chapter 2 - State of Practice." National Academies of Sciences, Engineering, and Medicine. 2005. Aggregate Properties and the Performance of Superpave-Designed Hot-Mix Asphalt. Washington, DC: The National Academies Press. doi: 10.17226/13844.
×
Page 43
Page 44
Suggested Citation:"Chapter 2 - State of Practice." National Academies of Sciences, Engineering, and Medicine. 2005. Aggregate Properties and the Performance of Superpave-Designed Hot-Mix Asphalt. Washington, DC: The National Academies Press. doi: 10.17226/13844.
×
Page 44
Page 45
Suggested Citation:"Chapter 2 - State of Practice." National Academies of Sciences, Engineering, and Medicine. 2005. Aggregate Properties and the Performance of Superpave-Designed Hot-Mix Asphalt. Washington, DC: The National Academies Press. doi: 10.17226/13844.
×
Page 45
Page 46
Suggested Citation:"Chapter 2 - State of Practice." National Academies of Sciences, Engineering, and Medicine. 2005. Aggregate Properties and the Performance of Superpave-Designed Hot-Mix Asphalt. Washington, DC: The National Academies Press. doi: 10.17226/13844.
×
Page 46
Page 47
Suggested Citation:"Chapter 2 - State of Practice." National Academies of Sciences, Engineering, and Medicine. 2005. Aggregate Properties and the Performance of Superpave-Designed Hot-Mix Asphalt. Washington, DC: The National Academies Press. doi: 10.17226/13844.
×
Page 47
Page 48
Suggested Citation:"Chapter 2 - State of Practice." National Academies of Sciences, Engineering, and Medicine. 2005. Aggregate Properties and the Performance of Superpave-Designed Hot-Mix Asphalt. Washington, DC: The National Academies Press. doi: 10.17226/13844.
×
Page 48
Page 49
Suggested Citation:"Chapter 2 - State of Practice." National Academies of Sciences, Engineering, and Medicine. 2005. Aggregate Properties and the Performance of Superpave-Designed Hot-Mix Asphalt. Washington, DC: The National Academies Press. doi: 10.17226/13844.
×
Page 49
Page 50
Suggested Citation:"Chapter 2 - State of Practice." National Academies of Sciences, Engineering, and Medicine. 2005. Aggregate Properties and the Performance of Superpave-Designed Hot-Mix Asphalt. Washington, DC: The National Academies Press. doi: 10.17226/13844.
×
Page 50
Page 51
Suggested Citation:"Chapter 2 - State of Practice." National Academies of Sciences, Engineering, and Medicine. 2005. Aggregate Properties and the Performance of Superpave-Designed Hot-Mix Asphalt. Washington, DC: The National Academies Press. doi: 10.17226/13844.
×
Page 51
Page 52
Suggested Citation:"Chapter 2 - State of Practice." National Academies of Sciences, Engineering, and Medicine. 2005. Aggregate Properties and the Performance of Superpave-Designed Hot-Mix Asphalt. Washington, DC: The National Academies Press. doi: 10.17226/13844.
×
Page 52
Page 53
Suggested Citation:"Chapter 2 - State of Practice." National Academies of Sciences, Engineering, and Medicine. 2005. Aggregate Properties and the Performance of Superpave-Designed Hot-Mix Asphalt. Washington, DC: The National Academies Press. doi: 10.17226/13844.
×
Page 53
Page 54
Suggested Citation:"Chapter 2 - State of Practice." National Academies of Sciences, Engineering, and Medicine. 2005. Aggregate Properties and the Performance of Superpave-Designed Hot-Mix Asphalt. Washington, DC: The National Academies Press. doi: 10.17226/13844.
×
Page 54
Page 55
Suggested Citation:"Chapter 2 - State of Practice." National Academies of Sciences, Engineering, and Medicine. 2005. Aggregate Properties and the Performance of Superpave-Designed Hot-Mix Asphalt. Washington, DC: The National Academies Press. doi: 10.17226/13844.
×
Page 55
Page 56
Suggested Citation:"Chapter 2 - State of Practice." National Academies of Sciences, Engineering, and Medicine. 2005. Aggregate Properties and the Performance of Superpave-Designed Hot-Mix Asphalt. Washington, DC: The National Academies Press. doi: 10.17226/13844.
×
Page 56
Page 57
Suggested Citation:"Chapter 2 - State of Practice." National Academies of Sciences, Engineering, and Medicine. 2005. Aggregate Properties and the Performance of Superpave-Designed Hot-Mix Asphalt. Washington, DC: The National Academies Press. doi: 10.17226/13844.
×
Page 57
Page 58
Suggested Citation:"Chapter 2 - State of Practice." National Academies of Sciences, Engineering, and Medicine. 2005. Aggregate Properties and the Performance of Superpave-Designed Hot-Mix Asphalt. Washington, DC: The National Academies Press. doi: 10.17226/13844.
×
Page 58
Page 59
Suggested Citation:"Chapter 2 - State of Practice." National Academies of Sciences, Engineering, and Medicine. 2005. Aggregate Properties and the Performance of Superpave-Designed Hot-Mix Asphalt. Washington, DC: The National Academies Press. doi: 10.17226/13844.
×
Page 59

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

14 CHAPTER 2 STATE OF PRACTICE 2.1 INTRODUCTION The Superpave mix design method, a product of SHRP, was introduced in 1994 (1). The Superpave method included binder, aggregate, and mixture specifications. Many of these specifications used new test methods. One goal of the Super- pave method was to provide uniform test methods and spec- ifications to be used across the United States. Aggregates were addressed through gradation, consensus aggregate properties, and source aggregate properties. Defi- nitions were provided for NMAS and maximum aggregate size. A limited number of gradation control points were estab- lished for each NMAS. A restricted zone was recommended along the maximum density line to prevent the use of large quantities of natural sand and to help ensure adequate VMA. Four consensus aggregate properties were specified: coarse aggregate angularity, flat and elongated particles, uncom- pacted voids content in fine aggregate, and sand equivalent. Specification levels for the consensus property tests depend on design traffic level and depth in the pavement structure (for tests related to permanent deformation). Specifications for consensus aggregate properties are based on the blend of materials used in a given HMA mix and not on individual aggregates. The consensus aggregate properties were to be uniformly implemented across the United States regardless of local geology. Source aggregate properties include LA abrasion, soundness, and deleterious materials. It was felt that these properties would need to be adjusted depending on the local geology; therefore, agencies were to set the specifica- tion levels for source aggregate properties for Superpave- designed mixes. There has been controversy regarding some of the consen- sus aggregate property tests and specification levels. All four tests are empirical in nature, and there is very little data avail- able to support the establishment of specification require- ments. Their relationship to performance has been questioned. In some cases, the implementation of the consensus aggregate properties and Superpave gradation bands have prevented the use of materials or mixes that have been historically used to provide good-performing mixes. This has led to research relating the consensus aggregate properties and aggregate gra- dations to the performance of HMA and to efforts to develop or use alternative tests. A discussion of research related to each of the Superpave aggregate tests, Superpave gradation bands, and alternatives follows. Since the aggregate crushing process affects the resulting shape and texture of the aggre- gate particles, a brief discussion on crushing is also included. 2.2 COARSE AGGREGATE ANGULARITY 2.2.1 Background Prior to the development of Superpave, several studies indicated increased resistance to permanent deformation with increasing fractured faces in coarse aggregate (4–12). The fourth questionnaire used in the Delphi process to identify the consensus aggregate properties ranked coarse aggregate angu- larity second to gradation limits in terms of importance (1). However, no test method was identified. FHWA’s Office of Technology Applications recommended Pennsylvania DOT Test Method 621 (13). ASTM D5821 was based on the Penn- sylvania test method and was later adopted as the method for measuring coarse aggregate angularity within the Superpave mix design method (14, 15). The fractured face count of a representative sample of coarse aggregate is determined by visual inspection. ASTM D5821 (16) defines a fractured face as “an angular, rough, or broken surface of an aggregate particle created by crushing, by other artificial means or by nature.” A fractured face is only counted if its area is greater than 25% of the largest projection (cross-sectional area) of the particle. A fractured particle is “a particle of aggregate having at least the minimum number of fractured faces specified (usually one or two)” (16). To run the test, a representative sample is washed over the 4.75-mm sieve and dried to a constant mass. The size of the sample is dependent on the nominal maximum aggregate size of the aggregate. The aggregate particles are visually inspected and divided into piles of particles with no fractured faces and one or more fractured faces. Prior to 2001 when the ASTM D5821 was revised, the separation included a questionable pile for particles where the tester was uncertain whether a fractured face met the definitions (14). The questionable pile was eliminated from the ASTM standard in 2001 (16); how- ever, the use of a questionable pile is still included in a pro- visional AASHTO Standard TP-61 (15). After all of the par- ticles are sorted, the mass of each pile is determined. The percent fractured particles are expressed as the mass of par- ticles having a given number of fractured faces divided by

15 the total mass of the samples (result expressed as a percent- age). For Superpave specifications, after the percent of parti- cles with one or more fractured faces is determined, the aggre- gates are re-examined for two or more fractured faces. 2.2.2 Relationship Between Percent Coarse Aggregate Fractured Faces and Performance Cross and Brown (17) reported on the selection of aggre- gate properties to minimize rutting. The study was based on testing conducted on 42 pavements in 14 states; 30 of the 42 pavements had experienced premature rutting. Rut-depth measurements and cores were taken at each site. The cores were tested for density, asphalt content, and gradation. The percent with two crushed faces was determined separately for the material retained on the 4.75-mm sieve and the mate- rial passing the 4.75-mm sieve and retained on the 0.600-mm sieve. The uncompacted void content was determined accord- ing to the National Stone Association flow test, Method A (the basis for AASHTO T304). Some of the cores were recom- pacted using the U.S. Army Corps of Engineers (USACE) gyratory compactor or Marshall hammer method (75-blow). The measured rut depth at each site was converted to a rut- ting rate by dividing the rut depth by the square root of accu- mulated equivalent single axle loads (ESALs). Data analysis indicated that none of the aggregate proper- ties were related to the rutting rate when all of the data were included. The authors felt that when air voids were less than 2.5%, rutting is likely to occur regardless of the other mix properties. Using the data from pavements with in-place air voids greater than 2.5%, a relationship shown in Equation 1 (17) between the percent with two crushed faces in the coarse aggregate and the rutting rate was developed. The relation- ship produces an R2 = 0.42. Analysis of variance indicated the relationship was significant (α = 0.01). (1) Kandhal and Parker evaluated the properties of nine coarse aggregate sources (2). Nine tests were performed to evaluate coarse aggregate shape, angularity, and texture including the following: • Index of Aggregate Particle Shape and Texture (ASTM D3398), • Image Analysis (Georgia Institute of Technology), • Flat and Elongated and Flat or Elongated Particles by ASTM D4791, • Flakiness Index (British Standard 812), • Elongation Index (British Standard 812), • Percent of Fractured Particles in Coarse Aggregate (ASTM D5821), Rut Depth (mm) ESAL 0.03138 0.0025 (Percent 2 Crushed Faces) ÷ = − × • Uncompacted Voids in Coarse Aggregate (Currently AASHTO TP56), and • Uncompacted Voids in Coarse Aggregate—Shovel Tech- niques (AASHTO T19). ASTM D5821 was only performed on the three gravel sources included in the nine sources. Because of the limited data col- lected, ASTM D5821 was excluded from correlation matri- ces with other test methods and the rutting performance of mixes produced with each of the coarse aggregate sources. Rut testing was performed on the nine mixtures using the Superpave Shear Tester and Georgia Loaded Wheel Tester (GLWT). The uncompacted voids in coarse aggregate test (AASHTO TP56) produced the best relationships with the rutting parameters from all nine mixtures and a reduced data set that had unusual mix properties (2). The results from AASHTO TP56 and ASTM D3398 were highly correlated. The authors recommended uncompacted voids in coarse aggregate (AASHTO TP56) and flat or elongated particles on the 21 ratio to characterize coarse aggregate shape, angu- larity, and texture. Hand et al. (13) conducted round-robin testing to deter- mine the precision of ASTM D5821. The study was initiated because of concerns that insufficient fractured faces in the original crushed gravel source used at WesTrack may have contributed to the premature failure of the coarse-graded sec- tions. The materials were collected from cold feed samples taken during the construction and reconstruction of WesTrack. Four materials were included in the study based on the coarse aggregate fractions of (1) coarse blends of Dayton crushed gravel produced in 1994 and placed in the tangents; (2) fine blends of Dayton crushed gravel produced in 1994 and placed in the tangents; (3) coarse blends of Dayton crushed gravel produced in 1995 and placed in the curve sections; and (4) the crushed andesite from Lockwood, Nevada, used in the coarse-graded replacement sections. The percent fractured faces of the fine and coarse mixtures placed on the tangent sections were found to be equal. The actual values (98% one fractured face and 96% two or more fractured faces) exceeded the Superpave requirements for 10 to 30 million design ESALs. The study concluded, “CAA [coarse aggre- gate angularity] did not have an effect on the rutting perfor- mance of Superpave mixtures at WesTrack” (13). A Canadian study was conducted in Saskatchewan to inves- tigate the effect of the percent fractured coarse aggregate par- ticles on rutting performance (18). The majority of Saskatch- ewan’s coarse aggregate comes from glacial gravel deposits. Aggregates with high fractured face counts are more expen- sive. Ten pavements ranging in age from 2 to 9 years were evaluated. Rut depths were measured and cores were recov- ered within and between the wheel paths. Cores were tested for density, voids filled, asphalt content, coarse aggregate fractured face count, and uncompacted void content in fine aggregate. The fractured face count was determined accord- ing to Saskatchewan Standard Test Procedure 204-4. The

fractured face counts ranged from 54% to 93.7%. It was not reported whether the fractured face count represents one or two fractured faces. A stepwise regression was performed to identify the factors most related to the in-place rut depth. The regression identified traffic (accumulated ESALs), between wheel path asphalt content, between wheel path voids filled, and between wheel path density (R2 = 76.6) (18). The rut depths and accumulated ESALs provided in the paper were converted to a rutting rate as recommended by Cross and Brown (17). Regression analysis between the reported frac- tured face counts and rutting rate indicated no relationship. Overall, the rutting rates were higher than those reported by Cross and Brown. 2.2.3 Precision of ASTM D5821 ASTM D5821 for fractured face count of coarse aggregate reports a multilaboratory standard deviation of 5.2% for well- trained observers (16). Thus, the acceptable range between two properly conducted tests by two well-trained observers would be 14.7%. This precision is based on an Ontario Min- istry of Transportation study that included 34 observers’ evaluations of two samples of partially crushed gravel. Hand et al. (13) reported the precision statement shown in Table 1 based on 10 laboratories’ tests of four aggregates used at WesTrack. 2.2.4 Alternative Methods of Measuring Coarse Aggregate Angularity Alternative methods to ASTM D5821 have been investi- gated that combine shape, angularity, and texture into one measure (2, 19–21). Two alternatives that have received atten- tion are ASTM D3398, “Index of Aggregate Particle Shape and Texture,” and AASHTO TP56, “Uncompacted Voids in Coarse Aggregate.” The literature indicates that angular, rough-textured aggregates have a particle index value greater than 14, whereas rounded and/or smooth aggregates have a particle index value less than 12 (19). Ahlrich (19) developed the uncompacted voids in coarse aggregate test based on ASTM C1252, “Uncompacted Void Content in Fine Aggre- 16 gate.” The two tests are highly correlated, producing an R2 = 0.94 (2, 19). AASHTO TP56 is preferable to ASTM D3398 because of the significant time required to perform ASTM D3398. One drawback to both tests is that the effects of par- ticle shape, angularity, and texture cannot be separated. Kandhal et al. (20) indicated that the particle shape and angularity index obtained from ASTM D3398 increase sharply as the percentage of two fractured faces from ASTM D5821 increases above 80%. However, the authors note (20): Theoretically, 100% crushed particles would be preferable to use when employing gravel coarse aggregates in an HMA mix. However, the benefits that might be achieved by requir- ing the 2-face crushed count to be 100% should be weighed against the additional cost involved in the crushing operation. Ahlrich (19) investigated 11 aggregate blends meeting the Federal Aviation Administration’s P401 gradation. The blends were produced by combining different percentages of crushed limestone, crushed gravel, uncrushed gravel, and natural sand. The blends were combined to produce 0%, 30%, 50%, 70%, and 100% crushed coarse aggregate particle counts. Opti- mum asphalt content was determined for each blend using the USACE Gyratory Testing Machine. The resulting mix- tures were tested for rutting resistance using a confined repeated-load permanent deformation test. Coarse aggregate shape, angularity and texture were eval- uated using the USACE test for fractured face count (CRD- C 171), ASTM D3398, and the uncompacted voids in coarse aggregate test (AASHTO TP56). Testing indicated a strong correlation (R2 = 0.98) between the uncompacted voids con- tent of the as-received material and the fractured face count. Table 2 shows the correlations between individual tests and three parameters from the confined repeated-load permanent deformation test. The combined (coarse and fine aggregate) particle index value (PI Composite) from ASTM D3398 appears to provide the best overall correlation. The particle index value on the coarse material also provides good corre- lations. The percent crushed face count (PCP Composite) as measured by CRD-C 171 for the composite coarse and fine aggregate as well as the uncompacted voids in coarse aggre- gate are also good predictors. Property and Index Type Standard Deviation, % Acceptable Range of Two Results One or More Fractured Faces Single-Operator Precision 1.1 3.0 Multilaboratory Precision 1.8 5.1 Two or More Fractured Faces Single-Operator Precision 1.8 5.1 Multilaboratory Precision 2.9 8.2 TABLE 1 Precision statement for both one or more and two or more fractured faces (13)

As described previously for the aggregates studied as part of NCHRP Project 4-19, Kandhal and Parker (2) identified AASHTO TP56, “Uncompacted Void Content in Coarse Aggregate,” as being the test most related to the rutting per- formance of a coarse-graded mix produced using nine differ- ent coarse aggregate sources and a single natural sand source. Hossain et al. (21) studied the results from ASTM D3398, “Index of Aggregate Particle Shape and Texture,” and the uncompacted voids in coarse aggregate to measure coarse aggregate angularity. The effects of gradation and the per- centage of flat and elongated particles were considered on the test results. Two standard gradations were developed for the uncompacted voids test. The gradations were based on the maximum density line for gradations with maximum particle sizes of 12.5 or 19.0 mm. The particle index value and uncom- pacted voids tests were run on both the individual size frac- tions and the proposed blended gradations. Testing indicated a good relationship between the calculated index using the results from the individual size fractions and the measured values from blended samples representing the two standard gradations. The authors recommended the use of a standard gradation for relative comparison of coarse aggregate sources (21). The standard gradations established by Hossain et al. (21) were adopted by AASHTO TP56. For comparing grada- tions, the authors recommended testing the individual size fraction and then calculating the result for a target gradation. Flat and elongated particles tend to increase the measured uncompacted voids content and aggregate particle index. However, flat and elongated particles are not desirable in HMA (21). Relationships between percent flat and elongated particles and both uncompacted voids and particle index were obtained for the limited materials used in the study. The relationships were non-linear and were different for gravel and crushed stone (based on the differences in angularity and texture between the two groups) (21, 22). The relationships were developed for both the 31 and 51 ratios for flat and elongated particles. Hossain et al. (22) evaluated the rutting performance of 11 mixes produced with blends from four crushed gravel sources, a limestone source, a granite source, and a natural sand. Alabama DOT 416 Mix 4 specifications were used to estab- 17 lish a single gradation for all of the mixes. The gradation is a 12.5-mm NMAS mixture that approximately follows the Superpave maximum density line. Rut testing was performed with the GLWT and a confined repeated-load permanent deformation test. No single aggregate test provided a strong relationship with the performance of all the mixes. The high- est correlation coefficient (R = −0.70) was for percent flat or elongated particles by particle count at the 51 ratio. Ongoing research by Rismantojo (23) as part of NCHRP Project 4-19(2) evaluated five coarse aggregates using accel- erated loading. The aggregates included a dolomite, lime- stone, natural gravel, granite, and traprock. Coarse aggregate tests performed as part of the study include • Flat and Elongated Particles (ASTM D4791) at the 21, 31, and 51 ratio; • Uncompacted Voids in Coarse Aggregate (AASHTO TP56) Method A (standard grading) and Method B (indi- vidual size fractions); • Micro-Deval (AASHTO TP58); • Magnesium Sulfate Soundness (AASHTO T104); • LA abrasion (ASTM C96 Type C); • Bulk Specific Gravity (ASTM C127); and • Water Absorption (ASTM C127). A correlation matrix was developed among the aggregate tests. Several strong relationships were indicated between the various forms of ASTM D4791 used in the study. A fair rela- tionship (R = 0.786, p-value = 0.064) was indicated between flat or elongated particles and uncompacted voids in coarse aggregate Method A. Figure 1 presents the combined data from NCHRP Projects 4-19 and 4-19(2). The regression line in the figure excludes the slag and sandstone sources tested as part of NCHRP Project 4-19 as outliers. Regression analy- sis for the combined data (including the outliers) produces an R2 = 0.24. The relationship is not significant at the 5% level with a p-value = 0.06. This indicates that although particle shape has a strong influence on uncompacted voids results, texture—such as that found on the slag and sandstone—can also have a strong effect. Rutting models developed by Kandhal and Parker (2) indicated that high percentages of flat TABLE 2 Rankings for correlations between aggregate characterization tests and permanent deformation results (19) Rank Permanent Strain (R2) Creep Modulus (R2) Slope of Deformation Curve (R2) 1 PCP Composite (0.87) PI Composite (0.73) PI Composite (0.71) 2 PI Composite (0.78) PI Coarse (0.69) PCP Composite (0.65) 3 PI Coarse, (0.68) PCP Composite (0.63) PI Coarse (0.52) 4 UV Coarse (0.65) UV Coarse (0.56) UV Coarse (0.43) 5 PCP Coarse (0.60) PCP Coarse (0.46) ASTM C1252 Method A (0.41) Note: PCP = percent crushed particles (CRD-C 171) PI = Particle Index Value (ASTM D3398) UV = uncompacted voids in coarse aggregate (AASHTO TP 56) Composite = both coarse and fine aggregate

or elongated particles at the 21 or 51 ratio were undesir- able. Rismantojo (23) notes that Kandhal and Parker’s (2) conclusion that high percentages of flat or elongated particles are undesirable from a rutting stand point is unrealistic based on the relationship with uncompacted voids in coarse aggre- gates. Higher percentages of flat or elongated particles tend to produce higher uncompacted voids and higher uncom- pacted voids tend to produce mixtures with less rutting. Rismantojo (23) performed correlations between coarse aggregate properties and both mix volumetric properties and rutting performance. Flat or elongated particles at the 21 ratio were positively correlated with optimum asphalt con- tent. This indicates that for the aggregates tested, higher asphalt contents resulted for mixes with higher percentages of flat or elongated particles. Thus, mixes containing flat and elongated particles may be more durable. Flat and elongated particles at the 31 ratio were negatively correlated with the density at Ninitial. As the percentage of flat and elongated par- ticles at the 31 ratio increased, the density at Ninitial decreased. Thus, flat and elongated particles also make it easier to meet the Ninitial requirements. VMA was correlated positively with the uncompacted voids in coarse aggregate for both Methods A and B. Overall, Method B, which is the average of the uncompacted voids for three individual size fractions, pro- duced better correlations. Full-scale rutting tests were performed at the Indiana DOT APT Facility in West Lafayette, Indiana. Five mixes were tested in the APT facility. The rounded gravel mix produced 29.5 mm of rutting after 5,000 passes, at which time testing was terminated. The other four sections containing quarried 18 stone were tested to 20,000 passes. A strong relationship was identified between the uncompacted voids from both Methods A and B and the total rut depth at 5,000 passes (R = −0.947 and R = −0.983, respectively). This relationship is strongly influenced by the uncrushed gravel mixture. When the gravel mix is excluded and only the four mixes that were tested to 20,000 passes are analyzed, the uncompacted voids in the coarse aggregate (Method A) performed on the plant stock- pile material produces the best correlation with R = −0.758 (23). The relationship is not significant at the 5% level. 2.2.5 Summary of Research Related to Coarse Aggregate Angularity Numerous research studies have indicated improved rut resistance with increased percentages of fractured faces in coarse aggregate. However, the current test method, ASTM D5791, is subjective, requiring the technician to visually determine the presence and number of fractured faces. The Superpave fractured face count specifications are based on the consensus of an expert panel and not on laboratory test results. Research completed since the implementation of the Superpave method has focused on alternative tests that are more quantitative and objective. Several studies have evaluated the relationship between both the particle index value (ASTM D3398) and the coarse aggre- gate uncompacted voids test (AASHTO TP56) and rutting per- formance. Trends indicate that higher particle index values or uncompacted voids contents produce more rut-resistant pave- ments. Relationships have been identified between both tests y = 0.2287x + 38.809 R2 = 0.70 40 45 50 55 5 10 15 20 25 30 35 40 45 50 Flat or Elongated Particles 2:1 Un co m pa ct ed V oi ds , % NCHRP 4-19 NCHRP 4-19 (2) Blast Furnace Slag Sandstone Figure 1. Relationship between flat or elongated particles and uncompacted voids (2, 22).

