CONCEPTS USED FOR DEVELOPMENT OF BITUMEN SPECIFICATIONS

CONCEPTS USED FOR DEVELOPMENT OF BITUMEN SPECIFICATIONS M.F.C. van de Ven 1, K.J. Jenkins 2 and H.U. Bahia 3 1 Delft University of Technology The Netherlands. E-mail: M.vandeVen@citg.tudelft.nl 2 SANRAL Chair in Pavement Engineering University of Stellenbosch, South Africa. E-mail: kjenkins@sun.ac.za 3 University of Wisconsin Madison. E-mail: bahia@engr.wisc.edu ABSTRACT The evolution of bitumen specifications has been dramatic in the last 15 years. Simple empirical tests based on mechanical properties formed the first types of binder tests. More recently however, fundamental, visco-elastic and damage characterization methods have been adopted. Changes in specifications have been motivated by various factors and have had various consequences in different parts of the world. In this paper, the authors bring together their experience of working in three different continents, to evaluate various bitumen specifications and discuss merits of different concepts used from as early as 1950 to 2003. The paper shows how the Bitumen Test Data Chart (BTDC), and the van der Poel nomograph, which is based on penetration and softening point, could be used to link the empirical test results to performance related properties such as stiffness and strength and how these derived measures could be linked to traffic volume, traffic speed and pavement design temperatures similar to the Superpave PG system. The approach is used to discuss the current South Africa Specification and to propose a possible performance grading system for South Africa using actual climatic and traffic data. Finally, this paper covers the most recent changes proposed in the USA and Europe for the original Performance Grading concept, aimed at including Polymer Modified Bitumen. 1. INTRODUCTION During the past 15 years bitumen specifications have seen dramatic changes from simplistic, empirical tests to fundamental, visco-elastic and damage characterization methods. Changes in specifications are motivated by various factors and have various consequences in different parts of the world. The development of the specifications up until 1990 was based on simple concepts of hardness and viscosity. In many countries, including SA, the specifications were dominated by two simple empirical tests, namely the Penetration and the Softening Point, related to in-service temperatures and the (Newtonian) viscosity related to mixing and compaction temperatures. These specifications are not directly related to performance, because they are based on experience and are only valid for penetration grade bitumen. In the early 1990 s, the SHRP project in the USA resulted in the introduction of Superpave binder specifications. As a result, the bitumen specification world was changed dramatically by two developments: (1) the introduction of new (for the bitumen industry) rheological testing methods, and (2) the concept of Performance Grading (PG) to relate pavement, temperature and traffic conditions to fundamental properties. This resulted in the development of binder-blind specifications (regardless of whether modified or not). Both developments were successfully Proceedings of the 8 th Conference on Asphalt Pavements for Southern Africa (CAPSA'04) 12 16 September 2004 ISBN Number: 1-920-01718-6 Sun City, South Africa Proceedings produced by: Document Transformation Technologies cc

