Models for rolling resistance In Road Infrastructure Asset Management systems

Transcription

1 MIRIAM 2 Models for rolling resistance In Road Infrastructure Asset Management systems Rolling Resistance Measurement Methods for Studies of Road Surface Effects Authors: Ulf Sandberg, Swedish National Road and Transport Research Institute (VTI) Anneleen Bergiers, Belgian Road Research Centre (BRRC) Jerzy A. Ejsmont, Technical University of Gdansk (TUG) Luc Goubert, Belgian Road Research Centre (BRRC) Marek Zöller, The Federal Highway Research Institute (BASt) Deliverable # 2 in MIRIAM SP Document type and No. Sub-project Author(s) Authors' affiliations (acronyms) Contact data for main author Report MIRIAM_SP1_02 SP 1 Measurement methods and source models Ulf Sandberg (a), Anneleen Bergiers (b), Jerzy A. Ejsmont (c), Luc Goubert (b), Marek Zöller (d) (a) VTI, (b) BRRC, (c) TUG, d (BASt) ulf.sandberg@vti.se Document status and date Deliverable Version Dissemination level File Name Public MIRIAM_SP1_Measur_Methods_Report

2 Rolling Resistance Measurement Methods for Studies of Road Surface Effects / Report MIRIAM_SP1_02 Main author: Ulf Sandberg. Swedish Road and Transport Research Institute (VTI) II

3 Foreword MIRIAM, an acronym for "Models for rolling resistance In Road Infrastructure Asset Management systems", is a project started by twelve partners from Europe and USA. They have collectively contributed internal and external funding for this project. The managing partner is the Danish Road Institute. The overall purpose of MIRIAM is to provide information useful for achieving a sustainable and environmentally friendly road infrastructure. In this project, the focus is on reducing the energy consumption due to the tyre/road interaction, by selection of pavements with lower rolling resistance and hence lowering CO 2 emissions and increasing energy efficiency. MIRIAM has been divided into five sub-projects (SP). The work reported here has been made within SP 1 "Measurement methods and surface properties model". A first phase of the project has included investigation of pavement characteristics, energy efficiency, modelling, and raising awareness of the project in order to secure economical and political support for a second phase. The second phase will focus on development and implementation of CO 2 controlling models into the road infrastructure asset management systems. The website of MIRIAM is where extensive project information can be found. The order of authors on the title page, following the main author Ulf Sandberg, is alphabetical and is not related with the extent or importance of the co-authors' contributions. This report is the second Deliverable of SP 1. The Deliverables of Phase 1 are the following: Deliverable 1: Rolling Resistance Basic Information and State-of-the-Art on Measurement methods Deliverable 2: "Rolling Resistance Measurement Methods for Studies of Road Surface Effects" Deliverable 3: Comparison of Rolling Resistance Measuring Equipment - Pilot Study" Deliverable 4: Road surface influence on tyre/road rolling resistance" These are all represented by written reports. See the MIRIAM website to download the reports, or to check where the reports may be downloaded. Rolling Resistance Measurement Methods for Studies of Road Surface Effects / Report MIRIAM_SP1_02 Main author: Ulf Sandberg. Swedish Road and Transport Research Institute (VTI) III

4 Acknowledgements and disclaimer It is gratefully acknowledged that the studies reported here and the production of this report have been funded by the following organizations (in alphabetical order only): Belgian Road Research Centre (BRRC) Pooled funds of project MIRIAM Swedish National Road and Transport Research Institute (VTI) Swedish Transport Administration (STA) Technical University of Gdansk (TUG), Gdansk, Poland The Federal Highway Research Institute (BASt) The funding organizations have no responsibility for the contents of this report. Only the authors are responsible for the contents. Any views expressed are views of the authors only. Rolling Resistance Measurement Methods for Studies of Road Surface Effects / Report MIRIAM_SP1_02 Main author: Ulf Sandberg. Swedish Road and Transport Research Institute (VTI) IV

5 Table of contents SUMMARY IX 1 INTRODUCTION 1 2 PURPOSE, LIMITATIONS AND CONCEPT 2 3 MEASURES RELATED TO ROLLING RESISTANCE OF VEHICLES Direct measures of rolling resistance Indirect measures - Road surface parameters influencing rolling resistance useful for prediction purposes 4 4 DISCUSSION ABOUT MEASURING VERSUS PREDICTING ROLLING RESISTANCE Availability of predicting parameters Prediction of rolling resistance Measuring rolling resistance but predicting unevenness influence Measuring rolling resistance versus predicting it 6 5 MEASUREMENT METHODS General A few historical remarks Existing standards Laboratory drum Trailer method Coastdown Steady State Wheel Torque Fuel consumption Fuel consumption versus road parameter mapping Overall evaluation and recommended method 14 6 EXISTING TRAILER EQUIPMENT General BASt Test vehicle (trailer) The BASt measurement method Calibration procedure Introduction Static calibration Introduction Operating point Functional correlation between longitudinal force and difference of height BRRC Test vehicle (trailer) The BRRC measurement method Calibration procedure TUG Test vehicle (trailer) The TUG measurement method Rolling Resistance Measurement Methods for Studies of Road Surface Effects / Report MIRIAM_SP1_02 V Main author: Ulf Sandberg. Swedish Road and Transport Research Institute (VTI)

6 6.4.3 Calibration procedure Other trailers described in earlier MIRIAM report Other trailers supplement 27 7 REPEATABILITY AND REPRODUCIBILITY Short-term repeatability according to TUG repeated measurements in Sweden Belgian measurements in project Artesis Short- and long-term repeatability Calibration issue for the BRRC trailer Short-term (within-day) repeatability according to the MIRIAM RRT project Introduction Results for the BASt trailer Results for the BRRC trailer Results for the TUG trailer Comparison of the trailers Short-term (day-to-day) repeatability according to the MIRIAM RRT project Introduction Results for the BASt trailer Results for the BRRC trailer Results for the TUG trailer Reproducibility according to the MIRIAM RRT project Introduction Comparison of results obtained with the BASt and TUG trailers Comparison of results obtained with the BRRC and TUG trailers Trailer-related differences 43 8 INFLUENCE OF TESTING PARAMETERS Tyre Tyre condition (quality, wear, rubber hardness) Tyre load Artesis project - background Artesis project effect of tyre load Tyre inflation Tyre warmup Speed Temperature Road curvature, ruts, slope and cross-slope The influence of wind shielding 53 9 REFERENCE TYRES Introduction Reference tyres for noise measurements SRTT and AAV SRTT and AAV4 as rolling resistance reference tyres? IDEAS FOR FURTHER WORK Fuel consumption versus road parameter mapping New Zealand steady state wheel torque method 57 Rolling Resistance Measurement Methods for Studies of Road Surface Effects / Report MIRIAM_SP1_02 Main author: Ulf Sandberg. Swedish Road and Transport Research Institute (VTI) VI

7 10.3 Further studies of the coastdown method Drum method Improvement of the BRRC trailer Use of reference surface Use of megatexture and enveloped MPD parameters Study how orientated textures affect rolling resistance Study the effect of drum curvature on rolling resistance measurements compared to testing on a flat surface FURTHER WORK IN MIRIAM: MAJOR PROBLEMS TO SOLVE Background Unevenness Heavy vehicles Conduct a new RRT, including the trailer as well as coastdown method and use at least one heavy vehicle in coastdown 11.5 Reproducibility and stability of calibrations Stability of reference tyres Enveloped MPD Temperature correction International standard measurement method RECOMMENDATIONS REGARDING EQUIPMENT AND OPERATION CONCLUSIONS REFERENCES Rolling Resistance Measurement Methods for Studies of Road Surface Effects / Report MIRIAM_SP1_02 Main author: Ulf Sandberg. Swedish Road and Transport Research Institute (VTI) VII

8 Rolling Resistance Measurement Methods for Studies of Road Surface Effects / Report MIRIAM_SP1_02 Main author: Ulf Sandberg. Swedish Road and Transport Research Institute (VTI) VIII

9 SUMMARY MIRIAM has established a sub-project (SP), designated SP 1, to deal with measurement methods for rolling resistance and related issues. This subject forms the most fundamental basis for the MIRIAM ambition to consider rolling resistance in pavement management or other types of infrastructure systems. Without robust measurement methods and equipment that can use them there will be no reliable data as input to such systems and the end result will be most uncertain, if useful at all. This report is intended to provide basic knowledge about measurement methods which relate to the road surface influence on rolling resistance, as well as to provide some information about equipment that are useful for collecting rolling resistance data. The methods used were literature studies, compiling experience from recent research projects (not yet reported), using other Deliverables in MIRIAM SP 1 and, above all; a rather extensive field experiment within MIRIAM called the round robin test. The work in MIRIAM is the first time one has looked at rolling resistance measurement methods applicable to road surface effects systematically, and evaluated various measurement equipment, both in some detail and from an overall point of view. As an overall conclusion it may be stated that rolling resistance measurement methods are still in their infancy. This applies to testing of tyres, where the many standards produced so far totally miss that the rolling resistance is a tyre/road interaction and therefore have serious drawbacks. This also applies to testing road surfaces where the tyre/road interaction has indeed been a major concern from the beginning, but where field testing appears to be extremely difficult. In the MIRIAM round robin test (RRT), conducted in June 2011, the results of using three trailers for measurement of the rolling resistance of 11 test sections on the IFSTTAR test track in Nantes have been analyzed. This experiment is the subject of a separate report, but many of the results are used in this report. It is the major source of information for this report. However, before the RRT was conducted the partners involved in SP 1 met and discussed the major measurement issues; including what equipment to use, and test parameters; such as tyre load, inflation pressure, test speeds, etc. This was a first attempt of "standardization". In the RRT, essentially, the trailer method was used, with some supplements by the drum method. Two drum facilities and three trailers have been available. The trailers came from BASt, BRRC and TUG. In addition, BRRC made texture measurements, and IFSTTAR supplemented with some other data, as well as taking care of practical matters. Comparison of all the rolling resistance equipment has shown partly inconsistent and questionable results. While repeatability within the same day was shown to be acceptable to excellent for the three trailers (between 1 and 3 % of the actual C r ), repeatability between consecutive days was unacceptably poor (between 7 and 25 % of the actual C r ). The latter did not include the TUG trailer which did not conduct such tests. Reproducibility was found to be totally unacceptable, with large differences between the trailers (up to 40 %). Only one of the trailers showed consistent measurement results with respect to repetitions, different speeds and ranking of surfaces with different tyres. Despite the partly good, partly poor results, the RRT was conducted in a careful and successful way, given the time and economic resources that were available. All test crews performed their tasks with great ambition and competence. The weather conditions were also favourable. Rolling Resistance Measurement Methods for Studies of Road Surface Effects / Report MIRIAM_SP1_02 Main author: Ulf Sandberg. Swedish Road and Transport Research Institute (VTI) IX

