A STUDY OF IN-PLACE RUTTING OF ASPHALT PAVEMENTS

E. R. BROWN 1 and STEPHEN A. CROSS2

I. INTRODUCTION In recent years, the amount and severity of rutting in asphalt pavements appears to have increased according to reports of engineers in many State Highway Departments. This apparent increase in rutting has been due to some extent to the increase in truck tire pressure, axle loads, and volume of traffic. Some studies have shown typical truck tire pressures to be approximately 120 psi (1). As a result of these higher loads and tire pressure, more attention must be given to selecting high quality materials, in designing the asphalt mixtures, and in quality control during construction. Concern for rutting and high truck tire pressure led to a National Symposium on the subject in March 1987 (2). The general feeling at this symposium was that the higher truck tire pressure and increased truck weights definitely have led to increased rutting but the feeling was that more attention to selection of materials and construction could minimize the rutting problem as well as other problems that might affect the performance of asphalt pavements. The objective of this study was to evaluate in-place pavements experiencing rutting and pavements experiencing no rutting to begin to classify asphalt mixtures that should perform satisfactorily and those that would likely rut under traffic. The information reported herein is part of a larger study to evaluate rutting in the field and in the laboratory and to develop information that would insure improved performance. This report summarizes the work accomplished during the first 1% years of the study. The entire study is projected to continue for a total of five years. Five pavements were selected for analysis and the results are reported in this portion of the study; four of the pavements had been identified as experiencing premature rutting and one pavement had been identified as having no rutting after more than 10 years of service. Rutting measurements were taken across the outside traffic lane and a trench was cut across the lane for each of the highways experiencing rutting. The trench was closely investigated to determine the extent of rutting in each layer of asphalt mixture. Cores were taken at approximately one foot intervals across the pavement lane, transported back to the laboratory, and analyzed to determine material and mixture properties. Laboratory tests included asphalt content, aggregate gradation, and analysis of in-place and recompacted mixture properties.

1 2

Assistant Director for Research, and

Senior Research Associate, National Center for Asphalt Technology. Auburn University. Alabama

The oral presentation was made by Dr. Brown

BROWN AND CROSS

SITE #1
Ion 99 96 91 96
Binder Course Pavement Surface

95 94 93 92 91 90 69 a8 07 86 85 0 2 4 6 8
Sand Asphalt

10

MSTANCE FROU CEJilERUNE (Flj

Figure 6: Relative layer elevation vs. distance from centerline for Site #l

IN-PLACE RUTTING

gregate gradation, Rice Specific gravity (ASTM D2041), unit weight, resilient modulus (ASTM D4123), Indirect tensile strength (ASTM D4123), Marshall stability and flow (ASTM D1559). Some of the mix was tested as received while some mix was reheated, broken-up, and recompacted to evaluate the mixture using standard compactive effort. Three compactive efforts were used to recompact most mixtures: 75-blow manual hammer; Gyratory Testing Machine (GTM) set at 120 psi, 30 revolutions, and 1 degree angle; and GTM set at 120 psi, 300 revolutions, and 1 degree angle. The height of the samples compacted with 300 revolutions with the GTM was measured over a range of revolutions to help evaluate voids as a function of revolutions. The recompacted samples were tested for unit weight, stability and flow. The voids of the recompacted samples were compared to the in-place voids and to the desired voids to evaluate original mix design. The Gyratory Shear Index (GSI) was determined for all samples compacted in the GTM. A GSI of 1.0 is normal for a mixture that is stable during compaction. A higher GSI has been shown to indicate more unstable mixtures. A plot demonstrating determination of GSI is shown in Figure 5. Compaction of asphalt in the GTM simulates densification and eventual plastic flow that is observed in the field. An asphalt mix is stable until voids are closed during com-