and flat and/or elongated particles. Increasing values of flat and/or elongated particles at 21 or 31 ratios tend to increase the particle index value and uncompacted voids. This indi- cates that the tests are highly influenced by particle shape. The uncompacted voids test is also sensitive to the texture of coarse aggregate particles. Both of these tests combine the effects of shape, angularity, and texture. Digital imaging meth- ods are being developed that can separately quantify these parameters. These will be discussed later in the report. 2.3 FLAT AND ELONGATED PARTICLES 2.3.1 Background The asphalt industry believes excessive flat and elongated particles (F&E) to be undesirable. Perfectly cubical aggre- gates may also be undesirable. Prior to the implementation of the Superpave method, several design procedures had speci- fications limiting the percent of F&E allowed in the mix. AASHTO M283 allowed no more than 15% combined F&E as tested in accordance with ASTM D4791. Roberts et al. (24) state: “Flat and elongated particles impede compaction and thus may prevent the development of satisfactory strength in HMA.” The Aggregate Handbook (25) states: “Specifica- tions requiring particle-by-particle measurements vary widely in terms of limiting values and allowable percentages of defec- tive particles. Research has largely been unsuccessful in estab- lishing criteria related to performance.” It is generally believed that high percentages of F&E were undesirable because they hinder compaction and, if broken under the roller, expose uncoated aggregate surfaces. The fourth questionnaire used in the Delphi process by SHRP to identify the consensus aggregate properties ranked thin, elongated pieces eighth in terms of importance (1). ASTM D4791, “Standard Test Method for Flat Particles, Elongated Particles, or Flat and Elongated Particles in Coarse Aggregate,” was recommended as the test method for thin, elongated particles by the expert panel. USACE (26) origi- nally developed the method. When the Superpave method was first implemented, ASTM specified that the test be performed on the +9.5-mm material by size fraction (14). The Superpave method specified that the test be run on the +4.75-mm material (1). ASTM later revised the standard to include the +4.75-mm material (16). The test is run by first performing a gradation on a represen- tative sample of the coarse aggregate. One hundred particles are split out for testing for each size fraction that has at least 10% retained. The Superpave method specifies that the par- ticle is considered flat and elongated if the particle’s maxi- mum dimension is five or more times the particle’s minimum dimension. The maximum and minimum dimension of each particle is measured or, alternatively, a proportional caliper may be used. If a proportional caliper is used, the largest dimension of the particle is used to set the caliper. If the par- ticle can pass through an opening that is one-fifth of the max- 19 imum dimension, the particle is considered flat and elon- gated. The percentage of flat or elongated particles may be reported by weight or by particle count. The overall percent of F&E is based on a weighted average determined from the sample gradation and the flat and elongated percentages for each size fraction. The Superpave method allows no more than 10% F&E exceeding the 51 ratio for the combined aggregate blend used in asphalt mixtures for pavements with >1 million ESALs in the design life (15, 27). The Superpave specifications are based on the blend of coarse aggregate used in the HMA, not on an individual stockpile. The Stone Matrix Asphalt Tech- nical Working Group guide specification allows 5% 51 and 20% 31 F&E (15, 28). 2.3.2 Relationship Between F&E and Performance A limited number of studies were completed to relate the effect of F&E on performance prior to SHRP (29–33). Huber et al. (34) evaluated the effect of F&E on the volumetric properties of two Superpave 19.0 NMAS mixes. The coarse aggregate was a crushed No. 57 stone produced from a lime- stone source. Coarse aggregate was produced using both a vertical shaft impact crusher and a cone crusher. Vertical shaft impact crushers tend to produce more cubical aggregate. Nei- ther crusher produced F&E that exceeded the 51 ratio. The vertical shaft impact crusher produced 9.0% and the cone crusher produced 19.4% particles exceeding the 31 ratio. HMA was produced with aggregate from both crushers to meet each of the two gradations for a total of four mixes. Two laboratories tested each mix at constant asphalt content in the Superpave gyratory compactor. Based on the sample density results, the authors concluded that F&E exceeding the 31 ratio do not negatively impact volumetric properties (34); however, this was a very limited study from which to make general conclusions about the effect of F&E on mixture vol- umetric properties. Brown et al. (35) evaluated the effect of five levels of F&E on the volumetric properties, aggregate breakdown, and mois- ture susceptibility of stone matrix asphalt (SMA). An Arkansas limestone source was crushed to provide two different levels of F&E. The F&E varied from 67 to 38 for the 21 ratio, 25 to 3 for the 31 ratio, and 1 to 0 for the 51 ratio. SMA was produced with both coarse aggregates and 75/25, 50/50, and 25/75% blends of the two aggregates. There was a slight trend of increasing VMA with increasing percentages of F&E. The VMA increased 1.2% from the cubical to the more F&E coarse aggregate. Gradation testing indicated a statistically significant increase in aggregate breakdown on the 4.75-mm sieve for higher levels of F&E. Breakdown increased by approximately 4% between the two extremes in particle shape for samples compacted with 50 blows of each face with a Marshall hammer (35).

Vavrik et al. (36) evaluated the effect of F&E on the vol- umetric properties and aggregate breakdown for gyratory- compacted samples. Aggregate was obtained from a dolo- mite and a gravel source. The coarse aggregate particles for each source were sorted into particles whose maximum-to- minimum dimension was less than the 31 ratio, greater than the 31 ratio but less than the 51 ratio, and greater than the 51 ratio. A modified Superpave mixture design was devel- oped using the cubical (less than 31 ratio) coarse aggregate from each source. The same manufactured sand, natural sand, and mineral filler were used for both designs. The dolomite mixture used 55% coarse aggregate with the fine aggregate being an 80% to 20% blend of manufactured and natural sand. The gravel mixture used 52% coarse aggregate with a 70% to 30% split of manufactured to natural sand for the fine fraction. The design asphalt contents were chosen using the locking point concept. The locking point is defined as the first occurrence of three gyrations at the same height preceded by two gyrations at the same height. The locking point is the number of gyrations at which the first of the three consecu- tive gyrations of the same height occur. The average locking point for the cubical dolomite mixture was 101 gyrations, and the average locking point for the cubical gravel mixture was 90 gyrations. Four samples were then compacted at the optimum asphalt content determined for the cubical coarse aggregate of each of the aggregate sources at each of four blends of F&E. The samples were compacted to the 110 gyrations. The volumet- ric results are summarized in Table 3. The data indicate a trend of increasing VMA with increasing F&E as indicated 20 by the previous studies. Vavrik et al. (36) evaluated aggre- gate breakdown during gyratory compaction and noted that for the dolomite mixture, the percent passing the No. 4 sieve (4.7-mm) increased by 3% to 5% and that the percent passing the No. 8 (2.36-mm) sieve increased by 1% to 4% for the non- cubical blends as compared with the cubical blends. Similarly, the percent passing the No. 4 sieve (4.75-mm) increased by 2% to 4% for the gravel mixtures. No corresponding increase was observed for the percent passing the No. 8 sieve for the gravel mixtures. The authors concluded that increased F&E resulted in increased aggregate breakdown and that the changes in the volumetric properties (including VMA) resulted from the changes in gradation caused by aggregate breakdown (36). Buchanan (37) evaluated the effect of six levels of F&E from two aggregate sources on the volumetric properties, rut- ting performance, and fatigue performance of a 12.5-mm NMAS Superpave mix design. The six levels of F&E con- sisted of the as-received aggregate from a limestone and a granite source as well as each of those aggregates crushed at two different rotor speeds in a scale vertical shaft impact crusher. The blend percents of F&E, volumetric properties, and rut depths are shown in Table 4. Some of the F&E results for the limestone aggregate appear anomalous. Higher rotor tip speeds should produce more cubical particles because the aggregate is thrown with more energy against cascading aggregate in the crusher. This is consistent with the data shown in Table 4 except for the limestone sample at 65 m/s. Based on Table 4, the volumetric properties for the lime- stone aggregate match the findings of Huber et al. (34) with Material Air Voids, % VMA, % Locking Point Dolomite Coarse Aggregate Cubical 3.75 14.7 101 50-50-01 15.1 4.24 97 30-50-20 15.4 4.48 102 70-0-30 15.1 4.16 101 Gravel Coarse Aggregate Cubical 14.6 3.55 93 50-50-01 15.3 4.37 113 30-50-20 15.6 4.61 120 70-0-30 15.6 4.62 119 1 50-50-0 are the percentages of cubical particles with shape ratios (maximum-to-minimum dimensions) >3:1 but less than 5:1 and particles with shape ratios >5:1. TABLE 3 Volumetric data for various levels of F&E for Illinois study (36) F & E Ratios Aggregate Type 2:1 3:1 5:1 Optimum AC% VMA, % APA Rut Depth (Dry), mm Limestone As-Received 69.2 29.5 3.8 4.2 Limestone @ 55 m/s1 58.6 21.8 0.2 4.5 13.7 13.9 5.9 6.6 Limestone @ 65 m/s 72.0 16.2 3.7 4.2 13.7 6.2 Granite As-Received 85.4 23.057.0 5.0 14.2 9.2 Granite @ 45 m/s 42.9 14.4 0.4 4.6 13.4 6.2 Granite @ 68 m/s 35.1 2.1 0.1 4.5 13.4 6.1 1 Indicates the rotor tip speed on the vertical shaft impact crusher. TABLE 4 F&E levels and resulting mixture properties (37)

little change over a moderate range of 31 particles. The gran- ite aggregate indicates decreasing VMA with decreasing F&E as noted by Brown et al. (35) and Vavrik et al. (36). Rut test- ing was performed in the Asphalt Pavement Analyzer (APA) at 64°C with a 100-lb vertical load and 100 psi hose pressure. There was not a statistically significant difference between the APA rut depths for the limestone mixtures. The rut depths for the granite mixtures produced in the vertical shaft impact crusher were significantly less than the rut depth for the as- received mixture; however, the granite mixes produced using the aggregate from the vertical impact crusher have failing VMA values and therefore lower asphalt contents. Constant strain fatigue tests were performed according to AASHTO T321 at two strain levels. There was not a statistically signifi- cant difference in the fatigue results for either aggregate at the three levels of F&E evaluated (37); therefore, there does not appear to be an effect of F&E on a mixture fatigue resistance. Oduroh et al. (38) evaluated the effect of three levels of F&E on compacted sample density, shear stiffness, and ten- sile properties. The goal of the study was to provide data to states that are considering the 31 ratio in lieu of the 51 ratio as the definition of F&E. The three levels of F&E investi- gated were 0%, 15%, and 40% particles exceeding the 31 ratio for maximum-to-minimum particle dimension. A single Kentucky limestone coarse aggregate and Ohio River natural sand were used to produce a 12.5-mm NMAS mixture. The samples were mixed with an unmodified performance grade (PG) 64-22 at an optimum asphalt content determined using Ndesign = 96. Performance tests were performed with the Superpave Shear Tester (SST) and Indirect Tensile Tester (IDT) on sam- ples prepared at optimum asphalt content with 0%, 15%, and 40% F&E. Testing included frequency sweep at constant height, simple shear at constant height, repeated shear at con- stant height, and IDT tests. Based on the results of the test- ing, adding 15% or 40% particles exceeding the 31 ratio did not affect the compacted mixture density, shear stiffness, or 21 tensile properties. Repeated shear at constant height testing indicated that the rutting susceptibility of the mixtures with three levels of F&E appeared to be similar at 58°C (38). Aho et al. (39) evaluated the relationship between the per- centages of F&E and aggregate breakdown during construc- tion. Samples of six surface mixtures, representing a range of F&E contents, were sampled at the plant; the mixtures were sampled behind the paver prior to compaction and in-place after compaction. The samples taken after compaction were taken 2 ft (0.6 m) from the corresponding sample taken behind the paver (assumed to be longitudinally). Aggregate samples were collected from both the quarry and the asphalt plant stockpiles for F&E and LA abrasion testing. The per- centages of particles exceeding the 31 ratio ranged from 8.1% to 54.1% and the percentages of particles exceeding the 51 ratio ranged from 0.4% to 21.1% based on the average of eight tests. In addition, for three of the five aggregates, gyratory samples were prepared to simulate aggregate break- down during mix design. Three samples each were com- pacted to 4% and 7% air voids. Statistical analyses were performed to compare the per- cent passing the No. 4 (4.75-mm) sieve from extracted sam- ples. Comparisons were made between the plant sample and the sample taken behind the paver prior to compaction and between the sample taken behind the paver prior to com- paction and the sample taken after compaction. Statistical differences were observed before and after compaction for two mixtures produced with dolomite aggregates and between the plant and paver sample for one of the mixtures. The LA abrasion values of the two dolomite sources were 25% and 26%. Breakdown was not observed when comparing samples taken before and after compaction for the remaining three mix- tures even though one mixture had 54.1% and 21.1% F&E, based on the 31 and 51 ratios, respectively (39). A relation- ship was observed between lift thickness and breakdown. The recovered samples from each sampling location were also tested for F&E. As shown in Figure 2, testing indicated Figure 2. Interaction among F&E, LA abrasion, and aggregate breakdown (39).

that for the mixtures that contained in excess of 30% parti- cles exceeding the 31 ratio, the percent F&E decreased from the quarry to the plant stockpile, from the plant stock- pile to the plant mix, and from the plant mix to the sample taken behind the paver. None of the five projects indicated a significant difference between the percent of F&E measured on samples recovered behind the paver to samples recovered after compaction (39). Comparisons were made between the amount of breakdown that occurred during construction and the amount of break- down that occurred during gyratory compaction. The data indicated practically no difference between the amount of breakdown that occurred for samples compacted to 4% or 7% air voids. The authors concluded that the amount of breakdown that occurs during gyratory compaction generally exceeds that which occurs during normal construction (39). Ongoing research as part of NCHRP Project 4-19(2) has examined the effect of F&E on volumetric properties and permanent deformation during accelerated loading. Signifi- cant relationships were observed between the percent of flat or elongated or the percent F&E and both LA abrasion loss and uncompacted voids in coarse aggregate, as shown in Table 5. Based on the six aggregates tested in this study, there was no correlation between F&E and VMA; however, the gradations differ slightly between the mixes. A strong pos- itive correlation (R = 0.993, p-value = 0.0007) was observed between percent flat or elongated particles at the 21 ratio and total binder content. No significant relationships were observed between F&E and the rutting performance of the mixes (23). It should be noted that the percent of F&E only ranged from 11.6% to 28.0% for the 31 ratio and 1.8% to 8.1% for the 51 ratio. Additional research on F&E is ongoing at the International Center for Aggregates Research. The study is examining the change in the percentage of F&E during laboratory mixing and compaction, aggregate breakdown during mixing and compaction, volumetric properties, tensile strength, and resis- tance to rutting (40). In addition to the proportional caliper typically used for ASTM D4791, the study examined the use of the multiple ratio shape analysis device. Using similar gra- dations, increasing F&E on the 31 ratio from cubical to 20% 22 and 30% of particles exceeding the 31 ratio increased the mixture VMA by 1% to 1.5%. 2.3.3 Precision of F&E Tests ASTM D4791 does not contain a precision statement. Prowell and Weingart (41) conducted a study to determine a precision statement for ASTM D4791. A total of five aggre- gates were tested in the study, but only three were used for determining a precision statement for ASTM D4791. The aggregates were igneous granite, diabase, and dolomitic limestone with percentages of particles exceeding the 31 ratio ranging from 8.2 to 45.8 and percentages of particles exceeding the 51 ratio ranging from 0.2% to 12.7%. Fifteen labs participated in the study. Each laboratory tested two replicates (100 particles each) of each of three particle sizes: − 3/4 in. to +1/2 in., −1/2 in. to +3/8 in., and −3/8 in. to +No. 4. Figure 3 shows the pooled within lab (W/L) and between lab (B/L) standard deviations versus the average percent for the 51 particles. From Figure 4, there appears to be a linear relationship between the standard deviation and average level of the test. A similar trend was observed for the 31 samples. In cases in which a linear relationship exists between the standard deviation and the mean of the test val- ues, ASTM C802 recommends that one use the coefficient of variation. Figure 3 shows the coefficient of variation versus the average percent particles exceeding the 51 ratio. The coefficient of variation sharply decreased with increasing mean test values for the 51 ratio until it reaches an asymp- totic level. This indicates that the test method (51 ratio) is subject to erratic variations. The test is variable, even at low levels of F&E. The coefficient of variation is inflated by dividing the standard deviation by the low (close to zero) mean levels of 51 F&E; however, the coefficient of variation for the 31 ratio was relatively constant. The precision statement for ASTM D4791 was written according to ASTM C670 (42). The single-operator coeffi- cient of variation for the 31 ratio was found to be 26.1% (41). The difference between two individual test results with a 95% confidence interval is determined by multiplying the standard deviation by 2√–2; therefore, results of two properly Tests Flat or Elongated Particles 2:1 and Uncompacted Voids Method A Micro-Deval and Magnesium Sulfate Soundness Flat or Elongated Particles 5:1 and LA Abrasion Flat and Elongated Particles 5:1 and LA Abrasion Micro-Deval and Bulk Specific Gravity LA Abrasion and Bulk Specific Gravity Micro-Deval and Water Absorption Magnesium Sulfate Soundness and Water Absorption 0.786, 0.0641 0.863, 0.027 0.832, 0.040 0.844, 0.035 –0.877, 0.022 –0.812, 0.049 0.961, 0.002 0.894, 0.016 1Not significant at the 5% level, significant at the 10% level. Correlation Coefficient, p-value TABLE 5 Significant correlations between different coarse aggregate tests in NCHRP Project 4-19(2) (23)

conducted tests by the same operator on the same sample using the same proportional caliper should not differ by more than 73.9% of their average. The multilaboratory coefficient of variation has been found to be 35.3%. Therefore, results of two laboratories on identical samples of an aggregate should not differ by more than 99.9% of their average. For comparison, the single operator and multilaboratory coeffi- cients of variation for the 21 ratio are 9.1% and 15.0%, respectively. A precision statement was not prepared for the 51 testing ratio because of the variability in the standard deviation and the coefficient of variation as a function of the mean. The high within- and between-laboratory standard deviations and coef- ficients of variation call to question the value of using ASTM D4791 for a specification limit unless the 21 ratio is used. 23 The AASHTO Materials Reference Laboratory has also compiled statistics for the precision of flat and elongated aggregate tests performed on proficiency samples No. 117 and 118 (43). The results are shown in Table 6. The results are based on testing at the 51 ratio. Similar to the research by Prowell and Weingart (41), the results indicate the tremendous variability of the test method. 2.3.4 Summary of Research Related to F&E A limited number of studies have been conducted to relate the percentage of F&E to performance since the implemen- tation of the Superpave method. None of the studies have addressed the relationship between F&E and performance 0 50 100 150 200 250 300 0 2 4 6 8 10 12 14 Average Percent F&E, % Co ef fic ie nt o f V ar ia tio n, % B/L W/L Figure 3. Pooled coefficient of variation for F&E tests versus average for 51 ratio (41). 0 1 2 3 4 5 6 7 0 2 4 6 8 10 12 14 Average Percent F&E, % St an da rd D ev ia tio n, % B/L W/L

near the existing specification level of 10% particles exceed- ing the 51 ratio of maximum-to-minimum particle dimen- sion. The research conducted to date generally supports the following: • Percentage of F&E changes with handling of the stock- pile and mixing. • Aggregate breakdown during compaction increases for higher percentages of F&E. • VMA generally increases with increasing percent F&E. • There does not be appear to be a relationship between the percentage of F&E exceeding the 31 ratio—in the range of approximately 10% to 40%—and performance. • ASTM D4791 is a highly variable test procedure. Alter- native methods of determining the percentage of F&E should be developed. This variability may mask relation- ships with performance. 2.4 METHODS OF MEASURING FAA AND THEIR RELATIONSHIP TO PERFORMANCE 2.4.1 Introduction It has long been recognized that the characteristics of the fine aggregate component of HMA can have a significant and sometimes dominant influence on mixture rutting and fatigue cracking resistance (2, 34, 44, and 45). Kandhal et al. (46) have classified the test methods to describe aggregate angu- larity into two broad categories: direct and indirect. Direct methods are defined as those wherein particle shape or tex- ture are measured and described qualitatively or quantita- tively through direct measurement of individual particles. In indirect methods, particle shape and texture are determined based on measurements of bulk properties. 2.4.2 Uncompacted Voids Content in Fine Aggregate The Superpave method specifies AASHTO T304 (ASTM C1252), “Uncompacted Void Content in Fine Aggregate, Method A,” to ensure that the blend of fine aggregates in an HMA mixture has sufficient internal friction to provide rut- 24 resistance in an HMA mixture (47). The amount of friction depends on the aggregate particle shape and texture. Higher internal friction is associated with increased rutting resis- tance. AASHTO T304 is commonly referred to as the FAA test. FAA levels used in the Superpave method are below 40, 40 to 45, and above 45. The higher values are specified for layers near the pavement surface and for higher traffic lev- els. AASHTO T304 was to be used in conjunction with the restricted zone to limit the amount of rounded natural sand in high traffic mixes. The angularity and texture of the fine aggregate also affect the packing characteristics of the HMA and, therefore, the VMA of the compacted HMA. More angular or poorly shaped particles or particles having a high degree of texture may not pack as tightly as rounded or smooth particles and, therefore, may provide greater VMA in the compacted HMA. The FAA test is an indirect measure of particle shape, angu- larity, and texture. The FAA test is based on the National Aggregate Association Flow Test (Method A) that is used to evaluate the effect of the fine aggregate on the finish ability of Portland cement concrete. The FAA value is defined as the percent air voids in a loosely compacted sample of fine aggre- gate. The FAA test assumes that more angular particles or particles with more surface texture will not pack together as tightly as rounded or smooth particles would. In AASHTO T304, a 190-g sample of fine aggregate of a prescribed gradation is allowed to flow through the orifice of a funnel and fill a 100-cm3 calibrated cylinder. Excess mate- rial is struck off, and the cylinder with aggregate is weighed. The uncompacted void content of the sample is then com- puted using the loosely compacted weight of the aggregate, the bulk dry specific gravity of the aggregate, and the cali- brated volume of the receiving cylinder. There are three methods for running AASHTO T304: Methods A, B, and C. The mass of the sample for all three methods is fixed at 190 g. Method A specifies a known gra- dation ranging from material passing the 2.36-mm sieve to material retained on the 0.150-mm sieve. Method B specifies that the test be run on three individual size fractions: 2.36 to 1.18 mm, 1.18 to 0.600 mm, and 0.600 to 0.300 mm. The reported void content for Method B is the average of the results from the three individual size fractions. In Method C, the test is run on the as-received gradation (48). Single Operator Precision Sample 1 Multilaboratory Precision Sample 2 Multilaboratory Precision Sample 1 Sample 2 Particle Size, mm Number of Labs Avg. 1S% D2S% Avg. 1S% D2S% 1S% D2S% 1S% D2S% 19.0 to 12.5 128 14.4 51.0 144.3 13.1 53.4 26.6 75.3 29.1 82.2 12.5 to 9.5 123 17.6 43.9 124.1 17.3 42.1 119.2 22.7 64.2 23.1 65.2 9.5 to 4.75 122 24.9 45.7 129.3 23.3 46.4 131.2 18.3 51.8 19.6 55.4 151.0 TABLE 6 Precision of ASTM D4791 F&E tests from AMRL Proficiency Samples 117 and 118 (43)