implemented in the USA but are lagging in other parts of the world. One of the logical reasons is the cost and complexity of the new rheological test equipment. What is surprising, however, is the reluctance to the implementation of the concept of Performance Grading. Although the concept is best implemented using rheological properties, it can be applied using various types of test results, including those derived from Penetration and Softening point. This subject will be discussed in this paper and methods of using simple test results to produce a Performance grading system will be presented. 2. SPECIFICATION DEVELOPMENTS This section discusses the developments in the three continents. First the developments in the Europe are given, then the American views are highlighted and finally the SA situation is compared to the others. Also the difference in accents are highlighted (mixes versus seals). 2.1 Europe Penetration grade bitumen specifications in Europe have always been seen by the industry as indirectly related to performance. Not only were these tests used as a specification for the binder and asphalt material, but they were linked to the structural design of pavement layers using the concept of estimated bitumen stiffness [Shell, 1991 ]. Low Temp Cracking Ave. Service Temperature Hot Summer Mixing & Compaction Consistency Pen Tr&b Visc @ 60 degc Visc @ 135 degc 0 25 50 75 100 125 150 Temperature (degc) Figure 1. Conceptual use of BTDC for windows in specifications. The Bitumen Test Data Chart (BTDC) as developed by Heukelom [Shell,1991 ] in fact was and is indirectly related to performance as can be seen in Figure 1. For specification purposes windows were created closely related to the BTDC concept. Figure 1 is an example of such an approach. A strong link to performance was based on the fact that the simple empirical tests like penetration and softening point can be used to make a reasonable approximation of the bitumen stiffness. A good example is the van der Poel nomogram. This bitumen stiffness can be related to the stiffness of the mix via volumetric properties i.e. using the Ugé nomogram. The mix stiffness and the volumetrics are used to estimate fatigue lines for mixes, which are indicators of pavement service life [Shell, 1991]. When the stiffness of the mix is estimated at any temperature and loading speed, it is possible to use this value in linear elastic multi-layer analysis to determine stress and strain distribution in a pavement structure. Using the estimated maximum tensile strain in the asphalt layer the layer fatigue life is estimated from the predicted fatigue lines. This methodology has been used extensively and is still used by some consultants and pavement designers in Europe. Figure 2 shows a flow chart of this methodology, which explains the possible indirect link between penetration grading and performance.

Figure 2. Flow chart of structural fatigue design of flexible pavements without doing any testing for ranking purposes. Since 1990, the European specifications for paving grade bitumens have been developed in two stages: First generation: production of harmonised specifications and test methods for paving bitumen. For use in Europe, based on national standards Second generation: produce specifications that are more directly Performance-Related, with existing or new properties and test methods where appropriate. 2.1.1 First generation specifications Three standards have been produced: Paving grade bitumen: unmodified bitumen with penetration ranges from 20 to 900 Pen. Hard Paving Grade Bitumen: unmodified bitumen with penetration ranges of 10/20 and 15/25 Polymer Modified Bitumen Together with these standards 23 associated test methods have been produced, in all cases harmonisation of existing tests between different national versions. The existing tests are mainly empirical, but they respond to the Essential Requirements from CEN being: Consistency at elevated service temperature (softening point or viscosity at 60ºC for high service temperature) Consistency at intermediate service temperature (penetration for intermediate service temperature, typically 25ºC) Durability (short term ageing)

Additional properties such as viscosity at high temperatures (135 ºC) related to handling and application, Fraass breaking point (low temperature cracking) and safety/ handling properties (density, flash point) show the link to performance. For the PMBs, additional tests namely elastic recovery at 10 or 25 ºC, cohesion tests, storage stability tests, have been included to demonstrate the influence of the presence of the polymer and indicate additional performance characteristics. All methods from the first generation can be considered for use in the second generation, although in most cases correlation to performance is not established and more fundamental tests may be required in future. 2.1.2 Second generation specifications To produce genuine performance-related specifications it was necessary to reconsider multiple performance measures. There are many complex factors that influence the relationship between the road-user requirements for a safe, comfortable and durable surface, and bitumen properties as measured in the laboratory. These factors include construction, mix properties, pavement structural and surface characteristics. In 1999, Eurobitume organised a global workshop to pave the way for development of second generation specifications. A set of requirements in a 3-stage process was identified: Define the key performance characteristics for the pavement and asphalt mixes Identify binder properties related to those characteristics Identify binder test methods to measure those properties Table 1. Outcome of 1999 Eurobitume workshop [TC336WG1-TG5 ]. Performance requirements for pavement and mix Relevant binder properties Binder test methods Friction Low relevance None Permanent deformation Rheological properties Viscometry, DSR Surface cracking (ageing) Ageing (short, long term) RTFOT, RFT, PAV Stripping Binder/aggregate Test on mixture Structural contribution Rheology (modulus,etc) DSR Noise emission Low relevance Not relevant Low temperature cracking Rheology, failure DSR, BBR, DTT Fatigue cracking (thin Failure (fatigue, healing) None layers) Manufacturing/laying Rheology, storage stability Viscometry, TUBE Following that workshop, a major exercise to update and refresh the thinking on the outcomes was done by the five key European Road Industry Stakeholders Group (Eurobitume, EAPA, FEHRL, WERD/CERD, IISPR). They produced a report [CEN/TC336 N 49, 2003 ] from which it became clear that the different parties agree on many aspects. A system with different levels of performance was recommended, including an option where the performance is not determined (class 0).