Figure 7: Typical Rutting at Site #l

IN-PLACE RUTTING

paction to the point that plastic flow begins to occur. An increase in GSI above 1 measured during compaction in the GTM simulates plastic flow in the field. III. FIELD MEASUREMENTS The relative surface elevation was measured across the traffic lane at one foot intervals using an elevated straight edge that had been leveled. The results of measurements to determine thickness of each layer in each core were combined with the surface measurements to plot the relative elevation of the top of each pavement layer. Measurements were also made at each trench to locate the limits of rutting (Figure 4). All significant rutting in the pavements investigated had occurred within 3-4 in. of the pavement surface. It is clear from Figures 5, 6, and 7 that a significant amount of rutting had occurred in the surface of the pavement at site 1. This particular pavement had experienced rutting in the past and had been milled, patched, and overlaid at various times to alleviate the rutting problem. The large amount of maintenance work explains why the various layers vary in thickness and grade. After observing the trench for site 1 and closely reviewing Figure 6 it is apparent

Figure 9: Typical Rutting at Site #2

IN-PLACE RUTTING that most of the rutting had occurred in the surface layers (the surface layers consisted of three thin layers of asphalt mix). Figures 8 and 9 show that a small rut had occurred adjacent to the shoulder for the pavement at site 2. Very little rutting was observed adjacent to the centerline. The shoulder paint stripe had moved in some locations which is often a result of stripping. The rutting at this site appeared to be partially related to stripping which explains why the rut is adjacent to the shoulder (source of water). Observations while cutting the trench and during the laboratory testing operation showed a significant amount of uncoated aggregate which was concluded to be stripping. Measurements taken at site 3 showed very little rutting (Figures 10 and 11) however there were locations where the material had shoved outward adjacent to the shoulder. These were localized areas and did not result in significant rutting. Site 3 consisted of an old PCC pavement which had been overlaid with asphalt concrete. The asphalt mix contained primarily uncrushed aggregate and had clearly stripped. During the trenching operation many aggregate particles were observed that had been completely stripped of asphalt. It is anticipated that the high amount of stripping and the localized shoving that has begun will very rapidly lead to a significant rutting problem.

Figure 11: Typical Rutting at Site #3

12

BROWN AND CROSS

Site 4 was selected as a pavement that had performed over ten years with no major performance problems (Figure 12). The plot in Figure 13 shows that this asphalt mix has a small rut but this did not appear to be a typical rut. The surface was perfectly straight except for a slight dip at 8 feet from the centerline. This depression was well within the allowable tolerance of construction variation and obviously had no effect on the traffic using the roadway. Site 5 had experienced some rutting primarily adjacent to the centerline. Figures 14 and 15 show that most of the rutting here had occurred in the friction and surface course. Cores were obtained during rain and it was clear that the rut adjacent to the centerline held water. IV. RESULTS OF THE LABORATORY TESTS Some of the asphalt concrete cores were tested in the laboratory to determine the asphalt content and aggregate gradation. The asphalt content and aggregate gradation results are shown in Table 2. The asphalt content was measured without correcting for ash and therefore the numbers reported are somewhat high. They are reported here for information but the measured asphalt contents were not used for calculating other mixture properties. The three

Figure 12: Asphalt Mix with Excellent Performance at Site #4

IN-PLACE RUTTING

15

top courses for site 1 were combined because of their varying thin thicknesses. The combined asphalt content for these three layers appears to be high which may be the result of high asphalt content in one or more of the three mixes or it may be the result of excessive tack coats between the layers. The aggregate gradations used for the 5 sites were approximately equal for a particular mix type. Site 3 was an exception. The mixes used for site 3 did not contain crushed aggregate and as a result had very little material passing Nos. 50, 100, and 200 sieves. Tests were conducted on cores as received to determine in-place properties. Tests included Rice Specific Gravity, Stability, Flow, Resilient Modulus, and Indirect Tensile Strength. The results of these tests are shown in Table 3 and Table 4. The amount of voids in total mix (VTM) is likely the most important physical property of asphalt mixtures that relates to rutting. The VTM varies at different points across the traffic lane. The VTM should generally be lower underneath the wheels but this is not always the case. Once rutting starts the VTM may actually increase with additional traffic. The various layers in the pavement will also have variations in the amount of VTM. Low VTM near the surface of the pavement can result in serious rutting problems. According to many engineers plastic flow of the asphalt mixture is likely to begin once the