The Superpave researchers chose Method A to limit the effect of gradation, particularly material passing the 0.075-mm sieve on the test result. For example, if one were to test a washed manufactured sand with a −0.075 mm sieve content of 8% and a crushed screening produced from the same aggregate and crushed with the same crusher settings with a –0.075 mm sieve content of 14% under Method C, the crushed screening would produce a lower FAA value than would the washed manufactured sand even though the two materials have identical particle shape and texture. Several studies have been conducted to compare Methods A, B, and C (49–53). The studies have indicated a strong relationship between Methods B and C, with Method B pro- ducing uncompacted void contents almost 5 points higher (49, 51–53). Hossain et al. (49) observed that the uncom- pacted void contents were generally higher for smaller sized particles. Hudson (50) stated that, based on visual observa- tion, particle shape appeared to be constant with size. Thus, particle texture may have a greater effect for smaller parti- cles. Roque et al. (51) noted the strong effect of texture in AASHTO T304 tests. Hudson (50) states: Test method C relates to the materials “as-is,” or “in-situ.” Little or no shape information can be determined from this method as the reduction in voids content that would be attrib- uted to improved particle shape cannot be separated due to the influence of the sample gradation. Researchers have also investigated the effects of alternate gradations. Hossain et al. (49) evaluated a gradation typical of dense-graded HMA that included material passing the 4.75-mm sieve and retained on the 2.35-mm sieve. Alternate gradations are strongly correlated with the Method A grada- tion (49, 51). The blended uncompacted voids contents were on average 2.4% lower when the material retained on the 2.36-mm sieve was included (49). Hudson (50) stated that the current AASHTO T304 equipment was not suitable for test- ing the material passing the 4.75-mm sieve and retained on the 2.36-mm sieve because the outlet orifice and the receiv- ing container were both too small. Virginia Test Method 5, which uses an enlarged version of the AASHTO T304 appa- ratus, produced identical uncompacted void contents when the Method A grading was tested in both devices (53). Based on the preceding research, altering the AASHTO T304 Method A gradation or fixtures would appear to shift uniformly the uncompacted void contents for all aggregates. Several concerns have been expressed regarding the use of the FAA test as a screening tool for rutting resistance of fine aggregate. There is concern that some 100% crushed particles do not meet the minimum requirements (>45) for mixes used in the upper 100 mm of the pavement structure with traffic lev- els in excess of 3 million ESALs during the design life (54). Typically, these particles are extremely cubical in nature. A second concern is that particles passing the 4.75-mm sieve but retained on the 2.36-mm sieve are not evaluated for angular- 25 ity or shape under the current Superpave aggregate proper- ties (52). Work by Hudson (50) indicates that the current AASHTO T304 apparatus may not be appropriate for testing particles in this size range. A third concern is related to the variability of the test procedure and its dependence on the fine aggregate dry bulk specific gravity (52). Finally, there is concern that the FAA test may not be related to the rutting propensity of the HMA mixture. These concerns led to numerous studies evaluating the FAA test as well as alternative tests to relate FAA, shape, and texture to the rutting performance of HMA mixtures. Commonly used alternative tests will be discussed prior to efforts to relate FAA to HMA performance. 2.4.3 Alternative Methods of Measuring FAA 2.4.3.1 Direct Tests (Digital Imaging Methods) In the past few years, digital image processing technique has been introduced into the HMA industry to analyze macro- and microstructures of HMA, aggregates, air voids, gradation, and so on. Several researchers have attempted to use image analysis to measure the FAA. Particle shape from image analysis, automated image analysis, and morphology analysis from profile images and from 3-D images are some of the image analysis methods being used actively in recent years. Particle Shape from Image Analysis: This automated tech- nique was developed at the University of Arkansas for FHWA (55). The fine aggregate is spread on a glass plate, and a high- resolution video camera is used to capture the image of each particle. Modern digital imaging hardware, image analysis techniques, and computerized analysis were used to quantify aggregate shape. EAAP (ellipse-based area of the object divided by the perimeter squared) Index and Roundness Index were found to have the most potential for predicting rutting performance. Automated Image Analysis: The automated image analy- sis approach was developed by Massad et al. (56). Two procedures—surface erosion-dilation technique and fractal- behavior technique—were used to quantify FAA. The surface erosion-dilation technique consists of subjecting the aggre- gate surface to a smoothing effect that causes the angularity elements to disappear from the image. The aggregate angular- ity is measured in terms of a surface parameter, which is defined as the area lost during the erosion-dilation process as a percentage of the total area of the original image. The fractal-behavior technique uses image-analysis tech- niques to capture the aggregate boundary. Fractal length of the boundary is the slope of effective-width-to-number-of- cycles relationship. The fractal length increases with aggre- gate angularity.

Morphology Analysis from Profile Images and 3-D Images: Similar techniques have been applied by Wang and Moham- mad (57) and Kecham and Shashidhar (58) to evaluate parti- cle size, shape, angularity, and texture of aggregate. 2.4.3.2 Indirect Tests Standard Test Method for Index of Aggregate Particle Shape and Texture (ASTM D3398): In this test method, the sample is first broken down into individual sieve frac- tions. Thus, the gradation of the sample is determined. Each size of material is then separately compacted in a cylindrical mold using a tamping rod at 10 and 50 drops from a height of 2 in. The mold is filled completely by adding extra mate- rial so that it levels off with the top of the mold. The weight of the material in the mold at each compactive effort is deter- mined, and the percent voids is computed. A particle index for each size fraction is then computed, and, using the gra- dation of the sample, a weighted average particle index for the entire sample is also calculated (16). Direct Shear Test (ASTM D 3080): The direct shear test (DST) method is used to measure the angle of internal frac- tion of a fine aggregate under different normal stress condi- tions. A prepared sample of the aggregate under considera- tion is consolidated in a shear mold. The sample is then placed in a direct shear device and sheared by a horizontal force while known normal stress is applied (16). DST is probably the most straightforward way to determine the stress-dependent shear strength of fine aggregate. Research conducted by Fernandes et al. (59) found that direct shear strength may provide a more relevant parameter to evaluate fine aggregates. The researchers also stated that the DST is significantly more complex and less repeatable than the FAA test, and its relation to the per- formance of fine aggregates needs to be further verified and developed. CAR Test: The CAR test method was developed to evaluate shear resistance of compacted fine aggregate (60, 61). It is similar to the Florida bearing ratio test (61). In this method, fine aggregates are compacted in a 100-mm mold following the Marshall hammer method using 50 blows applied to only one face of the specimen. The compacted sample height was maintained as 63.5 mm. The CAR stability was measured by applying a compressive load using the Marshall test machine. The compacted sample, while still in the mold, is placed in the Marshall test machine in the upright position. A load of 50 mm/min is transmitted through a 37.5-mm-diameter steel cylinder on the plane surface of the compacted sample. The highest load that one specimen can carry was reported as the CAR stability value. This test is believed to be a performance- related method of measuring FAA (61). 26 2.4.4 Relationships Between Fine Aggregate Shape, Angularity, and Texture and HMA Performance 2.4.4.1 Introduction The following section describes 12 studies relating FAA to HMA performance. Because of the controversy over the fine aggregate uncompacted voids test, the studies are dis- cussed individually and in some detail. 2.4.4.2 NCAT National Rutting Study by Cross and Brown Cross and Brown (17) reported relationships between aggregate properties and field rut depth obtained from a national rutting study. The study indicated the aggregate prop- erties had little relationship with rutting when the in-place air voids of the pavement section were less than 2.5%; however, relationships between aggregate properties and field rut depths were observed for pavement sections with in-place air void contents in excess of 2.5%. A relationship with an R2 = 0.67 was determined between the National Aggregate Association (NAA) Flow Test Method A, which is the basis of AASHTO T304, and the pavement rut depth divided by the square root of the applied ESAL. The relationship was developed from the analysis of data from 13 pavements. The pavement rut depth divided by the square root of ESALs was used to account for the fact that greater truck traffic was likely to produce greater pavement rut depths. A rutting model with an R2 = 0.77 was developed between rate of rutting and aggregate properties with data from pave- ments with in-place air voids in excess of 2.5%. The aggre- gate properties considered included coarse aggregate crushed faces, uncompacted voids in fine aggregate, gradation pa- rameters, and both nominal and maximum aggregate size divided by lift thickness (17). Only two factors—percent of coarse aggregate with two or more crushed faces and uncom- pacted voids in fine aggregate—were included in the model (Equation 1). (1) where P = predicted rate of rutting, rut depth (mm)/square root ESAL; CF = two or more crushed faces in coarse aggregate (%); and NAA = NAA uncompacted voids, (%). In 1992, Cross and Brown (10) reported that a rutting rate of 0.005842 mm per square root ESALs delineated good per- forming pavements from rutted pavements. Using this crite- P CF NAA= − −0 080038 0 00008 0 00151. . ( ) . ( )

rion and the relationship between the NAA flow test and rut- ting rate (17), Kandhal et al. (20) determined a minimum uncompacted voids content of 43.3%. Cross and Brown (17) developed several additional models relating uncompacted voids content and air void contents of recompacted speci- mens using various compaction methods. 2.4.4.3 Evaluation of Particle Shape and Texture of Mineral Aggregates Used in Pennsylvania by Kandhal et al. Kandhal et al. (20) evaluated 18 sources (8 natural and 10 manufactured) of fine aggregate from Pennsylvania using ASTM D3398 and both Methods A and B of the NAA uncom- pacted voids test. They observed an overlap between the nat- ural and manufactured sands in that one manufactured sand, a limestone, produced both a particle index (12.8) and an NAA uncompacted void contents (Method A = 43.1) that were lower than those of several natural sands. The authors concluded that a minimum particle index of 14 and NAA uncompacted voids content Method A of 44.5 separated between natural and manufactured sands with confidence lev- els of 86% and 82%, respectively. During the development of the Superpave method, an expert panel using a modified Delphi process determined the consensus aggregate properties (1). During the fifth round of questionnaires used as part of the Delphi process, the expert panel recommended minimum uncompacted voids of 42.8% for pavements with design traffic levels less than 300,000 ESALs and 44.2 for pavements with design traffic levels less than 10 million ESALs. These values represented the expert panel’s average recommendations for pavement layers in the top 50 mm of the pavement structure. The recommended uncompacted void levels were reduced to 41.4% and 42.8%, respectively, for layers at a depth of 127 mm. 2.4.4.4 Evaluation of Natural Sands Used in Asphalt Mixtures by Stuart and Mogawer Stuart and Mogawer (62) conducted a study to evaluate different methods of measuring fine aggregate shape and tex- ture. Twelve materials were evaluated in the study: five nat- ural sands with a poor performance history, four natural sands with a good performance history, and three manufactured (crushed) sands with a good performance history. Five meth- ods were used to characterize the sands: NAA uncompacted voids Method A, DST, ASTM D3398, Michigan Test Method 118-90, and a flow rate method. Michigan test method 118-90 is similar to the NAA uncompacted voids test in that the vol- ume of voids in a loosely compacted sample is used to deter- mine the air voids–to–solids ratio and, in turn, an angularity index. The volume of voids is determined in water in a grad- 27 uated cylinder, which eliminates the need for a bulk specific gravity; however, the results are affected by the aggregate absorption. Based on Michigan DOT’s survey responses, this test method has been replaced by AASHTO T304. The flow rate test uses the NAA apparatus. The flow rate is determined by dividing the volume of a 500-g sample of fine aggregate by the time it takes to flow through the NAA orifice. A shape- texture index is calculated from the flow time by dividing the flow time from a standard set of steel balls by the flow time for the fine aggregate. Standardized gradations were used for the study. Previous studies had evaluated the as-received gra- dations and recommended a standardized gradation for the NAA uncompacted voids test, Michigan Test Method 118-90, and flow rate test (63). The twelve sands were ranked by each of the test methods based on the average test value. The best method of differen- tiation was the flow time test. This was also the easiest para- meter to obtain. ASTM D3398 correctly differentiated all of the poor-quality sands from the good-quality sands. The weighted particle index that divided good- and poor-performing materi- als was between 11.7 and 13.9. NAA uncompacted voids Method A ranked one of the poor-quality materials the same as one of the good-quality materials. Both had an uncom- pacted voids content of 44.7%. However, the test procedure for the poor sand was violated because the sand did not have size fractions retained above the 0.600-mm sieve. Thus three size fractions were excluded from the standard. Mogawer and Stuart (63) concluded that 44.7% uncompacted voids would divide good- and poor-performing sands for high traffic lev- els. The remaining methods, Michigan Test Method 118-90 and the DST, did not differentiate the sands as well. The authors noted that the DST was time consuming. Attempts were made to differentiate between the rutting performance of HMA produced with four of the sands, two of good quality and two of poor quality. Twelve aggregate blends with levels of 10%, 20%, and 30% of each of the sands were tested with the GLWT, the French Laboratorie Central des Ponts et Chaussées (LCPC) Pavement Rutting Tester, and the USACE’s Gyratory Testing Machine. The remainder of the mix was made up of a good-quality traprock coarse aggre- gate and traprock crushed sand. Unfortunately, none of the rutting tests differentiated between the performance of the sands. This was most likely due to the high quality of the other aggregates (crushed traprock) used in the blend (62). Stuart and Mogawer (62) presented three additional impor- tant conclusions: 1. Methods for measuring shape and texture can only be expected to group sands into performance categories, such as high or low potential for rutting. The perfor- mance of a sand depends on its quality, the quantity used, the qualities of the other aggregates, and the traf- fic level.

2. Each sand should be tested to determine its rutting poten- tial. The methods are not sensitive enough to evaluate the blend of materials found in a job-mix formula gradation. 3. The discrepancies provided by the NAA and the Michi- gan DOT methods may be related to gradation. A sin- gle, standard gradation should be used in these methods so that the voids that they provide are only a function of shape and texture. 2.4.4.5 Investigation of the Influence of Aggregate Properties on Performance of Heavy-Duty HMA Pavements by Ahlrich Ahlrich (19) reported an investigation of aggregate parti- cle shape and texture on the permanent deformation proper- ties of HMA meeting the Federal Aviation Administration’s P-401 specification. Eleven blends meeting the P-401 grada- tion band were produced with varying amounts of crushed coarse aggregate (0%, 30%, 50%, 70%, and 100%) and vary- ing amounts of natural sand (0%, 10%, 20%, 30%, and 40%). The blends were produced using crushed limestone, crushed gravel, and uncrushed gravel. The fine aggregate portion of the blends was evaluated by visual inspection of the percent crushed particles according to CRD-C-171, ASTM D3398 (Particle Index Test), and ASTM C1252 Methods A and C (FAA test). The uncompacted voids contents of the fine aggregate portion of the 11 blends as measured by ASTM C1252 Method A ranged from 38.4% to 47.1%. ASTM D3398 and ASTM C1252 Method A both produced strong correla- tions (R2 = 0.98) with the percent crushed particles (mini- mum two fractured faces). ASTM C1252 Method A produced the best correlation with the percent of (rounded) natural sand in the blend (R2 = 0.94). ASTM C1252 Method C produced lower R2 values with both the percent crushed faces and per- cent natural sand (R2 = 0.66 and R2 = 0.71, respectively). A volumetric mix design was performed for each of the 11 blends using the USACE’s Gyratory Testing Machine. The samples were prepared with AC-20 (approximately PG 64- 22). Samples were tested using a triaxial (confined) repeated load creep test at 60°C. Three properties were used to evalu- ate the rutting propensity of the mixtures: permanent strain, creep modulus, and slope of the deformation curve. The com- posite (coarse and fine aggregate) particle index measured by ASTM D3398 produced the best correlation with all three parameters (R2 = 0.78, 0.69, and 0.71, respectively). ASTM C1252 Method A produced better correlations with all three parameters than the other two fine aggregate tests (ASTM D3398 and percent crushed particles). The R2 values ranged from 0.29 to 0.41. The better correlation with the compos- ite aggregate index from ASTM D3398 is not unexpected because the coarse aggregate fraction was also varied between the blends. Ahlrich (19) concluded, On the basis of the strong correlations and simple test proce- dure, the promising alternatives for specification require- 28 ments to characterize aggregate particle shape and texture instead of percent crushed particles are modified ASTM C1252 for the coarse aggregate fraction and ASTM C1252 for the fine aggregate fraction. 2.4.4.6 Study of the Contribution of FAA and Particle Shape to Superpave Mixture Performance by Huber et al. Huber et al. (34) conducted a study to assess the contribu- tion of FAA and particle shape to the rutting performance of a Superpave-designed HMA. Four fine aggregates were selected for the study: a Georgia granite; Alabama limestone; Indiana crushed sand (geology not identified, most likely limestone); and Indiana natural sand. The uncompacted void contents (AASHTO T304 Method A) of the four aggregates were measured as 48, 46, 42, and 38, respectively. A refer- ence mixture was prepared with the Georgia granite (coarse and fine aggregate) and a PG 67-22 binder. The other three aggregates were sieved into size fractions and substituted for the granite fine aggregate to produce four mixtures, keeping the gradation constant. All four blends were mixed at the optimum asphalt content determined for the granite blend. No adjustment was made for variances in asphalt absorption between the fine aggregates. The resulting mixtures were tested in the Couch Wheel Tracker (a modified Hamburg Wheel Tracker), the APA, and the SST using the frequency sweep test. The rutting tests did not appear to differentiate between the blends in a consistent manner or at all in some cases. The authors concluded that the choice of coarse aggregate may have masked the effect of the fine aggregate (34). There was not a correlation between any of the tests and the uncompacted void contents. This finding is not unexpected because there were not significant differ- ences between the rutting results. 2.4.4.7 NCHRP Project 4-19 by Kandhal and Parker NCHRP Project 4-19, “Aggregate Tests Related to Asphalt Concrete Performance in Pavements,” (2) evaluated fine aggregate tests related to rutting performance. Three tests were used in the study: ASTM D3398, AASHTO T304 Method A, and particle shape from image analysis (the University of Arkansas Method). Used in this study were nine fine aggre- gate sources with a range in uncompacted void contents of 40.3% to 47.5%. Three of the materials were natural sands. The fine aggregates were mixed with an uncrushed gravel coarse aggregate. All of the mixes were produced using the same gra- dation, above the maximum density line. The coarse aggregate and gradation were chosen to emphasize the response of the fine aggregate. The aggregate was mixed with a PG 64-22 binder. A mix design was conducted for each mixture using

an Ndesign level of 119 gyrations to determine optimum asphalt content. The resulting mixtures were tested using the GLWT and the SST. Simple shear at constant height and frequency sweep at constant height were performed using the SST. Poor correla- tion coefficients were observed between all three fine aggre- gate tests and the SST results. The index of aggregate shape and particle texture from ASTM D3398 produced the best correlation with the GLWT rut depths (R2 = 0.67). The uncompacted void contents produced a slightly lower corre- lation (R2 = 0.60). The authors noted that the uncompacted voids were highly correlated with the aggregate index (R2 = 0.99) and that the uncompacted voids test was much simpler to run. They therefore recommended AASHTO T304 to quan- tify fine aggregate particle shape, angularity, and surface tex- ture. The Roundness Index from the University of Arkansas digital image analysis produced a fair correlation with the GLWT rut depth (R2 = 0.56). 2.4.4.8 Study of the Effect of FAA on Asphalt Mixture Performance by Lee et al. Lee et al. (64) conducted a study on the effect of FAA on HMA performance for the Indiana Department of Transpor- tation. The study included six fine aggregate sources, which were used to produce 18 9.5-mm NMAS mixtures using dif- ferent gradations and blends of the fine aggregate. Only one of the fine aggregate sources was a natural sand. The coarse aggregate used for all 18 mixtures was a partially crushed (80% one crushed face) gravel. The angularity and texture of the fine aggregate sources were evaluated using ASTM C1252 Method A (FAA test), CAR test, and Florida Bearing Value (Indiana Test Method 201-89). The Florida Bearing Value is a precursor to the CAR test. Instead of using a Mar- shall press, the sample was loaded through the flow of lead shot into a receiving container. The uncompacted voids con- tent of the fine aggregate ranged from 38.7 to 49.0. Blends of the six sands were prepared to produce uncompacted void contents of 46, 45, and 43. Regression analysis indicates an R2 = 0.70 between the uncompacted voids and CAR peak load. The trend indicated an increase in CAR peak load with an increase in uncompacted voids. Volumetric mix designs were conducted for each of the 18 mixtures. The first nine mixtures were produced one each with the six sands and three blends of those six sands. Nine additional mixtures were produced, five using a slag sand with varying percentages of natural sand and mineral filler and four with a limestone sand (S gradation mix) and differ- ent percentages of natural sand. Rut testing was performed on the mixtures using the PurWheel Laboratory Tracking Device and the SST. The PurWheel device applies loads to the slabs of HMA with a rubber wheel having a contact pres- sure of 620 kPa. PurWheel testing was conducted on dry slabs at 60°C. SST testing for frequency sweep at constant 29 height and repeated shear at constant height were performed according to AASHTO TP7-94. Correlation analysis between the three fine aggregate tests and rutting performance based on both repeated shear at con- stant height and the PurWheel rut depths indicated that the uncompacted voids content was most correlated with rutting performance (64). A stepwise regression was performed to predict the rutting performance of the mixtures using the Pur- Wheel. The independent variables considered were uncom- pacted voids content, asphalt content, air voids content (of the PurWheel samples), dust to asphalt ratio, gradation parame- ters, the interaction between uncompacted voids and asphalt content, and the number of loading cycles to 2% shear strain from the repeated shear at constant height test. Six of the eight variables were included in the model by the stepwise regression: uncompacted voids, asphalt content, air voids, the interaction between uncompacted voids and asphalt con- tent, cycles to 2% strain in the SST, and gradation. The uncompacted voids content was the most significant param- eter (F-value = 41.00). Comparing the aggregate properties individually to the rutting results from the PurWheel device and repeated shear at constant height, FAA had the highest correlation with the PurWheel results (R2 = 0.40) and the Florida Bearing Ratio had the highest correlation with the repeated shear at constant height (R2 = 0.29). The authors concluded that uncompacted voids alone may not be suffi- cient to evaluate the fine aggregate contribution to mixture rutting performance. It was observed that a mixture having an uncompacted voids content of 43 performed as well as a mixture with an uncompacted voids content of 48. The authors note that this may be due to the confounding effects of gra- dation and compactability (the uncompacted voids content of 48 represents the slag mixtures). 2.4.4.9 Pooled Fund Study 176 One of the goals of the National Pooled Fund Study No. 176, “Validation of SHRP Asphalt Mixture Specifications Using Accelerated Testing,” was to examine the effect of FAA on the rutting performance of Superpave mixtures. Two coarse aggregates—a limestone and granite—and three fine aggregates—a natural sand, limestone sand, and granite sand— were used in the study (65). The fine aggregates had uncom- pacted void contents of 39, 44, and 50, respectively. The aggregates were combined with a neat PG 64-22 to pro- duce 21 mixture designs: 9 of 9.5-mm NMAS and 12 of 19.0-mm NMAS. A trend was observed between the design asphalt content and the uncompacted voids content. The rela- tionship indicated that for a given gradation shape (above, through, or below the maximum density line), optimum asphalt content increased with increasing uncompacted voids. The rutting propensities of the mixes were tested with the PurWheel, the SST, and Triaxial Tests and in the APT facil- ity. The APT facility is a full-scale, indoor accelerated load- ing facility managed by Indiana DOT and Purdue University.