Table 2. Binder properties to consider in a second generation binder specification system [CEN/TC 336 N 49, 2003 ]. Identification properties (for quality and delivery control) Performance related properties HSE properties Other information for the mix, seal, etc, producer Consistency (measurement must be performed quickly) It should preferably be linked to grading, to facilitate checking of conformity Including minimum quality requirements: Perceptible properties Solubility Volatility (loss on heating) Stiffness or rheological property at elevated service temperature Stiffness and/or fracture at low service temperature Ageing for durability assessment Stiffness at intermediate service temperature (maybe binder fatigue, etc) Flammability property (flash point) Information on classification and labelling Information on recyclability Density Storage stability (for PMB) Viscosity versus temparature (pumpability) Specific requirements (for special applications) Can be anything ( pigmentability, fuel resistance, de-icing resistance, etc). Innovation function. A multiple level binder specification system was selected for flexibility reasons such as: Allow user and producer to agree on level of sophistication of specification A system applicable to all binders No over-specification In relation to the framework of the future system of specifications, Table 2 includes suggestions for the selection and organisation of the different properties in a second generation specification. At the moment a task group is working on selecting the key P-R requirements and developing a framework for the structure of the new specification. 2.2 USA The first bitumen specification implemented by the American Association for State Highway and Transportation officials (AASHTO) in the USA dates back to 1931 (Roberts et al, 1996). These specifications recognised bitumen as a semi-solid and therefore consistency measures were used to develop the Pen-grading system. Tests included penetration and ductility, both were used at 25ºC originally developed for natural bitumens, which are more solid-like than petroleum refined bitumens. Petroleum asphalts were significantly less vicsous (less solid) and in comparison to natural asphalts could be considered oils. In fact until today asphalt contractors in the USA call bitumen the oil in the mix. These specifications were flawed for two main reasons: these tests are empirical in nature and can give only index properties, there is no real control on temperature sensitivity as related to pavement conditions.

Although the penetration grading includes testing at 25 ºC, there were no tests at high or low temperatures and the limits on penetration and ductility were so relaxed that many bitumens of various temperature sensitivity and performance could fit in the same grade. The bitumen industry tolerated these specifications because of the lack of more scientific tests for approximately 30 years. In the early 1960s the viscosity grading system was introduced. Testing included penetration at 25ºC, the viscosity at 60ºC, and the kinematic viscosity at 135 ºC. Ageing was also considered and the thin film oven test was used to measure relative ageing of bitumen. The viscosity grading (asphalt-cement, AC, grading) were considered a major improvement compared to the pen grading system. It included two tables, Table 1 and Table 2 (Roberts et al, 1996). Due to the nature of the specifications, Table 1 permitted higher temperature sensitivity for bitumens than Table 2. The acceptance of the viscosity grading system by the various state agencies was not unanimous. Several states exchanged the pen grading fully with the viscosity grading, while others kept the pen grading and claimed that viscosity grading had it s own problems. The claims were based on scientific reasons such as the continuous lack of low temperature properties, the wide acceptance limits and the lack of long term ageing procedure. In the early 1970s the oil embargo created more challenges for the bitumen industry. Crude oil sources diversified and refining procedures were significantly varied due to focus on maximizing fuel production. This made the use of viscosity grading even more difficult and various highway agencies started amending the viscosity grading to prevent early failures that started to show up on various projects. The amendments of specifications included temperature susceptibility, shear susceptibility, primitive rheological properties, and other compositional characteristics. These amendments were based on important research studies, mainly done in Europe and the USA. While some of these amended specifications were successful in different local areas, the total number of specifications used across the USA increased exponentially. In the mid 1980s there were more than 35 different bitumen specifications, which created major problems for bitumen suppliers limited by storage capacity and localised markets. In 1987 the Strategic Highway Research Program (SHRP) was initiated to include significant focus on developing national uniform bitumen specifications. While initially it was expected that the national bitumen standards would include compositional properties it was very quickly found that only physical properties that describe the visco-elastic nature of bitumen will be included. In 1991 the first version of performance grading system based on fundamental rheological properties was introduced. It took 5 to 6 more years to implement what is called today the Superpave PG grading throughout the USA. The PG grading system is considered unique for the following reasons: 1. It recognizes that there are three main distress mechanisms affecting pavement performance, rutting, fatigue cracking and thermal cracking. Bitumen plays a role in resisting each of these distress mechanisms. 2. Bitumen properties that contribute to resistance of these distresses are not the same. They are also measured at different temperatures. Bitumen contributing to rutting resistance is measured at maximum pavement temperatures and it favours a more elastic and more stiff binder. Bitumen contributing to fatigue resistance is most critical at average pavement temperatures and while it still favours an elastic binder it requires a softer bitumen. Bitumen contribution to thermal cracking is critical only at minimum pavement temperatures and it favours a less elastic and soft bitumen. 3. Acceptance limits are the same but they have to be met at specific pavement temperature and traffic conditions. In other words the grading is based on pavement conditions (temperature and traffic) but all grades meet the same bitumen property criteria.