Figure 15: Typical Rutting at Site #5

30

BROWN AND CROSS

required for satisfactory mix design and quality control. The 75-blow manual Marshall hammer used for recompaction is equal to or slightly higher than the in-place density. Another cause of low voids is lack of control of the asphalt mixture during construction. Many states arbitrarily increase the asphalt content to meet specification requirements for in-place voids. This adjustment in asphalt content will result in low voids under traffic. This adjustment of asphalt content is often made when paving in cold weather. When satisfactory density is not being obtained during construction, additional compactive effort should be provided instead of increasing the asphalt content. The asphalt content is very critical to satisfactory performance of a mixture and hence should only be modified by those familiar with the mix design process. Some mixtures are simply designed to have low voids to insure minimum cracking during cold weather and to insure other desirable properties. The data from the pavements investigated show that rutting occurs below approximately 3 percent voids, therefore, the mix design should be selected so that the void content in-place never decreases to 3 percent. The void content should be designed between 4 and 5 percent (using the proper laboratory compactive effort) for very high traffic volume roads. Low temperature cracking can be. minimized by compacting this mixture to approximately 6-7 percent air voids during construction. The Marshall flow in the recompacted samples appears to be an indicator of rutting potential with an R2=0.25 (Figure 21). A flow above 16 for the pavements tested resulted in rutting equal to approximately 10-40 percent of layer thickness. A flow below 16 resulted in only one of the mixes having more rutting than 10 percent of layer thickness. The rutting at site 3 was likely the result of stripping of the asphalt from the aggregate. The amount of rutting was small but the roughness caused by the rutting had resulted in loss of ride quality. This stripping and rutting would have been prevented or minimized if a high quality crushed aggregate had been use (5). CONCLUSIONS The results of this study show that mixes can be produced to support todays traffic. The pavements evaluated which had rutted under traffic in most cases appeared to have rutted due to low air voids (in recompacted samples and/or in the field). Only two of the pavements investigated had rutting sufficiently high to require rehabilitation. One of the best indicators of rutting is low air voids in the laboratory compacted asphalt mixture. Satisfactory laboratory compaction effort (providing density approximately equal to that under traffic) must be utilized when compacting these samples. The GSI determined during compaction with the Gyratory Testing Machine was shown to be a good indicator of mixes that had rutted under traffic. Based on the results of this study, a maximum GSI value of 1.1 is recommended when compacting samples with 1 degree angle, 120 psi, and 300 revolutions. The Marshall flow appears to be a good indicator of rutting potential. A

