VMA: ONE KEY TO MIXTURE PERFORMANCE by Donald W. Christensen, Jr., Ph.D., P.E. Senior Engineer Advanced Asphalt Technologies, LLC 444 E. College Ave., Suite 510 State College, PA 16801 Ramon F. Bonaquist, Ph.D., P.E. Chief Operating Officer Advanced Asphalt Technologies, LLC 108 Powers Court, Suite 100 Sterling, VA 20166 Submitted to the South Central Superpave Center for Publication in the National Superpave Newsletter February 2005 INTRODUCTION Over the past 5 to 10 years, highway engineers have become increasingly concerned that the composition of hot mix asphalt concrete (HMAC) being placed on our roads may not be providing adequate durability under all conditions. A recent increase in surface distress in many asphalt concrete pavements has lead the National Cooperative Highway Research Program (NCHRP) to sponsor research specifically aimed at top-down cracking. Because of the perceived increase in surface cracking and raveling, many pavement engineers feel that asphalt concrete mixtures have become too lean—that is, the voids in mineral aggregate (VMA) and effective binder content by volume (VBE) are too low for adequate fatigue resistance and durability. Unfortunately, there is substantial evidence that increasing VMA and VBE can decrease the rut resistance of asphalt concrete pavements. The purpose of this article is to describe recent research providing guidelines for establishing design VMA values high enough to ensure good durability while still achieving superior rut resistance. The information presented in this article has largely grown out of research performed under NCHRP Project 9-25: Requirements for Voids in Mineral Aggregate for Mixtures Designed according to the Superpave System, and Project 9-31: Air Void Requirements for Superpave Mix Design (1). However, the findings, conclusions and recommendations presented in this article are those of the authors and may or may not reflect the position of the NCHRP or the Federal Highway Administration. NCHRP Projects 9-25 and 9-31 are nearing completion, and will probably be concluded by the time this article is published. However, some refinements in the findings and recommendations presented in this article are possible as the results of this research are implemented over the next few years. As is always good practice, highway engineers should use 2 their judgment and experience with local materials and conditions when evaluating and applying the information presented in this article. HISTORICAL PERSPECTIVE ON VMA Using VMA to evaluate and design asphalt concrete mixtures first gained popularity in the 1950’s, largely due to the efforts of Norman McLeod, a Canadian engineer and researcher who has made many other contributions to hot-mix asphalt technology. He was the first to propose the use of minimum VMA values for mixtures and was also the first to suggest that minimum VMA should increase with decreasing aggregate size (2, 3). Prior to McLeod, Marshall mix design specifications had only two limitations—air voids should be between 3 and 5 % and voids filled with asphalt (VFA) should be between 75 and 85 %. McLeod felt that VMA provided better control over many aspects of mixture composition and believed that establishing minimum VMA values would help ensure good durability. At about this same time, a controversy began concerning the best way to ensure good mixture durability. Like McLeod, many pavement engineers believed that VMA and effective binder content—meaning binder not absorbed into the aggregate—were the critical factors in achieving durable asphalt concrete. Campen, Goode and Lufsey, among others, argued instead that the amount of asphalt binder relative to aggregate surface area was critical to achieving good mixture durability (4, 5, 6). Some engineers chose to express the ratio of asphalt binder to aggregate surface area as “film thickness,” while others spoke in terms of “bitumen loading factor (6).” Pavement engineers and researchers still argue over this terminology today. Some would like to see mix design specifications address the issue of film thickness, while others vehemently oppose this idea on the grounds that asphalt cement binder simply does not exist as thin films 3 within asphalt concrete. These engineers instead believe—correctly—that asphalt concrete is a particulate composite material made up of an intimate mixture of asphalt and aggregate. It is probably best to go beyond the