TITLE: by. Designing asphalt mixes to last A B Sterling and K A Zamhari Overseas Centre Transport Research Laboratory Crowthorne Berkshire RG45 6AU United Kingdom PA3305197 PA33305/97 STERLING, A Band KA ZAMIARI (1997). Designing asphalt mixes to last. 3rd Annual Bitumen Conference, Singapore, 25-2 7 June, 199 7 Republic of Indonesia Ministry of Public Works Agency for Research and Development Isiueof Road Engineering Road Research Development Project Published Paper PA 8 DESIGNING ASPHALT MIXES TO LAST by Dr A B Sterling Dr K A Zamnhari Paper presented at the 3rd Annual Bitumen Conference, Singapore, 25-27 June 1997. C TANSPORT RESEAR04 LMORAIO#Y Transport Research Laboratory, in association United Kingdom. with PT Yodya Karya, Indonesia. '11. 1 DESIGNING ASPHALT MIXES TO LAST List of Contents Page No. ABSTRACT 1 1. A BRIEF HISTORY OF ROADS IN INDONESIA 2 2. WHY DID THE ASPHALTIC CONCRETE FAIL ? 3 2.1 Causes of cracks in road pavements 3 2.2 Where cracks start 4 2.3 Design to prevent premature failure by cracking 4 3 . WHY DO WE HAVE PLASTIC FLOW IN OUR ROAD PAVEMENTS 5 3.1 The cause of plastic flow in asphalt 5 3.2 Mix design to prevent plastic flow 5 4. IS IT POSSIBLE TO DESIGN AN ASPHALT MIX THAT HAS ACCEPTABLE RESISTANCE TO BOTH CRACKING AND PLASTIC RUTTING? 8 5. UNDERSTANDING GRADING CURVES ESPECIALLY GAP GRADING 10 6. HOW TO MAKE A GAP GRADED MIX 1 1 7. SO WHAT DOWE DO? 12 7.1 Gap Graded Mixes 12 7.2 Continuously Graded Mixes 13 7.3 Split Mastic Asphalt 1 3 8. HOW DO WE REPAIR ROADS THAT HAVE FAILED BY PLASTIC RUTTING? 14 ACKNOWLEDGEMENTS 14 REFERENCES 11 5 DESIGNING ASPHALT MIXES THAT LAST Dr Tony STERLING Consultant Engineer to the UK Transport Research Laboratory cdo Institute of Road Engineering, Bandung, Indonesia. Tel: +62-22-780-2252 Fax: +62-22-780-7182 E-mail: STRLNGI1@ibm.net Dr. K ZAMIA I Research Engineer Institute of Road Engineering, Bandung, Indonesia. Tel: +62-22-780-2252 Fax: +62-22-780-7182 E-mail cdo: STRLNGl~ibm~net ABSTRACT There is a worldwide tendency for traffic to get heavier. This has been accompanied by an ever increasing number of road failures caused by plastic ruts and corrugations. Those who have sought .to avoid this problem, by making leaner, stiffer mixes, have often found their asphalt failing, very prematurely, by cracking. . This paper discusses the mechanisms of both cracking and plastic flow failures. It offlers a simple mix design procedure and a set of mix design crite ria that can help avoid plastic flow without becoming susceptible to premature cracking. 1. i DESIGNING ASPHALT MIXES THAT LAST 1. A BRIEF HISTORY OF ROADS IN% INDONESIA Routes from village to village always start as earth tracks. On steep hills, especially those with clay soils, it is necessary to cut steps and this is the first level of "engineering." As soon as modes of transport develop from the use of legs (whether two or four) to the use of wheels it becomes necessary to provide some form of surfacing. In Java the most common has been the "Telford'l type pavement. This needs no machines. A road Figure I Telford Type Surfacing bed is excavated by hand and large "kerb" blocks are set into the ground at the edges. Between these the road is made ~~ &"~~$ '~~~' ~~*~~' of hand placed, hand broken ~~'* .A4~~~ stones. These are preferably slighitly wedge shaped, with the ~ ~ ~ ~ ~ .~~~~~ ~~wider face at the bottom. Key stones are then used to lock the larger pieces together. You can still see these roads being built on hillsides throughout Java. They provide cheap and effective (but slow and uncomfortable) all weather access to remote communities. 'When traffic levels grow, the uneven Telford surfaces become unacceptable. In Java, the next stage, of development was Figre Peetrtio Maada Tye SrfaingPenetration Macadam. T1he Figre Peetrtio Maada Tye SrfaingTelford was covered with smaller hand broken stone that were sealed with a coat of hot bitumen. As the name "Penetration Maadrn -~, indicates the bitumnen did 4.....