IN-PLACE RUTTING

31

maximum flow of 16 is often specified for mix design and construction control and that appears to be a reasonable number from the data presented in this study. Mixes having flow values above 10 tended to have higher amounts of rutting. Based on the test results obtained in this study, it appears that the Resilient Modulus and Indirect Tensile Strength values are not significantly related to rutting. Stripping of the asphalt mixture had caused rutting to some extent at two of the sites. The amount of rutting at these two sites was small and at the time of sampling these pavements were performing satisfactorily. Most of the rutting observed in this study had occurred in the top layers of asphalt concrete. These layers often contained fine aggregate gradations and high asphalt contents. Asphalt mixes can be designed and constructed to carry todays traffic as shown by these mixes at site 4. Steps must be taken during mix design to ensure that the asphalt content is correctly selected for the mix being produced and that sufficient quality control tests be conducted to verify mix design and to provide data to make adjustments in mix proportions if needed. RECOMMENDATIONS One of the biggest causes of rutting is excessive asphalt content in asphalt mixtures. Steps should be taken to insure proper asphalt content is selected and provided during mix production. Compactive effort should be selected to provide a density equal to that which will be obtained under traffic (75 blow with manual hammer or Gyratory Testing Machine have been shown to be sufficient). The asphalt content should be selected to provide a void content of 4-5 percent in laboratory compacted mixtures for high traffic volume roads. Asphalt content should not arbitrarily be increased to facilitate compaction, to minimize segregation, or for any other reason except to provide satisfactory voids in the laboratory compacted asphalt mixture. The maximum Marshall flow should be specified to be 16. If a Gyratory Testing Machine is used the GSI should not exceed 1.1. REFERENCES
1. Stuart W. Hudson, and Stephen B. Seeds. Evaluation of Increased Pavement Loading and Tire Pressure, Paper presented at 67th Annual Transportation Research Board Meeting, January 1988. Federal Highway Administration, Proceedings of a Symposium/Workshop on High Pressure Truck Tires, Austin, Texas, 1987 G. A. Huber, and G. H. Heiman. Effect of Asphalt Concrete Parameter on Rutting Performance: A Field Investigation, Proceedings, Association of Asphalt Paving Technologists, Volume 56, 1987, pp. 33-61. Miller C. Ford, Pavement Densification Related to Asphalt Mix Characteristics, Paper presented at 67th Annual Transportation Research Board Meeting. January 1988. E. R. Brown, J. L. McRae and A. Crawley. Effect of Aggregate on Performance of Bituminous Concrete, ASTM STP 1016, 1987.

2. 3.

4. 5.

BROWN AND CROSS Discussion MR. WILLIAM A. PHANG: Ray in your correlation work, just for clarification, did you only use the mixes that were in the upper layers or did you include all of the different layers in the correlation? PROF. E.R. BROWN: In the correlation we included the new layers that had been placed if it was an overlay. If it was pavement that had not been overlaid, we used the top two to three layers. Most of the projects that we evaluated were overlays. PROF. THOMAS W. KENNEDY: At least youve clarified now when you say the misuse of the Rice Specific Gravity. I dont want anyone leaving here thinking that we should go back to some other type of maximum theoretical density or laboratory density because its not the Rice that is causing the problem, its somebody abusing the mix by putting in asphalt to get density. I would not want to leave that unsaid at this point in time. PROF. BROWN: Thank you, Tom. You are exactly right. The Rice Specific Gravity Test is an acceptable way to measure the theoretical maximum density, but we have to be careful to insure that we dont play games with the mix to fill voids and make it appear that compaction requirements are being met. MR. JAMES A. SCHEROCMAN: I completely agree with the comments that Tom Kennedy just made. I was very pleased to hear you say that the problem with using theoretical maximum specific gravity to determine the degree of compaction is not that value itself but the misuse of the Rice test method that is used to determine the theoretical maximum specific gravity (density) value. Thus I would echo exactly what Tom said. The Rice test method, properly applied, provides the air void content of the asphalt concrete mix directly, which is an extremely important variable in terms of pavement performance. We have talked previously about the problems of using density values which are based on a percent of lab density because we dont know what actual air void content is that is associated with the percent lab density. Further, the lab density determined at the time of the initial mix design is different than the percentage of lab density determined on plant produced asphalt concrete mix. Thus the target value that a contractor is shooting at can change. For this reason, I was very pleased to hear your comments here in regard to the Rice test method and the misuse of that procedure. The Rice method, used properly, is as valid in predicting the amount of rutting that will occur in the mix as is the use of the percent of laboratory density procedure. Use of the percent lab density requires that the target density value be calculated using asphalt concrete mix manufactured in a batch or drum mix plant. PROF. BROWN: You are absolutely correct. It is the misuse of the Rice. What we need to realize about the Rice, is that it is a way to determine the initial in-place air voids. It has absolutely nothing to do with the final in-place air voids. To insure good performance we have to be concerned about both. My