This was the first ever true and direct performance related grading system for bitumen implemented in such a large continent. Figure 3 gives an overview of the specification issues and the used testing methods. Performance Grading Specification Thermal Cracking Cooling rate Strength Glass Transition Fatigue Cracking PAV Permanent Deformat ion Traffic speed Traffic volume Pav. Structure RTFO Pumping Storage Stability Additive type Mixing & Compaction 2.3 South Africa -20 20 60 135 Pavement Temperature, C Figure 3. Overview of performance grading system. The South African bitumen specifications date back to 1951, with the first publication of penetration requirements. Thereafter a revision was made in 1966 and the Rolling Thin Film Oven Test (RTFOT) introduced in 1972. Early bitumen specifications in South Africa were based predominantly on bitumen produced in the Middle East that was very similar, containing highly aromatic crude oils. Even with a relatively stable source, intermittent problems were experienced with the bitumen. Since the democratisation of South Africa in 1994 and rejoining of the international market, crude sources for SA bitumen have diversified making it more difficult to produce bitumen of consistent quality. In 1989, the Bituminous Materials Liaison Committee (BMLC) set up a task group to investigate problems associated with bitumen supplies and to make proposals regarding bitumen specifications. This lead to specification changes in 1994 and 1995 that were significant (SABS 307 1995). Requirements for properties after conditioning in RTFOT were added to account for changes e.g. ageing, during plant mixing, placement and compaction of asphalt. The 1995 SA binder specifications did not address compositional balance of the bitumen directly. This was a point of concern and subsequently van Assen and van de Ven (1996) investigated this aspect and reported on it in at CAPSA (1999). Following the recommendations of a panel of international experts, the Bitumen Specification Task Group agreed on certain changes and rulings: The viscosity at 60ºC be specification limits The viscosity at 135ºC should include upper limit specification