~~., .~~,. ~ penetrate to some extent and ~~~ ~~~ ~~~ ~ this held the surface together. P'enetrasi surfacings are still very common on Indonesian roads that carry low traffic levels. They can be maintained by periodic surface seals uising bitumen emulsion, blinded with sand or grit. H-owever, band made roads are not very smooth. Fast driving is usually uncomfortable, sometimes unsafe. As traffic levels increased it became an economic necessity to re-surface main roads with machine laid asphalt. In many cases this work was accomnpanied by widening of the old road. 2 Figure 3 AC Laid over Penetration Macadam Type Surfacing Tlhc first material used for this purpose was Asphaltic Concrcto. '1ids is a stilriiiatcrial that is rather sensitive to variations in mix proportions. It was not very suitable for use over the Telford/Penetrasi constructions, some of which were rathcr flexible. The result was that many of the Asphaltic Concrete surfaces failed, by cracking, very prematurely. The currcent range of Specifications used hin Indonesia (DGIi1 1986 & 1992) were developed to solve the problem of premature cracking. They achieved that aim, by substantially increasing the bitumen content of asphalt mixes, but we now have a new problem. The majority of new road surfacings still fail prematurely, this time by plastic rutting. The purpose of this paper is to explain what is going wrong. It will answer questions such as: * How much of the problem is the fault of the current Specifications? * What can we do to correct themn? * IHow much of the problem comes from peoplecignoring the Specifications? To answer these questions we will look at: * The earlier cracking failures in a little more detail. * The causes of the plastic rutting that we now suffer froin. • What wecan doto avoid premialture failure by both these mechanisms. 2. WIl1Y DID) TILE ASPRALTIC CONCRETE ]FAIL ? The simple answer is that the mix was too stiff for the high levels of strain that it experienced. H-owever, that simple answer hides many mis-conceptions and we need to look at these if we are not to be in danger of repeating our mistakes. 2.1 Causes of cracks in road pavements Cracks are caused by tensile stresses or strains. These can result from Traffic or the Environment. Cot apression Tra ffi c: Standard structural theory says that the largest tensile strains occur at the bottom of the asphalt, directly under the vehicle wheel. Smaller, but significant, tensile strains also occur at the top of the asphalt, before and after the wheel. H~owever, there are also strains all around the contact 3 area betwecn thc tyre andl tlhe road, causedl by localiscil derormation of the surface, and immediately below the wheei, caused by traction, braking and steering forces. All these strains occur predominantly at the upper surface of the asphalt. Environment: Changes in temperature, from day to night and from hot to cold seasons, produce tensile strains in the asphalt, especially at the upper surface. This mechanism is probably less important in Indonesia than in countries farther from the tropics. 2.2 Where cracks start At the surface of the road the bitumen looses its lighter oils, by evaporation, and is progressively oxidised. Thlese changes lead to it becoming hard and brittle. The penetration can drop to between 10 and 20 at the top while it may be 50 or above in the body of the layer. Classical pavement design theory assumed that cracks would start at thse bottom of the asp~halt because that is where the tensile strains, caused by flexure of the pavement, are largest. However, cores taken from cracked roads have shown that most cracks start at bthe top of the asphalt. This observation has been verified in many parts of the world, including UK, Africa, Arabia and Indonesia. Apparently, the embrittlement makes the tensile strains at the top of the layer more damaging than those at the bottom. 