IN-PLACE RUTTING

33

point is that too many people are looking at initial in-place air voids and forgetting about the final in-place air voids. the method used to predict the final in-place air voids is to compact samples during production using the required number of blows with the Marshall apparatus and to measure the voids of this compacted mix. Too many asphalt projects are being constructed where samples are not being compacted in the laboratory during production resulting in very little control of the final in-place air voids. This approach often results in mix changes to lower initial in-place air voids. MR. SCHEROCMAN: I have no argument with you at all. I think that the work that Harold Von Quintus showed earlier, regarding what we have found in the AAMAS project concerning the ultimate level of compaction, will be extremely important in that regard. MR. RICHARD DAVIS: I became worried about the rutting problem about 10 years ago. I was asked by the highway officials of one of our states to come and look at their pavements. They said that their greatest problem was with stripping. We looked at a lot of pavements and, of course, there was rutting, but they felt the cause was stripping. As I looked at the pavements they mentioned that they thought that someone had taken the goodies out of the asphalt, and they also thought that several other things might have changed. I couldnt see the changes that bothered them. I looked for other causes. I became aware of the fact that truck tire pressures had increased. I came to the conclusion, after some study, that their stripping was due to the pavements not being able to resist the stresses in shear. That they were shearing. In this shearing action the asphalt was scrubbed off the aggregate. This increased the air voids in the pavement, because after the shearing failure begins, the air voids are increased. I think, in rutting pavements, we should be careful about characterizing the problem as stripping. Im not saying that there are no stripping problems, but if the major causes of the problem is failure in shear, you are wasting your time looking for a way of increasing the adherence of the asphalt to the aggregate. I think that when you look at problems of this type, particularly when there is rutting, you ought to give some thought to whether the actual failure is primarily due to the inability of the pavement to resist the stresses in shear. PROF. BROWN: I dont really have anything to add. MR. ROBERT DUNNING: I have two comments. On the graph that you had correlation coefficient with the gyratory of 0.65. It looked to me like you have two separate lines there. A straight line up to about 1.3 that is horizontal, and then another separate line that shows the problem. I wondered if you made a break there, and had more data, you might find a much better correlation coefficient. PROF. BROWN: I think you are correct. What we are trying to do at this point is to simply show trends. That is the reason we used the straight line in the regression analysis. We have a lot more data that we are collecting and that will be included at a later date. The analysis will be more sophisticated when sufficient data is available.

34

BROWN AND CROSS

MR. DUNNING: The second comment I wanted to mention was with respect to the Rice specific gravity. We have found, especially on airport jobs, a bias between the laboratory compaction and field compaction. Assume that you do a mix design with 4 percent void in the laboratory. Perhaps a drummixer will be used and you end up with a mix that has moisture in it in the field. The field laboratory control samples then maybe compact to say 1.5 percent void. The contractor is required to make 95 percent of that 1.5 percent void sample. The reason we have the change is that the effective specific gravity as calculated from the Rice Specific Gravity decreases because of decreased absorption of the asphalt into the aggregate as a result of moisture in the aggregate. I have run across this several times. You can then end up getting a bleeding mix and it is being required because of a problem with field testing not a problem with the mix itself. When using the Rice we need to recognize that it can change as a result of moisture content. PROF. BROWN: You are right. Moisture will affect the measured Rice Specific Gravity and thus affect the calculated voids. We have to insure that the moisture contact is within an acceptable range, otherwise poor performance will result. MR. JOHN D. ANGELO: We have a mobile laboratory that does mix verification in the field and we have tested five sites so far. We have found that typically samples taken from the plant and tested with a Marshall compaction hammer consistently came out much denser than the original mix design. It goes back to your statement about design compactive effort not giving the same density as achieved on the roadway. I was wondering if the production process is the cause of this difference not the compaction effort? PROF. BROWN: Yes I think it is a complicated problem. We have seen that mixes compacted in a field laboratory during production produce higher densities than that measured during mix design. There are several reasons for that, I suspect. The biggest reason, I would imagine, is the increase in dust caused by abrasion of the aggregate when the aggregates are fed through the asphalt plant. The point is, if you do not modify the mix to account for that additional dust, it is going to lead to premature problems such as rutting. The mix needs to be modified based on actual production at the plant. We have become accustomed to doing mix designs and telling the contractor to meet the design without consideration to inherent changes in plant produced materials. We should modify the design during construction as needed to meet actual conditions. Only qualified individuals should have the authority to change the mix during construction. Another item that goes along with what you were talking about is the use of different types of hammers which result in different densities. The density that we are getting during mix design is lower than the density we are getting in place. We need to correct that by increasing the density obtained during mix design. MR. JOHN MCRAE: I would like to congratulate Ray and Steven on this fine practical work. Tying the field in with the lab is so important to us. I would like to mention something that I have picked up on. I havent shared this