The primary goal of the Phase I testing was to evaluate the sensitivity of the various test methods to the study factors (66). Based on screening tests performed with the PurWheel device in Phase I of the study, four mixtures were selected for APT facility testing. A limestone coarse aggregate was used to produce 19.0-mm NMAS mix designs using all three sands. The natural sand (FAA 39) and limestone sand (FAA 44) were used to produce coarse-graded mixes (below the maximum density line). The limestone sand and granite sand were used to produce fine-graded mixes (above the maximum density line). These four mix designs were placed at both low and high in-place densities. The results of the APT facility testing are shown in Table 7. It is apparent that both mixtures produced with the limestone sand (FAA 44) had design asphalt contents that were approx- imately 1 percentage point less than the mixtures produced with the natural or granite sand. For the low-density sections, the crushed limestone sand (FAA 44) produced both the best and worst rutting results in the APT facility; however, the dry PurWheel results ranked both of the limestone sand mixtures as performing the best. For the high-density (low air void) sections, the limestone sand mixtures performed best in both 30 the PurWheel and the APT facility. However, it should be noted that the air void contents of the natural sand and gran- ite fine aggregate sections were close to the 2.5% level iden- tified by Cross and Brown (10, 17) below which mixtures were less sensitive to aggregate properties. The air void con- tents of the limestone fine aggregate sections (FAA 44) were approximately 2.5 percentage points higher than the natural sand and granite fine aggregate sections. These variations were not planned but are part of the variation associated with full-scale test sections. Thus, although the limestone fine aggregate indicated the best rutting performance for the high- density sections, this result may be more related to the higher in-place air voids and lower asphalt contents of those mix- tures than to the performance of the fine aggregate. This emphasizes the fact that screening tests for FAA and texture cannot by themselves ensure mixture performance. In Phase II of Pooled Fund Study 176, an additional 6 mix- tures were tested in the APT facility for a total of 10 mixtures and in excess of 20 sections (considering varying densities and asphalt contents). Stiady et al. (67) discussed the findings relative to aggregate. Based on Figure 5, the rounded natural sand (FAA 39) produced the worst rutting performance; Figure 5. APT facility rutting versus uncompacted voids content by gradation type (67). Mixture (FAA, Gradation) Design Asphalt Content, % Average As- Constructed Wheel Path Air Voids, % APT Rut Depth, mm (Adjusted for 76-mm layer thickness) PURWheel Dry Test Ranking Low Density Sections 44 ARZ 4.6 8.8 5.3 1 50 ARZ 5.9 6.4 6.3 3 39 BRZ 5.5 5.2 9.4 4 44 BRZ 4.6 6.4 11.8 2 High Density Sections 44 ARZ 4.6 5.3 4.3 1 44 BRZ 4.6 5.7 8.0 2 50 ARZ 5.9 2.9 9.3 3 39 BRZ 5.5 2.6 15.7 4 TABLE 7 INDOT/Purdue APT facility results from Phase I of Pooled- Fund Study 176 (65)

however, the limestone fine aggregate (FAA 44) performed as well or better than the granite fine aggregate (FAA 50). The mix designs produced with the granite fine aggregate had consistently higher asphalt contents. Analysis of variance (ANOVA) performed on the triaxial shear strength test results from the 21 mixtures indicated that the uncompacted void contents for the fine aggregates in the mixtures were a sig- nificant factor (66). 2.4.4.10 Evaluation of Superpave FAA Specification by Chowdhury et al. Chowdhury et al. (54) conducted a study to evaluate vari- ous measures of FAA and texture and their relationship to rut- ting performance. The study was conducted for the Interna- tional Center for Aggregate Research. The study evaluated 23 fine aggregates using seven different procedures: uncom- pacted voids content (AASHTO T304), DST (ASTM D3080), CAR test, three different methods of digital image analysis, and visual inspection. The image analysis techniques included the Hough Transform by the University of Arkansas, which was discussed previously; unified image analysis by Wash- ington State University; and the VDG-40 Videograder con- ducted by the Virginia Transportation Research Council. The samples tested by Washington State University were sieved, and only the material passing the 1.18-mm sieve and retained on the 0.600-mm sieve was used for analysis. The aggregates were stained black to improve their contrast with the background prior to capturing the images. An optical microscope linked to an image analyzer was used to capture images of the fine aggregate. Three techniques were used to analyze the binary image: surface erosion-dilation, frac- tal behavior, and form factor (56). Surface erosion-dilation involves removing layers of image pixels on the fringe of the 31 object (i.e., erosion) followed by replacement of these pixels (i.e., dilation) to simplify the form. The surface parameter is believed to be a measure of angularity and is calculated as the percentage of the particle area lost after six cycles of erosion followed by six cycles of dilation (56). Fractal behavior is defined “as the self-similarity exhibited by an irregular boundary when captured at different magnifications” (56). Fractal length increases with an increase in aggregate angular- ity. Form factor describes an object’s dimensions, particularly surface irregularity. The form factor of a perfectly circular object is 1; therefore, form factor decreases with increasing surface irregularity. The VDG-40 Videograder was developed by LCPC, the French national road and bridges laboratory (68). The device was developed primarily to measure aggregate grading of particles larger than 1 mm (No. 16 sieve), but it can also mea- sure shape properties. Aggregates are backlit as they fall in front of a linear charged couple device camera, which pro- duces a line scan image of the aggregate. The aggregates fall off a rotating wheel, which prevents them from tumbling as they fall in front of the camera. An ellipse having the same length and area is fit to each particle. The ratio of the length to the width of each particle is reported as the slenderness ratio (SR). The SR may be determined as a distribution or an average. The flatness factor is a property for the group of aggregates tested; it is related to the ratio of the average width to average thickness of the particles. Based upon the data presented in the paper (54), a correla- tion matrix was developed between the indices for angularity determined with each test method (Table 8). (See Chowdhury et al. for some of the correlations [54, 69].) Regression analy- sis was performed using Minitab statistical software. The upper number in the cell is the coefficient of determination (R2) and the lower number is the significance level (p-value) based on the ANOVA. Test Procedure UV, AASHTO T304 Method A Angle of Internal Friction (AIF) ASTM D3080 Log CAR Stability University of Arkansas K-index University of Washington Surface Parameter (SP) VDG-40 Slenderness Ratio (SR) UV 1 0.0002 0.07 1.00 0.222 0.17 0.050 0.76 0.000 0.72 0.000 0.47 0.000 AIF 1.00 0.000 0.53 0.000 0.06 0.244 0.05 0.292 0.22 0.028 Log CAR 1.00 0.000 0.20 0.031 0.16 0.061 0.72 0.000 K-index 1.00 0.000 0.69 0.000 0.50 0.000 SP 1.00 0.000 0.43 0.001 SR 1.00 0.000 1Coefficient of determination (R2) 2ANOVA level of significance (p-value) TABLE 8 Correlation matrix for fine aggregate test results using data from Chowdhury et al. (54)

The uncompacted voids content correlated well with two of the digital imaging methods, K-index (R2 = 0.76) and sur- face parameter (R2 = 0.72), and had a fair correlation with the SR (R2 = 0.47). The relationships between uncompacted voids and all three direct measures of fine aggregate particle shape were significant based on the ANOVA. The authors noted that four crushed limestone aggregates that have good field performance histories showed high values of K-index even though their uncompacted voids contents were less than 45 (54). Kandhal and Parker (2) also found a good relation- ship between the EAPP (i.e., ellipse-based area of the object divided by the perimeter squared) and uncompacted voids content (R2 = 0.76) as measured by the University of Arkansas Hough Transform. Uncompacted voids content cor- related poorly with both angle of internal friction (AIF) and the Log of CAR stability. There was a fair correlation between the two shear mea- surements, AIF and Log of CAR stability (R2 = 0.53). AIF did not correlate well with any other test, although the rela- tionship with the VDG-40 SR was significant (p-value = 0.028). Log of CAR stability correlated well with the VDG-40 SR (R2 = 0.72). There was a fairly good correlation between K-index and surface parameter (R2 = 0.69); both methods had moderate correlations with the VDG-40 SR. The authors noted (54): The CAR test appears to separate uncrushed and crushed aggregates much better than the FAA test. This could be, in part, due to the high filler content of the crushed materials as compared to the sands. A laboratory rutting study was conducted with four of the fine aggregates: three crushed materials and one natural sand. Two blends of materials were also produced using two of the crushed materials, one with 15% and the other with 30% of the natural sand. A single limestone coarse aggregate and a coarse 19.0-mm NMAS gradation were used for all of the mixtures. The binder grade was not reported. Superpave mix designs were performed for each of the six blends. The mixtures pro- duced using the natural sand and blend with 30% natural sand did not meet the Superpave minimum VMA requirements. Cylindrical samples at 4 ± 1% air voids were tested in the APA at 64°C with a 445-N (100-lb) vertical load and 694 kPa (100 psi) hose pressure. Regression analysis indicated a fair to poor relationship (R2 = 0.37) between uncompacted voids and APA rut depth (54). The mix with 100% natural sand fines (FAA = 39.0) had the highest rut depth (9.2 mm) fol- lowed closely by the mix with the crushed river gravel fines (FAA = 44.3, rut depth = 9.1 mm). The mix containing the crushed river gravel had the highest asphalt content of all of the mixes evaluated (tied with granite/natural sand blend). The mix with the granite fines (FAA = 48.0) had the least amount of rutting (4.0 mm), followed closely by the mix with the limestone fines (FAA = 43.5, rut depth = 4.4 mm). This illustrates the concern with the current uncompacted voids 32 specifications. Based on laboratory results, it is possible to design mixes using fine aggregate that fails the uncompacted voids criteria but produces acceptable rutting performance. Regression analysis using data provided by Chowdhury et al. (54) did indicate a good relationship between uncompacted voids and VMA (R2 = 0.70). This suggests that uncompacted voids may also identify fine aggregates that will assist in meeting minimum VMA requirements. Angle of internal friction, as tested by ASTM D3080, pro- duced the best relationship (R2 = 0.69) with the APA rut depths (54). Log of CAR stability and the VDG-40 SR pro- duced fair correlations (R2 = 0.46 and 0.42, respectively). No correlation (R2 = 0.07) was found with the Washington State University surface parameter discussed previously, but a fair correlation (R2 = 0.58) was found with a second parameter, fractal length. 2.4.4.11 Evaluation of Superpave Criteria for VMA and FAA for Florida DOT by Roque et al. Roque et al. (51) conducted a study on FAA for the Florida DOT. A total of nine fine aggregates were included in the study: six limestone sources, two granite sources, and a gravel source. The fine aggregates were evaluated using AASHTO T304 Methods A, B, and C as discussed previously; using the ASTM D3080 (DST); and visually. Two alternative grada- tions, other than that specified in AASHTO T304 Method A, were also evaluated (51, 59). These gradations were selected to represent the range of fine aggregate gradations used in the study. The authors concluded that “material type had a far greater effect on FAA than did gradation. Furthermore, all three gradations appeared to result in the same relative FAA rankings for the fine aggregates tested” (51). A poor correla- tion (R2 = 0.32) was observed between the uncompacted voids content and direct shear strength when both tests were conducted using the AASHTO T304 Method A gradation. The trend indicates decreasing shear strength with increasing uncompacted voids content. This may be due to the packing characteristics of the fine aggregates with higher uncom- pacted voids contents. The authors conclude that “although FAA had some influence on the shear strength, aggregate toughness and gradation appeared to overwhelm its effects, confirming that FAA alone was not a good predictor of fine aggregate shear strength” (51). Five of the fine aggregate sources, three limestone sources, a granite source, and a gravel source were used to evaluate the effect of fine aggregate on mixture performance. A sin- gle limestone source was used as the coarse aggregate and to develop a reference coarse and fine gradation commonly used in Florida. The four other fine aggregates were used to volu- metrically replace the reference aggregate. The material pass- ing the 4.75-mm sieve was replaced for the coarse gradation,

and the material passing the 2.36-mm sieve was replaced for the fine gradation. Volumetric replacement was done to account for any differences in specific gravities between the materials. Superpave mixture designs were performed for each of the 10 blends using Ndesign = 109 compaction level. The binder grade was not reported. Six of the ten mix designs failed one or more Superpave criteria. Two of the three limestone sources failed minimum VMA (14% minimum for 12.5-mm NMAS). The granite source failed the voids filled with asphalt (VFA) requirements on the high side because of a high VMA (16%). The authors noted that “the FAA did appear to identify substandard VMA mixtures” (51). Rutting tests were performed with the APA. Test tempera- ture and loads used in the APA were not reported. The results for the fine mixtures are reported by Roque et al. (51). The authors state that the rutting results agree with the direct shear results, aggregate toughness, and known field performance. The trend between uncompacted voids and APA rut depths indicated decreased rutting with increasing uncompacted voids. Two fine aggregates with uncompacted voids less than 45 and high toughness (LA abrasion < 35%) exhibited a rut depth equivalent to a fine aggregate with an uncompacted voids content in excess of 45. Roque et al. (51) recommend including aggregate toughness as part of the AASHTO T304 acceptance criteria. Aggregates with uncompacted voids between 42 and 50 would be acceptable with LA abrasion values of the parent rock less than 35%. If the LA abrasion of these fine aggregates were to exceed 35%, their rutting performance may not be adequate. 2.4.4.12 Evaluation of the Effect of FAA on Compaction and Shearing Resistance of Asphalt Mixtures by Stackston et al. Stackston et al. (70) conducted a study to evaluate the effect of FAA on compaction effort and rutting resistance. Three aggregate sources were used in the study. Twenty-four Superpave mix designs were developed using blends of the three materials and two gradation shapes: fine and s-shaped. The response of the mixtures was evaluated using Superpave volumetric properties and the gyratory load plate assembly. The gyratory load plate assembly measures the force on the sample at three points. This force is converted to a force per cycle. Testing indicated that the density at Ninitial decreases with increasing uncompacted voids content. This indicates that mixes with higher uncompacted voids contents would be less likely to be tender mixes. Data from the gyratory load plate assembly indicated that mixes with higher uncompacted voids contents are harder to compact. The authors reported that the effect of uncompacted voids content was not consis- tent in terms of rutting resistance as measured by the gyra- tory load plate assembly (70). 33 2.4.4.13 NCHRP Project 4-19(2) Ongoing research as part of NCHRP Project 4-19(2), “Validation of Performance-Related Tests of Aggregates for Use in Hot-Mix Asphalt Pavements,” is examining the rela- tionship between uncompacted voids tests and rutting through accelerated testing using the Indiana prototype APT facility. Six fine aggregates were initially selected for the fine aggre- gate characterization portion of the study: crushed gravel, granite, dolomite, traprock sands, and two natural sands. The uncompacted void contents (Method A) for these sands ranged from 40.3 to 49.1 (23). Later, alternative dolomite and traprock sands were included that produced HMA mixtures with better volumetric properties (uncompacted void con- tents of 46.8% and 49.2%). The study tracked the measured uncompacted void con- tents from the HMA mix design through field construction. On average, a 1.8% reduction in voids was observed between the HMA mix design value and material recovered from HMA samples taken at the asphalt plant. Rismantojo states that “the degradation was significantly correlated with the initial UVA [uncompacted voids] values. Fine aggregates with high ini- tial UVA values appeared to degrade more than those with low UVA values” (23). Mixture designs were performed with all eight fine aggre- gates using a single uncrushed gravel coarse aggregate to amplify the effect of the fine aggregate. The original dolomite and traprock sources produced VMA values that were exces- sively high (17.4% and 18.0% at Ndesign = 100 gyrations). This resulted in failing VFA values (exceeding 75%). The mix- tures produced using the other original fine aggregates and two replacement aggregates met all of the Superpave crite- ria. Correlations were performed between the volumetric prop- erties and measured fine aggregate properties. Uncompacted voids produced a significant correlation (R2 = 0.59) with den- sity at Ninitial (23). A model was developed to relate uncom- pacted voids and dust proportion to VMA. As expected, VMA increased with increasing uncompacted voids and decreasing dust proportion (23). The six mixtures with passing Superpave volumetric prop- erties were tested in the full-scale Indiana APT facility. The results indicate that uncompacted voids Methods A and B as well as the uncompacted voids from Virginia Test Method 5 (VTM 5) were significantly related to the total rut depth after 1,000 passes. The R2 = 0.65 for Method A was slightly less than for the other two methods. AASHTO T304 Method A produced the best relationship with the total rut depth after 20,000 passes (R2 = 0.51); however, the relationship was not significant (p-value = 0.286) (23). The author noted that the decrease in rut depth with increasing uncompacted voids occurs to a lesser extent above 45% voids. Rismantojo (23) concludes that the results of the current study are similar to those reported by Kandhal and Parker (2), including that fine- graded mixtures with uncompacted voids contents (Method A) between 42% and 46% demonstrate similar levels of rut- ting resistance.

2.4.5 Precision of AASHTO T304 AASHTO T304 reports a single-operator standard devia- tion (Std) of 0.13% voids and a multilaboratory standard devi- ation of 0.33% voids (71). This means that two properly con- ducted tests should not differ (D2S) by more than 0.37% and 0.93% voids, respectively, for a single operator and between two different labs. AASHTO T304 testing is included as part of the AASHTO Materials Reference Laboratory (AMRL) proficiency samples testing program. The precision results for the four latest proficiency samples are shown in Table 9. The average uncompacted voids contents for the samples tested in Table 9 ranged from 42.7% to 44.7%. The data in Table 9 indicates that AASHTO T304 is more variable in practice than reported in the test method. The Southeast Asphalt User/Producer Group conducted a round-robin for AASHTO T304 Methods A, B, and C. The study included seven aggregate sources from the southeastern United States: two natural sands, two granite sources, two limestone sources, and standard graded sand. The standard graded sand had been previously used to establish the precision statement for AASHTO T304. Sixteen laboratories participated in the study, although not all of the data were returned for all of the samples. The results indicated that Method C was more vari- able than Methods A and B, which had similar variability. For Method A, the single operator standard deviation was 0.57% voids and the multilaboratory standard deviation was 0.75% voids, which correspond to D2S limits of 1.61 and 2.12, respectively (72). The variability of the bulk dry specific grav- ity measurements (72) used in the calculations to determine the uncompacted void content significantly increases the test variability. The AMRL results and Southeast Asphalt User/ Producer Group Study indicate that the AASHTO T304 pre- cision statement may need to be revised. 2.4.6 Summary of Findings on Fine Aggregate Texture and Angularity The findings on fine aggregate texture and angularity are as follows: • The results of AASHTO T304 Methods A and B are highly correlated, with Method B producing larger uncompacted void contents. Tests using alternative gra- 34 dations other than Method A were also highly correlated to the Method A results and maintained the same rank- ing of fine aggregates. The results from AASHTO T304 Method C are affected by the fine aggregate gradation and are not recommended for comparing particle shape and texture. • The current Superpave consensus aggregate properties do not address the angularity of the material that pass the No. 4 sieve but are retained on the No. 8 sieve. It is doubtful that the current AASHTO T304 apparatus could accommodate material of this size fraction. • Numerous test procedures are available to assess fine aggregate texture and angularity. Several of the imaging techniques and the CAR test appear to be promising. Researchers using the DST (ASTM D3080) have indi- cated that it is difficult to obtain consistent results; how- ever, to date, the majority of the work to correlate fine aggregate shape and texture to performance has been completed using AASHTO T304 Method A. • The results of studies relating the uncompacted voids content from AASHTO T304 Method A to performance are mixed. Generally, studies indicated a trend between uncompacted voids content and improved rutting per- formance, but in some cases the trend was weak. Subtle differences in uncompacted voids content can be over- whelmed by the effect of the coarse aggregate or other HMA mixture properties. Several studies supported the 45% uncompacted voids criteria for high traffic, but several also indicated performance was unclear between 43% and 45% (or higher) uncompacted voids. There is clear evidence that good-performing mixes can be designed with uncompacted voids contents between 43% and 45%, but evaluation of these mixes using a rutting performance test is recommended. • Higher uncompacted void contents generally resulted in higher VMA and lower densities at Ninitial. • The variability of AASHTO T304 method A appears to be larger than reported in the test method. Much of this variability appears to be related to variability in the fine aggregate specific gravity measurements used to calcu- late the uncompacted voids. Ongoing research to improve fine aggregate specific gravity measurements may also benefit AASHTO T304. Multilaboratory Precision First Sample Second Sample Single Operator Precision Sample Numbers Number of Labs Std. D2S Std. D2s Std. D2s 119 120 136 0.937 2.651 1.012 2.863 0.358 1.103 123 124 183 1.129 3.194 1.149 3.250 0.406 1.147 127 128 211 1.291 3.651 1.349 3.815 0.377 1.066 131 132 242 0.917 2.594 0.858 2.428 0.381 1.077 TABLE 9 AMRL AASHTO T304 proficiency sample results (71)