2.3 Design to prevent premature failure by cracking Any bitum-inous mix, even pure bitumen, will fail by cracking if large enough tensile strains are applied to it, often enough. There is usually an approximately linear relationship between the logarithm of tensile strain amplitude and that of the number of strain repetitions the material can stand before cracking. Well-designed mixes will. have a higher strain tolerance than bad ones but all will eventually fail. Even a good mix may fail prematurely if the road is too weak and the strains too high for the number of vehicles that must be carried. Hence, prevention ofpremature cracking involves both: * mix design and * pavement design. Mix Design Parameters that are likely to correlate well with good resistance to cracking include: * Binder Properties, such as Penetration or Viscosity and maybe Penetration Index. * Effective Binder Content (EBC). * Voids in Mineral Aggregate and % Voids Filled with Bitumnen (VMA/VFB). * Binder Film Thickness (I3FT). In mixes with a continuous stone matrix, only partially filled with sand asphalt, the atmosphere has access to the individual, coated particles of aggregate. In such mixes is possible that Binder Film Thickness (B3FT) will be usefuil to ensure there is "enough" bitumen to make the mix durable. Mixes that fall into this class include coarse DBM mixes as well as SMA open graded asphalts. The grading curves for all these mixes lie well below the Fuller Curve. In mixes that have a continuous sand asphalt matrix, with discontinuous stone particles, it is the sand asphalt that is exposed to the atmosphere. Because this is a continuum, the concept of individual jparticles, each coated by a finite thickness of binder, is not relevant. For such mixes, which include 4 contintuous (AC) and gap) graded (liRkS) mixes, thc percentage Voids Fillcd with Bitumen (VFI) is likely to be a better criterion for durability than DVIL To be successful aia ash alt inix mutst hate "eitosgh " bitunren to bind thre aggregate particles and "enough " air voids (VIM) to avoid failuire by plastic flowv. In ail asphalt that has low Voids in th e Min eral1 Aggregatre (VMlA,), thtese two requ iremtentis are con tradictory. If youl add en ough b itu m en fo r duir a b ility thse re w ill notg b e su ffi c ie nt VIM;', ify ou l ea ve s uffi c i e n VI M thIse re w ill no t be enough bitunmen to mtake a durable mix.i Therefore, it is essential to achieve a high VMA that allows rooms for both "en ough"1 bitumtenl amird "en ough"` VIMl. 3 . WHlY DO WE H1AVE PLASTIC FLOW IN OUR ROAD PAVEMENTS 3.1 The cause of plastic flow in asphalt Studies in several countries have shown that when the VIM drop below 3% asphalt concrete mixes are very likely to fail by plastic flow. Figure 3 Plastic Rutting and Refusal VI Data from roads observed in Indonesia Pr abili Som mie uri under some traffic confrm that the probability of failure' by of 1 ~ conadit~ionse,swith 2v% VIM. Others don't plastic deformation increases greatly as VIM of, 1lr he safe threshold is 3% VIM drop beyond 3%. (TARP 1993) See Figures byl lsic AEIB TRAFICKING. 7, 8 and 9 for evidence from the Middle East and from Malaysia. (TRI, 1997) There is Plastic rutting does not happen some evidence to suggest that gap graded when mixes retain VIM of more mixes may have a slightly lower failure than 3Yo. ~threshold (2% VIM) than continuously VM graded mixes (3% VIM). 