IN-PLACE RUTTING with very many people yet. I am finding that more and more if I look at the gyratory trace on a stripping mix, it pulls in and gives you less than unity. In other words the ratio goes down. I think that this may be a helpful way of looking for stripping mixes when you run those in the laboratory.

35

PROF. BROWN: The one pavement that we tested that was badly stripped, showed exactly what you said. The GSI was lower than 1. MR. MCRAE: We didnt see this until recent years, although we have been using this a long time. I think the fact that were sampling in the field and finding these conditions is significant. MR. GAIL KING: Ray I enjoyed your presentation very much. I cant resist pointing out that our French compatriots here claim to have essentially solved their rutting problems with two laboratory tests. First, they have implemented exactly what you recommended here. Using gyratory compaction, they monitor changes in density with compactive effort, and hence reject numerous mixes that are susceptible to densification under traffic. As you pointed out, these decisions are being based upon data with correlation coefficients of about 0.6. Perhaps thats not bad for laboratory data, but consider the statistics. How many miles of pavement might fail after passing laboratory designs valid to this level of significance? The Europeans have gone one step further by also subjecting mixes to rutting simulator tests. These studies also eliminate many potential problems. It bothers me that we, in the U.S., seem to have rejected wheel tracking tests in favor of static creep tests as our predictors of rutting performance, especially since the Europeans have seemingly had such good success with their various traffic simulators. Most people here are familiar with the fine work being done at the University of Nottingham, but their wheel track test is admittedly a bit large for everyday mix design use. The laboratory scale wheel track tester developed by LCPC in France is expensive, but ideal for simulating various field conditions. It is well suited for evaluating mix designs on a daily basis. The German device is much less expensive and all tests are run with the specimen submerged under water. This unit has the ability to simulate rutting as caused by stripping as well as load induced plastic movement. I wish the U.S. research community would take a closer look at some of the commercially available devices. PROF. BROWN: That may be something that we look at further down the road, but, again, the first part of our study was to determine what is wrong with the pavement. How bad is the rutting. That is basically what we are working on now. The second part is how do we solve that. We are working on that as we go obviously. But I still believe that we could solve most of our rutting problems just using what we know now. That is to do more and better quality control and use better mix design techniques. When I say better, just use the procedure we now have and follow them more closely. MR. B.A. VALLERGA: Ray, you did a pretty good job. I wonder why you didnt extend it a little further by using other tests like, for example, the Marshall Test device and the Hveem Stabilometer, the two instruments that 90 per-