Low temperature ductility was considered important, although the specification of 100cm ductility requirement was considered conservative. The minimum Softening Point value after RTFOT and maximum change value were adopted. At the time of the investigation of van Assen and van de Ven (1996), it was apparent that the South African SABS 307-1995 specification was similar to those of Australia, France, Kenya, Spain, United Kingdom, USA and the European Community (CEN). However, the viscosity requirements of the SABS specification after RTFOT was different from that of the other countries. At that time SABS 307 relied strongly on ductility specifications at low temperature, however, in 2002 the ductility test was removed from the specifications. Other typical properties not included in the SABS 307 specification include Fraass Breaking Point, Density, Wax Content, Solubility and Flash Point. All of these properties can be related to the chemical composition of the bitumen. Surface dressings i.e. seal coats, are the predominant type of surfacing used in South Africa, particularly on rural roads. Specifications for the binders used in these applications need to take account of adhesion, shear susceptibility, durability and setting rate expressed as viscosity at highest service temperature after simulated laboratory ageing. According to the bitumen models, lower asphaltene contents would, by implication, be an indication of higher ratios of smaller molecular mass aromatics present in bitumen that has slower ageing and slower increase in viscosity. The SABS 307 specification takes account of setting or hardening potential of a binder through the increase in Ring and Ball Softening Point after RTFOT. Values that are too high, however, necessitate upper limits that could compromise durability in favour of setting. 3. THE PG CONCEPT LINKED TO THE STANDARD SA SPECIFICATIONS As outlined above, SA bitumen specifications have gone through a significant change in the mid 1990 s. In addition to adding the requirements for properties after the rolling thin film oven ageing, the sensitivity of bitumen to temperature is controlled by specifying limits for properties at 10ºC, 15ºC, 25ºC, 60ºC, 135ºC, and softening point. It is also important to notice that the ranges in penetration values and the ranges in softening point values selected for the grades control the Penetration Index (PI) to a narrow range of values, thus ensuring favourable temperature susceptibility. Such an approach is certainly better than the traditional penetration specifications in which properties are controlled at mainly 25ºC. It is however an indirect method to controlling sensitivity of bitumen to temperature, to ageing and to traffic conditions. It is indirect in the sense that grades are based on the relative change in properties at a given set of temperatures, and, a collection of empirical or index properties are used rather than a fundamental engineering property. It is thus difficult to claim that the current specification is performance based. It can be best described as bitumen-production based. To use the specification, pavement technologists have to choose the best bitumen grade for a given application based on experience and engineering judgment. A potential improvement is the use of a direct approach in which grades are based on actual performance factors at which common criteria are specified using performance related properties, similar to the approach used in the Superpave specifications. The difficulty in the direct approach is the need for directly measured performance related properties such as G* and phase angle. This approach requires expensive and complex equipment that is out of reach of many pavement engineers.

The best compromise, however, is not keeping the indirect (index) specifications but rather using performance grading based on empirically (indirectly) estimated performance related properties. In other words, establishing a performance-grading framework in which grades are truly selected based on application conditions including pavement temperature and traffic. The criteria for accepting bitumen in these grades are based on engineering properties that are performance related, but derived from simple index properties such as penetration, softening points, or viscosity. In this approach we achieve a good compromise in which: Grades are performance based and directly related to pavement and traffic conditions, and The expensive rheological testing equipment is not necessary and could be replaced by numerical procedures for estimating rheological properties from results of simple tests (for unmodified binders). This is not totally a novel approach. In the 1980 s Shell Bitumen company introduced a complete pavement design manual based on starting with using Penetration and Softening point and estimating rheological and damage resistance properties of asphalt mixtures using numerical models as given in the Shell Handbook (1991). In Figure 2 the approach is described. What is novel in this paper is the use of the concept to derive a Bitumen Performance Grading system using the commonly measured index properties. In the following section the approach is used to demonstrate how it could be implemented for South African conditions. 3.1 The Ground Rules for Performance Grading Performance grading should include three basic elements. Should rely on bitumen specific constitutive models. In simple terms, models describing stress-strain relationships under loading conditions experienced in the field. Consider the pavement conditions including temperature, traffic speed, and traffic volume. Should include acceptance limits derived from experience and documented field performance. To achieve the first element it is necessary to define the models that can be used to derive bitumen constitutive models based on Penetration and Softening Point measures. It is well known that such a model exists, namely the van der Poel nomograph, which allows estimating creep stiffness at a range of temperatures and loading times using the PI. The PI is calculated from Penetration and Ring and Ball softening point. Figure 4 shows the nomograph which includes a scale of temperature (T) relative to the softening point, a scale for loading time (t) and a set of contours that allow estimating stiffness (S(T,t)) based on the PI value of the bitumen. To achieve the second element the distribution of pavement temperatures, such as the maximum and minimum design pavement temperatures could be taken from weather data bases in SA and used to define the Performance Grades needed for various regions in the country. Figure 5 shows the distribution of maximum pavement temperatures in various regions and also depicts the range (minimum and maximum) for selected stations in the country. Therefore a set of bitumen grades could be selected. Based on discussions with pavement engineers and experts in SA at a workshop in May of 2003, it appeared that one can work with 3 main high temperature grades: (PG 52, PG 58, PG 64). To consider high traffic volume and/or slow moving traffic, a PG70 grade could be added. To avoid creating an unmanageable number of grades and to simplify the system, it is believed that a 3 high temperature grade system with PG 58, PG 64, and PG70, is a good concise system.