argument of whether or not asphalt films actually exist in asphalt concrete mixtures and instead agree that the amount of binder relative to aggregate surface area—perhaps best referred to as apparent film thickness—plays an important role in mixture performance. RUT RESISTANCE AND MIXTURE COMPOSITION Recent research suggests that engineers on both sides of the VMA vs. film thickness controversy have been partly right and partly wrong. The ratio of asphalt binder to aggregate surface area does in fact significantly affect mixture performance, but not in the way that many early researchers believed. Instead of affecting durability, this factor is probably more closely linked to rut resistance. All else being equal, as the amount of binder relative to aggregate surface area increases, the rut resistance of a mixture will decrease. An even better indicator of this characteristic is the ratio of VMA to aggregate surface area. In practical terms, this means that technicians and engineers responsible for mixture design should be wary of mixtures that have high VMA relative to the amount of fines in the aggregate gradation. Such mixtures will tend to exhibit poor rut resistance compared to mixtures with more aggregate fines and lower VMA. A good example of this was the extreme early rutting exhibited by many of the mixtures placed at WesTrack several years ago (7). A team of investigating engineers determined that this failure was in part due to the excessive film thickness of many of the mixtures. To ensure that asphalt concrete mixtures have adequate levels of rut resistance, it is necessary to establish minimum values for aggregate fineness at different levels of VMA. 4 Furthermore, it must be understood that rut resistance is a function not only of VMA and aggregate fineness, but also of design compaction level, field compaction level, and binder grade relative to both climate and design traffic level and speed. An effective model relating these factors to field rutting rate for data from WesTrack, MN/Road and the NCAT test track project was developed as part of NCHRP Projects 9-25 and 9-31 (1, 7, 8, 9, 10): 12,500 VMA3.24 RR = G* 1.08 0.65 S a2.16 Gb2.16 N design [(100 − VF ) (100 − VD )] 18.6 (1) Where RR = rutting rate (mm/m/ESALs1/3) = total rut depth in mm, divided by thickness of asphalt concrete in meters (to 0.10 m maximum) divided by the cube-root of traffic in ESALs |G*| = shear complex modulus of the binder (Pa), measured at 10 rad/s and at the yearly, 7-day average maximum pavement temperature at a depth of 50 mm Sa = specific surface of the aggregate in m2/kg Gb = bulk specific gravity of the aggregate Ndesign = design compaction level (gyrations) VF = air void content of the pavement as constructed VD = air void content of the mix as designed It should be pointed out that Equation 1 is based not on the concept of film thickness, but upon the concept of resistivity. When an asphalt concrete mixture is deformed under loading, the asphalt binder must flow through the aggregate structure. Conceptually, resistivity represents how difficult it is to force the asphalt binder to flow through the aggregate structure of a mixture. As resistivity increases, rut resistance also increases since energy consumed in forcing binder to flow through the aggregate not available to damage the aggregate structure permanently. 5 Resistivity increases with increasing aggregate surface area, increasing binder viscosity and decreasing VMA. Although closely related to apparent film thickness, resistivity is related to the ratio of VMA3 S z2 , and not to the ratio of VBE S a . Furthermore, the concept of resistivity does not depend in any way upon the existence of binder films within a mixture. In Figure 1 below, observed rutting rates and those predicted using Equation 1 are compared. The 90 % prediction interval for new observations for this model is equivalent to a factor of 2.0. Therefore, if using Equation 1 for design purposes, a factor of safety of 2.0 will provide 95 % confidence that the selected rut rate will not be exceeded (the 95 % confidence in this case is for a one-sided limit, while the 90 % confidence interval in Figure 1 is two-sided). For example, if the maximum allowable rut rate for a given project is 12 mm, designing for an ultimate rut depth of 6 mm will result in only 1 chance in 20 that the ultimate rut depth will exceed 12 mm. It should be pointed out that in analyzing