BROWN AND CROSS cent of those in the asphalt field believe are indicative of a stability measurement. Could you answer that question first, and then I have another? PROF. BROWN: Yes, we are extending it. We did not have enough data in time for this report. We are just starting to work on some of the mechanical tests. We will conduct creep tests, for one thing. We have performed Marshall tests, but I did not report it. With the limited number of pavements that we have tested to date, the Marshall stability doesnt show any relationship to rutting. But we do have that information. We have performed flow tests along with the stability. We have divided stability by flow to see if it shows anything. It is just too early at this point to try to correlate these tests to rutting. MR. VALLERGA: I agree with you that resilient modulus has no relation to rutting. Tensile strength also has little, if any relation to rutting. Volume of the air voids may have some relation to rutting, but not necessarily indicative of rutting. Flow, possibly. GSI number, potentially. I think that those are all in the right direction, but I dont think you have gone far enough. My last question is related to the matter you have brought forth of stripping contributing to rutting. Can you briefly describe what the mechanism is? What in the world is happening to cause this rutting as a result of stripping of the asphalt from the aggregate surface? PROF. BROWN: My feeling is that generally if we have stripping and if it is confined to 3 inches or more beneath the surface, there is a good chance that the stripping wont cause any problems on the surface. If we have stripping near the surface, then what happens is we lose cohesion of the mix and it can actually begin to shove outward. We have seen rutting in this case next to the shoulder. The cause of the rut is the outward movement of the material. I might also add that in the one case where we observed this problem, the aggregate was rounded which certainly didnt help. PROF. R.A. JIMENEZ: Ray, no question with what you have said, except for the presentation, where you express the rut depth as a percentage of layer thickness. That bothers me a little bit because I will accept a l/4 in. or even a 3/8 in. rut. if you will, in a 2-in. layer, which is about 18 percent. But I cannot accept an 18 percent rut on a 4-in. layer. You are expressing the rut as a percent of the layer thickness. PROF. BROWN: You disagree with expressing it as a percent of the layer thickness? PROF. JIMENEZ: Right. Because then 18 percent doesnt tell me how deep the rut is. If the 18 percent is about a l-in. layer, I can accept that rut. If you are talking about a 5-in. layer, then I cannot accept that rut. PROF. BROWN: That is correct. It doesnt tell you anything as a user or doesnt tell you anything about the rut depth. But to evaluate the problem for the various mixes, we have to look at each mix individually. We have to know how much of the rutting is occurring in each mix in order to evaluate the adequacy of that particular mixture. I believe the best way to quantify the rutting in each mixture is as a percent of layer thickness.

IN-PLACE RUTTING PROF. JIMENEZ: Well, you usually dont know how thick the layer is.

37

PROF. BROWN: For this study we cut cores and trenches and we measured the layer thicknesses. If we just look at rut depth, we have no idea which layers are causing the problem. But when we use percent of layer thickness, that allows us to compare one asphalt mix to another asphalt mix in terms of rutting performance. We can then develop criteria for the asphalt mixes to minimize rutting. MR. VAUGHN MARKER: Somewhere back in my misspent youth I heard that every force has to have an equal and opposite force resisting it or else something would be moving. I havent heard anybody say anything about the internal resistance of this mix as a function of aggregate surface texture. This, in my opinion, may be the most important consideration we need to take into account where rutting is concerned. I dont think there is any question that if you get excess asphalt forcing the particles apart from each other so they cant develop inter-particle friction you are going to get movement because the asphalt becomes a lubricant. I think that one of the things we ought to think about, in conjunction with void volume and what not, is what kind of aggregates we are putting into the mix in the first place. I had recent experience with specifications which permit up to 30-35 percent of naturally rounded sand. With that kind of a mix, I dont care what kind of voids you have, under some of the truck traffic we are getting today youre going to get a rut. And so I dont think that we can say that everything is a function of the voids alone. We must pay attention to the voids, but we must be conscious of aggregate surface texture as well in order to be treating the cause instead of trying to treat the effect. Are you going to be looking at this type of thing? PROF. BROWN: Aggregate properties are involved and I agree 100 percent with what you said. The aggregate has to carry the load. If the aggregate doesnt carry the load, rutting will occur. But what we are finding, in the few pavements that we have looked at so far is that we are putting so much asphalt in the asphalt mixes that it doesnt matter what the aggregate quality is, it is not going to carry the load. The point is if a mix is over asphalted, the properties of the aggregate then become unimportant. But in a properly designed mix, the aggregate is extremely important and it has to carry the load. PROF. B.M. GALLAWAY: Barney raised a question about the mechanism of stripping within the body of a pavement structure. Often times this occurs as an emulsification process. Most of the time when you have stripping within the mat, you have all the elements of emulsion manufacture. Jack Dybalski might want to comment on this, but you have thin films. You have water. You have a compatible chemical system, that is the chemistry is right, usually very often a high ph at least about 8. And you have work or shear energy? The work being furnished by the traffic. When you put all that together, you can create an emulsion from the thin film of asphalt that coats the aggregate particles. This is possible only if there is a low affinity of asphalt film for the aggregate. But we build a lot of mixes with a lot of sands, siliceous and rounded sands in them that are of that type. In other words they love water better than they do as-