10 Some of the reasons for the problems are identified and can be corrected (such as the lack of cover over the test tyre for one trailer). Other error sources are identified, such as temperature variations, but it is not yet known what the most relevant temperature correction is. Also, some of the test sections were a little too short; especially for the testing by the BASt trailer. It must be concluded that one of the most serious problems is the instability from day to day of calibrations and this must be studied much more. The tyres used seem to have been well chosen. They performed approximately as expected and the SRTT and AAV4 tyres so far seem to be useful as reference tyres in the next few years; at least they may serve as such until better tyres are identified. Whereas Phase One provided fair possibilities to study rolling resistance of car tyres, there was very little possibility to study measurements with heavy vehicles and their tyres. As the latter are the sources for a major part of the transportation fuel consumption and air pollution (perhaps up to half of it), this means that Phase One of MIRIAM SP1 had resources to address only a limited part of the measurement problem. This report includes comprehensive lists of research needs, both in general and within the next Phase of MIRIAM. Phase One of MIRIAM gave the partners a lot more knowledge about rolling resistance measurements, which will be most useful in further work. The work and its result have created a good platform from which further steps can be taken in Phase Two of MIRIAM. This report should provide major progress towards the goal of producing a useful measurement method and standard. Rolling Resistance Measurement Methods for Studies of Road Surface Effects / Report MIRIAM_SP1_02 Main author: Ulf Sandberg. Swedish Road and Transport Research Institute (VTI) X

11 1 INTRODUCTION Other parts of project MIRIAM have demonstrated that tyre/road rolling resistance contribution by road surfaces are influenced substantially by texture and unevenness and maybe other fundamental road surface properties as well. Despite the influence this has on energy consumption and CO 2 emissions, rolling resistance is hardly ever given any importance in road management, whereas the tyre contribution is given a lot of attention; even regulations. This is due to a general ignorance about the effect of road surfaces on rolling resistance which, in turn, is due to the lack of practical measurement methods by which one could collect data. Standard rolling resistance measurement methods are indeed available but only with respect to testing of the tyres. These are measured on laboratory drums using ISO and SAE methods but taking these methods out on the road is virtually impossible. In order to develop and study measurement methods, there must be a basic understanding of the influencing parameters as well as what energy losses that should be included in the concept of rolling resistance. A report, which intended to provide basic knowledge about the influence on rolling resistance of various parameters, suggest a definition of rolling resistance and provide some detailed state-of-the-art knowledge about the measurement methods and equipment that are useful for collecting rolling resistance data, is already published within project MIRIAM [Sandberg (ed), 2011]. In parallel to this work, another Deliverable in MIRIAM has explored the road surface influence on rolling resistance; the work of which has several issues in common with measurement methods [Sandberg et al, 2011]. This present report focuses on the measurement methods suitable to determine the road surface influence on rolling resistance and attempts to bring the subject to a new level of development. It explores the major measurement problems and suggests how to solve or get around them, recommends the most suitable method and finally presents a first draft for a standard for such a method. 1

12 2 PURPOSE, LIMITATIONS AND CONCEPT The overall purpose of project MIRIAM is to study the potential for saving energy and CO 2 emissions by adding rolling resistance data in road surface management systems. The particular purpose of this report is to provide basic and up-to-date knowledge about the measurement methods which are suitable to measure the influence of road surfaces on vehicle rolling resistance. The concept behind this report is the following: Rolling resistance is one of the most important functional properties of road pavements, applicable to the entire road network, which means that road authorities need to have information about it and to be able to control it The direct measurement of rolling resistance is very difficult and requires the use of rather advanced equipment and methodology, operated by very skilled and experienced staff. Consequently, direct measurement of rolling resistance is possible only on a very small part of the road network Measurement methods and equipment for rolling resistance related to road surface influence are underdeveloped and need further development, testing and validation. Especially, development and testing must focus on a number of difficult-to-control influencing parameters (temperature is one of them) as well as how to make measurements relevant for heavy vehicles without having to use heavy vehicles. Also selection, availability and stability of reference tyres are crucial problems. With regard to limitations, it is important to note the following: Rolling resistance is an interaction between tyre and road; although for the purpose of serving MIRIAM, this project has its focus on the road surface contribution and deals with tyres only as reference tyres for measuring purposes Although it contributes significantly to driving resistance, air resistance of the tyres is not a parameter intended to be included in the relations studied here as it is not a road-related property. 2

13 3 MEASURES RELATED TO ROLLING RESISTANCE OF VEHICLES 3.1 Direct measures of rolling resistance The most relevant measure of rolling resistance, as well as practical to use, seems to be something based on the force (F rr ) that is required to move the rolling tyre in the desired direction. However, it is clear that this force will depend on the load that is applied to the wheel and thus the tyre. Studies have found an approximately linear relation between rolling resistance and wheel load, represented by a force F z (equal to the mass m of the load times the gravitational constant g). Since wheel loads can vary under different conditions, a near-constant coefficient, rolling resistance coefficient, RRC or Cr (both terms are commonly used), has been created to represent the characteristic of tyre/road rolling resistance: the dimensionless ratio of rolling resistance to wheel load: RRC = Cr = F rr /F z. ISO uses the notion Cr. Thus: Rolling resistance coefficient (RRC) = Cr = F rr /F z where the forces F rr and F z (see the preceding paragraphs) are magnitudes and not vectors. This coefficient in turn depends on several tyre and road surface parameters as well as (to some extent) the vehicle speed. It is important to note that the RRC or Cr is a relative measure. It can be used to compare tyres and pavements. Typical values of Cr for tyres in new condition are in the range to for passenger car tyres, and to for heavy truck tyres [FEHRL, ][FEHRL, ]. It is substantially lower for heavy vehicle tyres than for light vehicle tyres, not the least due to the higher inflation. However, the problem is that rolling resistance of a full vehicle is not identical to that of the tyre/road interaction (times the number of tyres), and its measurement includes or may not include some components which are arguable. An attempt to analyze these various contributions to driving and rolling resistance of a driving vehicle and a logical terminology was made in [Sandberg (ed), 2011]. A graph from that report is reproduced in Figure 3.1. It is known from [Sandberg et al, 2011] that the road surface causes not only energy losses in the tyres but also in the suspension system which provides the link between the tyres and the vehicle chassis. To some extent, losses occur also in parts of the transmission and in the wheel bearings, but the road surface does not have any obvious influence on those. It is important to remember that when it comes to the road surface influence of vehicle rolling resistance one must include both the tyre and the suspension losses. These two contributions may be measured as one integrated "vehicle rolling resistance" parameter or as a pure "tyre rolling resistance" parameter plus a suspension contribution by two different measuring systems. For wheel bearings and transmissions in good condition (which it is assumed that any measuring system should be) present knowledge suggests that one may neglect these contributions as they do not change the relative influence of the road surface on the measured resistance. It follows that the rest of this report is focused on tyre rolling resistance and the suspension losses related to the road surface. 3

14 (Vehicle) Propulsion resistance Inertial resistance Gravitational resistance Engine resistance Auxiliary equipment resistance (Vehicle) Driving resistance (Vehicle) Aerodynamic resistance Body air resistance Tyre air resistance Vehicle rolling resistance Tyre/road rolling resistance Bearing resistance Transmission resist. (churning & mech.) Suspension resistance Level 1 Level 2 Level 3 Figure 3.1: Illustration of the suggested terminology structure (only the upper 3 levels). See further explanations in [Sandberg (ed), 2011]. 3.2 Indirect measures - Road surface parameters influencing rolling resistance useful for prediction purposes At least at present and in the foreseeable future, the direct measurement of rolling resistance is very difficult and requires the use of rather advanced equipment and methodology, operated by very skilled and experienced staff. Consequently, direct measurement of rolling resistance is practical only on a very small part of the road network and for research purposes. For larger networks, an alternative possibility is to predict rolling resistance from other road surface parameters; especially from such parameters that are often collected routinely by road surveying vehicles. Such a method would require as a minimum a collection of the Mean Profile Depth (MPD) measured according to ISO , and the International (Road) Roughness Index (IRI) measured according to the international standard ASTM E Once these parameters have been measured one may use the source model suggested in [Sandberg et al, 2011] to estimate the Cr values of road surfaces, with or without including also the suspension losses. However, the present report will deal mainly with the direct measurement of rolling resistance. 4

15 4 DISCUSSION ABOUT MEASURING VERSUS PREDICTING ROLLING RESISTANCE 4.1 Availability of predicting parameters MPD and IRI are probably the two most commonly and widely measured parameters by the kind of road surface surveying vehicles that are used by most national and/or regional road authorities in industrialized countries in order to collect data for their pavement management systems. Some countries collect them year by year; some do it at less frequent intervals. MPD is often collected both in wheel tracks and between wheel tracks. Therefore, the MPD and IRI parameters are suitable for use in a prediction scheme for rolling resistance, since they would allow, in principle, a nationwide estimation of rolling resistance properties. Other parameters that are widely collected include rut depth (transverse profile), megatexture and skid resistance. 4.2 Prediction of rolling resistance In another MIRIAM report it is suggested that Cr may be predicted as follows [Sandberg et al, 2011]. The best source model for the road surface influence is currently proposed to be: Cr = Rolling resistance coefficient = C MPD + X IRI where MPD is Mean Profile Depth in mm, measured according to ISO and X is a constant yet to be determined and "C0" is a constant unique to a certain tyre and several other circumstances; usually in the range to A reasonable "best average" may be C0 can be considered to be the average (for the test tyres) Cr when run on a completely smooth texture (which should never exist on roads but is the ISO standard surface on drum facilities). In this case the equation would be: Cr = MPD + X IRI This simple model is useful over a speed range of at least km/h for the rolling resistance part of the driving resistance. Suspension losses are included only if the IRI term above is specified by assigning a number to its constant "X". The equation is based on light vehicle data. For heavy vehicles, one may use the same model, scaled to representative values of C r for heavy vehicle tyres, as long as no better model is available, but one must be aware that it is still very uncertain for this category. 4.3 Measuring rolling resistance but predicting unevenness influence Present practical measurement equipment, i.e. rolling resistance trailers, seem to be taking road unevenness effects (IRI) into no or questionable account [Sandberg et al, 2011]. At least the TUG trailer seems to be insensitive to at least low and medium unevenness. To measure the unevenness effect on rolling resistance separately is presently very difficult, with no apparent equipment available for this particular purpose. Until such equipment is available, a possibility that should be considered is a "hybrid solution" consisting of measuring the Cr by trailers but adding the IRI contribution by a prediction. This of course is possible only when the "X" in the equation above has been determined; something which should be a primary objective of the next phase of MIRIAM. 5