1 2 3 4 5 6 We can off~er an explanation for the above, It seems reasonable but we need to remember that it is only hypothetical. To resist plastic flow, asphalt mixes depend on both internal friction, between the aggregate particles, and binder viscosity. If, at any poin t in the mix, the local VIM approaches zero, the bitumen begins to separate one aggregate particle fr-om another. In other words, it ceases to be a binder and becomes a lubricant. Onice this happens the mix will fail by plastic flow. Continuously graded mixes are highly non-homogeneous. By the time the average VIM has dropped to 3% there may be a significant number of locations in the mix that have near zero voids. Gap graded mixes, especially those with lower stone contents, would be expected to be less inhomogeneous and this may account for their reaching average VIM as low as 2% before flow becomes significant. There is some evidence to suggest that in "free flow" conditions the VIM threshold is closer to 2% than 3% but this is not yet certain. Indeed, the American Asphalt Institute, in the 1994 issue of their mnix design manual (MS-2 1994), recommend the final void content after trafficking should be 4%. 3.2 Mix design to prevent plastic flow Figure 6 Loss of VIM Due to Compaction by Traffic Marshall mix design procedures have always 75 Blow Marshall After traffic required the VIM to be above 3%. The trouble is, the density achieved by 35, 50 or even 75 blows on each face of the sample is less than that which occurs in the wheel paths of roads carrying severe traffic loads. Thie final in situ dusity may be 3% or even 4% higher than the 75 blow Marshall density. 5 A:1 Thlis means that 3% or 4% of air voids can bc lost (luring the fiuther compaction caused by traffic. If the original design was for 4% or 5%, the residual VIM can have dropped to between 0% and 2%, in which case the road will be failing by plastic flow. To prevent this happening, the Asphalt Institute (MS-2 1994) now recommend that mixes are laid and compacted to air voids of 8%. This allows for 3% -5% of expected consolidation during trafficking. 'Me target for the "final` VIM is 4%. To be sure that in-situ VIM never drop below 3% we need an additional test procedure, in which samples are compacted to a refusal condrition. That is, until they "refuse" to become any more dense. We can then set an upper limit to thse design binder content, corresponding to 3% VIM at the refusal condition. 'There are two methods by which one can perform 'refusal density" teats. The method we recommend uses a vibrating hammer to compact samples in a CBR mould. It follows the procedures described in the Percentage Refusal Density (PRD) test in BS 598 (BSI 1989), with one major improvement. The hammer is surcharged by hanging 30 kg onto it rather' than using body weight. The other method is an extension to the Marshall test (AASHTO T245), using about 400 blows per face instead of 75. It is essential that the density achieved in these "refiisal density" tests really does correspond to that reached in the wheel paths under severe traffic. The vibrating hammer method is quicker than the extended Marshall procedure, is less likely to break the aggregate particles and it is very easy to do. The extended Marshall procedure is not difficult but it is unpleasantly noisy and does break down more aggregate than the vibrating hammer method. VIM as a1 Determinant for the Performance of Asphalt Concrete in the Middle East Crack initiation.......... Crack propagat ....... O.K. Flow .4No cracke-i4--.- May crack 4- ill crack ---.- Will Mas aMay No Flow Flo Flow 0 2 3 5 6 7 9 b~~~~~~~~% VIM Figure 7 Performance of Asphialt Concrete on "Middle East Road One" 6 No ~~~~~~Ma Y Will -Cratlkt- ~Crack Crack Will Ma ~~~No Flow Flow Flow '3 2 4 5 7 a % vim Figure 8 Perform~ance of Asphalt Con~crete on "Middle East Road Two" 25 20 E A 15 0~ 2.. 10 5 0 4 Air voids (~.) after trafficking 11 Fig~ure 9 0 VIM is Determinent of Freedom from Ruttiniz in Malaysia 2 7 4. IS iT POSSIBLE TO DESIGN AN ASPHA~LTMIX THAT HlAS ACCUTVABLE RESISTANCE TO BOTH CRACKING AND PLASTIC RUTTING ? The short answer to this question is, "Yes`. We need to look very careftilly at the requirements of such a mix. 1. We are considering heavy traffic so the mix will be IHRS, SMA or AC and the Marshall design will use 75 blows per face. 2. The Contractor will lay the asphalt at 98% of Marshall density. 