these data, the amount of age hardening that occurred in the three different projects was accounted for through application of the MirzaWitczak global aging system (11). This may not seem to be an important factor, but can be highly significant, particularly when comparing rutting data from relatively short test track projects—where there is little opportunity for age hardening—to data from long term test roads where there can be significant hardening of the binder and asphalt concrete over the length of the project. This phenomenon in fact probably contributed to the rapid rutting at the WesTrack facility. High temperature binder grades for test tracks should be increased by one-half to one full grade, depending upon climate, to provide similar rut resistance to long term pavements. In general, the hotter the climate, the greater the needed correction to account for lack of age hardening in test track pavements. 6 Equation 1 suggests that to maintain similar levels of rut resistance, the ratio of VMA3 S a2 should be kept more or less constant. By applying this principle and examining a range of asphalt concrete mixtures exhibiting reasonable levels of rut resistance, it is possible to establish guidelines for aggregate fineness as a function of VMA. However, aggregate fines not only affect the rut resistance of asphalt concrete mixtures, but also improve their durability by limiting permeability. During NCHRP Projects 9-25 and 9-31, a model was developed for estimating in-place permeability of asphalt concrete mixtures from aggregate specific surface and air void content (1). However, in order to use this model to establish reasonable minimum values for aggregate fines, maximum permeability values are needed for asphalt concrete pavements. Recently, pavement engineers in Florida have suggested such a limit after studying the permeability of mixtures designed according to the Superpave system (12). They concluded that wearing course mixtures should exhibit an average in-place permeability of 100 × 10-5 cm/s. By applying the model developed during NCHRP Projects 9-25 and 9-31 to this suggested permeability limit (and by assuming a gradual increase in permeability with increasing NMAS and VMA), it is possible to calculate a second set of aggregate fineness requirements, in addition to those based upon rut resistance. Fortunately, the two sets of requirements are quite close, so that for practical purposes a single set of aggregate fineness values as a function of VMA can be established. These values are given in Table 1, both in terms of aggregate specific surface and FM300, which is the sum of the percent passing the 75-, 150- and 300-micron sieves (1). FM300 is easy to calculate, and also relates very closely to specific surface; for aggregate gradations used in the NCAT test track, NCHRP Project 9-9 and National Pooled Fund Study 176, the correlation between FM300 and specific surface was r2 = 89 %. Furthermore, based upon these data aggregate 7 specific surface in m2/kg can be estimated simply by dividing FM300 by 5 (1). Table 1 also includes recommended maximum values for FM300, which were calculated by multiplying the minimum values by 1.7; these maximums are consistent with the range in FM300 values for typical dense-graded asphalt concrete mixtures, and will help ensure that mixtures will not become too difficult to mix or compact because of excessive fines. As mentioned earlier, in order to apply the recommendations in Table 1 with confidence, the proper high temperature binder grade must be used for a given climate, design traffic speed and volume. The LTPPBind Software has been developed for this purpose, though many highway engineers believe that this software does not provide realistic estimates of required binder grades. Analysis of the high temperature binder grades (including adjustments for traffic speed and volume) using Equation 1 suggest that care is needed in using LTPPBind, or a binder with inadequate stiffness at high temperatures might be selected (1, 8). When using the LTPPBind software, a good approach is to select high temperature binder grades using 98 % reliability and applying the “KMC” traffic adjustments. In combination with the recommended aggregate fineness and VMA values given in Table 1, this should provide a minimum effective reliability against excessive rutting of about 95 %, and because binder grades are usually rounded up, the actual