BROWN AND CROSS phalt. When that film strips, it will not be pure asphalt it will be contaminated with filler. These films with a little filler in them, too much some times in some of them, but almost always we have some filler in the asphalt film. The asphalt film and the filler conglomerate and this agglomeration is in the liquid form, in the emulsion form, under the stress of traffic that liquid form of asphalt water and emulsifier from the aggregate system will migrate to the surface. I have measured a number of mixes where the asphalt content down in the mat would be 2 percentage points lower than what went into the original mix. The asphalt content in the top half l/4 in. of that pavement would be 3 or 4 percentage points above design. One mix, for example, I examined on Highway 21 in Bryan went in at 5 percent and the top l/4 inch had 11 percent asphalt in it, all of which migrated from within the system up to the top. Barney, that may be a little bit, at least from my vantage point, of what takes place in the stripping operation. Once that material is stripped, the asphalt film leaves it, then you have an unstable system with a leathery surface on it and the material moves laterally and then you have a rut with a dark color to it. In other words it looks fat and it ruts. If you look at the composition vertically, you will find there is a lot more asphalt in the surface then there should be and where does it come from. Not like Manna from Heaven, but it comes up from the devil, the strip susceptible aggregate below. MR. MCRAE: I would like to say just a further word on the stripping. I think everything that has been said I certainly agree with and I think Bob gave Barney a good answer. Another way to explain this is to say you have lost your cohesion. Cohesion is very important in the surface layer. It serves a very important function there. In this process you have kind of got an unbound aggregate. In regard to Vaughn Markers statement we could measure the shear in the gyratory and this needs to be related. Youll find the shear going down (reduction in roller pressure) when the GSI increases. MR. V.P. PUZINAUSKAS: I wanted to share our experience in testing, New Jersey turnpike which rutted some in the truck traffic lanes. The New Jersey turnpike has three lanes each way dedicated to trucks only. Traffic is well over a million trucks per lane per year which is claimed to be the heaviest traffic in the world. We cored, some 100 cores from that road. I want to submit that that particular section would not rut anywhere in the United States but it did rut on the New Jersey turnpike. What we tested we found that air voids were about 2 percent. In some cases less than 2 percent. We cored the shoulder which was not subjected to traffic. It was also 2 percent. The original material was about 2 percent. We tested the recovered asphalt which was AC20 just as it was originally. It didnt age. It was all trap rock. All very angular, a very good type of mix. That is why it really performed so well for years. The stability of the mixture was about 3,000 lb, maybe a little bit more. What I am trying to point is that there are a number of factors which have to be considered. Not only voids. Not only asphalt. Not only aggregate. But it is a combination of things. I think that this is very important. Those contributory factors might lead to failure or might lead to success.

IN-PLACE RUTTING

39

PROF. BROWN: You are correct, rutting is not caused by any one property. Failure to meet one property can cause rutting, but there are a number of other properties that can also cause it. Rutting is also a function of traffic. For the most part we are looking at typical interstate pavements. These are interstates that generally have approximately the same amount of traffic. We have the traffic data that have been evaluated so far, and we eventually are going to try to relate this traffic to rutting. Again, only five pavements have been evaluated so far, hence we cant conclude too much. But I do think that we have furnished sufficient information to be helpful in beginning to solve the rutting problem. Certainly we are not saying that we have solved all the problems. What we are saying is in at least five or six pavements that we have looked at, they simply had too much asphalt in them for the purpose intended. We looked at some where I am sure the aggregate was the problem. We are not confining ourselves to only one property. We are looking at many properties. When we complete our studies, we will have a better handle on all these problems that relate to rutting and on ways to prevent these problems from occurring.