Figure 4. The van der Poel Nomograph. To cover the low temperature range, it can be seen that temperatures range between -5ºC and +8ºC. It is therefore reasonable to use a system of +6, 0.0, and -6.0ºC. For a full factorial system a 3 high temperature and a 4 low temperature grid would result in 12 grades, which is more than double the grades used today in SA. To minimize this multiplicity of grades, a partial system could be used to include 8 grades as shown in rows (1) and (2) of Table 3. Figure 5. Pavement design temperatures for South Africa.

To achieve the third element of a performance grading system, the performance related properties should be selected and the acceptable limits for acceptance of bitumen have to be defined. Following are the properties selected: For workability, the viscosity at 135ºC as used in the current specifications could be used. However, in a performance specification the limits should not be changed; they should remain the same for all grades since contractors are expected to use same practice regardless of the binder source or grade. A range of 0.12 0.65 Pa.s could be recommended based on the ranges listed in the current SA bitumen specifications (SABS 307-1972 Amended in 1997) For Rutting resistance, it is proposed that the penetration and softening point be used to calculate the PI for the grades currently used in SA. The PI values are used to calculate the stiffness modulus at speeds of traffic normally seen in the field. Based on a study by Barksdale and co-workers (Barksdale 1971) it is found that at a typical speed of 60 Km/hr, for a pavement surface layer thickness of 100mm, the loading time is 0.015 seconds. Based on road engineers it appears that the pen 40/50 grade has worked well for moderate traffic volume roads in the moderate climate (PG 58) zone. The average PI for such grade is assumed to be = -0.5 (PI=-0.5) with a softening point of 54 ºC. Using these values in solving the van der poel nomograph indicates that such bitumens give a S(0.15) value of approximately 100 KPa. It could therefore be assumed that for a bitumen to provide sufficient contribution to rutting, the criterion of S(0.15) at maximum pavement design temperature should be equal to or greater than 100 KPa. This can then be used in the new specification table for all grades. For higher grades (64 and 70 ) this stiffness minimum value should be met at 64ºC and at 70ºC respectively. This method has then allowed us to establish the second requirement in Table 3 (see row 4). Also, to consider effect of RTFO aging a separate row is added to require a minimum stiffness that is 2.5 times the unaged condition. Although it could be debated that we do not need the unaged criterion since pavements are never built with unaged bitumen, it is important to measure both since it gives a relative aging performance indicating that bitumen will not be too sensitive to short term, high temperature, aging nor insensitive to such aging conditions. Insensitivity to aging has been considered a possible cause of asphalt mixture tenderness. For Fatigue Resistance, it is recommended that average pavement temperatures be used because fatigue is most critical when bitumen is relatively stiff but subsurface layers are not frozen. Also there is more possibility of wet subsurface layers at spring and fall conditions and thus weaker layers. In fatigue we assume that stiffness should be below a certain maximum value so that bitumen can deform repeatedly without damage. To derive stiffness limits, the nomograph published by Shell Research (1991) for estimating the fatigue life of mixtures from the PI of bitumen and mixture stiffness can be used. A typical pavement structure is assumed with strain levels of 1.0 *10-4 mm/mm for stress controlled condition and 5.0 *10-4 for strain controlled conditions. Specifying a minimum fatigue life of 1.0 *10 6 cycles, the required mixture stiffness is approximately 3 * 10 9 Pa for a PI value in the range of (-0.5 to 0.0). Using a volume concentration of 13% bitumen in a typical asphalt mixture, the maximum allowable bitumen stiffness at 0.015 second loading time should be 50,000 Kpa. In this estimate it was assumed that PAV aging results in increasing the softening point by 25 C and thus the softening point was increased by this amount to simulate long term field aging. To insure proper application of this limit the PI value of the aged binder should be less than 0.0. In the specification system shown in Table 3, the S(0.015) is limited and also the PI value for the PAV aged material.