reliability will usually be greater than this. Many agencies already have established protocols for binder grade selection; if the established binder grading system is working well, there is no need to change it—it should work fine at higher design VMA values provided the aggregate fineness guidelines of Table 1 are followed. FATIGUE RESISTANCE AND VMA Durability of asphalt concrete mixtures is not just a function of permeability, but is also a function of fatigue resistance. Some researchers have proposed that fatigue resistance increases 8 with increasing VFA, while others have proposed that it improves with increasing binder content. Research performed during NCHRP Projects 9-25 and 9-31 strongly suggests that it is binder content by volume that best relates to fatigue resistance (1, 13). Because design air void content in mixtures designed according to the Superpave system is fixed at 4 %, it can also be stated that for these materials fatigue resistance increases with increasing VMA. Many paving engineers have suspected this and have attributed the surface distress observed in some recently constructed pavements to insufficient VMA and binder content. In fact, VMA values and binder contents have decreased significantly over the past 30 years, as illustrated in Figure 2. This plot summarizes VMA and binder contents from several different research projects and specifications covering asphalt concrete mixtures designed since the 1970s. In the National Rutting Study by Brown and Cross, the average VMA value for Marshall mixtures designed and placed in the 1970’s and early 1980’s was found to be 17.0 % (14). Marshall mix designs placed on the MN/Road mainline test section had slightly less VMA at 15.3 % (9). Some wearing course mixtures are now being placed with VMA values in the 14 to 15 % range, as seen in the recent Florida permeability study and in the mixtures placed at the NCAT Test Track (10, 12). As a comparison, SMA mixtures, which in general have been exhibiting significantly better durability than coarse-graded mixtures designed according to the Superpave system, have design VMA values in the 17 to 18 % range (15). Clearly a significant portion of the recent decrease in mixture durability can be attributed to the reduction in VMA and binder content over the past several decades. Given the success of SMA and its widespread adoption, there probably is not a need to increase VMA values for dense-graded HMAC mixtures to the 17 to 18 % range, but there is a need for modest increases 9 in design VMA and binder contents. Based largely upon the historical changes in VMA and binder contents, the superior durability of SMA mixtures, and the relationship between fatigue resistance and effective binder content, it is recommended that minimum VMA values (and implied effective binder contents) be increased 1 % for mixtures designed according to the Superpave system. These recommendations are summarized in Table 2. Although the design air void content in the Superpave system is currently set at 4 %, some agencies allow design air voids to vary up to 1 % from this value. In such cases, VMA and binder content are no longer directly related, and it becomes important to check minimum effective binder contents by volume, in addition to evaluating VMA. For this reason, recommended minimum VBE values are included in Table 2. Because wearing course mixtures are subject directly to tire loading, and also to air, water and sunlight, it is important that they be extremely durable. For this reason, Superpave wearing course mixtures should have a minimum VMA of 16.0 % and a minimum VBE of 12 %, regardless of nominal maximum aggregate size. Rich base course mixtures should meet similar requirements. Some engineers may be concerned that returning to richer wearing course mixtures might result in an increase in rutting. However, this will not be a problem if the changes are done carefully. As discussed at the beginning of this article maintaining proper levels of aggregate fineness while increasing VMA will help maintain rut resistance. It must also be recognized that the general quality of the materials used in producing asphalt concrete pavements has improved significantly with the implementation of the Superpave system. Many of the rut-prone Marshall mixtures placed in the 1970’s and 1980’s were designed at low levels of compaction and included substantial amounts of natural sand. Some