16 4.4 Measuring rolling resistance versus predicting it Direct measurement of Cr is associated with several uncertainty factors, such as temperature, poor reproducibility, questionable stability of test tyres, etc. It can be made only to a limited extent as driving around and measuring Cr on an extensive road network is very time consuming and expensive, and exposes the equipment and tyres to substantial wear. Prediction of Cr based on MPD and IRI has the problem of the correctness and representativity of the model on which the prediction is based; also on the uncertainties of the collected MPD and IRI values. On the other hand, as MPD and IRI values are probably available for a very extensive road network, estimation of rolling resistance for such a large network can be made. Weather and temperature will have no effect. Instead of doing this on-the-road it can be calculated easily, which is inexpensive and free from the hazards when doing road measurements. It should be obvious that for road management purposes the latter option is the preferable one. However, until the source models are complete and of high quality one will have to make direct measurements of rolling resistance. Even when the models are considered satisfactory, they must be validated now and then, and new surface types may come up which require further studies. Therefore, a limited extent of direct rolling resistance measurements will always be necessary. However, these measurements must be of high quality in all respects if they are going to serve as validation or calibration of prediction methods. The situation is somewhat similar with respect to environmental noise immission, where prediction methods are widely used as a rule, and more or less nationwide, but where direct noise measurements are also conducted, but to a very limited extent. 6

17 5 MEASUREMENT METHODS 5.1 General Measurement methods and equipment for rolling resistance are the primary focus of SP 1 of MIRIAM. The methods which are available can be grouped into five general categories: Laboratory drum (DR) method: Laboratory measurements made with test tyres rotating on drums. The drums may be equipped with sandpaper or replica road surfaces, apart from the "normal" steel surface of the drum. Trailer (TR) method: Measurements with test tyres in a special towed trailer rolling at constant speed. The trailer may be designed either for passenger car tyres or for heavy truck tyres. The Steady State Wheel Torque (WT) method: A test vehicle (car or truck) is driven at constant speeds km/h while the driving torque of one (driven) tyre, together with the relative wind speed and direction are continuously measured. The driving torque is divided by the dynamic tyre radius and corrected for ambient wind to obtain the driving force required to overcome all resistive forces with the exception of driveline losses. Coastdown (CD) method: Coastdown measurements are using any type of vehicle, measuring deceleration and a number of other parameters, from which rolling resistance may be calculated. The test vehicle is coasted (gearbox in neutral) from a higher to a lower speed. Fuel consumption (FC) method: Fuel consumption measurements are using especially instrumented (normal) vehicles, from which rolling resistance may be calculated by means of a fuel consumption/rolling resistance model. A variant of this would be when using a vehicle not powered by fuel, such as an electric vehicle. Then the method should be called Energy Consumption (EC) method instead. Note that driver influence might be high in this method. These methods can be supplemented by additional measurements of e.g. suspension forces. However it is necessary to realize that the choice of the measurement method implies also some assumptions and limitations concerning the model derived from the results. The main challenge in most cases consists in the difficulty of isolating the effects of rolling resistance from other contributors to the total driving resistance and fuel consumption. A State-of-the-Art report about rolling resistance measurements is produced by [Sandberg (ed), 2011]. This will give more detailed information than this chapter. This chapter is focused on measurement methods applied for measurement of road surface influence on rolling resistance. Tyre testing is treated only very briefly. 5.2 A few historical remarks Rolling resistance measurements on tyres were introduced essentially in the 1970's, partly pushed by the "oil crisis"; most commonly made with the drum method. Maybe a few were made even earlier than that. For example, a measurement in 1928 is reported in [Agg, 1928]; see further [Sandberg (ed), 2011]. In most of the early measurements (more than 30 years ago) the test objects were tyres rather than pavements. An example when it was realized that also the road surface was an important factor is reported in [Williams, 1981]. Work at Dunlop Tyres in the U.K. to reduce rolling resistance of tyres around 1980 involved the use of a drum facility on which a number of tyres were tested on a number of replica road surfaces. Figure 5.1 shows some of the replica surfaces at the Dunlop facility. The results of these efforts are summarized in [Sandberg et al, 2011]. 7

18 Figure 5.1: Sets of replica road surface segments for mounting on a drum. Photo by the main author, at the Dunlop Tyres Development Centre, in Fort Dunlop, Birmingham, U.K., in the 1980's (by permission). 5.3 Existing standards There are a number of international or industry standards intended for measurement of rolling resistance of tyres. No standard or common practice has been published for measurement of rolling resistance properties of pavements. All standards so far are based on laboratory drum measurements. The existing test standards include two set by the Society of Automotive Engineers (SAE), SAE J1269 and SAE J2452, which may be seen as industry standards, and two set by the International Organization for Standardization (ISO), ISO 18164:2005 and ISO 28580:2009. These standards are used extensively in the tyre and automotive industries for rolling resistance measurement. Table 5.1 compares the available standards which are all based on the laboratory drum principle. In addition, the United Nations Economic Commission for Europe (ECE) has issued a Regulation, ECE R117, which includes a measurement method for rolling resistance. This is not unlike the ISO standards and is expected to be revised for greater harmonization. These standards may be used for testing the influence of road surface on rolling resistance provided the drums are equipped with relevant road surfaces. Normally, it is not possible or at least not safe to construct normal road surfaces on drums as centrifugal forces may cause particles to be thrown off. This problem can be solved by producing replicas, in epoxy or similar strong materials, of real road surfaces and mount them on the drum; an example of which is shown in Figure 5.1. An exception is when using a so-called internal drum; i.e. when the tyre is rolled on the interior rather than the exterior periphery of the drum. The German PFF facility is an example of the latter; see [Sandberg (ed), 2011]. 8

19 Table 5.1: SAE and ISO standards for measurement of rolling resistance of tyres. 5.4 Laboratory drum A typical laboratory test for rolling resistance consists of a test drum, a cylinder aligned with the centre of the drum, and a tyre to be tested. The tyre is held against the drum, which is run by a motor coupled to it. The tyre s rolling resistance applies a braking effect to the drum s rotation, and this effect is translated into measurements of forces, torques, decelerations, etc. Rolling resistance is then calculated from these measurements. Figure 5.2 shows the configuration of this procedure [Gent & Walters, 2005]. The Society of Automotive Engineers (SAE) and the International Organization for Standardization (ISO) have both prescribed test standards for this procedure. The standards for rolling resistance testing include four common measurement methods (although not all of them are specified in each standard). These measurement methods include: measurement of the resistive force at the tyre spindle (force method) measurement of the resistive torque on the drum hub (torque method) measurement of the electrical power used by the motor to keep the drum rotating (power method) measurement of deceleration after the driving force at the drum is discontinued (deceleration method). During the measurement of rolling resistance, aerodynamic drag, which can account for up to 15 % of the laboratory measurement of rolling resistance [Gent & Walters, 2005], may need to be measured and subtracted from the result, and this is commonly though not always done. Because the magnitude of aerodynamic drag on the tyre in a laboratory test differs significantly from that on an actual vehicle, the inclusion of aerodynamic drag in laboratory tests makes the results unrepresentative of the real situation. However, because there are 9

20 practical limitations on the measurement of aerodynamic drag in some test methods, such as techniques to remove systematic errors associated with machine offset [Gent & Walters, 2005], some measurement methods (including the torque and power methods) do not mandate that aerodynamic drag be subtracted. Test drum Actuating cylinder Tire to be tested Motor From Gent et al., 2005, pp 516 Figure 5.2: Typical test configuration for tyre rolling resistance. From page 516 in [Gent & Walters, 2005]. In addition to the measurement methods, the number of testing points can also differ, depending on the test condition. A single-point test includes only one setting for tyre pressure and tyre load, while a multi-point test includes a series of settings of tyre pressure and tyre load. Rolling resistance is then calculated from the regression of the multi-point measurements. Different testing standards prescribe different numbers of testing points. SAE J2452, ISO 18164:2005, and ISO 28580:2009 all adopt the concept of energy loss per distance travelled as the definition of rolling resistance, but note that it is equivalent to a drag force with the unit Newton. For measuring the rolling resistance of car tyres, drums having a diameter of 1.7 m are the most common, while for truck tyres, drums with greater diameter (at least 2.0 m) are used. Major facilities for drum measurements of rolling resistance include, but are not limited to: The PFF at the BASt, Bergish-Gladbach, Germany, see [Sandberg (ed), 2011] Two drums at the TUG, Gdansk, Poland, see [Sandberg (ed), 2011] A great number of drums at TÜV-SÜD Automotive, Munich, Germany FAST at Karlsruhe Institute of Technology (KIT), Karlsruhe, Germany Three facilities at Standards Testing Labs (STL), Massillon, OH, USA In addition, most technical centres at the tyre manufacturers have drums for testing rolling resistance Among the above facilities, as far as the authors are aware, the only facilities which are equipped with road surfaces or replicas of such are the ones at BASt and TUG; see further description in [Sandberg (ed), 2011]. 10

21 5.5 Trailer method The trailer method is a method where rolling resistance measurements are made with test tyre(s) in a special towed trailer, while rolling at constant speed. The trailer may be designed either for passenger car tyres or for heavy truck tyres. The measurement may be made either of the torque on the test tyre, the towing force between towing vehicle and trailer, or the angle by which the tyre vertical support is displaced when a rolling resistance force is acting in the tyre/road interface (which is actually a measurement of the rolling resistance force on the tyre). See Figure 5.3 and further Chapter 10 in [Sandberg (ed), 2011]. Figure 5.3: The measuring principle of the BRRC and TUG trailers (simplified). The trailer may be either a single-wheel trailer or a three-wheel trailer. In the latter case the test wheel is in the middle and there are two supporting wheels at the sides. There will normally be some kind of suspension, which may or may not be deactivated during measurement. Trailers of this type are normally designed for being towed by a powerful car or a van. The trailers may or may not have enclosures over the test tyre in order to avoid measuring the air resistance of the test tyre. The air resistance is of course part of the driving resistance of a vehicle and the tyre influences this, but when road surface effects are studied the air resistance is just a disturbing factor. In principle, a heavy vehicle version may be constructed based on the same principle. However, the only known heavy vehicle trailer so far is just a regular full-size four-wheeled (or more) truck trailer towed by a truck with measurement of the towing force in the tow bar or in the hook. In order to avoid the air resistance, the towing speed is very low. The method is sensitive to all kinds of irrelevant influences, such as wind, cross-fall, gradient, and measurements on trafficked roads are not practical. A more advanced truck semitrailer device is under construction in Germany; see Chapter 10.9 in [Sandberg (ed), 2011]. One shall be aware that the rolling resistance force is extremely low in comparison to other forces. Typically it may be % of the vertical force and load on the tyre. If one wants to measure it with an inaccuracy of 1 % of the measured value, one must reduce errors and irrelevant influences to less than % of the tyre load (50 ppm); and this in a driving environment when speed is not exactly constant, where driving direction needs small correc- 11