2.5 IMAGING METHODS FOR THE ASSESSMENT OF AGGREGATE SHAPE, ANGULARITY, AND TEXTURE 2.5.1 Introduction The proceeding sections discussed some of the shortcom- ings of the indirect methods of measuring aggregate shape, angularity, and texture. For example, the uncompacted void tests for fine and coarse aggregate do not separate the effects of shape, angularity, and texture. Further, the indirect tests can be time consuming and are subject to testing variation based upon the experience of the technician. The sample size evaluated can be small in proportion to the quantity of mate- rial produced: for example, the percent F&E is only based on the shape of 100 particles of a given size fraction. The rela- tively poor precision statements for the uncompacted voids in fine aggregate (AASHTO T304) and F&E (ASTM D4791) demonstrate the magnitude of the test variability. By compar- ison, Maerz (73) outlines the advantages of digital systems: • Reduced unit testing cost, • Reduced technician subjectivity, • Faster results, and • Ability to test larger sample size to improve statistical validity. These advantages are somewhat offset by additional capital costs for the equipment. Digital equipment may also be more complicated, requiring a greater degree of technician train- ing. Finally, digital systems do not always provide measure- ments that are directly comparable to those of currently accepted techniques. For instance, there can be differences in gradations based on digital data as compared with wire-mesh sieves with square openings because F&E may fit through the sieve opening on the diagonal (e.g., a 1/2-in.-wide particle may fit through a 3/8-in. sieve. Several researchers have evaluated digital imaging methods to measure aggregate shape, angularity, and texture. Some of these methods have been introduced previously where they have been used in performance studies in conjunction with the currently accepted methods. NCHRP is currently sponsor- ing Project 4-30A, “Test Methods for Characterizing Aggre- gate Shape, Texture, and Angularity.” The objective of this research is to identify or develop test methods for both cen- tral and field laboratories to measure shape, angularity, and texture (74). These methods are to be applicable to HMA, hydraulic cement concrete, and unbound base materials. The following sections provide a brief overview of the major types of digital image or digital vision systems. 2.5.2 Video Imaging Systems 2.5.2.1 Early Imaging Systems The first attempts to use digital imaging to quantify aggre- gate shape involved a photocopy machine and a digitizing 35 tablet (75). The technique was viable for particles larger than the No. 8 sieve. Aggregate particles were first placed in clear trays in a “flat” orientation such that their minimum dimen- sion was orthogonal to the surface of the copier. A photocopy was then made of a group of 50 aggregates producing two dimensional images of each aggregate particle. The minimum (i.e., the third) dimension of the aggregates was then mea- sured with a vernier caliper. Both the photocopied image and the thickness determined with the vernier caliper were digi- tized by means of a digitizing tablet. Measurements on aggre- gates smaller than the No. 8 sieve were made with micro- photographs. The minimum dimension of these particles was measured by evaporating a thin film of metal onto the slide and measuring the shadow of the particle. Once the images were digitized, the data could be manipulated to determine shape factors such as elongation ratio, flatness ratio, shape factor, or surface roughness (75). A black-and-white charged couple device (CCD) camera coupled with an image analysis system replaced the use of dig- itizing tablets and photocopied images. Frost and Lai (76) cap- tured static images using a Sony black-and-white camera and a Cambridge Instruments Quantiment Q570 Image Analysis System. Coarse aggregate particles were adhered to two pieces of Plexiglas joined at a 90° angle. The Plexiglas fixture was placed on a light box, which backlighted the sample to pro- duce a high contrast between the particles and the back- ground (77). Two dimensions were acquired: the longest dimension, dL, and the intermediate dimension, dI. The Plexi- glas bracket was rotated 90°, and the shortest dimension, dS, was captured. From this data, the ratio of the principal dimensions—elongation and flatness—could be calculated along with several other measures of shape (76). Broyles et al. (78) used two black-and-white video cameras simultaneously to capture static images in three dimensions. Rows of aggregates were arranged on a stepped platform so that they could be viewed by two cameras at 90° to one another. Using this technique, the authors could complete 100 measurements of the principal dimensions of a particle in fewer than 10 min. This system could be used to calculate fre- quency distributions of flat or elongated particles for a range of ratios. In addition to shape parameters, analysis methods were developed for roughness and angularity (77). 2.5.2.2 VDG-40 Videograder LCPC developed a videograding device designed to rapidly provide a gradation analysis of a large sample (Figure 6) (68). The device is commercially available. Prowell and Weingart (41) and Weingart and Prowell (79) investigated the use of the VDG-40 videograder for determining aggregate shape. As dis- cussed previously, the device was primarily developed to mea- sure aggregate grading of particles larger than 1 mm (No. 16 sieve), but it can also measure shape properties. A sample of the aggregate (up to approximately 50 lbs) is loaded into a hop- per. A vibrating feed tray orients the aggregate particles such

that they lie flat (i.e., the longest and intermediate dimensions are visible). The aggregates fall off a rotating wheel; this pre- vents the aggregates from tumbling as they fall in front of the camera. Aggregates are backlit as they fall in front of a lin- ear CCD camera, which produces a line-scan image of the aggregate. An ellipse having the same length and area as the image is fit to each particle. The device produces a sample gradation and two estimates of aggregate shape. The ratio of the length to the width of each particle is reported as the SR. The SR may be determined as a distribution or average. The flatness factor is a property for the group of aggregates tested related to the ratio of the average width to average thickness of the particles. 2.5.2.3 WipShape The WipShape device, developed by Maerz (73, 80), uses two orthogonally mounted video cameras to capture aggre- gate images. The prototype used a vibrating feeder to pro- duce approximately a 2-in. separation between aggregate par- ticles on a black conveyor belt (80). The aggregate particles were lit from the side and above using two lamps. Problems were observed with the contrast between dark or mottled aggregates and the black feed belt (73). This led to the devel- opment of a final prototype with a circular rotating table (73). The table is translucent and allows backlighting of the aggre- gate particles. Images are captured at 60 frames/s using a pair of Sentech STC 1000 cameras. An Imaging Source DFG-BW1 digitization board captures the image from both cameras simultaneously. Custom software manages the data acquisi- tion. “Thresholding” or the identification of the grayscale pixel value that separates the aggregate particle from the back- ground, is accomplished automatically. The software fits a virtual “box” around the aggregate to determine the principal dimensions. The software determines aggregate size (grad- ing), aspect ratio (elongation or flatness), and angularity. Angu- 36 larity is determined by analysis of the average radius of cur- vature of the particle (73). 2.5.2.4 University of Illinois Aggregate Image Analyzer The University of Illinois Aggregate Image Analyzer (UI- AIA) is similar in concept to the first WipShape prototype. However, the UI-AIA, which is shown in Figure 7, uses three orthogonally mounted cameras: a top camera, side camera, and front camera. This allows an accurate determination of the volume of the particles, which in turn increases the accu- racy of mass-based calculations such as gradation and per- cent F&E by mass because the volume and mass of the par- ticle are related by the specific gravity (81). The aggregates particles are fed onto a conveyor belt moving at approxi- mately 8 cm/s with 25 cm spacing between particles. One of two sensors triggers the cameras to capture the image in sequence using LabView software. An imaginary box is fit- ted to the captured images to determine the principal dimen- sions of a particle. Then, the volume of the box not occupied by the aggregate is subtracted from the volume of the virtual box to obtain the volume and, from that volume, the mass of the particle. An angularity index was also developed for the device to supplement coarse aggregate angularity measure- ments (ASTM D5821) (82). 2.5.2.5 Aggregate Imaging System The preceding systems are primarily designed to evaluate coarse aggregate particles. The Aggregate Imaging System (AIMS) contains both a fine aggregate and a coarse aggre- gate module (83). These two modules allow the system to Figure 6. VDG-40 Videograder. Figure 7. University of Illinois Aggregate Image Analyzer.

capture measurements of shape (form), angularity, and tex- ture for both fine and coarse aggregates. The system (Figure 8) consists of a video microscope, video camera, data acquisition system, lighting system, auto- mated carriage, and associated software. The aggregate par- ticles are randomly spread on a tray. An Optem Zoom 160 video microscope is coupled with an LC-150 black-and- white CCD video camera to acquire the images. The camera is mounted on a carriage system that allows 250 mm of move- ment in the X and Y axes and 50 mm of movement in the Z axis. The Z-axis assembly can be manually moved an addi- tional 250 mm to switch from fine to coarse aggregate mea- surements. Fine aggregate measurements and coarse aggre- gate measurements of the longest and intermediate axis and information for coarse aggregate angularity are accomplished using backlighting of the aggregate tray. All other measure- ments are accomplished with top-lighting. The images are captured using a National Instruments PCI 1409 analog frame grabber. Image processing is conducted using LabView soft- ware (83). Fine aggregate analysis is based on 2-D images. The fine aggregate images are acquired to produce a resolution such that each pixel is less than 1% of the average aggregate diam- eter (84). At this resolution, the field of view includes 6 to 10 particles. Coarse aggregate analysis is based on a combination of measurements. First, aggregates are backlit, and 2-D images are captured to determine the largest and intermediate dimen- sions as well as angularity. One aggregate particle is captured in each image. The resolution of the image is set such that the pixel size is less than 1.0% of the average aggregate diameter. The third dimension of the aggregate is acquired during a sec- ond measurement pass. During this pass, the aggregates are top lit. The camera first focuses on a point on the tray. Then, the Z-axis is moved up until the top of the aggregate is in focus. The travel of the Z-axis is the third dimension of the aggregate. Gray scale images for texture analysis are cap- tured during this pass (84). 37 2.5.2.6 Laser-Based Aggregate Scanning System The Laser-Based Aggregate Scanning System (LASS) uses a laser line scanner mounted on a 2-D linear slide sys- tem and a data acquisition system to measure aggregate par- ticles between 1.0 mm and 100 mm in three dimensions (Fig- ure 9) (85). In the prototype laboratory version, aggregate particles are placed on a scanning platform. The laser scan- ner moves along the 1.5 m Y-axis on the overhead slide per- forming 25 scans/s. The X-axis scan width is 120 mm. The laser scanner projects a stripe on the scanning platform. The reflection of the laser stripe is captured by a CCD camera. Knowing the location of the laser source, the 3-D coordinates of the surface of the object can be calculated. LASS has been used to measure grading, shape, angularity, and texture (86). 2.5.3 Image Analysis The systems described above represent a sampling of the systems currently available. Two of those systems, UI-AIA Figure 8. Aggregate imaging system. Figure 9. Laser-Based Aggregate Scanning System (85).

and AIMS, are under investigation as part of NCHRP Project 4-30. WipShape, UI-AIA, AIMS, and LASS can all measure the principal aggregate dimensions, length, width, and thick- ness of coarse aggregate. The VDG-40 Videograder can only measure two dimensions—length and width—and produce a global average of the third dimension. Image analysis tech- niques are used to extract information about grading, shape, angularity, and, in some cases, texture. 2.5.3.1 Aggregate Grading The VDG-40 Videograder, WipShape, UI-AIA, AIMS, and LASS will all determine the size of the aggregate parti- cles being evaluated. The aggregate size is used to categorize particle angularity and texture measurements by size. In its current form, only the VDG-40 is designed to test large aggre- gate samples, representative of a gradation sample. Weingart and Prowell (79) compared sieve results with the output of the VDG-40 for production samples of a No. 8 material. The material passing the 1.18-mm sieve was not evaluated. The results agreed well except for the 9.5-mm sieve. The devel- opers of LASS also believe that their technology is adaptable to online measurements. Problems can occur when comparing the results from dig- ital imaging to wire mesh sieves. The size of an aggregate particle is generally taken to be the intermediate dimension (dI) or the particle width. Some flat or elongated particles can fit through square sieve openings that are smaller than their width on the diagonal (73). Rauch et al. (87, 88) completed a study to evaluate tech- niques to rapidly determine the gradation of unbound aggre- gates. Based on their review of potential technologies, digi- tal image analysis, and laser scanning were recommended for further research. A continuation of the study evaluated five automated gradation devices: LCPC VDG-40 Videograder, W.S. Tyler Computer Particle Analyzer, Micrometrics Opti- sizer PSDA 5400, John B. Long Co. Video Imaging System, and Buffalo Wire Works Particle Size Distribution Analyzer. Two of the systems—the VDG-40 and the Computer Parti- cle Analyzer—use line-scan cameras. These systems evalu- ate all of the particles (greater than a minimum size) that pass in front of the camera. The remaining three systems use matrix-scan technology. Matrix-scan devices typically sam- ple 10% to 20% of the aggregate stream. Five aggregate materials were used to prepare 15 test sam- ples that were tested in each device. The samples were assem- bled to provide diverse shape, color, and texture. Based on the analysis of the results, the two-line scan devices are more repeatable; however, the results from the devices using line- scan technology do not compare as well with the benchmark sieve results (88). The Micrometrics Optisizer PSDA and John B. Long Co. Video Imaging System appear to provide the best overall results (88). 38 2.5.3.2 Aggregate Shape Analyses of aggregate shape or form are generally aimed at replacing ASTM D4791, the F&E test. WipShape, UI-AIA, AIMS, and LASS all measure particles in three dimensions. From these measurements, the principal dimensions of a par- ticle are determined. The WipShape and UI-AIA systems both fit a virtual box around the aggregate particles to determine the longest, inter- mediate, and shortest dimensions (73, 81). Using these mea- surements, elongation is the ratio of the longest dimension to the intermediate dimension, and flatness is the ratio of the intermediate dimension to the smallest dimension. The Super- pave method’s specifications are based on the ratio of the longest to the smallest dimension or F&E as specified by ASTM D4791. Both WipShape and UI-AIA can produce fre- quency histograms of the percent of particles exceeding var- ious ratios of elongation, flatness, or flatness and elongation. Sphericity and form factor have been proposed as indexes of aggregate shape (76). Sphericity is described by Equation 2: (2) where ψ = sphericity, ds = smallest dimension (thickness), dI = intermediate dimension (width), and dl = largest dimension (length). Shape factor is described by Equation 3: (3) where SF = shape factor, and ds, dI and dl are defined as for Equation 1. Measures of aggregate shape using wavlets have been pro- posed for LASS and AIMS (83, 86, 89). Masad (83) states: The fundamental idea behind wavlets is to decompose a sig- nal or image at different resolutions. Wavlets are special func- tions, which satisfy certain mathematical conditions and are used in representing data, which could be one-dimensional signal (speech), or a two-dimensional signal (image). 2.5.3.3 Angularity and Texture Several researchers have proposed methods of analyzing aggregate angularity and texture using fractals (55, 57, 90, and 91). Maerz (73) uses the minimum average curve radius SF dd d s l I = × ψ = ×d dd s I l 23

to estimate angularity using WipShape. Both AIMS and LASS use wavlets to describe angularity and texture (83, 86). 2.6 TESTS FOR AGGREGATE PROPERTIES RELATED TO MOISTURE DAMAGE 2.6.1 Introduction Moisture is a key factor in the deterioration of asphalt pavement. Factors that influence moisture damage include aggregate, asphalt binder, type of mix, weather and environ- mental effects, and pavement subsurface drainage. The pres- ence of plastic fines in the fine aggregate portion of HMA may induce stripping in the mix when exposed to water or moisture. The following test methods are used to evaluate the contribution of aggregate to moisture damage. 2.6.2 Sand Equivalent Test The sand equivalent test is a consensus aggregate property specified in the Superpave mix-design method. The test was originally developed in 1952 as a rapid field test by Francis Hveem, whose original work suggested that low sand equiv- alents could indicate either clay or dust content. It is now used to determine the relative proportions of plastic fines or clay-like material in fine aggregates. Excessive clay-like par- ticles may cause the asphalt binder to debond from the aggre- gate in the presence of moisture. Fine aggregate (passing the 4.75-mm [No. 4] sieve) is placed in a graduated, transparent cylinder that is filled with a mixture of water and a flocculat- ing agent. After agitation and 20 min of settling, the sand sep- arates from the clay-like fines and the heights of sand and sand plus clay are measured. The sand equivalent is the ratio of the height of the sand to the height of sand plus clay ×100. Higher sand equivalent values indicate more sand and less clay and silt. Minimum specified sand equivalent values for fine aggregate in HMA range from 26 to 60 (92). The Super- pave method specifies a minimum requirement of 40 to 50, depending on traffic. This test has the advantages that it is quick to perform; requires very simple equipment, which can be used with min- imal training or experience; and has given reasonably good results. However, there are some concerns about this test. In 1997, Stroup-Gardiner et al. (93) evaluated the sand equiva- lent test using 29 aggregates from a wide range of aggregate sources in Minnesota. Only 3 of 29 sand equivalent tests for individual stockpiles were less than 40%. The researchers found that sand equivalent test values were not sensitive to either the general mineralogy or the percentage passing the 0.075-mm sieve. There was no significant relationship between the sand equivalent test and mixture moisture sen- sitivity (tensile strength ratio test results) or VMA. Alhozaimy (94) investigated the relationship between the sand equivalent and material finer than No. 200 sieve tests. 39 A total of 100 samples of natural silica sand and 100 samples of crushed sand were collected from the stockpiles of differ- ent ready-mixed concrete plants in Riyadh, Saudi Arabia. Also, the effects of silica and crushed sands—using different values of sand equivalents and passing a No. 200 sieve— on water demands of mortar were investigated. A strong cor- relation was found between the sand equivalent test and fine materials passing a No. 200 sieve for silica sand, whereas there was no correlation for the crushed sand. This indicates that the amount of fine materials in silica sand can be deter- mined by either the No. 200 sieve or the sand equivalent test. However, the sand equivalent test can be misleading for crushed sand. Kandhal et al. (95) conducted a study to determine the best aggregate test method that indicates the presence of detri- mental plastic fines in the fine aggregate, which may induce stripping in HMA mixtures. Ten fine aggregates representing a wide range of mineralogical compositions and plasticity characteristics were used. Their plasticity characteristics were evaluated by three test methods: sand equivalent test, plas- ticity index, and methylene blue value. Ten HMA mixtures were made using a common limestone coarse aggregate, with these ten fine aggregates. AASHTO T283 (Resistance of Compacted Bituminous Mixture to Moisture Induced Dam- age) and a Hamburg wheel-tracking device were used to eval- uate the stripping potential of the ten HMA mixtures. Statis- tical analysis of the aggregate test data and the mix validation test data showed that no significant relationship existed between sand equivalent test values and mixture validation tests results. 2.6.3 Plasticity Index Plasticity index (PI) is being used by several agencies to measure the degree of plasticity of fines. PI is the difference between the liquid limit and the plastic limit of the material passing 425-μm (No. 40) sieve. ASTM D1073 (Standard Specification for Fine Aggregate in Bituminous Paving Mix- tures) and D242 (Standard Specification for Mineral Filler for Bituminous Paving Mixtures) limit the PI of this fraction passing the 425-μm (No. 40) sieve (including the mineral filler) to a value of 4 or less. Some states specify a maximum PI for material passing the No. 200 sieve. A review of litera- ture indicated no reported correlation between the PI and the field performance of HMA (95). Precision data have not been established for liquid limit and plastic limit tests, which are based on subjective judgment and experience of the tester. 2.6.4 Methylene Blue Test The test method titled “Determination of Methylene Blue Adsorption Value of Mineral Aggregate Fillers and Fines” is recommended by the International Slurry Seal Associa- tion (ISSA) to quantify the amounts of harmful clays of the

smectite (montmorillinite) group, organic matter, and iron hydroxides present in fine aggregate (96). The principle of the test is to add quantities of a standard aqueous solution of the dye (methylene blue) to a sample until adsorption of the dye ceases. A representative sample of dry fine aggregate is screened through the No. 200 sieve. The portion of the sample passing the No. 200 sieve is tested for methylene blue adsorption value (MBV). Ten grams of the sample are dispersed in 30 g of distilled water in a beaker. One gram of methylene blue (MB) is dissolved in enough distilled water to produce 200 ml of solution so that 1 ml of solution contains 5 mg of MB. This MB solution is titrated stepwise in 0.5 ml aliquotes from the burrette into the continually stirred fine aggregate sus- pension. After each addition of MB solution and stirring for 1 min, a small drop of the aggregate suspension is removed with a glass rod and placed on a filter paper. Successive addi- tions of MB solution are repeated until the end point is reached. Initially, a well-defined circle of MB-stained dust is formed and is surrounded with an outer ring, or corona, of clear water. The end point is reached when a permanent light blue coloration, or “halo,” is observed in this ring of clear water. The MB value of a specific fine aggregate fraction is reported as milligrams of MB per gram of specific fine aggre- gate fraction, such as MBV = 5.3 mg/g, 0/No. 200. The MBV expresses the quantity of MB required to cover the total sur- face of the clay fraction of the sample with a mono-molecular layer of the MB. Therefore, the MBV is proportional to the product of the clay content times the specific surface of the clay (97). The MB test is simple and practical, and its cost is reason- able. Cross and Voth (98) used this test as a reference test for evaluating the APA’s suitability for predicting moisture sus- ceptible mixtures. Kandhal et al. (95) found that the MB test is the fine aggregate test that is best related to stripping of HMA (compared with sand equivalent test and PI) and then recommended the MB test to be used to indicate the presence of detrimental plastic fines, which may induce stripping in HMA mixtures. Aschenbrener and Zamora (99) evaluated several special- ized aggregate tests and their relation to HMA performance. The tests include MB, Rigden voids index, stiffening power, and dust coating on aggregates. The evaluation was con- ducted using aggregate sources from 20 projects with known field performance. It has been found that the MB, dust coat- ing on aggregates, and Rigden voids index, or stiffening power, when used with one another, accurately identified aggregate problems in the stripping pavements. The study indicated that the MB test, the other aggregate tests, or both can be used to isolate the potential problematic components of the HMA if an HMA fails a performance-related test, such as the Hamburg wheel-tracking device. The MB test was also used by Harders and Noesler (100) to address different surface activities (i.e., surface energies). 40 The surface energy theory will be discussed in the following section. Results from ongoing project NCHRP Project 4-19(2), “Validation of Performance-Related Tests of Aggregates for Use in Hot-Mix Asphalt Pavements,” showed a very signifi- cant relationship between the rutting performance of the wet pavements and the MBV (23). A high MBV may be associ- ated with a high amount of harmful material or a more active clay mineral type. 2.6.5 Surface Free Energy Theory The mechanism of moisture damage can be explained by the theories of adhesion. Four broad theories have been pre- sented to explain adhesion of asphalt binder to aggregate: the mechanical theory, the chemical reaction theory, the surface free energy theory, and the molecular orientation theory. The surface energy theory is being used by several researchers in evaluating the moisture potential of asphalt mixtures (100– 104). The surface energy theory primarily involves calculat- ing the surface energies of the asphalt binder and the aggre- gate. The bonding energy between asphalt and aggregate can then be calculated either between the two components alone or in the presence of a third liquid such as water. Cheng et al. (101) stated that the bonding strengths helped to select the most compatible mixtures, to improve the adhesive bond, and to reduce debonding potential in the presence of mois- ture. Several methods have been developed to measure the surface energy of an asphalt-aggregate system. Elphingstone (102) measured the surface energies of various kinds of asphalts using the Wilhelmy Plate technique and measured the contact angle of many asphalt samples. Unfortunately, he could not obtain the surface energies for a number of sam- ples with his technique because of errors in the contact angle measurements. Li (103) measured the surface energies of a variety of European aggregates. Cheng et al. (104) measured surface energies of some widely used aggregates and asphalt binders in the southern United States using the Universal Sorption Device (USD) and Wilhelmy Plate method, respec- tively. Later, Cheng et al. (105) developed the adhesion failure model. Comparison between mechanical test results (repeated load permanent deformation tests) and the adhesion failure model showed the same trends of moisture damage potential for the aggregates and asphalts evaluated. Although the surface energy theory is not new, methods to evaluate moisture dam- age potential based on this theory and testing protocols need additional study. 2.6.6 Net Adsorption Test The net adsorption test (NAT) was applied to the HMA industry by the SHRP A-003B contractor to predict moisture damage (stripping) in asphalt-aggregate mixes (106). It was developed to determine the adsorptive nature and the water