also contained poorly crushed gravel as coarse aggregate. The relationships among binder flow properties, rut resistance, climate and 10 traffic were not well understood, and as a result binders used in heavily trafficked wearing courses sometimes had insufficient stiffness for such severe applications. With the strict requirements for aggregate quality, compaction and binder grading now in use within the Superpave system, VMA and binder contents can and should be increased without fear of excessive rutting, as long as aggregate fineness is also increased slightly and care is taken in selecting the appropriate binder PG grade. VMA AND THE MIX DESIGN PROCESS Durable, rut resistant HMAC must be designed with a high level of compaction, reasonably high VMA and binder contents, and adequate fines. Although achieving proper VMA is in general an inexact, trial-and-error process, a few basic rules can help laboratory engineers and technicians increase VMA in overly lean mixtures. In general, VMA will increase as the aggregate blend moves further away from a maximum density gradation. This means that for a coarse blend—one that falls below the maximum density gradation—VMA can be increased by adding coarse aggregate. Similarly the VMA of fine blends can be increased by adding fine aggregate. Adding mineral filler to a mixture will sometimes increase VMA, and sometimes decrease VMA. Usually, mixing an additional aggregate into an existing blend will decrease VMA rather than increase it, because the new aggregate will tend to fill voids among the existing particles. Therefore, the number of aggregate sources in a blend should be decreased and not increased when higher VMA in a mixture is needed. Weak, friable aggregates with poor abrasion resistance will break down under moderate to high compaction, making it difficult to achieve high VMA levels. If this is the case, alternate aggregate sources should be evaluated. Sometimes aggregate particle shape or angularity can affect VMA, either increasing it or decreasing it compared to other aggregates of similar gradation. Highway agencies should consider 11 broadening aggregate gradation requirements when possible, to make achieving proper levels of VMA easier, since beyond selection of the proper aggregate size and fineness, there is little research to suggest that specific characteristics of aggregate gradation affect pavement performance in any significant way, while VMA, air voids and binder content are critical to proper durability and rut resistance. CONTROL OF VMA DURING FIELD PRODUCTION Many states allow VMA of asphalt concrete mixtures to fall below the minimum design value during field production. This is believed necessary because of the variability inherent in the production of asphalt concrete, and also because of the relatively poor precision of the specific gravity measurements underlying calculation of air voids, VMA and related parameters. These are certainly valid concerns, but engineers should understand the strong link between proper levels of VMA and aggregate fines, mixture durability and rut resistance. Although occasionally allowing VMA to drop up to 1 % below suggested minimum values during production is probably not a major problem, plant personnel should strive to achieve an overall average VMA close to the design value. Highway agencies should consider paying for asphalt binder as a separate item to remove what can otherwise be a strong incentive to produce mixtures with the lowest VMA and minimum binder content possible while still earning full contract price. SUMMARY Obtaining proper VMA is essential to producing durable and rut resistant asphalt concrete mixtures. To ensure adequate fatigue resistance, the minimum VMA values for mixtures designed according to the Superpave system should be increased by 1 % over current recommended minimums. In cases where design air void content is allowed to vary from 4 %, 12 the implied minimum effective binder content should also be increased 1 % and enforced in addition to minimum VMA values. Because of severe exposure to the effects of traffic loading and the elements, wearing course mixtures and rich base course mixtures should have a minimum VMA of 16 % and a minimum VBE of 12 %, regardless of nominal maximum aggregate size. Adequate rut resistance can be achieved regardless of VMA by making certain that the proper binder grade is selected for a given application and that the aggregate blend contains sufficient fines relative to the design VMA. Adequate fineness will also help keep the permeability of in-place mixtures low. Although some variability in VMA is inevitable during field production, plant personnel should strive to keep overall VMA levels close to design values. 