22 tions all the time and where there may be ambient wind gusts, wind gusts from other vehicles, pavement ruts, varying cross-fall and longitudinal gradients. It goes without saying that this requires the utmost of design and analysis considerations. In some cases, patented technical solutions take care of some of the unwanted influences. Figure 5.4 illustrates this by showing a close-up of the mechanical design of the TUG trailer. A problem with the trailer method is that it is unclear as to what extent road unevenness influences the measurements in a representative way; essentially the part which causes energy losses in the suspension. The wheel suspension may not be typical and it may even be blocked during measurements. Figure 5.4: The central mechanical design of the TUG trailer. Wheel cover partly removed. Photo by Hans Bendtsen, DRI, Denmark, Coastdown A coastdown measurement on a road section is performed by letting a selected test vehicle, equipped with selected test tyres, roll freely (clutch down, gear in neutral position) between defined start and end points. The speed is measured continuously along the road section. The acceleration is either measured directly or is derived from the speed curves. A possible alternative is to measure the speed only at the start and end points. The various resistive forces acting on the vehicle will make it slow down. The rolling resistance is one of these forces. Air resistance is another major force. The larger the rolling resistance the larger the retardation becomes. Normally, measurements are made in both directions on the road, in an attempt to "average out" influences such as gradient and wind, and to get highly varying conditions. Nevertheless, ambient winds may not be too high. By performing several coastdown measurements, under various conditions, it is possible to distinguish and separate the contributions of the different resistances acting on the vehicle. Thus, most of the air resistance can be separated out from rolling resistance and by clever 12

23 measurement and analysis design one can separate a number of other energy losses as well such as in transmission and wheel suspension. The coastdown method is equally sensitive to various influences as the trailer method, since the rolling resistance is such a small component of the total momentary forces (see above). It requires very careful measurement and analysis design, as well as experience and extensive competence. Its advantages are that it may equally well be performed with heavy vehicles as with light vehicles (albeit less practical and more expensive) and that also wheel suspension losses are measured. See further Chapter 11 in [Sandberg (ed), 2011]. 5.7 Steady State Wheel Torque In New Zealand a special measurement method was developed in the late 1980's and early 1990's, called the steady state torque method [Cenek, 1994]. It essentially involves a test vehicle (both a car and a truck have been used) being driven at steady speeds between 20 and 75 km/h. At each speed, the driving torque of one (driven) tyre, together with the relative wind speed and direction are continuously measured. The latter are parts of an on-board anemometry system by which air resistance is controlled. The driving torque is divided by the dynamic tyre radius (1.03 x Static Radius) and corrected for ambient wind to obtain the driving force required to overcome all resistive forces with the exception of driveline losses. It should be noted that this method measures rolling resistance including the contribution by suspension losses. The latter is one of the advantages with the method; another advantage is that it may be used also with a heavy vehicle. It shares the problems of the other methods, but in this case the elimination of the air resistance effect from the rolling resistance effect of the pavement is the major problem. For this reason, high speeds are not useful. 5.8 Fuel consumption An indirect way of measuring rolling resistance is to measure fuel consumption (FC) and estimate the rolling resistance contribution by means of some model. At least this is feasible if one only wants to study the road surface influence, in which case one needs to determine an approximate rolling resistance contribution to FC. Or, one is satisfied with studying the road surface influence on FC. Such measurements have been made several times both in Europe and USA [Sandberg et al, 2011]. In the FC method, a test vehicle is used which is equipped with precision fuel flow meters. Also, vehicle speed must be measured with high precision and ambient air, fuel and engine temperatures should also be measured. Wind speed and direction affect results and should be measured as well. In some cases the air speed at some point near the vehicle may be measured. FC measurements are made between two defined points on the road, in both directions, assuming a road is used with no or small gradients and bends with very high curve radius. This method is sensitive to the condition of all parts of the power unit, apart from all ambient and road parameters; especially wind conditions. The driver's skill to keep a constant speed, or the efficiency of a cruise controller, are also potential error sources. 13

24 5.9 Fuel consumption versus road parameter mapping A rather special method of relating FC to road parameters was first proposed by Mr J Elsander of the Swedish Transport Administration. It is based on measurements made by one or more vehicles equipped with fuel flow meters and travelling over a large and varying road network. Fuel consumption is measured and recorded per (say) km. These FC values per km are then regressed on corresponding road parameters per the same km. Such road parameters may be MPD, megatexture and IRI parameters which should be available for the road network or maybe even collected by the same vehicle at the same time as when making FC measurements. An obvious possibility is to use the road surveying vehicles which collect data for the national and regional networks annually. Of course, factors such as vehicle speed, acceleration, wind speed and direction, road topography, driving style, different climate, etc, influence each measurement point in a way which makes any correlation based on a small number of points insignificant and meaningless to try. However, if such data are collected over several months, perhaps a full year, and for a very large road network, maybe even using more than one vehicle, a multiple regression of FC on (say) MPD, megatexture and IRI may turn out to show interesting relations. The difficulty and danger of such a study would be if there is a systematic bias in the measurements, such as a correlation between MPD or IRI and road type. For example, MPD and IRI may systematically be higher on smaller than on larger roads, in mountain areas than in flatland areas, etc. Drivers may drive slower on rougher roads than on smoother roads. Temperatures may be higher in northern parts of the country than in southern parts, which may co-vary with MPD and IRI. One may imagine several such "uncontrolled" effects which may co-vary with the road parameters. Temperature may in fact be one of the measured parameters and also be included in the multiple regressions. Therefore, one may try to separate the data into boxes, each representing a uniform road type and maybe region. Then one can in some way try to combine the results per each of these boxes into a uniform model. An alternative way is classifying road type and region into some categories and introducing them as test parameters. The method is planned to be tested in Sweden. It is not feasible for studies on project or lower levels Overall evaluation and recommended method In order to be able to recommend one or a few methods for standard measurements of the rolling resistance properties of road surfaces, the methods described above have been evaluated according to ten criteria. Table 5.2 shows the criteria as well as the estimated marks of the five methods (the one in 5.9 is not considered useful for standard measurements). The drum method is a highly arguable method here, but it has been included since, in principle, it is possible to put road surfaces or replicas of road surfaces on drums. Marks are assigned in a scale from (three minus) signs to +++ (three plus) signs; where plus signs mean positive performance. The method which comes out the best is the trailer method. The coastdown method is not far behind but especially its complexity in measurement and required manpower are penalties which gives the trailer method an advantage. Complexity is an important criterion since it is a highly influencing factor on possibilities of making errors and mistakes and thus is related to data quality and risks of collecting poor data. When using high-competence staff such risks may be minimized, but that is expensive and it will not always be done. 14

25 Note that for trailers it is unclear as to what extent they measure the suspension losses; therefore, the cell is blank with a? for this option. Some trailers may be better than others in this respect. Following the evaluation, with present knowledge and experience, the authors recommend using the trailer method for further studies and for first standardization. Table 5.2: Evaluation of the methods suitable for measuring rolling resistance properties of road surfaces. See the text for information. Colour code: green = good, red = bad, yellow = neutral. Method Drum Trailer Coastdown Wheel torque Fuel consumption Purity of rolling resistance measurement Including / not including suspension energy losses Easy to conduct in the field Representativity of real road measurement Resolution Repeatability Reproducibility Cost of equipment Complexity of equipment, measurement and analysis Required time/manpower for measurement & analysis ? / / Overall mark

26 6 EXISTING TRAILER EQUIPMENT 6.1 General The previous chapter concluded that the trailer method presently is the recommended one for measuring the rolling resistance properties of road surfaces. This chapter describes the equipment that may conduct such measurements and which have demonstrated usefulness in this respect. The three trailers presented here have in common that they use single car tyres as test tyres (two of them may use also the smallest light truck tyres) mounted in the middle of the vehicle. Other trailers are mentioned only briefly at the end of the chapter. 6.2 BASt Test vehicle (trailer) The BASt rolling resistance trailer for passenger car tyres applies a separate wheel suspension for the test tyre/wheel combination (see Figures ). This wheel suspension is mounted in the same geometric axis as the supporting tyres of the two-wheeled trailer. For applying the desired tyre load a pneumatic cylinder in combination with a nitrogen reservoir is used. The test tyre/wheel combination is joined at point "B" and "C" with five links (three transversal and two longitudinal links in a parallelogram alignment) to the trailer chassis with a camber angle of 0. The lower longitudinal suspension link is equipped with a force transducer for the longitudinal force. The pressure accumulator based vertical force F Z is passed via a force transducer and a bearing towards point "C". The BASt trailer has been used for measurements since Figure 6.1: The BASt rolling resistance trailer for car tyres. 16

27 Figure 6.2: Principle of the test wheel suspension on the BASt trailer Figure 6.3: The suspension of the BASt trailer The BASt measurement method The method used is illustrated in Chapter ("two point method") in [Sandberg (ed); 2011]. The different tyre loads F Z high and F Z low are obtained by a dual pressure control unit with adjustable pressure values. Two electro-pneumatic valves supply the pressure for the tyre load cylinder. The signal from the force transducers for the forces F X and F Z is sampled at a rate of 500 Hz and the speed and temperature signals are sampled at 5 Hz. The rolling resistance value 17

28 RRC is calculated from the mean values of F X and F Z at high and low load, so there is only one single value for the whole test track length. It is assumed that the rolling resistance value does not change significantly along the test track. Besides the two force transducers for vertical tyre load F Z and longitudinal force F X (= F LU ) the trailer is equipped with a platinum temperature sensor (Pt100) to measure the ambient air temperature either in an enclosure around the test tyre or outside of it, as well as an infrared temperature sensor to measure the tyre temperature at its shoulder. Since there is an unavoidable influence from F Z towards F X if the trailer chassis is not parallel to the road, it is necessary to measure the position of the trailer platform with two laser displacement sensors mounted in the front and rear Calibration procedure Introduction Calibrations are made only in the laboratory before each measurement campaign, or when there is a change in the setup, such as change to a test tyre with different diameter. Calibrations are not made in the field as it would be too complicated and probably not needed. Besides the rolling resistance force R measured by the force transducer K1 in the lower suspension link B-E (Figure 6.2) there are other influences on the measurement value of F X. These may be such as bearing friction, air drag caused by wind turbulences of the tyre/wheel combination, influence on the longitudinal force caused by a suboptimal vertical position of the wheel suspension, electrical and mechanical offset or zero drift of the force transducer used. The general relation for the longitudinal force is: with R F B F O F X R F W F F B rolling resistance force F W air drag and wind turbulence loss bearing friction F XZ influence of vertical load F Z on longitudinal force F X (to be determined by a static calibration) zero drift of force transducer XZ F O Provided that there is a linear relation between tyre load (F Z ) and longitudinal force (F X ) for a certain tyre/road combination, a so-called "two point measurement" (see Chapter Fel! Hittar inte referenskälla.) is used to eliminate the additive parts of F X which are not causally associated to the rolling resistance force R. The two points are called "high" and "low". F F X high X low R F F W W F F B F F B XZ low XZhigh F O F O F X high F X low R F XZ (Eq.1) For a very low tyre load F Z low (~ N) the value of F XZ low can be neglected Static calibration Introduction The operating point of the trailer measurement mechanism (suspension linkage with force transducers); i.e. the position of the trailer in relation to the ground, plus the appropriate error 18