sensitivity of a wide range of typical paving-quality aggre- gates. This relatively fast and simple test was developed to provide a rapid, simple, quantitative measure of the amount of asphalt adhered to aggregate after exposure to water. It was used to evaluate the affinity of asphalt for aggregate and to determine the water sensitivity of a given asphalt-aggregate pair. The test is composed of three parts. First, asphalt is flowed over and adsorbed onto aggregate from a toluene solu- tion using a recirculating column. The adsorption step is allowed to run for 7 h. Second, a small amount of water is introduced into the toluene solution, and the adsorbed asphalt that is sensitive to the presence of water is desorbed from the aggregate. Third, the amount of asphalt remaining on the aggregate after the introduction of water is determined. This amount is termed “net adsorption”; it gives a measure of the “affinity” of the asphalt for the aggregate by water and serves as an indicator of the water sensitivity of the pair. The aggre- gate properties predominated in the test, showing a stronger influence than the asphalt on the initial amount of asphalt adsorbed, on the amount of asphalt desorbed by water, and on the amount of asphalt remaining—the net adsorption. After the NAT was proposed, it was validated using both laboratory and field data by Hicks et al. (107) and Terrel et al. (108). Both research teams used two accelerating rutting tests: Oregon State University (OSU) wheel tracker and SWK/ UN (SWK Pavement Engineering in Nottingham, UK) wheel tracker to evaluate water sensitivity in the validation of the NAT. The prediction of water sensitivity of the binder as pro- posed by the SHRP A-003B NAT shows little or no correla- tion to either these two wheel tracking tests or to the SHRP A-002 predications for permanent deformation. As a conclu- sion, the authors suggested that the NAT is a poor indicator of the moisture sensitivity of the binder. Woodward’s research in his Ph.D. dissertation (109) showed the ability of NAT to rank aggregate-asphalt pairings. He stated that the NAT test was able to highlight how optimum levels could be achieved in terms of predicating performance; however, this result has not been validated by a wide range of aggregate resources. 2.6.7 Other Aggregate Tests Related to Moisture Damage In addition to the NAT, the SHRP A-003B researchers developed two “specialty” tests (106). The limestone reactiv- ity test is a quick and reliable method for determining the amount of active sites present on the aggregate surface. It can be used to differentiate among limestone sources. Another “specialty” test assesses the reactivity of the asphalt-aggregate systems to the addition of anti-stripping agents. Woodward (109) used the Vialit Plate Test and the Instron Adhesion Pull-Off Test (INAPOT) along with the NET to predict the adhesion property in the laboratory. The INAPOT was developed as a method to quantify the effect of steady load and temperature conditions on the adhesive bond per- 41 formance of aggregate prisms pressed into a pot of bitumen (109). This test uses expensive test equipment and involves elaborate test sample preparation. The testing procedure con- sists of fixing a rectangular aggregate prism into the upper jaw of an Instron apparatus, pressing the prism into a layer of bitumen held in a container or pot, and extracting the prism from the bitumen under controlled conditions. The INAPOT requires that aggregate prisms be cut from lump rock sam- ples representative of the aggregate being assessed because of the inability of the Instron jaws to grasp individual aggre- gate particles. Instead, prisms 18 mm × 10 mm × 30 mm were cut; the face to be assessed was left as a natural uncut surface. Woodward’s study (109) showed that the INAPOT was able to quantify the variation in tensile adhesion characteristics for three individual constituents from the same greywacke quarry. The results agreed with in-service experience that the poor initial coating, adhesion, and premature stripping aggre- gate gave the worst INAPOT results. His study also indicated that the influence on adhesion of different rock types was much less than that of temperature. The French Vialit Plate test was originally developed in the early 1960s to simulate conditions experienced on-site with the use of chippings applied as surface dressing. Through his study, Woodward (109) concluded that this method offers the engineer a quick and simple means of predicting the in- service performance of aggregate used in surface courses under a wide range of simulated in-service conditions. 2.6.8 Summary of Aggregate Tests Related to Moisture Damage This section gives the state of practice for test procedures used to evaluate aggregate moisture damage potential. Many factors influence moisture damage: HMA characteristics (aggregate, asphalt binder, and type of mixture); weather dur- ing construction; environmental effects after construction; and pavement surface drainage. Aggregate tests related to moisture damage generally fall into two categories: tests to identify clay-like fines and tests that evaluate the surface prop- erties of the aggregate related to the adhesion of the binder to the aggregate. The Superpave method currently specifies the Sand Equiv- alent Test (AASHTO T176) to identify clay-like fines. Con- troversial results and findings exist for the sand equivalent test, PI, NAT, and other tests. In some cases, the sand equiv- alent test identifies crusher fines as harmful clay-like parti- cles. It appears that the MB test may be the best method to quantify the amount of harmful clays in fine aggregate. The NAT was developed during SHRP to evaluate the interaction between the asphalt binder and aggregate in the presence of water; however, validation work conducted as part of SHRP indicated a poor predictive ability for the test, and it has not been widely used since. At the present time, the surface energy techniques appear to be promising. The pro- cedures are relatively new. Results and efforts from NCHRP

Project 9-37, “Using Surface Energy Measurements to Select Materials for Asphalt Pavements,” can be used to apply the energy surface theory in the future. 2.7 TESTS RELATED TO AGGREGATE DURABILITY Aggregate durability generally encompasses two categories of tests: tests that measure aggregate abrasion resistance and breakdown during handling, mixing, laydown, and under traf- fic and tests that address aggregate weathering when aggre- gate is exposed to freezing and thawing or wetting and dry- ing. These tests are employed in concert to ascertain that the aggregate used in the production of HMA will be durable. Specifically, tests related to durability are selected to address the following: • Aggregate breakdown during handling, mixing, and placement. Such breakdown can alter the HMA grada- tion, resulting in a mixture that does not meet volumetric properties. This breakdown can generally be accounted for in the design process. • Abrasion or weathering of the aggregates in the pavement structure. Gross aggregate wear or weath- ering can occur in the form of raveling, popouts, or pot- holes. An example of extensive surface loss caused by popouts in a 2-year-old Superpave-designed pavement is shown in Figure 10. • Freeze-thaw durability, which is more closely associ- ated with the performance of aggregate base, Portland cement concrete, and surface treatments. This may be due to the fact that aggregate particles in HMA should be coated with asphalt. Popouts of surface aggregates may be related to freeze-thaw durability. • Other forms of abrasion on the pavement surface, such as polishing or loss of microtexture of coarse aggre- gate particles. Polishing is beyond the scope of this report and will only be treated briefly. 42 Some western U.S. states and European countries use basalts containing high-plasticity expansive clay minerals. 2.7.1 Aggregate Tests Related to Abrasion Resistance and Breakdown 2.7.1.1 LA Abrasion Aggregates must be resistant to crushing and abrasive wear to withstand handling during stockpiling, shipping, mixing at the HMA plant, laydown, and compaction. Once the HMA is in place, the aggregates need to be sufficiently hard or tough to transfer load through contact points. This is especially true of aggregates used in gap-graded mixtures such as SMA. Aggregates must also withstand surface abrasion and polish- ing from traffic. The SHRP aggregate expert task group identified the Los Angeles Abrasion Test, AASHTO T96 (ASTM C131), as the fourth most important aggregate property in both the first- and second-round questionnaires used in the Delphi process (1). The LA abrasion test was included as a source property in the Superpave mix design system. The specification val- ues for source properties were to be set by the agency to allow for variations in locally available aggregates. Based on the survey conducted as part of this study, 96% of the 48 U.S. states and Canadian provinces that responded use the LA abrasion test. The LA abrasion test was originally developed by the Municipal Laboratory of the city of Los Angeles in the 1920s. The LA abrasion test procedure requires that an aggregate sample be placed inside a rotating steel drum containing a specified number of steel balls or charge. As the drum rotates, a shelf inside the drum picks up the aggregate and steel spheres. The shelf lifts the aggregate and steel balls around until they drop approximately 27 in. on the opposite side of the drum, subjecting the aggregate to impact and crushing. The aggregate is subjected to abrasion and grinding as the drum continues to rotate until the shelf picks up the contents, and the process is repeated. The drum is rotated for a spec- ified number of revolutions, typically 500. Afterward, the aggregate is removed from the drum and sieved over a No. 12 (1.7-mm) sieve to determine the degradation as a percent loss. Kandhal and Parker (2) conducted a literature review on the early LA abrasion research as part of NCHRP Project 4-19. Their review indicated only a fair correlation with field performance for coarse aggregates; however, they did note that early developmental studies, most notably by Woolf (110) and Melville (111), indicated good correlations with perfor- mance. Testing conducted as part of NCHRP Project 4-19 indicated that LA abrasion loss was not related to historical pavement performance ratings (112). Limited recent research has been conducted on the LA abrasion test by Amirkhanian et al. (113). A survey con- ducted as part of the study indicated that the majority of state DOTs specified the LA abrasion test, similar to the currentFigure 10. Popouts and raveling in 2-year-old pavement.

study. The survey also determined that most agencies believed that the LA abrasion results were most related to breakdown during compaction and that the majority were satisfied with their current specifications. The study investigated the break- down of four granite aggregates with LA abrasion values ranging from 28 to 55. The study evaluated indirect tensile strength, resilient modulus, and aggregate breakdown for sam- ples compacted using 25, 50, 75, or 100 Marshall blows. The indirect tensile strength and resilient modulus tests were per- formed on both conditioned and unconditioned samples. The test results indicated that the indirect tensile strengths for the three mixtures produced with the granite having an LA abra- sion loss greater than or equal to 30% were significantly lower for both the conditioned and unconditioned samples than the strengths produced with the aggregate having an LA abrasion loss of 28%. Further, the tensile strength and resilient modu- lus ratios were generally lower for the mixture produced with an aggregate having an LA abrasion loss of 55%. Interest- ingly, for the dense-graded mixes tested, aggregate break- down was only significant on the 0.150 and 0.075 (No. 100 and No. 200) sieves. Unfortunately, the level of breakdown that occurs in the field was not investigated. As previously discussed, work by Aho et al. (39) indicated the interrelationship between F&E, LA abrasion, and expected breakdown in the field. The LA abrasion of the sources in this study represented a very narrow range (24% to 26%). This study concluded that breakdown in the gyratory compactor generally exceeded the breakdown that occurred in the field for dense-graded mixtures used in Illinois. In conjunction with their FAA study, Roque et al. (51) recommended includ- ing LA abrasion limits to differentiate between good- and poor-performing fine aggregates with borderline FAA values. 43 Xie and Watson conducted a study to evaluate the break- down of SMA mixtures during laboratory compaction (114). Five aggregate sources with a range of LA abrasion loss from 17% to 36% were selected for the study: crushed gravel, gran- ite (two sources), limestone, and traprock. Aggregates sizes were combined to produce 9.5-mm, 12.5-mm, and 19.0-mm NMAS mixtures. Samples were compacted with either a 50- blow Marshall or 100-gyration Superpave Gyratory Com- pactor (SGC) effort. Ignition furnace extractions were used to compare the batched, loose-mix, Marshall-compacted, and SGC-compacted gradations. Comparisons were made based on the critical or breakpoint sieve for the SMA mixtures. For the 12.5-mm and 19.0-mm NMAS mixtures, the 4.75-mm (No. 4) sieve was used; for the 9.5-mm NMAS mixtures the 2.36-mm (No. 8) sieve was used. As shown in Figure 11, the 50-blow Marshall compaction effort resulted in greater break- down than the SGC compaction. The data from this study correlated well with similar data from NCHRP Project 9-8 (36). Unfortunately, this study was not correlated to the actual breakdown that occurred in the field. Breakdown for larger sieve size is expected with gap graded mixes such as SMA because there are more coarse aggregate contact points dur- ing compaction. Increased breakdown for open-graded mix- tures was also noted in NCHRP Project 4-19 (2) when assess- ing dry aggregate breakdown in the SGC. 2.7.1.2 Other Tests Related to Aggregate Breakdown In Europe, a number of alternative tests are used to assess aggregate breakdown: the Aggregate Impact Value Test (BS SGC: y = 0.3732x - 3.3991 R2 = 0.8609 Marshall: y = 0.3192x + 1.6599 R2 = 0.6324 -2.0 2.0 6.0 10.0 14.0 18.0 22.0 26.0 10 20 30 40 50 60 LA Abrasion value, % 4. 75 m m S ie ve B re ak do w n, % SGC This Study SGC NCHRP 9-8 Marshall This Study Marshall NCHRP 9-8 Linear (SGC) Linear (Marshall) Figure 11. Critical sieve breakdown versus LA abrasion by compactor type (114).

812), the German Schlagversuch Impact Test (DIN 52115) and the Aggregate Crushing Value (BS 812). The aggregate impact value test uses a graded sample between 10.0 mm and 14.0 mm. The sample is subjected to 15 blows of a 100-mm (4-in.) diameter hammer weighing 13.5 to 14.0 kg that falls 380 ± 5 mm. The aggregate impact value is reported as the percentage of the initial sample weight passing the 2.36-mm (No. 8) sieve after impact. A lower value indicates a stronger aggregate. The German Schlagversuch impact test is similar to the British aggregate impact value test except that the equipment is more expensive and less portable (109). The aggregate crushing value test applies a compressive load to an approximately 2-kg sample in a steel container. A total load of either 400 kN is applied to a 150-mm-diameter piston or 100 kN is applied to a 75-mm-diameter piston in a 10-min period. The aggregate crushing value is expressed as the percentage of fines passing the 2.36-mm (No. 8) sieve based on the original sample mass. Woodward (109) notes that with the impact and crushing- type tests, if the sample consists of a blend of good- and poor- performing particles, the stronger particles can often carry the load masking the weaker particles. Conversely, with what Woodward describes as fragmentation tests like LA abrasion, weaker particles are equally exposed to fragmentation forces. Woodward (109) produced a correlation matrix among the aggregate impact value, German Schlagversuch impact, aggregate crushing value, and LA abrasion tests as shown in Table 10. The data in Table 10 is based on a range of aggre- gates commonly used in Great Britain and Ireland. There were 246 common aggregates tested with the aggregate impact value and aggregate crushing value tests, 75 to 91 common aggregates tested for LA abrasion, and 13 common aggregates tested with the German Schlagversuch Impact test. All of the relationships were statistically significant except the relation- ship between the LA abrasion test and the German Schlagver- such impact test. There appears to be a reasonable fit between LA abrasion and the aggregate crushing value tests. Senior and Rogers (115) also compared the results from the LA abrasion and aggregate impact value tests for use in Ontario. They found a correlation (R = 0.797) based on 98 aggregate sources. Senior and Rogers concluded that the aggregate impact value test could be a practical substitute for the LA abrasion test to assess the extent of breakdown expected during processing 44 and handling. The aggregate impact value test was stated to have the following advantages: it requires a small sample; it requires less expensive, portable equipment; and samples may be tested in a wet condition (115). Kandhal and Parker (2) found better correlations between LA abrasion loss and both the aggregate impact value and the aggregate crushing value (R = 0.932 and 0.934, respectively); however, no clear trends were noted between the results of the LA abrasion loss, aggregate impact value, or aggregate crushing value, and subjective field performance ratings. Both the USACE and Superpave gyratory compactors have been investigated as means of simulating aggregate break- down during construction and handling. Ruth and Tia (116) investigated the use of USACE Gyratory Testing Machine to simulate the breakdown that occurs in drum plants. Samples were tested with an initial angle of gyration of 3°, ram pres- sure of 690 kPa, and air roller pressure of 62 kPa. Samples were tested to either 25, 50, 100, or 200 revolutions. Initial results indicated that the majority of the breakdown (greater than 50%) occurred in the first 25 revolutions. Three aggre- gate sizes were tested: No. 67, No. 89, and screenings. Good correlations were observed between LA abrasion loss and the percent passing the No. 10 sieve from the Gyratory Test- ing Machine for the coarse aggregates (R2 = 0.893 to 0.985). Relatively good correspondence was found between the breakdown of samples blended to meet actual HMA job mix formulas tested at 25 gyrations and the breakdown that occurred when the same gradings were processed dry (no asphalt cement) through a drum plant. Two key advantages were observed for the Gyratory Testing Machine procedure: the distribution of fines passing the No. 12 sieve is investi- gated and samples may be tested wet. The use of the SGC to evaluate aggregate toughness was evaluated by Kandhal and Parker (2). AASHTO No. 8 stone and a limited number of AASHTO No. 57 stone sources were tested in the SGC. The actual number of gyrations used to sim- ulate breakdown is not reported (2, 112). Aggregate break- down was assessed in two ways: gradation change for a single sieve, such as the 4.75-mm (No. 4), or the sum of the grada- tion changes on all sieves before and after compaction. Kand- hal and Parker concluded that the SGC can be used to differ- entiate between tough and weak aggregates (2). They also noted that breakdown is greater for open graded mixtures, Test German Schlagversuch Impact Aggregate Crushing Value LA Abrasion Aggregate Impact Value 1.0 0.607 0.822 0.731 German Schlagversuch Impact 1.0 1.0 0.683 0.403 Aggregate Crushing Value 1.0 0.861 LA Abrasion 1.0 Aggregate Impact Value TABLE 10 Correlation (R) matrix for aggregate strength tests (109)

which would tend to have more contact points. Comparisons between the degradation in the SGC and field performance indicated no obvious breakpoints or correlations (2). 2.7.1.3 Aggregate Tests Related to Abrasion Resistance A number of studies have evaluated alternative tests to mea- sure aggregate degradation and abrasion resistance. Senior and Rogers summarize some of the concerns with the LA abra- sion test that led to the investigation of alternatives: “The Los Angeles test is not always appropriate because the steel balls impart a severe impact loading on the test sample, overshad- owing interparticle abrasion, which is the predominant process in pavement subject to traffic stress” (115). Senior and Rogers note that some coarse grained granites and gneisses tend to be brittle resulting in high LA abrasion losses but adequate field performance. By contrast, some softer aggregates such as carbonates and shales will absorb the impact of the steel balls, resulting in acceptable performance. These types of “soft” aggregates tend to be susceptible to slaking and to par- ticle degradation when wet. Woodward (109) emphasizes the concerns that road surfaces are frequently wet and that there can be a significant reduction in the wet strength of some aggregates. Samples cannot be tested wet in the LA abrasion machine because it would be difficult to remove the fines that would cake along the shelf and drum. 45 The micro-deval test has been proposed as an alternative to LA abrasion in North America (2). The micro-deval test was developed in France in the 1960s (115). To perform the micro-deval test on coarse aggregate, a 1500-g sample is first soaked in 2 liters of water for at least 1h. A 5-kg charge of 9.5-mm (3/8-in.) diameter balls is placed in a jar along with the sample and the water it was soaked in. The jar is then rotated at 100 rpm for 2 h. The sample is then washed and oven dried. The micro-deval abrasion loss is the percent of material passing the 1.18-mm (No. 16) sieve expressed as a percentage of the original sample mass. A reference material is available for periodic calibration of the loss. Senior and Rogers investigated alternative tests for assess- ing coarse aggregate toughness and durability in Ontario (115). The alternative tests included the unconfined freeze- thaw test for coarse aggregate, micro-deval abrasion test, aggregate impact value test, polished stone value test and aggregate abrasion value test. Results were compared with LA abrasion loss, magnesium sulfate soundness loss, and water absorption. The micro-deval test produced similar results to the sulfate soundness test (R = 0.85 for 106 samples) with greater precision (115). Parker and Kandhal (2) also reported a reasonable correlation between magnesium sulfate sound- ness loss and micro-deval loss (R = 0.848, p = 0.0001). The improved precision is indicated by comparing the single operator standard deviations as a function of test value shown in Figure 12. Senior and Rogers recommended the micro- deval test, polished stone value, and unconfined freezing and Figure 12. Standard deviation versus magnesium sulfate soundness or micro-deval abrasion (115).

thawing in concert to ascertain the performance of aggre- gates for HMA (115). Recommendations for modifications to the micro-deval test were developed for testing fine aggre- gates (117). The goal of these modifications was to replace the sulfate soundness test. Modifications included a smaller sample (500 g), small charge (1250 g), less water (750 ml), and a shorter test time (15 min). Loss was reported as the per- cent passing the 0.075-mm (No. 200) sieve. Based on the results of this study, micro-deval losses of less than 20 and 25 were established for high-quality and lower-quality HMA surfaces, respectively (109). Woodward compared the abrasion results from the micro- deval test and aggregate abrasion value test (109). In the aggregate abrasion value test (BS 812), aggregate particles are held in a mold and the exposed aggregate is placed on a flat, rotating steel plate. A standard weight is placed on the mold, and silica sand is metered onto the steel plate. The test is generally performed dry. The aggregate abrasion value is based on the loss determined from the sample mass before and after abrasion normalized for the density of the aggre- gates. Based on testing of 133 samples, Woodward indicated a significant correlation (R = 0.799) between the aggregate abrasion value and micro-deval loss. Senior and Rogers did 46 not report the correlation between aggregate abrasion value and micro-deval. However, based on the figure in Senior and Rogers (115), the relationship appears reasonable. Senior and Rogers (115) recommend the micro-deval test over the aggregate abrasion value because the micro-deval test is less expensive and less time consuming. NCHRP Project 4-19 recommended both the micro-deval test and magnesium sulfate soundness as the two tests most related to HMA performance in terms of popouts, raveling, and potholing (2). This recommendation was based on single variable correlations between both test results and subjective field performance rankings. Figures 13 and 14 indicate the relationships, respectively, between both micro-deval loss and magnesium sulfate soundness and field performance. An 18% maximum loss was recommended for both test methods. The correlation between micro-deval loss and magnesium sulfate soundness was not addressed in NCHRP Project 4-19. Based on the recommendations from NCHRP Project 4-19, Cooley et al. (118) evaluated the micro-deval loss, LA abra- sion loss, and sodium sulfate soundness of 72 aggregate sources from the southeastern United States. No statistically significant results were found between either LA abrasion or sodium sulfate soundness and micro-deval loss. Of the 72 Figure 13. Pavement performance and micro-deval abrasion loss (2).