3. In the wheel paths, the final density will be 102% to 103% of Marshall density. 4. To avoid plastic rutting we will need to retain at least 3% VIM after secondary compaction by traffic. 5. To avoid premature cracking we will need at least 65% VF13 at Marshall density. It follows that: (Pbe)x(Gmb) A. Since VFBŽ 65% = 100 (VMA)x(Gb) 6xJM~(b the ujinimum acceptable effective bitumen content (Pbe) is 1 OO(Gmb) B. Since Refuisal VIM 3% = (VMA -Vol. of Bitumen -Loss of VIMtunder traffic) Gb the maximum acceptable effective bitumen content (Pbe) is (VMA -dVIM -3)x Gm Thec design binder content mu st be between the minimum and maximum values, calculated from these formulae. The charts in Figure 10 show the relationship between VMA and the possible ranges of design binder contents. They consider two values of dVim-, the loss of VIM uinder traffic. These are 2% and 3%. As you can see, for the normal range of VIMA achieved in an AC mix, 13% - I 5%, there is no binder content that will satisfy both A. and B. above. We need at least 16% VM (preferably 17%) before we can have a design that can avoid premature failure, under severe loading conditions. To have a safe margin that will allow for typical variations in mix proportions, we want a VMA of 20% or more. Thfis is not achievable with an asphalt concrete type of mix. It can be achieved with a gap graded mix or with a coarse stone matrix mix such as an SMA grading. 8 How Much VMA Do We Need Fl~gure 10 Data Setil Wbe mm wne max tilt VMA /b Range -u.j 15J b I -- I - 4- - . I. -U.L 4.4 4.1 J5.8 vro 111191 d VIM vim rer J I ralffcked VJ.0ou T81 W. 4. ____ b103(S ~~ ~~~ ~~ ~~ange of Bitumen Vs VMI 7 ~~~W.F* 6 ~~~~0.4 Wbe min.=VFBminVMA¶3b/Gmb . . -.-- --- 1 Wbe max =(VMA -dVIM-VIMr min)G(b/Gmb 0 _______I____________ Range =Wbe max- Wbe mm 02 -- n 14 16 18 20 22 24 Data Set2 VMA (%) VMA /b Wbe min Wbe max -- -t- e n 17 rlick1] Range -VI-Bmi ~U5 15 d ~VIM~ W ~~~~U~ 3- -Vfl~T-e f ~~17 4 . ~~~~~~~~~ 1T3b .4 'Sb b.b U.b 29 f. -5 5W 5: b2 010 zT -55 I 3 0. Wbe min =VFBmin'VMA'Gb/Gmb Wbe ma x =(VMA -dVIM-VIMr min) 3bIG mb Range =Wbe max- Wbe min range of Bitumen Vs VMI 1.2 ~0.4 ~0.2 ....... . ..- ............ .....---- A........... 0 14 16 ~~~ ~~~ ~~~ ~~~~~~~~~~~~~18 2 22 4 VMA (%) To achieve a practicable range into which we can fit a design binder content with a realistic tolerances, say ± 0.5 %, we need a VMA of 20% to 22%. This is achievable with a gap graded mix, which is what H-RS was always meant to be or with a coarse stone matrix mix. With a continuous grading above the Fuller curve, the best we can do is about 16% VMA. So, we need to look snore closely at grading curves. hi jpalliclllar, we sieed to see what a Gap Gradcd Mix really is. 9 ... .. --- . ... -- 1 Wbe min Wbe max U.DD .115 5. UNDERSTANDING GRADING CURlVES ESPECIALLY GAP GRADING Lets start with the Fuller curve. It is named after the man FulerCu~e] who devised it and it represents the aggregate grading that 100 will have the lowest possible VMA. We can represent it by: so ( 045 40 - ~~~~~Percent Passing =100 Yp O' ~~~~~~If you are making concrete you want to minimise the 0.01 0.1 1 10 0 weve '1k* (M4volume of cement paste, both to save money and to reduce shrinkage, and the Fuller curve is the ideal grading. In asphalt, where we have a binder that can also act as an excellent lubricant, we must have enough air voidc to prevent the mix going plastic. We have to stay as far away from the Fuller curve as we realistically can. Aggregate that follows the Fuller curve has just enough fines to fill all the spaces left by the coarse aggregate. If the grading