29 of the longitudinal force F XZ, and the functional correlation between longitudinal force and difference of height, are determined by two separate measurements in a so-called "static calibration". The position of the trailer chassis platform is determined by two laser displacement sensors (mounted at the front and at the rear of the trailer chassis) during a rolling resistance measurement and during the static calibration Operating point The "static calibration" is also carried out by using the two point measurement method. For that reason the tested tyre/wheel combination is mounted on the trailer which is placed on flat and even ground of a hall. The tyre is loaded with the same nominal vertical load (F Z high ) and a reduced load (F Z low ) as when operated on the road. This calibration has to be redone for each tyre-wheel combination that has a different diameter compared to the previous one, for each change in vertical tyre load, and for each adaption of the suspension links to the current tyre size used. F F X X high S low S F F XZ XZ high S low S F F O S O S F X high S F X low S F XZ S (Eq.2) with F XZ high S F O S influence on longitudinal force during static condition at vertical load F Z high S zero drift of force transducer during static conditions F XZ low S influence on longitudinal force during static conditions at vertical load F Z low S For a very low tyre load F Z low S (~ N) the value of F XZ low S can be neglected. If the conditions during a road measurement are largely the same as during a static calibration then F XZ S can be equated with F XZ (Eq.1) Functional correlation between longitudinal force and difference of height In order to determine the influence F X high = f( h) of the clearance of the coupling point (between towing vehicle and trailer at a certain operating point of the trailer) to the ground the boot of the towing vehicle is loaded stepwise with weights. For each step of weight the tyre load F Z high is applied. with F X high = f([h front - h rear ] S - [h front - h rear ] C ) = f(δh S - Δh C ) = f( h) (Eq.3) h front Δh S F X high clearance between front of trailer chassis and ground difference of height during static calibration longitudinal force at high vertical tyre load F Z high h rear Δh C h clearance between rear of trailer chassis and ground difference of height during determining f( h) difference between difference of height at the static calibration and difference of height when determining the function f( h) 19

30 With the functional correlation between longitudinal force and difference of height, together with the calibrated operating point, it is possible to correct F X at all subsequent measurements. 6.3 BRRC Test vehicle (trailer) In 2009 BRRC decided to continue research started in the early 1980 s and refurbished the original trailer. New sensors were added and calibration procedures were tweaked. Figure 6.4 shows the new trailer and its towing car. Figure 6.4: The new BRRC trailer and its towing car. The BRRC trailer is designed as a quarter-car with an ordinary car suspension. The suspension dates from the 1980 s and was originally designed for a small car. Technological developments over the last years lead to larger tyres and heavier vehicles. As the trailer of BRRC was originally designed for tyres and cars commonly used in the 1980 s, it now encounters some limitations. Only tyres with a maximum diameter of 14 inches and a maximum width of 195 mm can be mounted. A maximum load of 2000 N is imposed in order not to force the suspension system. In 2011 a new tyre was mounted on the trailer: Michelin Energy Saver 195/70 R14 91T (see Figure 6.5). Until then a slick tyre had been used: Michelin SB-15/63-14X (see Figure 6.5). A tyre inflation pressure of 200 kpa was used for the old tyre, while for the new test tyre a pressure of 220 kpa is used. Figure 6.5: The new BRRC test tyre on the left (Michelin Energy Saver) and the old test tyre (Michelin SB) on the right. 20

31 The trailer presently has no enclosure that can prevent the tyre air drag from affecting the results. This means that speed influences the measured results; i.e. the measured rolling resistance is higher for higher speeds The BRRC measurement method The measuring principle is shown in Figure 5.3. The rolling resistance force causes the wheel to incline backwards with an angle θ with respect to the frame of the trailer. The rolling resistance coefficient is defined as the ratio of the rolling resistance force and the load on the wheel and equals the tangent of θ. For small angles, θ is equal to the rolling resistance coefficient provided it is expressed in radians. A symmetric, friction-free pneumatic damper of bellows damps fluctuations of θ. Different parameters are recorded continuously during measurement: Inclination θ of the wheel carrier with respect to the frame of the trailer (Figure 6.6) Inclination μ of the frame of the trailer with respect to the horizontal plane (Figure 6.6) Inclination α between the trailer and the towing vehicle (Figure 6.6) An external infrared sensor is directed at the sidewall near the shoulder of the tyre to record tyre temperature Speed Acceleration Figure 6.6: The BRRC rolling resistance test trailer and illustration of the three inclination angles measured. A software tool for data acquisition has been developed in LabVIEW. During monitoring, the graphs of the different parameters are shown on the laptop screen. In this way the operator may be notified of possible errors during measurement. All data are registered in a file. Corrections are applied afterwards according to this formula: C r = θ + ε 1 * μ + ε 2 * α where ε 1 and ε 2 are experimentally determined coefficients. So far, only a correction of the inclination α 0 at standstill has been applied and measurements have been performed only at constant speed. In general, the following measurement procedure is applied: 21

32 - Cold tyre inflation is adjusted to 220 kpa. - The height of the trailer with respect to the towing vehicle is measured to determine α 0 at standstill. - A calibration round is made to adjust μ 0. and to eliminate the influence of differences in the car load (e.g. by a different number of passengers). - A test tyre warm-up procedure is carried out, consisting of driving about 15 minutes at approximately 80 km/h. - The test section is measured three times. - Ambient air and road surface temperature are measured. - Corrections of the data are applied in the laboratory using an Excel sheet. - Data are corrected for tyre temperature following this formula [Descornet, 1990]: C r (T) = C r (T 0 ) * e ( (T 0 - T) / T 1 ) where T = 30 C, T 1 = 50 C - Average RRC and corresponding standard deviation are calculated Calibration procedure Two angles in the trailer system influence the measured RRC (where RRC is measured as the angle θ expressed in radians); see Fig These are: The longitudinal inclination μ (slope) The angle α between the shaft of the towing vehicle and the shaft of the trailer Small deviations in these two angles have a significant influence on θ and, therefore, the trailer setup regarding these are checked before every measurement. A deviation of only 0,1 leads to a deviation of 10 % of the RRC and a deviation of 0,1 on μ to an error of 3 %. Figure 6.7: Angles μ and α in the BRRC rolling resistance trailer influencing the measured RRC (which is the angle θ expressed in radians). 22

33 The angle μ is checked every time the trailer is attached to the towing vehicle. This is done by measuring the angle μ while the trailer is run "full circle" around a paved area. The average value of the angle, which is denoted μ 0 at calibration, must then be adjusted to be zero. Normally this procedure is carried out at the BRRC premises in Sterrebeek, before driving to the test sites. The vehicle starts at a given point in front of the BRRC main building and the measurement of μ 0 is started. The vehicle drives around the building over a distance of about 320 m until it arrives again at the starting point where the measurement is stopped. The sensor is readjusted to show zero if the average of μ 0 was found to be outside the interval [-0.1, +0.1 ]. If a recalibration is necessary, it is checked by driving a second round. When there is a non-zero slope μ during a road measurement, it is used as a correction to the RRC value according to the equation given in As the second part of the calibration, the angle α is measured, which can be deduced from height measurements of the frame of the non-moveable part of the trailer with respect to the pavement; see Figures 6.7 and 6.8. The calibration value of α is denoted α 0. Moveable part of frame with load Non-moveable part of frame Figure 6.8: Moveable and non-moveable parts of the BRRC rolling resistance trailer. The towing car and trailer are, therefore, first parked perfectly aligned and on a perfectly horizontal surface, which can be checked with a manual inclinometer. The heights of the nonmoveable part of the trailer frame are measured at the four corners. From the two heights at the left hand side the left hand α 0 is calculated and the right hand α 0 is calculated and determined in an analogue way. The average of the left hand and right hand angle α 0 is then further denoted as the angle α 0. Calculations are done by means of a simple Excel spread sheet. Important is that α 0 is determined with the vehicle loaded in the same way as during the RRC measurement. If the operator is not in the vehicle during the measurements of the heights of the frame, it has to be indicated in the Excel sheet, which will take this into account for the calculation of α 0. If larger distances are driven between measurement locations, α 0 is measured before RRC measurements take place and recalculated to compensate for the loss of load of the consumed fuel. If α during an RRC measurement is found to be different from zero it is used as a correction α to the RRC value according to the equation given in After a series of road measurements, this calibration round is repeated to verify that the angle is still inside the accepted interval around zero. If this is not the case, the road measurement is considered as incorrect and the calibration and measurement have to be redone. 23

34 6.4 TUG Test vehicle (trailer) The Technical University of Gdańsk (TUG) designed and built a test trailer for rolling resistance measurements of passenger car tyres in the late 1990's, but the first "production" measurements were made around The idea with this construction was taken from the original BRRC trailer from the 1980's shown in Fig in [Sandberg (ed), 2011], but TUG developed it further and improved the concept in several ways. Some of the constructions are patented, and it has been improved continuously during the past 10 years. The TUG trailer, called the R 2 trailer, in its condition of 2010 is shown in Figure 6.9. Figure 6.9: The tyre/road rolling resistance measurement trailer from TUG in the shape and condition of The trailer is designed to be towed by a reasonably powerful passenger car. The construction of the trailer is self-supporting (three-wheeler) which means that the trailer may be easily connected/disconnected from the towing car. The front wheels that stabilize the trailer have self-aligning properties. The hydraulic brake system of the trailer is operating on front wheels only and provides efficient braking of the trailer during transportation and tests. Trailer construction assures good stability of the trailer. Independent front suspension, based on double transverse arms was constructed as an adaptation of passenger car suspension. During tests the suspension is blocked by removable bars. Blocking of the suspension ascertains proper leveling of the trailer. The test wheel is supported by a vertical arm (4) that is an important element of the force measuring system (Figure 6.10). The front and rear suspension are connected to the horizontal arm 1. Rotation axis 3 is placed directly in the geometrical centre of the rotation axis of the front wheels. The load is provided by arm 2 that has a common rotation axis with arm 1. The load block (6) is resting on arm 2. 24