For low temperature cracking, the stiffness at 60 seconds is estimated at temperatures that are 10 C higher than the minimum grade temperature. The shift in temperature is used to offset the effect of short loading time of 60 seconds as used in the Superpave specifications. In addition, the logarithmic creep rate (m (60)) should also be controlled to minimize stress build up in bitumen due to high elasticity. Using the properties of the Pen 150-200 and assuming that the bitumens of this grade performed well in cold climates in SA, the limits of S(60) = 400000 KPa and m(60 ) = 0.300 could be derived. These limits have been derived for application in SA and differ from the 300MPa used by SHRP. Similar to the fatigue requirement the effect of long-term aging was considered by increasing the softening point by 20 points. To control brittleness of bitumen at minimum pavement temperature, the elongation at break estimated from chart developed by Heukelom and co-workers could be used. The minimum strain at break using the chart shown in Figure 6 should be more than 0.02 estimated using a 60 second loading time and the softening point of the PAV aged material. Table 3 includes the proposed performance grading system based on the concept of using index properties to derive engineering criteria that are performance related. The only element missing is the effect of traffic volume and traffic speed. Since for conventional asphalts most of the deformation at high temperatures are viscous and limited elasticity is expected, stiffness could be used as a surrogate for accumulated permanent strain. Figure 6. Estimating elangation at break (F Th de Bats and G van Gooswilligen, 1995. 3.2 Disadvantages of the Proposed Grading System Although the proposed grading system shown in Table 3 appears reasonable and does not depend on using expensive sophisticated testing devices, one should not ignore some of the important shortfalls of such a system. These shortfalls could be resolved by gradual replacement of the estimated engineering properties, such as S(0.015) and λ (60) with actual measured values using rheometers and extensiometers.

Table 3. Proposed performance grading for South Africa. 1 High Temperature Grade (HT) PG 58 PG 64 PG 70 2 Low Temperature Grade (LT) -10-16 -4-10 -16 +2-4 -10 Performance Related Property Performance Criteria For Workability Viscosity, Pa-s 0.12 < η <0.65 For Rutting Resistance Estimated Creep S(.015) >= 100 58 64 70 Stiffness (Unaged) @ HT, KPa Estimated Creep Stiffness (RTFI- Aged), KPa S(.015) >= 250 58 64 70 For Fatigue Resistance Estimated Creep Stiffness (PAV- Aged), Kpa Estimated PI value (PAV-Aged) S(.015) <= 75000 PI > 0.0 For Thermal Cracking Resistance 24 21 30 27 24 39 36 33 Estimated Creep Stiffness (PAV- Aged), KPa Estimated Creep Rate (PAV Aged) Elongation at break from Figure 2 (OAV aged) S (60) <= 400000 0.0-6 6 0-6 12 6 0.0 M(60) >= 0..310 0.0-6 6 0-6 12 6 0.0 λ (60) >= 0.02 0.0-6 6 0-6 12 6 0.0 * For Slow Traffic (10 Km/hr), S(.015) limit should be multiplied by factor of 10 * For High Traffic Volume, S(.015 ) limit should be multiplied by a factor of 10 Note: Assumptions or an order of magnitude (10x) difference between slow versus average speed and heavy versus normal traffic volume. The reasons this will ultimately be needed are the following: Nomographs are only approximate and show general trends rather than accurate values of stiffness or elongation at break. The nomographs are based mostly on conventional unmodified bitumen. It is well know that modified bitumens are more complex rheologically than conventional bitumens and thus these nomographs could be misleading. Accumulation of damage and sensitivity of service life to traffic conditions are not directly considered and using a multiplication factor of 10 to consider speed or volume is only an approximation. Such system will need continuous verification and calibration to establish predictability and build confidence in estimated engineering properties.