is above the Fuller curve it has more fines than the spaces between the coarse particles can hold. It follows that the sand asphalt matrix will be the continuous medium and the stone will be enclosed in this. Let's call this a "sand matrix" material. Below the Fuller curve there are not enough fines to fill all the spaces between the coarse particles. Therefore the coarse aggregate forms a skeleton that is the continuous medium. The fines fll some, but not all, of the spaces between the stones. We can call this a "stone matrix" material. There are three families of asphalt mixes. They work like this: 1. You can be above (finer than) the Fuller curve and can use a continuous grading (AC). 2.' You can be above (finer than) the Fuller curve and can use a gap grading (IRUS). 3. You can be well below the Fuller curve, with gradings that do not provide enough fine material to fill all the gaps between the coarse aggregate particles. With an AC, the farther you move the grading above the ~sphalt Concretel Fuller curve the more VMAA you will get, for a given 100 particle shape. If you use more angular fines you will get 8O - ~~~~~~a higher the VMA but the mix will be "harsher". 60 -Moving farther above the Fuller curve means more fines 40 and less stone. Thiat is not ideal for carrying heavy 20 - traffic. Also, no matter how little stone you use, you can ~~~~~10 never get the VMA as high as you really want. This 0.01 . 0 10 mk A a relatively sensitive mix. Unintended Sieve Sze (mm)changes in mix proportions are more likely to cause problems than in other mixes. BEitumen Maca~dami Dense Bitumen Macadam is coarser than the Fuller _______________ curve. H-aying fewer fines than spaces for them, mix with too much bitumen. You can get a higher VMA 60 with this sort of mix by reducing the fines even more. If 40 - 20 you go too far the mix becomes liable to segregation; to 0 ~~~~~~~prevent that you have to reduce the maximum size of the 0.01 0.1 0 10 sti.Tk this process far enough and you no longer Slave Sze (mm)have a dense mix at all; you have SMA (as shown in the lowest curve). 10 HRS has been blamed for many failur grading curve like the one on the righi will work if you call it AC and add, n call it HRS and put in 7.5% bitumei another excpensive failure. Finally we conic to the "gap graded' miuxes. As you can see, they lie well above the Fuller curve. The essential thming is the "Gap". Betwveen the 600 microns aiad 2.36 ninn sieves there is, aind inust be, a reduced anmount of inateriaL This is what gives these mixes the extra VMA. The figure on the. left is gap graded; that on the right is not. It doesn't s matter that the one on the right also fits inside a ec . dog-legged envelope; it isn't gap graded and it s will not have the VMA it needs. 40 .-- ,es when the culprit was a 20 - . --- t. The grading on the right iiaybe, 6% bitumen. If you4 ai you just have caused a' 01 sm(m In fact, 7.5% bitumen may be too much even for the mix on the left. That is why compaction to refusal is essential. We have to stop guessing what the correct bitumen content is and start measuring it. If you lititityour bitumten content so that there are 3% VIMl at refusal (making sure that you have done the refusal compaction properly) your mixes will not fail by plastic rutting. If you have enough VMA so that you can also get 65% VEB, at the 75 blow Marshall density, your mix will also be safe from premature cracking. 6. HO0W TO MAKE A GAP GRADED mix 100 0 The first thing to understand is that you 80 ... III;: I ' ~~~~cannot miake a gap. graded inix unless you have th rgt sand to bles Wit your 60 ..... , .~~~~~ ~ crushed fin es. Just any sand wvill niot do. 60 ... ... 1 .4. 4 . ..11 ~ :~ I' I'[ It may help to see how gap graded mixes Iijjii:: ~~~~~eastwards from London to the sea, is full of 0~~ ~~ ~~ j: I I' ~~~sand deposits. Te uppermos lie represents 20 ~ ~~~. ' ~~~~ a fairly typical grading. Note how little .. ~~11h. material there is between 600 microns and 20 ... 