35 Suspension element 7 carries the load from arm 2 to arm 1. The rear end of the horizontal arm 1 is connected to the vertical arm 4 which is equipped with the test wheel hub. Undesirable vibrations of the vertical arm 4 that may be induced during tests are suppressed by Foucault currents electromagnetic brake (not shown in the Figure 6.8). Inflation pressure in the test wheel is maintained by remote controlled release valve and pressure sensor. Direction of travel f = P f /F z P f LOAD Balast F z 7 Figure 6.10: The measurement principle of the TUG R 2 rolling resistance trailer. During tests the rolling resistance force acting on the test wheel pulls (deflects) the vertical arm 4. The deflection rate is measured by the laser sensor installed on arm 1 and sending the laser beam towards arm 4. Rolling resistance coefficient is defined as a ratio of rolling resistance force P f and vertical load F Z. The trailer is equipped with a patented compensation device that eliminates influence of factors such as road inclination and longitudinal acceleration that otherwise very substantially would disturb the measurements. The position of arm 1 in relation to the road surface is monitored by two laser sensors. A data logger (DaqBook/200) which is installed in the car receives signals form three laser sensors and two rotation sensors installed in the wheels. All measurements are controlled via a notebook computer (see Figure 6.11). The test wheel and the vertical arm, as well as the inclination compensation system are covered by a protecting enclosure (Figure 6.9). Both theoretical considerations and practical experience show that such an enclosure is necessary to reduce the influence of the air drag on the test results. During measurements the front suspension system is blocked by special bars to ascertain stable position of the trailer frame in relation to the plane of the road. Steel loads that load the test wheel are mounted on the arm (2 in Figure 6.10) that has its own suspension based on a motorcycle spring and damper unit. This suspension unit is in operation both during transportation and testing. 25

36 Figure 6.11: The trailer operator seat with its computer for data collection The TUG measurement method Before each measuring session the trailer is calibrated in the laboratory on a flat, horizontal surface. See below. During the measurements the test tyres are in warmed-up condition (warmed-up at least for 15 minutes, but always long enough to stabilize the inflation pressure). When the tyre is warm to allow testing the pressure is regulated. At least two tests runs at each speed are performed. When possible the number of runs is higher and the runs are made in both road directions. Generally, a measurement distance per run of 400 m or more is required, but measurements over distances as short as 100 m are sometimes performed. Short distances require more runs in order to maintain an acceptable uncertainty. The longitudinal resolution is high and repeatability is excellent, see Figure 7.1, but limitations are made by the very high variations that occur in the longitudinal direction, the most of which are averaged out by the data processing software. The data are analyzed in laboratory Calibration procedure The measuring system of the R 2 trailer is calibrated in the TUG or another laboratory before each measuring campaign. The calibration procedure is illustrated in Figure The trailer is placed on a level, flat, horizontal surface (4). The measuring wheel is replaced by a steel arch (1) that has known radius and very smooth outer surface that rests on a polished steel plate (2). The measuring arm is pulled by the test loads (8, 9) via a textile band (6). The band is supported by a steel bar construction (5, 7). During calibration the relation between the output signal from the laser sensor, representing the measuring position of the test arm, and the load is established. The calibration is repeated for 3 different steel arches (1) that simulate tyres of different radius. This makes it possible to establish the relation between laser sensor output signal and tyre radius. The front supporting wheels of the trailer are also placed on two different spacers to establish the potential influence of road surface rutting. No calibrations are made in the field. 26

37 Figure 6.12: The mechanical setup for calibration of the TUG R 2 trailer. See the text for explanations Other trailers described in earlier MIRIAM report Two other trailers are described in Chapter 10 of [Sandberg (ed), 2011]. These are: Heavy 4-wheeled trailer towed by a flatbed heavy truck owned by IPW Automotive in Hanover, Germany. Mobile test trailer at the Institut für Kraftfahrzeuge (ika), at the RWTH Aachen University, Aachen, Germany. These trailers are still being tried and validated. They are not described further here; instead the reader is referred to the reference given above. 6.6 Other trailers supplement There are a few other trailers for rolling resistance testing of car tyres in Germany and elsewhere which were not described in [Sandberg (ed), 2011] but which may be of interest in this report. The first one is a vehicle owned by "Das Forschungsinstitut für Kraftfahrwesen und Fahrzeugmotoren Stuttgart" (FKFS). It is a rolling resistance trailer designed for car tyres, which is based on the same measuring principle as the BASt trailer: two longitudinal [one of them equipped with a force transducer] and three transverse suspension links for the test tyres suspension. The load is applied by a pneumatic cylinder. The entire trailer is covered by an aerodynamically optimized chamber in order to provide an air-drag-free environment of the testing devices. See Figures The FKFS trailer was built approximately It has been used for both road surface and tyre tests. 27

38 Figure 6.13: The FKFS trailer. The interior is covered by a chamber in order to eliminate air drag of the test tyre which sits inside the trailer. The second one is a vehicle which was made by Opel AG in Rüsselsheim am Main in Germany; see Figure It is a towed axle with a long drawbar and a force transducer at the linkage point to the chassis (a gimbal-mounted bearing). The rolling resistance of two identical tyres was measured. They conducted several rolling resistance measurements on different road surfaces many years ago, but due to shortage of money the project was cancelled and only drum measurements were performed after that. Mr Peter Krehan from Opel did a lot of research on real roads with this equipment and is the author of a chapter about rolling resistance in a German book [Drews et al, 1991]. Figure 6.14: The FKFS trailer, showing the test tyre when a side cover is opened in order to give access to the test equipment. The third one is a vehicle owned by "IPW Automotive" in Hanover, Germany. This is a trailer for car tyres, which is constructed according to the towed axle principle used by Opel, although here it is not inside a front-driven car but is a separate trailer. See Figure It has been used or at least it can be used - for both road surface and tyre tests. 28

39 Figure 6.15: The Opel rolling resistance test vehicle a modified Opel Kadett car. The car does not exist any more. Figure 6.16: The IPW rolling resistance trailer a trailer version of the Opel Kadett car in the previous figure. Finally, another trailer has come to the attention of the authors recently. It is the "NAC Dynamic Friction Tester (DFT)", from Neubert Aero Corp in Clearwater, Florida, USA ( This is an "All-Inclusive Continuous Friction Measuring Equipment (CFME)". See Figure It is claimed to have the ability to "accurately measure Mu (Roll) rolling resistance of the pavement" but no actual measurements have been published so far known to the authors. It is equipped with two small-car-sized tyres and two much smaller ASTM E1551 standard tyres. 29

40 Figure 6.17: The "NAC Dynamic Friction Tester (DFT)", from Neubert Aero Corp in USA, primarily designed for airport friction measurement, but also measuring rolling resistance according to NAC. Picture from a pamphlet from NAC. 30

41 7 REPEATABILITY AND REPRODUCIBILITY 7.1 Short-term repeatability according to TUG repeated measurements in Sweden Repeated measurement runs on a Swedish SMA 0/8 surface by the TUG trailer gave the rolling resistance time history shown in Figure 7.1. The diagram shows four consecutive runs covering approximately a 40 m distance; first two with a tyre having a relatively high Cr, then two runs with a tyre having a relatively low Cr. The starting point was the same for all runs, indicated by a painted line on the road shoulder. The diagram shows that repeatability is excellent; not only between the two pair of runs using the same tyre, but also when comparing the runs of the two tyres. The time history pattern is almost identical in all four cases. The diagram also shows that longitudinal resolution is high. Since the speed is approx 14 m/s, the distance between the markers on the time scale (one per 0,1 s) is only 1.4 m. Thus one can see repeatable features over just 1-2 m longitudinal distance. The major limitations are caused by the very high variations that occur along the longitudinal direction. However, most of the variations can be averaged out by the data processing software. 0,016 Tyre W6c 50km/h 1st run RR coefficient [-] 0,014 0,012 0,010 Tyre W6c 50km/h 2nd run Tyre W3c 50km/h 1st run Tyre W3c 50km/h 2nd run 0,008 0, Time [ms] Figure 7.1: Illustration of repeatability of measurements. The TUG trailer was run two times on a Swedish SMA surface, for two different tyres (W6c and W3c), starting measurements at the same position (within approx. 1 m) and running at the same speed (50 km/h, approx. 14 m/s). On the horizontal scale, 100 ms corresponds to approx. 1.4 m distance. 7.2 Belgian measurements in project Artesis Short- and long-term repeatability In there had been a cooperation project between BRRC and Artesis University College in Antwerp, in which Artesis had access to the BRRC trailer for making rolling 31

42 resistance measurements [De Bie & Hofmans, 2011][Bergiers & Goubert & Vuye, 2012]. Earlier, they had also made coastdown measurements of rolling resistance, using texture equipment and test vehicles from BRRC. Some of the activities dealt with issues important to road surface influence on rolling resistance [Sandberg et al, 2011]. Activities related to measurement methods are reported below. Research performed in the Artesis project about long-time repeatability can be separated into two parts. In Part I, two measurement campaigns on ten test tracks were performed by different researchers. Between these measurements there was a period of eight to eleven months. In Part II, two measurement campaigns were performed by the same researchers on nine other test tracks than in part one. Between these measurement campaigns there was a period of three months. The results of Part I and Part II measurements are shown in Figures All data points, even potential outliers, were included in these graphs and analyses. Part II also included some tests of short-time repeatability (five days between tests). This was performed on the same nine test tracks as the long time repeatability. Before comparing the results a temperature correction was applied, according to [Descornet, 1990], which is further explained in a later section of this report. The result is shown in Figure 7.4. The results of long-time repeatability show that the repeatability of the rolling resistance measurements is good, because the slope of the regression lines approaches 1. The correlation factor of the Part II is good (Figure 7.3). However, the correlation factor of the first part of long-time repeatability is not so good (Figure 7.2). One of the possible causes is that the first measurements were performed by other researchers and communication errors occurred. Another cause can be changes in the road surface over the 8-11 months. Most probably, however, a calibration error was made by one of the two groups of students, as the students were not so experienced. The slope of the short-time repeatability differs more from 1 (Figure 7.4), but is still acceptable. Also, the correlation is good. These conclusions assume that one is looking at the bulk of data rather than at individual measurements. If one looks at individual measurements (points), all three diagrams display very poor repeatability. Figure 7.2: Long-time repeatability, measured 8-11 months apart (Part I) 32

43 Figure 7.3: Long-time repeatability, measured 3 months apart (Part II) Figure 7.4: Short-time repeatability, measured 5 days apart Calibration issue for the BRRC trailer In general it can be concluded that most of the results in the Artesis project are logical, but the values of the measured RRC are higher than expected in comparison with other researches. Normally the RRC measured with this equipment has a value between and 0.02 [Descornet, 1990]. Meanwhile the reason for these high values is already found. It appeared to be a calibration error of the hook, in the form of an error in the angle θ by 0.4. This issue has since then been solved. But it illustrates how easy it is to make substantial errors by just small adjustments. 33