sources studied, six were characterized as poor performers. Two of these six aggregates produced micro-deval losses less than 6%. Also, 33% of the aggregate sources characterized as fair exceeded the 18% micro-deval loss criterion recom- mended by NCHRP Project 4-19 (118). The authors recom- mend consideration of specifications for micro-deval loss based on parent aggregate type. Woodward (109) also rec- ommended specifications based on rock type. Ontario has implemented a specification for micro-deval loss based on aggregate type (119). For high-volume roads, the maximum micro-deval loss for igneous gravel is 5%; for dolomitic sandstone, 15%; for traprock, Diabase, and andesite, 10%; and for meta-arkose, meta-gabbro, and gneiss, 15%. Rismantojo (23) tested six aggregate sources for micro- deval loss, LA abrasion loss, and magnesium sulfate sound- ness as part of NCHRP Project 4-19(2). Figure 15 shows the relationship between the micro-deval value and magnesium sulfate soundness loss for 22 aggregates representing the combined results from NCHRP Projects 4-19 and 4-19(2). Figure 15 indicates a good correlation (R2 = 0.76) between the two tests. This matches a similar finding by Senior and Rogers (115). In theory, the two tests should indicate differ- ing modes of deterioration. Woodward suggests that the 47 micro-deval test represents inter-particle abrasion within the HMA (109). Sulfate soundness tests are meant to represent the degradation that may occur because of freezing and thaw- ing; however, several studies representing a large range of aggregate types have indicated that the tests are strongly cor- related. Therefore, as proposed by Senior and Rogers (115), it appears advisable that only one such test be used to screen aggregates. Rismantojo (23) also indicated correlations between the micro-deval loss and both LA abrasion loss and water absorption. The relationship with LA abrasion broke down when the NCHRP Project 4-19 and 4-19(2) data sets were combined. The strength of some of the aggregates tested as part of NCHRP Project 4-19 appear to be greatly affected by water. Eighteen aggregate sources in Oklahoma were evaluated by Tarefder et al. (120) to develop a baseline specification for micro-deval loss. The aggregates tested were predomi- nantly limestone and sandstone. LA abrasion loss, freeze- thaw soundness, water absorption, and aggregate durability index were also performed. A fair correlation (R2 = 0.63) was indicated between micro-deval loss and LA abrasion loss. The authors noted that different micro-deval abrasion loss values may be required, depending on the parent aggregate Figure 14. Pavement performance and magnesium soundness loss (2).

type. A micro-deval abrasion loss of 25% was proposed for Oklahoma. 2.7.2 Aggregate Tests Related to Weathering and Freeze-Thaw Durability There is some question as to whether aggregates used in HMA need to be resistant to freezing and thawing. Sound- ness tests such as magnesium or sodium sulfate soundness have commonly been used to assess degradation from freez- ing and thawing as well as from wetting and drying. Sulfate soundness tests were developed in the early 19th century to simulate the expansion of water within stone that resulted from freezing and thawing water (117). However, it has pre- viously been shown that the magnesium sulfate soundness test is correlated with micro-deval loss. This may indicate that the magnesium soundness test better simulates the slak- ing caused by wetting and drying rather than freeze-thaw deterioration. To perform the sulfate soundness test (AASHTO T104), a graded sample of coarse aggregate is prepared based on the NMAS of the aggregate being tested. A graded fine aggregate sample is prepared with at least 100 g of material retained on each of the 4.75-mm, 2.36-mm, 1.18-mm, 0.600-mm, and 0.300-mm sieves. The sample is soaked in a saturated solu- tion of sodium of magnesium sulfate for 16 to 18 h. The sam- ples are then briefly drained and dried in a 110°C oven to a constant mass. Upon rewetting, the sulfate crystals expand in the aggregate pores, simulating the expansion of water upon freezing. The cycle of wetting and drying is typically repeated five times. After the final cycle, the sample is rinsed to remove the sulfate solution and dried at 110°C to a constant mass. 48 The sample is then sieved by hand over a smaller sieve appropriate for the size fraction. For 9.5-mm to 19.0-mm aggregates, the sample is sieved over an 8-mm (5/16-in.) sieve. The weighted average percent passing the smaller sieve sizes expressed as a percentage of the original sample weight is the sulfate soundness loss. As discussed previously, there is a reasonable correlation between magnesium sulfate sound- ness and micro-deval loss. The relationship is fair to poor for sodium sulfate soundness (2, 118). Currently, equipment is available to perform actual freeze- thaw testing. AASHTO T103, “Soundness of Aggregates by Freezing and Thawing,” describes three procedures to con- duct freeze-thaw testing. A sample is initially washed and dried to a constant mass, after which it is sieved into size frac- tions. The three procedures for immersion and freezing are summarized in Table 11. The ethyl alcohol used in Procedure B is to aid the penetration of water. After the final cycle of freezing and thawing, the samples are dried to a constant mass and are sieved. The resulting weighted average loss for each size fraction is the soundness loss. Iowa, Ontario, and Michi- gan currently use test methods for freeze-thaw testing. Senior and Rogers (115) investigated the use of the uncon- fined freeze-thaw test that is similar to AASHTO T103 for coarse aggregates. Individual size fractions retained on the 13.2-mm, 9.5-mm, and 4.75-mm sieves are placed in separate 1-liter jars. The samples are soaked for 24 h in a 3% NaCl solution. The samples are drained and sealed, frozen for 16 h, and then thawed at room temperature for 8 h. The freezing- and-thawing cycle is repeated five times after which the sam- ples are dried and sieved similar to AASHTO T103. Testing by Senior and Rogers (115) suggests that the Ontario Uncon- fined Freeze-Thaw test is “to be preferred because it shows y = 0.5392x + 7.8803 R2 = 0.7646 0 5 10 15 20 25 30 35 40 45 0 10 20 30 40 50 60 70 MgSO4 Soundness, % M ic ro D ev al V al ue Figure 15. Relationship between magnesium sulfate soundness and micro-deval loss (2, 23).

better discrimination than the sulfate test and is more precise.” Freeze-thaw losses of less than 6% for high-volume roads and of less than 30% for low-volume roads are recommended. 2.7.3 Aggregate Properties Related to Polishing and Frictional Resistance The friction of a pavement surface is a function of the sur- face textures that include the wavelength ranges described by microtexture and macrotexture. Microtexture provides a gritty surface to penetrate thin water films and produce good fric- tional resistance between the tire and the pavement. Macrotex- ture provides drainage channels for water expulsion between the tire and the pavement, thus allowing better tire contact with the pavement to improve frictional resistance and to pre- vent hydroplaning. Initial macrotexture is a function of gra- dation, although tests for abrasion resistance—such as the aggregate abrasion value and micro-deval abrasion loss—are related to the change in macrotexture with time (119). Polishing is the loss of microtexture with traffic wear. Pol- ishing is common in many carbonate aggregates. No test meth- ods were included in the Superpave mix design system to assess aggregate polishing. Agencies with aggregate sources prone to polishing have developed a number of means for qualifying or grading aggregate sources for various traffic levels. Polish Stone Value (BS 812) may be the most widely used test. Coarse aggregate particles having certain dimen- sions are first cemented into a curved mold. A group of the curved molds are attached to the outside of a steel wheel. The steel wheel rotates bringing the samples into contact with a rubber wheel treated with coarse and then with fine emery powders, which abrade the aggregate samples. Water is con- tinually applied to the aggregate surface. Upon completion of the polishing, the mold with cemented aggregate particles is mounted on a British Pendulum Tester. A pendulum arm with an attached rubber pad swings across the mold so that the rubber pad drags across the samples. A scale measures the height of the pendulum swing after contact with the sample. Lower numbers represent greater microtexture. Some aggregates are resistant to polishing. Other aggre- gate types maintain their microtexture by the continual abra- sion of mineral grains exposing fresh nonpolished grains. This renewal is desirable as long as the loss is not so great as to represent an aggregate that is not durable. Senior and Rogers (115) recommend polish stone values in excess of 50 49 for high-volume roadways. Woodward (109), who has per- formed extensive skid testing, notes a quandary for British aggregate sources between desirable levels of polish stone value to maintain adequate skid resistance and desirable lev- els of micro-deval loss to ensure durability against raveling and popouts. 2.7.4 Summary of Tests Related to Aggregate Durability The following is a summary of the tests related to aggre- gate durability. • Aggregates are subject to breakdown during stock- piling, mixing and compaction. Excessive aggregate breakdown can alter in-place gradations and can affect the volumetric properties of the HMA. In the United States, the LA abrasion test (AASHTO T96) is the most commonly used test to assess aggregate breakdown dur- ing construction. It is correlated to other tests for aggre- gate breakdown. There is no evidence to suggest that the LA abrasion test should be replaced by another impact test for the purpose of assessing aggregate breakdown during construction. • Aggregates in HMA are subject to weathering and abrasion in situ. Although originally intended to assess degradation from freezing and thawing, sulfate sound- ness tests (AASHTO T104) have been widely used to assess aggregates’ resistance to weathering. Several studies indicated good correlation between magnesium sulfate soundness loss and micro-deval abrasion loss (AASHTO TP58). Several studies have also indicated that the strength of some aggregates is significantly lower when wet. The micro-deval test offers improved precision over sulfate soundness. The micro-deval test also indicates abrasion resistance. This suggests that the micro-deval test may be more suitable to predicting aggregates performance in relation to weathering and abrasion than is sulfate soundness. However, data sug- gests specifications for micro-deval loss may have to be based on aggregate type. • There is debate as to whether aggregates used in HMA need to be resistant to freezing and thawing. Historically, sulfate soundness tests have been used to assess aggregates ability to resist weathering. A limited Procedure Solution Immersion Number of Cycles A 24 hours 50 B 0.5% solution of ethyl alcohol and water Vacuum Saturation (25.4 mm Mercury) 16 C Vacuum Saturation (25.4 mm Mercury) 25 Water Water TABLE 11 AASHTO T103 Procedures A, B, and C

number of states and provinces have adopted freezing and thawing tests similar to AASHTO T103. Where freezing-and-thawing is a concern, a test that actually reproduces freezing and thawing may be preferable over a sulfate soundness test. • Aggregate polish resistance is of concern to agencies with a predominance of carbonate aggregates. The polish stone value test is the most widely used test to assess polish resistance of aggregates. When setting specifications, agencies need to consider the interaction between tests for abrasion resistance and durability— such as the micro-deval test and tests for polishing— because some aggregates with high polish stone values may not be durable. 2.8 EFFECT OF AGGREGATE GRADING ON HMA PROPERTIES Gradation is perhaps the most important property of an aggregate. The link between aggregate gradation and asphalt mixture performance was recognized early in the develop- ment of mix design methods (121). Gradation affects almost all the important properties of HMA, including stiffness, sta- bility, durability, permeability, workability, fatigue resistance, frictional resistance, and resistance to moisture damage. The mixture volumetric properties including asphalt content, VMA, and VFA have been identified as important parameters for durability and performance. However, the VMA is consid- ered the most important parameter and is used in the Super- pave mixture design specifications to eliminate use of poten- tially poor-performing mixtures. 2.8.1 Methods for Analyzing Gradations Prior to the establishment of formal mixture design meth- ods, gradation was evaluated to determine asphalt demand. Formulas were applied to the gradation, and asphalt require- ments were calculated to provide satisfactory durability with minimum amount of asphalt binder (121). By the 1920s, the Hubbard–Field method of mix design recognized the impor- tance of air voids as a key parameter controlling field perfor- mance of mixtures (122). The Hubbard–Field mix design is based on the need for air voids and for a minimum amount of asphalt binder. Voids in total mix and voids in aggregate mass were both specified. Early mixture design methods were based on a belief that a “gradation law” existed that controlled asphalt mixture properties. Considerable research effort was expended to discover this law that controlled aggregate packing. Associated with the gradation law was the belief that an “ideal” gradation existed that would provide adequate space for minimum amount of asphalt and air voids and adequate stability under traffic. Today, aggregate gradations are commonly evaluated using a “0.45 power chart” (123). Despite the chart’s usefulness, 50 some confusion exists regarding its practical application. One use of the 0.45 power chart is to estimate available VMA of compacted mixtures. Increased VMA is obtained by mov- ing further from the maximum density line, but several meth- ods exist for drawing maximum density lines. The packing characteristics of coated aggregate particles in an asphalt mixture are related to aggregate surface character- istics and gradation. Aggregate surface characteristics of the particles include angularity and surface texture. Gradation is the size distribution of the particles. When selecting aggregate for a project, surface characteristics may not be selected to obtain VMA. Conversely, VMA of a mixture is essentially obtained by default. If additional VMA is required, changes are usually made to the aggregate gradation. In some cases, natural sands, which are predominantly −600μm sieve mate- rial, are added. Natural sands have been identified as a cause of decreased resistance to permanent deformation and of tender mix problems during construction (121). As a result, limits have been placed on sand content and increases in VMA must be achieved by overall adjustment of gradation. Unlike natural sand addition, gradation adjustment can sometimes produce confusing results. Moving away from the maximum density line sometimes causes decreases rather than increases in VMA. In the recent years, the Bailey method was developed to select aggregate gradations for HMA mixture design. The Bailey method was originally developed by the Illinois DOT and has become a systematic approach to aggregate blending that is applicable to all dense-graded asphalt mixtures, regard- less of the maximum size aggregate in the mixture (124–126). The Bailey method uses two principles that are the basis of the relationship between aggregate gradation and mixture volumetrics: aggregate packing and definition of coarse and fine aggregate. In the Bailey method, aggregate interlock is selected as a design input. Gradation selection considers the packing char- acteristics of aggregates. The parameters in the method are related directly to VMA, air voids, and compaction properties. The definition of “coarse” and “fine” is more specific in order to determine the packing and aggregate interlock provided by the combination of aggregates in various sized mixtures: • Coarse Aggregate: Large aggregate particles that when placed in a unit volume create voids. • Fine Aggregate: Aggregate particles that can fill the voids created by the coarse aggregate in the mixture. The primary steps in the Bailey Method are (1) compare aggregates by volume and (2) analyze the combined blend. Aggregate is blended by volume. The combined blend is bro- ken down into three distinct portions: coarse aggregate, coarse portion of fine aggregate, and fine portion of fine aggregate. Each portion is evaluated individually. Figure 16 shows a schematic of how the gradation is divided into the three por- tions. A factor of 0.22 was used to determine a primary con-

trol sieve (PCS), secondary control sieve (SCS), and tertiary control sieve (TCS). Table 12 lists the control sieves for var- ious asphalt mixture sizes. The design and analysis of an aggregate blend using the Bai- ley method of gradation selection is built on four parameters: 1. Chosen unit weight, which describes the interlock of the coarse aggregate; 2. CA Ratio: coarse aggregate ratio, which describes gra- dation of coarse aggregate; 3. FAc Ratio: fine aggregate coarse ratio, which describes gradation of coarse portion of fine aggregate; and 4. FAf Ratio: fine aggregate fine ratio, which describes gra- dation of fine portion of fine aggregate. Changes to any of these parameters will affect the air voids, VMA, constructability, and performance of the resulting asphalt mixture. These changes are the same whether the change is made in the laboratory during design or in the field during construction. 2.8.2 Effect of the Restricted Zone on HMA Performance During SHRP asphalt research, an Aggregate Expert Task Group (ETG) was formed to develop recommendations for aggregate properties and gradations for HMA. The final rec- ommendations for gradations included a restricted zone that lies along the maximum density line between the intermediate 51 sieve size (2.36-mm or 4.75-mm, depending on the maximum aggregate size) and the 0.3-mm size. The restricted zone was recommended to reduce the incidence of tender or rut-prone mixes. The origin of the Superpave-defined restricted zone is documented in a SHRP report (1, 127). From a historical perspective, the restricted zone is some- thing new: not until the Superpave method was there a for- mal guideline for aggregate gradation called the “restricted zone.” However, the industry has been aware of potential performance problems with gradations that pass through the Superpave-defined restricted zone-region. In 1940, Hveem (128) described a number of HMA gradations that showed a hump between the 0.6-mm and 0.15-mm sieve sizes. Hveem indicated that the hump was caused by an excessive amount of sand in this size fraction. He said that the hump is indica- tive of wind blown sand (smooth-textured, rounded sand) within the aggregate blend and that in his experience, the hump resulted in HMA mixes with low stability. The initial concept of a restricted zone around the maximum density line can probably be indirectly traced back to Goode and Lufsey (123). Based upon work by Nijboer (129) to iden- tify a maximum density line, Goode and Lufsey presented a 0.45 power grading chart for plotting aggregate gradations. To utilize the newly developed gradation chart, Goode and Lufsey evaluated 24 gradations to observe the effect of sand con- tent on the stability of HMA mixes. What prompted their study were reported cases in which tender mixes were encountered with gradation humps between the 0.6-mm and 0.3-mm sieve sizes. Goode and Lufsey found that, in general, gradations that Coarse Aggregate Coarse Portion of Fine Aggregate Fine Portion of Fine Aggregate PCS SCS Figure 16. Overview of the divisions in a continuous gradation that allows an analysis of gradation (124). NMAS, mm 37.5 25.0 9.5 4.7519.0 12.5 Half Sieve 19.0 12.5 9.5 ** 4.75 2.36 PCS 9.5 4.75 4.75 2.36 2.36 1.18 SCS 2.36 1.18 1.18 0.60 0.60 0.30 TCS 0.60 0.300.60 0.30 0.150 0.150 0.075 ** The nearest “typical” half sieve for a 12.5-mm NMAS mixture is the 4.75 mm; however, the 6.25- mm sieve actually serves as the breakpoint. Interpolating the percent passing value for the 6.25-mm sieve for use in the coarse aggregate ratio will provide a more representative ratio value. TABLE 12 Control sieves for various asphalt mixes

show appreciable humps above the maximum density line at about the 0.6-mm sieve produced higher VMA and lower Marshall stabilities than do gradations that plot as a more dense gradation. The recently published Transportation Research Circular E-C043: Significance of Restricted Zone in Superpave Aggre- gate Gradation Specification (130) reviewed results from research relevant to the performance of mixtures with grada- tions passing above the restricted zone, below the restricted zone, cross the restricted zone or S-curve, and through the restricted zone. Independent results from the literature indicate that no relationship exists between the Superpave restricted zone and HMA rutting or fatigue performance. Mixes meet- ing Superpave and FAA requirements with gradations that violated the restricted zone performed similarly to or better than the mixes having gradations passing outside the restricted zone. Results from numerous studies (3, 127, 131–140) show that the restricted zone is redundant in all conditions when all other relevant Superpave volumetric mix and FAA require- ments are satisfied. Based on this research, the restricted zone is no longer included in AASHTO M323 (Superpave Method). 2.9 EFFECT OF AGGREGATE FINES AND FILLERS ON HMA PERFORMANCE Mineral fillers were originally added to dense-graded HMA paving mixtures to fill the voids in the aggregate skeleton and to reduce the voids in the mixture. When asphalt binder is mixed with aggregate, the fines mix with the asphalt binder to form a fines-asphalt mortar. The additions of fines to the asphalt binder can have three main effects: extend the asphalt binder, or stiffen the asphalt binder, or both. This modifica- tion to the binder that may take place because of the addition of fines could, in turn, affect HMA properties. 2.9.1 Research on Fines and Fillers Extensive research efforts on mineral filler and baghouse fines have been made by many researchers throughout the world. Kandhal and Parker (2) and Kandhal (141) summa- rized the influences mineral filler can have on the perfor- mance of HMA mixtures as follows: • Depending on the particle size, fines can act as a filler or an extender of asphalt cement binder. In the later case, an over-rich HMA mix can lead to flushing and rutting. In many cases, the amount of asphalt cement used must be reduced to prevent a loss of stability or pavement bleeding. • Some fines have a considerable effect on the asphalt cement, making it act as a much stiffer grade of asphalt cement compared with the neat asphalt cement grade and, thus, affecting the HMA pavement performance including its fracture behavior. 52 • Some fines make HMA mixtures susceptible to moisture-induced damage. Water sensitivity of one source of slag baghouse fines has been reported in the United States, and the water sensitivity of other stone dusts has been reported in Germany. Stripping of HMA mixtures as related to the properties of filler–asphalt combinations has been reported in Japan. In NCHRP Project 4-19, “Aggregate Tests Related to Per- formance of Asphalt Concrete in Pavements,” Kandhal and Parker (2) conducted dynamic sheer rheometer (DSR) tests on filler-asphalt mortars to determine the rutting and fatigue properties. Fines passing a No. 200 sieve (P200 material) obtained from six different mineral aggregates were included in the study. The P200 materials were characterized by Rig- den voids, particle size analysis, methylene blue test, and a German filler test. HMA specimens containing different P200 materials were tested in the Superpave shear test device for rutting and fatigue cracking. AASHTO T283 (modified Lottman test) was used to evaluate moisture susceptibility. It was found that the D60 size (the particle size that 60% would be passing or smaller than) and methylene blue values were related to rutting, whereas the D10 size (the particle size that 10% would be passing or smaller than) and methylene blue values were related to stripping. No performance-related test was identified for fatigue cracking. Anderson and Goetz (142) evaluated the stiffening effect of a series of one-sized fillers in one of the first studies that focused on determining the mechanical properties of asphalt filler mixtures. They concluded that both the size of the filler and asphalt binder composition had a significant influence on the stiffening effect. Rigden (143) conducted experiments to study the relationship between filler properties and the vis- cosity of mineral filler–asphalt cement mixtures. As much as a 1,000-fold increase in viscosity of neat asphalt cement was measured when certain fillers were added to the cement in ratios similar to ratios used in a typical HMA. Rigden showed a strong correlation between the voids content of dry com- pacted filler and the amount of stiffening produced by the filler. In 1992, Anderson et al. (144) conducted a study to deter- mine whether the addition of baghouse fines affects the fail- ure or fracture properties of HMA mixtures. No conclusions with respect to fatigue were drawn because of the problems encountered with the test procedures. With respect to frac- ture properties, it was concluded that the mineral filler frac- tion could have a significant effect and that • Gradation does not necessarily relate to stiffening—the finest dust acted in much the same manner as the coarser dust—and • Fracture toughness, J1c, appeared to be sensitive to the source of the aggregate as well as to the amount of added baghouse dust. In general, the addition of the dust increased the fracture toughness of the HMA mixture.