5~~4 1..... 4.* 4 ....... I ~2.36 mm, less than 15%. The lowest lie I.::: , ~~ represents a crushed stone or a Thames 0 fiffi uw Valley gravel. If you combine the two 0.01 0.1 1 1 D gradings in approximately equal proportions Sieve size (mm) you will have a gap graded mix. You couldn't get a continuous mix if you tried because there is a gap between the top end of the sand and the bottom end of the stone. These mixes are specified in 13S594. (1992) 11 J Now look at the grading of th~is sand, that 100 ~ ~ ~ ~ 11is typical of many Javanese river deposits. iflilil ~~40% of its content is coarser than 600 III ii Jill IImicrons. It is impossible to find a 80 II ihl I 11111111 II I Ij ~ combination of this sand and this coarse ~j1 MU ]. 1I~ rV~r II~ aggregate that will give a gap grading III15'[ 111 because of the overlap between the sand ~Ifi~U I~j[ Ii I ~i anid the stone. II IIP 111 111 Java has sand deposits along the north 40 :il~: ' 1 11 :1 coast that were once beaches. In a 231. 1 ;:i previous geological age, the sea washed ~I ii :~z:. out the silt. Then, as plate tectonic action I 1111:1 Z slowly lifted the island of Java and caused 20 ;.~p' r'~,ir III ii! ~1 ill the sea to recede, the rain had a million IIII IIII t ~ :':: years (or two or three) to wash ou't the o JJ 1 liii I'll I II salt. Some of these deposits are suitably - ~~~~~~graded for making HRS. So Indonesia 0.01 0 iee sze 1 m does have the sand it needs, just where it most needs it, along the Northern Corridor route. In UK we carry almost all our heaviest traffic on gap graded mixes. Indonesia has a section of road between Cirebon and Losari that has been doing the same thing for the past eight years. There are no cracks and the mix is just about to lose its voids and go plastic. However, it has survived for longer that all the AC mixes around it. If we design similar mixes, using compaction to refulsal to make sure they never go plastic, they should still be there in ten or twenty years time. 7. SO WHAT DO WE DO? 7.1 Gap Graded Mixes Thiis is probably the best answer, if you have a suitable sand available, because: 1. gap graded mixes are less sensitive than continuously graded mixes. 2. They are easier to compact. 3. They are more resistant to oxidation/embrittlement/cracking. 4. If you ensure 3% VIM at refuiial, they are very resistant to plastic flow. Blend 20% to 40% of fine sand with 60% to 80% of coarse and fine graded crushed stone. A Mix Design Spreadsheet is a usefuil tool to help) you get the blend right. Such a spreadsheet is not essential; highway engineers were making good asphalt mixes long before there were comnputers. 1-lowever, computers are now available; so, why not make mix design easier and more reliable? Remember thle 3 key issues: A> It is not 1IRS unless it is Gap) Graded B> For heavy traffic, you mutst have 3% VIM at refusal density. C> For durability, you must have 65% VFB at Marshall density. 12 7.2 Continuously Graded Mixca If you do not have the right sand to inakc a gal) graded mix, design one that is continuously graded asfarfroii tite Fuller curve as you can reasoun ably get. Again, use of a Mix Design Sprcadshect will make life quickcr and will lprobably increase the probability of getting the first trial mix to work. 1. Decide whether you want to make a `stone matrix" material, below the Fuller curve, or a "sand matrix"' material, above the Fuller curve. 