44 7.3 Short-term (within-day) repeatability according to the MIRIAM RRT project Introduction In this section of the Chapter, measurement runs performed on the same test section one after the other within the same day are analyzed. These tests were made within the RRT of MIRIAM SP1. The test sections are listed in Table 7.1. Further descriptions, including surface photos, are presented in [Bergiers et al, 2011]. Table 7.1: Summary of test sections used in the RRT Pavement designation Description M1 Very thin asphalt concrete 0/10, class 1 F Colgrip: Surface dressing, 1/3 bauxite (high skid resistance) L1 Epoxy resin (smooth section) L2 Sand asphalt 0/4 E1 Dense asphalt concrete 0/10 (new) E2 Dense asphalt concrete (old) M2 Very thin asphalt concrete 0/6, class 2 C Surface dressing 0.8/1.5 A Surface dressing 8/10 A Porous asphalt concrete 0/6 N Porous cement concrete Results for the BASt trailer BASt repeated measurements on test sections M1 and L2 with the AAV4, ES16 and SRTT tyres at 50 and 80 km/h. For each combination of tyre, speed and direction, several measurements of C r were carried out and the average and the standard deviation were calculated. The standard deviation was divided by the mean value and expressed as a percentage of this. The mean value of these percentages is then calculated; see the results in Table 7.2, which are expressed: - for all combinations - for each direction - for test sections M1 and L2 separately - for speeds 50 and 80 km/h - for each tyre separately The overall short-term repeatability of the BASt trailer was found to be 2.6 %, which appears to be independent of tyre and surface. 34

45 Direction Test section east 2.3 % west 3.0 % M1 2.6 % L2 2.6 % Table 7.2: Average value of the relative standard deviations for various tested combinations, as obtained with the BASt trailer. Speed 50 km/h 2.2 % 80 km/h 3.1 % Results for the BRRC trailer AAV4 2.6 % Tyre ES % SRTT 2.9 % All 2.6 % Measurements on test sections M1 and L2 were repeated by BRRC with the ES14 tyre at 50 and 80 km/h. Eight runs were made for each direction (west and east). The same average value as in the previous section is calculated for all cases; see results in Table 7.3. Direction east 3.5 % west 2.0 % Test section M1 3.1 % L2 2.3 % Speed 50 km/h 2.1 % 80 km/h 3.3 % All 2.7 % Table 7.3: Average value of the relative standard deviations for various tested combinations, as obtained with the BRRC trailer. The overall short-term repeatability of the BRRC trailer was found to be 2.7 %. There appears to be some speed and direction dependency, which may be related to wind effects, since the BRRC trailer lacks proper shielding of ambient air; i.e. it is sensitive to air drag Results for the TUG trailer TUG performed several runs on ten test sections with all tyre types at 50 and 80 km/h. The same average value as in the previous section is calculated for all cases; see results in Table 7.4. Direction Test section Speed Tyre east 1.2 % west 1.0 % M1 1.2 % L2 1.1 % 50 km/h 1.0 % 80 km/h 1.2 % AAV4 0.5 % ES % ES % SRTT 1.8 % Table 1.4: Average value of the relative standard deviations for various tested combinations, as obtained with the TUG trailer. The overall short-term repeatability of the TUG trailer was found to be 1.1 %. There is no significant influence due to speed, test section or direction. However, the tyre type seems to influence the repeatability. It is difficult to find a plausible explanation for this effect. All 1.1 % 35

46 7.3.5 Comparison of the trailers Table 7.5 shows a comparison of the short-time repeatability between the three trailers. The repeatability is expressed as relative standard deviations between repeated measurements in % of the average RRC value. It is interesting that the repeatability is substantially lower at 80 than at 50 km/h. The reason might be that, as the test section length is fixed, the test time is shorter at higher speeds. It means that for the lowest uncertainty, one should make tests at rather low speeds. However, at longer test sections than used in the RRT, it may be that the speed is less important for the standard deviations. Table 2.5: Average value of the relative standard deviations for the three tested trailers. Speed BRRC BASt TUG 50 km/h km/h Short-term (day-to-day) repeatability according to the MIRIAM RRT project Introduction In this section of the Chapter, measurement runs performed on the same test section on two consecutive days are analyzed. Also these tests were made within the RRT of MIRIAM SP Results for the BASt trailer Measurements were performed on various test sections with the SRTT tyre at 50 and 80 km/h on 6 and 9 June Results at 50 km/h are shown in Figure 7.5 (also in Figure 7.6 for 80 km/h). The overall relative RMS variation, σ, was calculated as follows: σ² = [ (C r,i,6 June C r,i,9 June )/ C r,i,6 June ]² / N for all tracks i, where - N is the number of test tracks - C r,i,x is the rolling resistance coefficient measured on track i on day x The RMS variation, σ, was then found to be 7 % for both speeds Results for the BRRC trailer BRRC performed measurements on several test sections on 6 and 9 June Since the trailer hit an object on the 9 th, partly damaging the device, only parts of the result are relevant. The results are shown in Table 7.6, where it appears to be a systematic increase of 10 and up to 25 %; probably due to a calibration error. 36

47 SRTT/BASt - 50 km/h 0,0160 0,0140 0,0120 Cr 0,0100 0,0080 0, June 9 June 0,0040 0,0020 0,0000 A-East A-West A'-East A'-West C-East C-West E1-East E1-West F-East F-West L1-East L1-West L2-East L2-West N-East N-West Test track + direction Figure 7.5: Measurements made on 6 and 9 June 2011 with the SRTT tyre on the BASt trailer at 50 km/h. Table 7.6: Day-to-day repeatability measurements using tyre ES14/BRRC on the BRRC trailer, at 80 km/h, made on 6 June (blue colour) and 9 June 2011 (no colour). Test track Direction Date C r Change between 6 and 9 June F E 6 June F W 6 June F E 9 June % F W 9 June % L1 E 6 June L1 W 6 June L1 E 9 June % L1 W 9 June % L2 E 6 June L2 W 6 June L2 E 9 June % L2 W 9 June ,4 % A E 6 June A E 9 June % C E 6 June C E 9 June % A' E 6 June A' E 9 June % Results for the TUG trailer No measurements were repeated by TUG on different days. 37

48 7.5 Reproducibility according to the MIRIAM RRT project Introduction In this section of the Chapter, measurement runs performed on the same test section by different measurement devices operated by their respective crews are analyzed. Also these tests were made within the RRT of MIRIAM SP Comparison of results obtained with the BASt and TUG trailers Measurements performed by BASt and TUG with the SRTT, AAV4 and ES16 tyres at 50 km/h are shown in Figure 7.6. Graphs representing BASt measurements are drawn with a full line, while the TUG graphs are drawn with a dashed line. The graphs show rather similar patterns with respect to the effect of the road surface. The C r values for the ES16 and SRTT tyres are much higher for BASt than for TUG, while the BASt values are lower than for the TUG for the AAV4 tyre. There is not much "sense" in these data, except that there is a similarity concerning how the different surfaces are ranked. Figure 7.6: Measurements performed by BASt and TUG with SRTT, AAV4 and ES16 tyres at 80 km/h. The results are only slightly better at 80 km/h; see [Bergiers et al, 2011]. Table 7.7 shows that the differences between the BASt and TUG values are indeed significant. The values were calculated as the average difference BASt TUG expressed as a percentage of the average C r (measured by TUG) for the tyre in question. 38

49 Table 3.7: Average deviation of C r values between the BASt and TUG trailers (BASt-TUG), expressed as percentage of the average C r value, for each tyre type. No temperature corrections were applied (but temperature differences were small). Speed SRTT AAV4 ES16 50 km/h 41 % - 10 % 43 % 80 km/h 33 % - 6 % 28 % It can be noted that: The average differences between BASt and TUG values are up to approx 40 % of the C r value. The average difference is "reasonable" only for the ES16 tyre The differences are smaller at 80 km/h than at 50 km/h Such differences of course mean that reproducibility is totally unacceptable. One would like them to be about 10 times smaller. The authors cannot offer good explanations for the deviations. Of particular concern is why there are such high differences for the SRTT and ES16 tyres, while it is relatively small for the AAV4 tyre. Note that each organization used its own test tyres. Can tyre differences be responsible for the problem or is it the trailers or poor calibrations that are responsible? TUG conducted measurements with AAV4/TUG and AAV4/BASt, and with SRTT/TUG and SRTT/BASt, in order to investigate the differences related to the tyres. The largest differences were found for the SRTT tyres, which is probably due to the fact that they came from different batches and had different rubber hardness. However, hardness can only explain a part of the large difference; there must be also some other factor; such as possibly some undetected tyre manufacturing problem. Applying a tyre correction to the SRTT results improves the comparison significantly, although the relative difference remains quite high. Applying a tyre correction to the AAV4 results does not improve the comparison much [Bergiers et al, 2011]. Another way of comparing the TUG and BASt results is to check the correlations between each pair of results (tyre by tyre). Then, very good correlations were found between the results of the ES16/BASt and ES16/TUG tyres at both speeds (Figure 7.7). However, there appeared to be some difference between the regression line and an assumed 1:1 relation (red colour), indicating the poor reproducibility. Good correlations were also found between the results of the SRTT/BASt and SRTT/TUG tyres for both speeds (Figure 7.8). The correlation at 80 km/h is even excellent (R 2 =0.984). However, the difference to the 1:1 line (red colour) is, again, substantial; especially at 50 km/h, indicating a poor reproducibility. Measurements with the AAV4/TUG and AAV4/BASt tyres at 50 km/h show a very good correlation, while those at 80 km/h fail to show any correlation (Figure 7.9). Therefore, also in this case, there appears to be a poor reproducibility. 39

50 Figure 7.7: Relation between C r of ES16/BASt and ES16/TUG at 50 km/h (left) and at 80 km/h (right). Figure 7.8: Correlation between C r of the SRTT/BASt and the SRTT/TUG at 50 km/h (left) and at 80 km/h (right). 40

51 Figure 7.9: Correlation between C r of tyres AAV4/BASt and AAV4/TUG at 50 km/h (left) and at 80 km/h (right) Comparison of results obtained with the BRRC and TUG trailers Results of measurements made with the ES14 by BRRC and TUG are shown in Figure 7.10 for both speeds. Graphs representing BRRC measurements are drawn with a solid line, while the TUG graphs are drawn with a dashed line. A large difference appears between the TUG and BRRC results at 80 km/h. This is most probably due to the lack of wind shielding of the BRRC trailer; thus the test tyre is exposed to air drag. When discarding the outlier M2, all graphs show a similar shape. The results at 50 km/h are situated closely together, although a larger difference was expected since TUG uses a higher load (4000 versus 2000 N). Drum measurements in the TUG laboratory also revealed differences between the ES14/TUG and ES14/BRRC tyres, which are speed and surface dependent (C r is influenced by ) [Bergiers et al, 2011]. Figure 7.10 reveals an outlier, namely the point for test section M2. The C r values measured by BRRC are too high for both speeds. This surface was measured separately after turning the vehicle sharply and after a full-throttle acceleration over a small distance. At a certain time these manipulations even caused an impact between vehicle and trailer. The acceleration may have caused higher C r values for M2. This problem will be verified by BRRC in the near future. According to Figure 7.11 there is a fair correlation at 50 km/h between the BRRC and TUG measurements for this tyre, while almost no correlation is visible at 80 km/h (probably due to the influence of tyre air drag and wind). However, if point M2 would have been discarded, better correlations would have been obtained (R² = at 50 km/h and R² = at 80 km/h). Again, reproducibility appears to be poor as the BRRC values are consistently higher. 41