In another study, Anderson et al. (145) stated that the importance of mineral filler fraction was often overlooked even though it is one of the most important components of HMA. Two mineral fillers, quartz and calcite, were added to four asphalt cements, and the rheological properties and fail- ure properties of the resulting mastics were determined using the test methods developed by SHRP. DSR, flexural creep, and direct tension were found to be applicable to void- less filler–asphalt cement mastics. Based on the study, it was found that • The addition of the mineral filler does not affect the tem- perature shift factors of the rheological response but does change the frequency dependency by lengthening the relaxation times, thereby stiffening the asphalt. • The presence of the mineral filler did not significantly affect the rate or level of oxidative or physical hardening. • At low temperature the mineral filler imparts a leathery- like behavior to the mastic, enhancing the strain and energy-to-failure characteristics of asphalt cement. “Leathery-like behavior” is used to describe a tempera- ture region for polymers between the glassy and rubbery state and is also referred to as the glass-transition region. In this state, deformation is time dependent and not totally recoverable. Normally, binder might be expected to have a glassy behavior at low temperature, resulting in a brittle failure. The authors concluded that asphalt mastics can play a major role in defining the performance of HMA. The data also led the authors to conclude that voidless mastics, similar in vol- ume concentration to the mineral filler–asphalt fraction in typical HMA, can be characterized with the same test meth- ods as those developed for neat asphalt cement. Gubler et al. (146) conducted DSR and bending beam rheometer (BBR) tests on a series of asphalt and binder com- binations. Two asphalt binders and three fillers with varied free volumes and ageing conditions were included in their study. The authors believed that stiffening is only one way in which the addition of mineral filler changes the properties of asphalt binder. In fact, mastics behave quite differently than does a binder that is simply stiffer binder. These differences include changes in the material properties with aging and the time and magnitude of loading that are of practical and sci- entific interest. Unlike asphalt binder, mastics are susceptible to shear; their mechanical properties are changed by the application of stress during the test itself. Their mechanical properties are also dependent on the amplitude of the applied stress and the time the stress is applied and are thus a function of the test- ing history. An important decrease in complex modulus (up to a 50% decrease) during testing at intermediate strains, fol- lowed by a partial recovery of the modulus during a subse- quent period with low strain, was demonstrated in the study. 53 Their results indicated that filler can promote the oxidation and hardening of asphalt binders. Since it is generally accepted that fatigue is related to hardening of the binder, fatigue must also be related to this phenomenon. With the increased use of SMA mixtures in the United States, the importance of the mortar in SMA mixtures had been recognized by many researchers. In SMA, the mortar is composed of fine aggregate, filler, asphalt cement, and a sta- bilizing additive. The mortar is an important component of SMA. It needs to be stiff to help prevent draindown and flushing during production and placement and to resist rut- ting during in-service life; it must also be flexible enough to resist fatigue and thermal cracking. Brown et al. (147) conducted a comprehensive study to determine whether SMA mortar can be evaluated by the Superpave system binder tests and to determine the manner in which each of the mortar components affects the overall mortar performance. In the study, for testing purposes, the fine mortar fraction was considered to be a binder and was tested in the Superpave binder equipment before and after aging. The total mortar fraction was considered to be more like a mixture and was tested at low, intermediate, and high temperatures using the BBR, resilient modulus, indirect ten- sile test, and Brookfield viscometer. Test results indicated how each of the mortar components affects the mortar prop- erties. Most of the stiffening effect comes from the mineral filler. It is believed that the finer the filler, the more stiffen- ing it will provide. However, in this study, the coarser bag- house fines stiffened the mortar more than a limestone dust did. This suggests that filler size is not the only important parameter in specifying fillers. The parent material from which the filler comes as well as filler particle angularity may also be important to SMA. The authors found that with minor modifications, the DSR and BBR appear to offer viable test methods for determining SMA mortar properties. The direct tension test also may be applicable, but will require more modification to testing procedures. The Brookfield viscome- ter does not seem to be applicable in its present form. A study was conducted by Mogawer and Stuart (148) to determine whether mastic and mixture properties can distin- guish good mineral fillers from poor ones. Eight mineral fillers with known performance were obtained from three European countries. Mastics were tested for stiffness using the BBR, DSR, and ring-and-ball softening point. The authors found that none of the tests distinguished among mastics with good and poor mineral fillers. Mixtures were tested for draindown of mastic using the NCAT draindown test, for rutting using the LCPC pavement rut tester, for low temperature cracking using the indirect ten- sile test, for workability using the USACE gyratory testing machine, and for moisture susceptibility using the ASTM D4867 method. None of the tests distinguished among SMA mixtures with good and poor mineral fillers. There was a good correlation between the free binder con- tent and the stiffness of the mastics measured by the BBR and

the stiffening power measured using the ring-and-ball appa- ratus. The ramification of this is unknown since the tests did not distinguish between good and poor fillers. The following three conclusions were drawn: 1. A poor correlation existed between the stiffening power measured by the DSR and the free binder content. 2. A poor correlation existed between the percent rut depth measured by the LCPC pavement rut tester and the free binder content. 3. A poor correlation existed between the tensile strengths measured by the indirect tensile strength test and the free binder content. Cooley et al. (149) conducted a study to evaluate the stiff- ening potential of baghouse fines using conventional and Superpave binder tests and to establish a reasonable upper limit on the percent bulk volume of dry compacted baghouse fines, as determined by the Pennsylvania State University– modified (Penn State–modified) Rigden void test, which would limit the stiffening potential. This limit could then be used in lieu of dust proportion to more accurately reflect the influ- ence of baghouse fines or filler. Variables included 10 bag- house fines or fillers, 2 grades of asphalt cements, and 4 dust proportions. Tests conducted on unaged filler-asphalt mor- tars included ring-and-ball softening point, Brookfield vis- cometer, and DSR at 64°C. Mortars aged in the pressure aging vessel (PAV) were also tested by the DSR at 22°C and the BBR at −18°C. It was concluded from this study that Penn State–modified Rigden voids test can be used to characterize the stiffening potential of baghouse fines as measured by the softening point, the Brookfield viscometer, and the DSR. Because of the set- tling of fines during testing with the DSR, only the softening point test and the Brookfield viscometer gave test data that had excellent correlation with the Rigden voids. Buttlar et al. (150) used particulate composite micro- mechanics models to investigate three reinforcement regimes in asphalt mastics: volume filling, physiochemical effects, and particle interaction. Consistent definitions were given for the three reinforcement mechanisms: • Volumetric-filling reinforcement: The stiffening caused by the presence of rigid inclusions in a less rigid matrix. This stiffening level was assumed to be adequately described by the generalized self-consistent scheme (GSCS) model or by the simplified GSCS-based pre- diction equations. • Physiochemical reinforcement: The stiffening caused by interfacial effects between asphalt and filler particles, including absorption, adsorption, and selective sorption. The altered asphalt effectively forms a rigid layer, which leads to a higher net volume concentration of rigid mat- ter, which in turn leads to increased mastic stiffness. 54 • Particle-interaction reinforcement: The stiffening beyond volume filling and physiochemical reinforcement. This effect increases with increasing filler content, as rigid mat- ter comes into contact and forms a skeletal framework. Ishai and Craus (151) summarized a long-term research effort conducted in Israel concerning aggregate and filler properties that have significant influence on the behavior and durability of bituminous paving mixtures. Six types of filler were used in the evaluation of the physicochemical properties of the fillers, the rheological characterization of filler-asphalt mastics, and the strength and durability tests on sand-asphalt mixtures and HMA mixtures. Parameters such as specific sur- face, shape factor, specific rugosity, and surface texture were evaluated for each filler type. The surface activity of the fillers, as related to interaction with bitumen, was character- ized by either adsorption intensity or selective adsorption. Criteria serving locally (in Israel) as a tool for accepting and rejecting fillers with respect to durability were suggested. They are based on the properties of the filler, the initial properties of the mixture, and the durability behavior of the mixture. Shashidhar and Romero (152) introduced two intermedi- ate measurable parameters—the maximum packing fraction, φm, and the generalized Einstein coefficient, KE—to charac- terize the asphalt-filler system. This introduction enables a better understanding of the influence of various factors such as average particle size, gradation, particle shape, presence of agglomerates, degree of dispersion, and the asphalt-filler interface on the stiffening potential of asphalt. The following conclusions were drawn from the study: • The stiffening effect of the fillers increased with decreas- ing particle sizes below 10 µm. Above 10 µm, such dependencies were not significant. • The asphalt-filler interface was shown to have a signifi- cant effect on stiffening. The interface properties changed from asphalt to asphalt, and the interface can be engi- neered to yield desired properties. • Fillers in asphalt had low φm, indicating that they were poorly dispersed in asphalt. • Agglomerates were shown to increase KE, decrease φm, and therefore to increase stiffening power. Asphalts with agglomerated fillers were shown to have stiffness many times the stiffness of asphalt with unagglomerated fillers. • An increase in the aspect ratio of the filler particles low- ers φm and increases KE. Both of these effects increase the stiffening power of fillers. • Rigden’s fractional voids concept does not take into account the agglomeration, degree of dispersion, and asphalt-filler interface contributions. • The stiffness of asphalt mastic in a specific system can only be predicted accurately with the measurement of parameters φm and KE. Shashidhar et al. (153) later developed methods evaluating the parameters φm and KE. The volume-filling contribution to stiffening was captured by the parameter φm and the physico-

chemical contribution was captured by KE. These parameters had a physical basis and were a function of many factors that affected the stiffening potential of fillers. It was found that the volume-filling contribution to stiff- ening (φm) was able to better distinguish between fillers with “good” and “bad” performance in SMA in the cases studied. It was able to predict the performance accurately even in cases in which prediction by Rigden voids had failed. The authors concluded that the stiffness of asphalt by a filler can be fully characterized by measuring the maximum packing fraction, φm, and the generalized Einstein coefficient, KE, of the system. The data obtained show that φm is a better pre- dictor of the performance of filler in SMA than Rigden voids. The φm denotes the maximum filler one can put into the system and the volume-filling contribution to stiffening. Mathematically, it is an asymptote at which the stiffening is infinite. φm is analogous to bulk density in many ways. KE is the physicochemical contribution to stiffening. It is a mea- sure of the rate of increase in stiffness ratio with the addition of fillers: the higher the KE, the higher the slope of the curves. 2.9.2 Summary of Research Related to Fines and Fillers It is widely believed that depending on the particle size, fines can act as a filler or as an extender of asphalt cement binder. Some fines have a considerable effect on the asphalt cement, making it act as a much stiffer grade of asphalt cement compared with the neat asphalt cement. Early work indicated that both the size of the filler and the asphalt binder composition had an impact on the stiffening effect. As much as a 1,000-fold increase in viscosity of the neat asphalt cement was measured when certain fillers were added to asphalt cement. Some fines may also make HMA mixtures more sus- ceptible to moisture-induced damage. Numerous studies have evaluated the effects of fines, filler, and mortar on HMA performance in the laboratory and in the field. Efforts to characterize fillers have generally fol- lowed three paths: 1. Characterization of particle size or packing. Several research studies have been conducted to develop suitable test parameters related to particle size or packing to eval- uate the fines and fillers. D60 (the particle size of P200 at 60% passing) and methylene blue values were found to be related to rutting, and D10 and methylene blue val- ues to stripping. The modified Rigden voids test has been used to characterize the stiffening potential of baghouse fines. 2. Binder tests performed on a mortar. Superpave binder tests, BBR, DSR, flexural creep, and direct ten- sion tests have been used by several researchers to char- acterize the fine mortar or voidless mastics properties. 3. Modeling of the overall interaction between the filler and binder. Recent efforts involve modeling the physical-chemical interaction between fillers and binder. 55 2.10 EFFECT OF CRUSHING OPERATIONS ON AGGREGATE PROPERTIES Barksdale (25) states, “Rock is broken or crushed when a force is applied with sufficient energy to disrupt internal bonds or planes of weakness that exist within the rock.” For quarried aggregates, the crushing process begins with the blast that turns solid rock into particles of a size range that can be accepted by the primary crusher. The resulting parti- cles from the initial blast are called “shot rock.” Additional crushing of shot rock or gravel is performed (1) to reduce the aggregate to product size; (2) to improve the aggregate shape; and (3), in the case of gravel for HMA, to create frac- tured faces. Aggregate is produced in a variety of sizes and for a variety of purposes besides HMA. In some cases, the properties desired for HMA may conflict with those desired for another product. The ratio between the sieve size representing 80% passing of the crusher feed stock and the sieve size representing 80% passing for the product of the crusher is termed the “reduc- tion ratio” (25). When processing aggregate, a 21 reduction ratio will result in at least one fractured face (154). Therefore, when determining the possible number of fractured faces for gravel sources, the feed size and the resulting product size must be considered. It is impossible to create a 12.5-mm crushed gravel having 100% of the particles with one frac- tured face if the feed stock is only 19.0 mm. Crushers reduce the size of aggregate particles through three mechanisms: abrasion, cleavage, and impact (Figure 17) (25). Abrasion occurs in localized areas when insufficient energy is applied to the particle to cause significant fracture. Abra- sion results in limited size reduction and the production of fines. Cleavage results when the compressive forces applied to Figure 17. Mechanisms of rock fracture (Figure from Kelley [155] published in Barksdale [25]).

aggregate particles cause a limited number of fracture planes when the aggregate particle is trapped between two crushing surfaces, between other aggregate particles, or between a com- bination thereof. As shown in Figure 18, cleavage produces a few large particles. Impact-type crushers cause the particle to shatter as it is propelled at high speeds against an anvil or against other aggregate particles. Impact-type crushers pro- duce the widest distribution of particle sizes. Barksdale (25) noted that when producing the same size coarse aggregate product, impact crushers tend to produce a greater percent- age of particles passing the 4.75-mm sieve, but compres- sion type crushers produce a greater percentage of material passing the 0.075-mm sieve (dust) within the fine aggregate fraction. 2.10.1 Types of Crushers There are four major types of crushers used to produce aggregate for HMA: jaw, gyratory, cone, and impact. Jaw, gyratory, and cone crushers are all forms of compression crushers. Compression-type crushers apply a compressive force to the aggregate that is trapped between crushing sur- faces. A common characteristic of these machines is that the aggregate must pass through a fixed opening. The fixed open- ing is adjustable and is referred to as the “close-side” setting (25). Jaw and gyratory crushers apply the crushing force slowly, producing cleavage and abrasion. Cone crushers, a subclass of gyratory crushers, apply their energy approxi- mately twice as fast, producing fracture by shatter as well as by cleavage (25). Examples of jaw and cone crushers are shown in Figures 19 and 20. A complete description of crusher types is provided in Barksdale (25). Typical reduction ratios for jaw-type crushers are 71. Gyratory or cone crushers can produce reduction ratios from 21 through 101. The use of high-reduction ratios tends to 56 Figure 18. Size distributions resulting from various fracture mechanisms (Figure from Kelley [155] published in Barksdale [25]). Figure 19. Schematic of a jaw-type crusher (156). Figure 20. Cutaway view of Symons cone crusher (156, originally from Nordberg, Inc., Milwaukee, WI). produce excessive fines by overcrushing. Overcrushing occurs when rocks of the desired size are recrushed before they can pass out of the crusher and be removed from the crushing stream by screening. There are two types of impact crushers, horizontal shaft and vertical shaft. Horizontal shaft impact crushers use one or more rotors, hammers, or rotating pins mounted on a cage. The rotors or hammers directly impact the rock as well as propel the rock against aprons, anvils, or other aggregate par- ticles where further impact occurs. Horizontal shaft impact crushers can produce a high reduction ratio, from 15 through 20 to 1. Horizontal shaft impact crushers are only suitable for low-abrasion aggregate feeds. In a vertical shaft impact crusher, the aggregate feed is introduced into a shoe or pump spinning on a vertical axis. The aggregate feed is thrown cen- trifugally against a series of anvils, pockets of aggregate par- ticles (i.e., autogenous), or a combination thereof (25). Ver- tical shaft impact crushers produce a small reduction ratio

and are often used for crushing fines. An example of an auto- genous vertical shaft impact crusher is shown in Figure 21. Crushers are often referred to by the order in which they crush the aggregate. For instance, the “primary” crusher is the first crusher into which the aggregate feed, either shot rock or gravel, is introduced. Following the primary crusher are the secondary, tertiary, and, in some cases, quaternary crushers— that is, the second, third, or fourth crusher in the crushing cir- cuit. Generally each crusher is used to produce a progressively smaller product, although in some cases, a crusher may be used more for shape improvement and less for size reduction. 2.10.2 Factors Affecting Aggregate Shape The geology of the aggregate is probably the most signif- icant factor affecting crushed aggregate shape. Fine-grained (i.e., aphinitic) aggregates, such as limestone, tend to be more brittle and therefore fracture into more F&E (157). Slates, injected quartzites, and basalts are examples of other aphinitic aggregates. Quartzites, basalts, and cherts also tend to frac- ture in a concoidal manner—that is, they produce curved frac- ture surfaces like glass (157, 158). The following factors tend to improve the shape of parti- cles crushed with compression crushers (159–161): • The crusher should be run with a full or choked feed cavity to promote interparticle crushing. • Crushers should be operated in closed circuits where a recirculating feed can be used to fill the crusher cavity. • The reduction ratio should be reduced. Reducing the feed size or increasing the circulating load can accom- plish this. • The close-side setting should be approximately equal to the desired product size. When single layer (a single aggregate particle trapped between the crusher jaws or crushing head and bowl liner) occurs, the particles are more likely to split in a flat and elongated man- 57 ner as shown in Figure 22a. This occurs because the stresses are essentially concentrated at two points, causing long cracks in between the contact points. Multilayer crushing produces a greater number of contact points as shown in Figure 22b. Multilayer crushing produces more cubical aggregate shape. Multilayer crushing is achieved by keeping the upper portion of the crusher cavity full or chocked. This requires that the crusher be operated as part of a close circuit. A closed circuit provides a recirculating load and surge piles to supply a rel- atively constant feed rate to the crusher. The feed rate must be adjusted (increased) as the crusher liner wears open or more particles will pass through the open-side setting with- out being crushed. The crushing head of gyratory or cone-type crushers oscil- lates around the central axis of the crusher as shown in Fig- ure 23. As discussed previously, the close-side setting is the smallest distance between the crusher head and the bowl liner. The open-side setting is the largest distance between the crusher head and the bowl liner. The open-side setting occurs at a point opposite the close-side setting. Increasing crusher speed decreases the number of particles that pass through the open-side setting of a cone or gyratory crusher without being crushed. Most older gyratory and cone crushers operate at a fixed speed. Newer high-pressure cone crushers tend to oper- ate at a higher or variable speed. Figure 21. Schematic of Barmac autogenous vertical shaft impact crusher (156, originally from Barmac). Figure 22a. Single-layer crushing.

58 Figure 23. Vertical view of gyratory action of gyratory/cone type crusher. per day (162). Besides the reduced production rate, the other drawback of this type of production process is the generation of excess crushed fines. Many crushing operations, including this German example, produce more crushed fine aggregate than they can sell. A related concept to the reduced reduction ratio is the con- cept of producing product close to the close-side setting of the crusher. A typical crushing operation in the United States might produce ASTM No. 57 stone in combination with ASTM No. 8 stone (159). In this case, the close-side setting of the crusher would be set to approximately 29 mm. Obser- vations indicate that the No. 57 stone would tend to have the best particle shape and that the smaller No. 8 stone would tend to be more flat and elongated. This phenomenon is illus- trated in Figure 24, which is produced with data from Jahn (163) for a granite No. 57 stone. The aggregate was tested using the multiple shape ratio method. Producing the No. 8 stone separately from the No. 57 stone could eliminate the effect. However, this process would increase the production of fines and would require additional equipment or reduced production. Particle shape can also be improved with the use of impact crushers. Shergold (164) states: It is thought that the good particle shape obtained with impact breakers can be explained with the assumption that, in impact breaking, the stresses have a more or less random distribution, whereas in compressive crushing the stresses are concentrated in relatively closer spaced planes near the surface. Huber et al. (34) demonstrated the improvement in shape when the same aggregate was crushed with a cone and a ver- tical shaft impact crusher. No. 57 stone was produced from the same Indiana limestone source in each crusher. Neither crusher produced flat and elongated maximum to minimum particle ratios greater than 51. The cone crusher produced 19.4%, and the vertical shaft impact crusher produced 9.0% particles exceeding the 31 ratio. One potential drawback of using impact-type crushers to produce coarse aggregate is the resulting effect on the fine aggregate. Iowa DOT is conducting research to compare the aggre- gate properties and properties of the resulting aggregate pro- duced with a cone or hammermill (a form of horizontal shaft impact crusher) crusher (165). The study evaluated three aggregate sources. Preliminary results indicate no significant difference in aggregate shape between particles crushed with the cone or hammermill crusher. However, the VMA of the HMA produced with aggregate crushed in the hammermill crusher was consistently 0.7% higher than the VMA of the HMA produced with the cone crusher. 2.10.4 Influence of Shape on Performance The production of more cubical coarse aggregate can produce more cubical fine aggregate (166, 167). Studies have Figure 22b. Multi-layer crushing. 2.10.3 Applications of Crushers German specifications require that all coarse aggregate used in HMA have less than 20% 3 to 1 particles (162). Good particle shape that meets this criterion can be produced through the used of compression-type crushers, even with granite-diorite aggregates. One German quarry uses a jaw primary crusher and 5.5-ft cone secondary crusher, after which aggregate for HMA passes through a series of four to five short head cone crushers to improve shape (162). Storage is provided to maintain a consistent feed rate. The resulting product is an 8-mm to 11-mm aggregate with typical percent F&E greater than 7% to 9% 31 ratio. This is an example of using smaller reduction ratios to improve aggregate shape. Using this method, the quarry produces 3,000 to 4,000 tons per day. However, when specialty stone was produced for an open-graded friction course requiring less than 5% 31 par- ticles, production levels were reduced to less than 150 tons

indicated that cubical fine aggregate particles may pack similarly to round fine aggregate particles, producing low percent uncompacted voids as measured by AASHTO T304 Method A (127, 54). Vertical shaft impact crushers have been used to produce more cubical fine aggregate for use in Port- land cement concrete (157, 168, 169). A Virginia stone pro- ducer investigated the effects of vertical shaft impact crush- ers to produce more cubical fine aggregate. The resulting fine aggregate packed more tightly than that previously produced by the quarry. This resulted in an undesirable increase in the unit weight of concrete block produced with the aggregate, resulting in a loss of sales (157). In North Carolina, a quarry separated sand-size particles prior to crushing with a vertical shaft impact crusher to sell to HMA and concrete block plants (168). This illustrates the sometimes divergent requirements of aggregates between different industries. Aggregate is produced for numerous applications including base, HMA, and Portland cement concrete. The needs of 59 these products often compete. Cubical coarse aggregates are believed to be desirable for the production of HMA. However, the production of cubical coarse aggregates may result in • Aggregate base that packs more tightly, reducing drain- age capacity; • More cubical fine aggregate, resulting in lower uncom- pacted voids; • Reduced LA abrasion loss; • More cubical aggregates may pack closer in HMA, resulting in lower VMA; • More cubical fine aggregate, which packs closer, result- ing in higher density block; and • More cubical fine aggregate that reduces water demand for Portland cement concrete, resulting in higher strength. Although aggregate shape could seemingly be customized for a given application, in practice this is not possible or, at least, is cost prohibitive. 0 5 10 15 20 25 30 35 4.759.512.519 Sieve Size (mm) Particles Retained On Pe rc en t b y M as s of P ar tic le s Ex ce ed in g Sp ec ifie d Ra tio 3:1 Ratio 5:1 Ratio Granite #57 Stone Figure 24. Example of shape variation with particle size.

Next: Chapter 3 - Survey of Current State Agency Specifications »
Aggregate Properties and the Performance of Superpave-Designed Hot-Mix Asphalt Get This Book
×
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF

TRB’s National Cooperative Highway Research Program (NCHRP) Report 539: Aggregate Properties and the Performance of Superpave-Designed Hot-Mix Asphalt examines technical literature available since the conclusion of the Strategic Highway Research Program in 1993 on the impact of the aggregate properties specified by the Superpave mix design method on the performance of hot-mix asphalt. The performance of hot-mix asphalt (HMA) is largely determined by the characteristics of its constituents: asphalt binder and aggregate. In developing the Superpave mix design method, the Strategic Highway Research Program (SHRP, 1987–1993) targeted the properties of asphalt binders and HMA and their effects on pavement performance.

  1. ×

    Welcome to OpenBook!

    You're looking at OpenBook, NAP.edu's online reading room since 1999. Based on feedback from you, our users, we've made some improvements that make it easier than ever to read thousands of publications on our website.

    Do you want to take a quick tour of the OpenBook's features?

    No Thanks Take a Tour »
  2. ×

    Show this book's table of contents, where you can jump to any chapter by name.

    « Back Next »
  3. ×

    ...or use these buttons to go back to the previous chapter or skip to the next one.

    « Back Next »
  4. ×

    Jump up to the previous page or down to the next one. Also, you can type in a page number and press Enter to go directly to that page in the book.

    « Back Next »
  5. ×

    To search the entire text of this book, type in your search term here and press Enter.

    « Back Next »
  6. ×

    Share a link to this book page on your preferred social network or via email.

    « Back Next »
  7. ×

    View our suggested citation for this chapter.

    « Back Next »
  8. ×

    Ready to take your reading offline? Click here to buy this book in print or download it as a free PDF, if available.

    « Back Next »
Stay Connected!