A pictorial comparison between the highly advanced Superpave system (Figure 3) and the proposed system of using index properties to estimate engineering properties (Figure 7) is apparent. One can only hope that the move in South Africa would be for direct measurements of engineering properties as shown in the Superpave system. Figure 7. In-direct performance grading systems. 4. CONCLUSIONS AND RECOMMENDATIONS It is apparent that there are tangible benefits in using a performance grading (PG) system for binders compared with solely empirical testing and reliance on experience for binder selection. The PG system offers: A systematic approach to taking account of climate, traffic speed and degree of traffic loading in the area that the binder is to be used thus minimising subjective and non-optimal binder selection, and A potentially unified approach for adjudicating binders regardless of whether they are unmodified or modified (although more developmental work is required before the Superpave PG system is fully functional in this respect). It is possible to establish an indirect PG system without the use of costly binder testing equipment e.g. rheometers etc. However, there are differences between a direct PG system e.g. Superpave, and indirect PG system as proposed in this paper. Certain assumptions need to be made to develop the proposed indirect grading system with regard to creep stiffness and elongation at break. It is recommended that actual measured values for such engineering properties be obtained for the relevant binders using rheometers. The indirect PG system is based on the assumption that so called S-type pen grade bitumen is used. Other bitumen types can not be used in this system without additional testing. The climatic regions in South Africa used for development of this grading system required certain approximations in converting to the Superpave temperature classification system for the proposed South African PG system, see examples in Figure 5. With more data, these

temperature ranges should be refined and updated. Ongoing verification and calibration of the entire system would be necessary to establish predictability and build confidence in estimated engineering properties. Important aspect from the performance related approach is that it forces developers of the specifications and the users to better understand the link between material and performance. 5. REFERENCES Shell Bitumen U.K., 1991. The Shell Bitumen Handbook. Riversdell House, Surrey, Reprint of Septemebr 1991. TC336WG1-TG5 Paving grade bitumens. 2003. Development of European specifications: test methods selection and development. CEN/TC 336 N 49, 2003. New framework of specification Synthesis of stakeholders needs and expectations. van de Ven MFC and van Assen EJ. 1999. South African Bitumen Specifications and Compositional Balance: An International View. 9 th Conference on Asphalt Pavements for Southern Africa CAPSA 99. van Assen EJ and van de Ven, 1996. Review of South African Bitumen Specification to take cognisance of compositional balance relative to long term behaviour. Report No. CR-96/034, CSIR, Pretoria. Roberts, F. L, Kandhal P.S., Brown, E.R., at al. Hot Mix Asphalt Materials, Mixture Design, and Construction, NAPA Educational Foundation, Second Edition, Maryland, USA, 1996. J.Y. Welborn, and W.J. Halstead, Testing of Asphalts and Asphalt Mixtures, Proceedings of the Association of Asphalt Paving Technologists, Vol.43A, 1974. Anderson, D.A., Christensen, D.W., Bahia, H.U. Physical Properties of Asphalt Cements and the Development of Performance-Related Specifications, Paper presented at the 1991 AAPT meeting in Seattle, Washington March 2-4, 1991. Heukelom, WJ, An Improved Method of Characterizing Asphaltic Bitumens with the Aid of their Mechanical Responses, Proceedings of the Association of Asphalt Paving Technologists, Vol. 42, pp.67-98 (1973). Van der Poel, C., A General System Describing the Visco-Elastic Properties of Bitumens and Its Relation to Routine Test Data, Journal of Applied Chemistry,Vol. 4,1954,221. F Th de Bats and G van Gooswilligen, Practical Rheological Characterisation of Paving Grdae Bitumens, Shell Research, Amserdam, 1995. Braksdale, R.G., 1971, Compressive Stress Pulse Times in Flexibel Pavements for Use in Dynamic Testing, Highway Research Record 345, Highway Research Board, pp 32-44.