2. Provided you can get 16% to 17% VMAA, either of these will work. 3 . Above all else, if you are designing for heavy traffic, make sure you have 3% VIM at refusal density and 65% VFB at 75 blow Marshall density. Above the Fuller Curve. .sand matrix Below the Fuller Curve. .stone matrix 7.3 Split Mastic Asphalt It would probably be more helpfu if we thought of SMIA as meaning Stone Mastic Asphalt. As we said in Section 5, above, SMA is an extreme stone matrix material that can have 30% , or even less, passing the 4.75 mm sieve. Using so little fines would tend to allow segregation of the larger sizes of coarse aggregate. To prevent this, the maximum stone size is limited to between 8 and 12.5 mm. The sand content is so low that the mix is no longer dense graded. This means that air has access to the surface of most aggregate particles. Oxidation and embrittlement of the binder, followed by cracking, are obviously potential problems. What saves the mix from rapid failure is that the binder films are very thick. In the current specifications, cellulose fibres are added to reduce the tendency for the binder to drain fr-om the mix. An alternative version of the specification, Butonite Mastic Asphalt, does niot use any cellulose fibres. Rubberised bitumen would probably work even better. These specifications ought to include VIM at refusal density to ensure the miixes can carry heavy traflic without fear of plastic rutting. We don't yet know what VIM they need at refusal density. Until we know better, 3% is a sensible working figure. 13 8. HO0W 1)0 WE REPAIR ROADS THAT 1HAVE FAILED BY PLASTIC RUITING? DO NOT OVERLAY I!!!! This is going to be difficult but it is essential. If you lay sound asphalt on top of plastic material it will fail very quickly. SO WHAT DO WE DO 1 . Leave it as long as you reasonably can. 2. Do any local repairs that are necessary to keep traffic flowing. 3. If ruts become large enough to interfere with traffic, yet the road remains structurally sound, plane off the ruts. 4 Wh~en you have to resurface, plane right back to sound material and then rebuild from there. WVE ARE GOING TO NEED 1. More planing machines 2. To learn how to recycle successfiflly and economically: (a) in-situ, (b) using plant based hot mixes (needs modified plant), (c) using cold mixes. RECYCLING SHOULD BE A PROFITABLE, GROWTH INDUSTRY FOR THlE NEXT TEN YEARS IN INONESIA... and anywhere else that has a lot of plastic rutting or a lot of cracking. ACKNOWLEDGEMENTS The authors gratefuilly acknowledge the facilities provided by: Dr Patana Rantetoding M.Sc. Director of IRE, Ministry of Public Works. lr Soeharsono Martakim, Director General of Highways, Ministry of Public Works. lr Joeflanto Hendro Moelyono, I-lead of Agency of Research and Development, Ministry of Public Works. 14 REFERENCES AASHTO(T24 5) Resistance to Plastic Fllow of Bituminous Mixtures Using Mirshall Apparatus. BS594 (1992) Hot Roiled Asphalt for Roads and Other Paved Areas. British Standard 135 594: Part 1 : 1992. Specification for Constituent Materials and Asphalt Mixtures. London : British Standards Institution. BSI (1989) Sampling and Examination of Bituminous Mixtures for Roads and Other Paved Areas. BS 598 : Part 104. Methods of Test for the Determination of Density and Compaction. London : British Standards Institution DGH (1986) LBRD) Road Betterment Programme: Specification for High Durability Asphalts. Central Design Office, Directorate General of Hlighways. Jakarta: Indonesia DGH (1992) Specification for Dense Graded Asphalt Designed by the Marshall Method. Jakarta: Directorate Gencral of 1 ighways MS-2 (1994) Mix Designi Methods for Asphalt Concrete and Other Hot-Mix Types, Asphalt Institute, Lexicon, Kentucky, USA. Manual Series Number 2 (MS-2), Sixth Edition. TARP (1993) The Evaluation of Asphalt Mixes in Laboratory Trials and A Review of Mix Design methods and Specifications for Asphalt Surfacings. Technical Assistance and Research Training Project at the Institute of Road Engineering, Banding, Indonesia. Technical Report, 1993 TRL (1997) Dense Bituminous Surfacings for Heavily Trafficked Roads in Tropical Climates, HRl Smith & CR Jones, TRI, Annual Review, 1996 is 1