13 REFERENCES 1. D. W. Christensen and R. F. Bonaquist, NCHRP Projects 9-25 and 9-31 Draft Final Report, Sterling, VA: Advanced Asphalt Technologies, LLC, November 2004. 2. N. W. McLeod, “Relationships Between Density, Bitumen Content, and Voids Properties of Compacted Bituminous Paving Mixtures,” Proceedings, Highway Research Board, Volume 35, 1956. 3. N. W. McLeod, “Voids Requirements for Dense-Graded Bituminous Paving Mixtures,” in ASTM STP-252, American Society for Testing and Materials, 1959. 4. W. H. Campen, J. R. Smith, L. G. Erickson and L. R. Mertz, “The Control of Voids in Aggregates for Bituminous Paving Mixtures,” Proceedings of the Association of Asphalt Paving Technologists, Vol. 26, 1957. 5. W. H. Campen, J. R. Smith, L. G. Erickson and L. R. Mertz, “The Relationship Between Voids, Surface Area, Film Thickness, and Stability in Bituminous Paving Mixtures,” Proceedings of the Association of Asphalt Paving Technologists, Vol. 28, 1959. 6. J. F. Goode and L. A. Lufsey, “Voids, Permeability, Film Thickness Vs. Asphalt Hardening,” Proceedings of the Association of Asphalt Paving Technologists, Vol. 34, 1965. 7. E. R. Brown, L. Michael, E. Dukatz, J. Scherocman, G. Huber, R. Sines, J. D’Angelo and C. Williams, Performance of Coarse-Graded Mixes at WesTrack—Premature Rutting, Final Report, FHWA-RD-99-134, June 1998, 20 pp. 8. D. W. Christensen and R. F. Bonaquist, “Rut Resistance and Volumetric Composition of Asphalt Concrete Mixtures,” Preprints, Journal of the Association of Asphalt Paving Technologists, 2005. 9. R. Mulvaney and B. Worel, MnROAD Mainline Rutting Forensic Investigation, St. Paul, MN: Minnesota Department of Transportation, October 2002, 91 pp. 10. E. R. Brown, B. Prowell, A. Cooley, J. Zhang and R. B. Powell, “Evaluation of Rutting Performance on the 2000 NCAT Test Track,” Preprints, Journal of the Association of Asphalt Paving Technologists, March 2004. 11. M. W. Mirza and M. W. Witczak, “Development of a Global Aging System for Short and Long Term Aging of Asphalt Cements,” Journal of the Association of Asphalt Paving Technologists, Vol. 64, 1995, pp. 393-424. 12. B. Choubane, G. Page and J. Musselman, “Investigation of Water Permeability of Coarse Graded Superpave Pavements,” Journal of the Association of Asphalt Paving Technologists, Vol. 67, 1998, p. 254. 13. D. W. Christensen and R. F. Bonaquist, “Practical Application of Continuum Damage Theory to Fatigue Phenomena in Asphalt Concrete Mixtures,” Preprints, Journal of the Association of Asphalt Paving Technologists, 2005. 14. E. R. Brown and S. A. Cross, “A National Study of Rutting in Hot Mix Asphalt (HMA) Pavements, Journal of the Association of Asphalt Paving Technologists, Vol. 61, 1992, pp. 535-573. 14 15. E. R. Brown, J. E. Haddock, C. Crawford, C. S. Hughes and T. A. Lynn, Designing Stone Matrix Asphalt Mixtures, Volume II (a)⎯Research Results for Part 1 of Phase I, NCHRP 9-8 Final Report, Prepared for the National Cooperative Highway Research Program, Auburn University, July 1998. 15 TABLES Table 1. Minimum Aggregate Fineness Requirements for Rut Resistance and Permeability. VMA Vol. % 20.0 19.0 18.0 17.0 16.0 15.0 14.0 13.0 12.0 Minimum Sa m2/kg 7.4 6.8 6.4 5.8 5.4 4.8 4.3 3.9 3.4 Minimum FM300 Maximum FM300 38 35 32 30 27 25 22 20 18 64 59 55 50 46 42 38 34 30 Table 2. Suggested Minimum Values for VMA and VBE for Dense Graded Asphalt Concrete Mixtures. Aggregate NMAS mm 9.5 12.5 19.0 25.0 37.5 Minimum VMA Vol. % 16 15 14 13 12 16 Minimum VBE Vol. % 12 11 10 9 8 FIGURES Observed Rut Rate 1.E+01 1.E+00 NCAT MN/Road WesTrack Equality 90 % PI 1.E-01 1.E-02 1.E-02 1.E-01 1.E+00 1.E+01 1.E+02 Predicted Rut Rate Figure 1. Observed Rut Rates for Three Test Roads and Values Predicted Using Equation 1 (includes line of equality and 90 % prediction interval for new observations). VMA, Volume % 0 5 10 15 20 25 Marshall Designs 1980's (1) Marshall Designs ca. 1992 (2) Superpave Designs 1996 (3) Binder Air Voids Superpave Designs 2000 (4) SMA Mixtures (5) Figure 2. VMA, Effective Binder and Air Void Contents (by volume) for Different Mix Types over the Past 20 to 30 years (1, 2, 3, 4, 5). 17