52 Figure 7.10: Comparison of measured C r of the ES14/BRRC and ES14/TUG tyres at 50 and 80 km/h. Figure 7.11: Correlation between the C r of the ES14/BRRC and ES14/TUG tyres at 50 km/h (left) and 80 km/h (right). 42

53 7.5.4 Trailer-related differences Measurements were performed by BASt and TUG with exactly the same tyres; first with AAV4/BASt and then with SRTT/BASt. See results in Figure At 50 km/h there appears to be a certain offset. For the AAV4 graphs the difference should have been the opposite, as TUG measured with a somewhat higher tyre inflation pressure (210 versus 200 kpa). The offset is opposite for the SRTT graphs. This offset does not appear at 80 km/h. The results are inconsistent as the differences depend on tyre and speed, which is difficult to explain. It may be an indication of a different speed dependency of the two trailers. However, Figure 7.13 gives more insight into what happens. In this pair of diagrams the TUG measurements are shown in the left diagram and the BASt measurements are shown in the right diagram. While the TUG data at 50 and 80 km/h appear to be very consistent, despite being totally different measurement sets, the BASt data appear to have some kind of random effect distorting the data from what should be a "consistent" shape. The same applies to the ranking of the surfaces and the shape of the curves of the two tyres. The conclusion must be that the BASt trailer measurements seem to have been subject to some problem which has made the reproducibility test quite meaningless. C r 0, km/h C r 0, km/h 0,018 0,018 0,016 0,016 0,014 0,014 0,012 0,012 0,010 0,008 0,010 0,008 0,006 0,004 0,002 0,000 AAV4/BASt_TUG AAV4/BASt_BASt SRTT/BASt_TUG SRTT/BASt_BASt M1 F M2 L1 L2 E1 C A' A Test sections 0,006 0,004 0,002 0,000 AAV4/BASt_TUG SRTT/BASt_TUG AAV4/BASt_BASt SRTT/BASt_BASt M1 F M2 L1 L2 E1 C A' A Test sections Figure 7.12: C r measured by TUG and BASt trailers, using the same tyres AAV4/BASt and SRTT/BASt, at 50 (left) and 80 km/h (right). C r 0,020 TUG C r 0,020 BASt 0,018 0,018 0,016 0,016 0,014 0,014 0,012 0,012 0,010 0,008 0,010 0,008 0,006 0,004 0,002 0,000 AAV4 50 km/h SRTT 50 km/h AAV4 80 km/h SRTT 80 km/h M1 F M2 L1 L2 E1 C A' A Test sections 0,006 0,004 0,002 0,000 AAV4 50 km/h AAV4 80 km/h SRTT 50 km/h SRTT 80 km/h M1 F M2 L1 L2 E1 C A' A Test sections Figure 7.13: C r measured by TUG (left) and BASt (right) trailers, using the same tyres AAV4/BASt and SRTT/BASt, at 50 and 80 km/h. 43

54 8 INFLUENCE OF TESTING PARAMETERS 8.1 Tyre The Swedish rolling resistance data reported partly in [Sandberg et al, 2011] would be a good source to study the tyre influence on classification of rolling resistance of road surfaces. These measurements were made with three different tyres: SRTT, AAV4 and Michelin Primacy. However, these data have not yet been analyzed separately tyre by tyre; instead the average C r of all three tyres has been used consistently so far. Instead, the data collected in the RRT is the main source of information here. The tyres tested in this experiment are described in Table 8.1. Some of them are just duplicates, and one of them is a duplicate with a corrupt tread pattern (AAV4_CT/TUG). The tread patterns of the three major types are shown in Figure 8.1. Table 8.1: Overview of tyres used in the RRT. Symbol Producer Tyre line Tyre size Load Index DOT mark Hardness AAV4/BASt Avon Supervan AV4 195R14C 106/104N ATJ8 PC Sh AAV4/TUG Avon Supervan AV4 195R14C 106/104N ATJ8 PC Sh AAV4_CT/TUG Avon Supervan AV4 195R14C 106/104N - 64 Sh ES16/BASt Michelin Energy Saver 225/60R16 98V HC 3V 00KX Sh ES16/TUG Michelin Energy Saver 225/60R16 98V HC 3V 00KX Sh ES14/BRRC Michelin Energy Saver 195/70R14 91T F1 J9 681X Sh ES14/TUG Michelin Energy Saver 195/70R14 91T F1 J9 681X Sh SRTT/BASt SRTT/TUG Uniroyal Tiger paw M+S P225/60R16 97S ANX0 EVUU Sh Uniroyal Tiger paw M+S P225/60R16 97S Sh Figure 8.1: The three major tyre types used in the RRT: SRTT, AAV4 and ES16 (from left to right). The widths are approximately to scale. Figure 8.2 shows how the tyres ranked the test sections used in the RRT. In this case, only the TUG data are shown, as the previous parts of this report has indicated that the TUG data 44

55 seems to be the most reliable set. The diagram is based on the measurements at 50 km/h but the ones at 80 km/h are very similar, except that the corrupt AAV4 has significantly lower Cr values. It appears that the two passenger car tyres, SRTT and ES16, give very similar ranking of the 11 surfaces. Also the smaller car tyre, ES14, give a similar ranking, but at a higher rolling resistance level. The latter is mainly due to a lower test load, since the tyre is smaller. The AAV4 tyre (which is a light truck tyre but at a dimension and test load more typical of a car) shows a similar ranking pattern, but with relatively less influence on the C r values. The second AAV4 tyre, the corrupt one, gives very similar values to the uncorrupted AAV4. 0,020 0,018 Speed 50 km/h 0,016 0,014 RR coeff. 0,012 0,010 0,008 0,006 0,004 0,002 0,000 AAV4/TUG AAV4/TUG_CT ES14/TUG ES16/TUG SRTT/TUG M1 F M2 L1 L2 G E1 E2 C A' A Test sections Figure 8.2: Ranking of 11 test sections used in the RRT by the tyres used in the RRT. Data from the TUG measurements, using the TUG tyres. Table 8.2: The average rolling resistance coefficients and the variation due to road surface, expressed as standard deviation, for the 5 tyres used by TUG in the RRT. Tyre SRTT ES16 ES14 AAV4 CT AAV4 Average Cr Standard deviation of Cr Std. dev. as % of Cr average Table 8.2 shows that the sensitivity to the road surface is approx. 40 % greater for the SRTT than the AAV4 tyres, when expressed as absolute values, but three times higher if expressed in relation to the average Cr values. This is in line with expectations/speculations regarding how sensitive truck tyres may be to road texture compared to car tyres. 45

56 Table 8.3 shows the correlation between each tyre's C r values and the corresponding MPD values of the test sections. It appears that all tyres show similar high correlations (R2 = ) but the slope differs between and The SRTT and the AAV4 tyres seem to be the best tyres according to these data. Table 8.3: The correlation between each tyre's C r values and the corresponding MPD values of the test sections, for the 5 tyres used by TUG in the RRT. Also the slope coefficient in the regression is listed. The data are averages for the results at 50 and 80 km/h. Tyre SRTT ES16 ES14 AAV4 CT AAV4 Slope of Cr vs MPD R 2 in the regression Cr vs MPD Tyre condition (quality, wear, rubber hardness) The previous section shows that the exact design of the tread pattern does not seem to be critical to rolling resistance, based on comparison of the AAV4 and the AAV4_CT tyres. A picture comparison of the tread patterns of these tyres appear in Figure 8.3. Figure 8.3: The tread patterns of the two AAV4 tyres. The one with corrupt tread pattern is in the right picture. Note the displacement of the left and right parts. When it comes to wear; i.e. the tread depth, a special study about this effect was made in an earlier EU project called SILENCE. Figure 8.4 shows that the rolling resistance of a typical car tyre becomes lower when the tread pattern is worn down, whereas the difference in Cr between a rough and a smooth texture is unchanged [Sandberg & Glaeser, 2008]. This means that the relative contribution of the texture effect is increasing substantially when the tyre is worn-down. It also means that tyres for testing purposes should be worn maximum 1 mm from the new condition; after this they will not be useful as reference tyres any more. Regarding the influence of rubber hardness (measured as Shore A), a laboratory drum test in the 1990's by TUG was made; the results of which are shown in Figure 8.5. Here, the tyre was a STOMIL "D164" 145/70R13, an OE tyre used by Fiat, which normally had a tread with a hardness value of 71 Sh, but for this test the same tyre was produced with the only difference being that the rubber compound had a hardness of 62 Sh. 46

57 Tyre W5 0,020 ISO, APS ISO, SW 0,015 CR 0,010 0, Tread depth [mm] Figure 8.4: Rolling resistance coefficient C r versus remaining tread depth of the tyre, for a very smooth-textured surface (lower curve) versus a rough-textured surface (upper curve). The tyre was a car summer-type Pirelli PZeroNero 225/45 R17. Similar results were obtained for other tyres. From [Sandberg & Glaeser, 2008]. When these two tyres were compared concerning rolling resistance, the results in Figure 8.5 were obtained. It appears that at km/h, the increased hardness from 62 to 71 Sh resulted in an increase in C r of approx 8 %; i.e. approx. 1 % increase per unit of hardness in Shore A. 0,0150 0,0145 0,0140 0,0135 Influence of tread rubber hardness on the rolling resistance coefficient (RRC) Hard rubber (71 Shore) Soft rubber (62 Shore) RRC [ - ] 0,0130 0,0125 0,0120 0,0115 0,0110 0,0105 0, Speed [km/h] Figure 8.5: Rolling resistance coefficient C r (= RRC) versus speed for two tyres being identical except for different tread rubber hardness. Previously unpublished data measured by TUG in Gdansk in the 1990's on a laboratory drum covered with sandpaper. 47

58 There are also data available from the MIRIAM RRT, namely from laboratory drum measurements made by TUG on a number of tyres, the hardness of which differed a little. First, three SRTT from TUG and one from BASt, plus three AAV4 and one from BASt were tested. Shore hardness within each tyre group varied by 9 units. Yet, the correlation between Cr values and Shore values were insignificant for both tyres. Secondly, all the 9 tyres from the RRT field experiment (see Table 8.1), plus one used in the Belgian so-called Artesis project, were tested on the 1.7 m laboratory drum of TUG. These tyres were two SRTT, three AAV4, two ES16 and three ES14. These were treated as four different series in a regression, shown in Figure 8.6. The figure indicates that the slope of Cr vs Sh for all four tyres is rather consistent. The average slope was calculated to be 0, Of course, only two or three points in each series is too little, but the interesting thing is that it results in such a consistent slope. For comparison, the slope from the results shown in Figure 8.5 is also included. One can see that the slope derived from the RRT tyres is approx. three times that of the TUG experiment in the 1990's. One may conclude, with great caution, that it seems that the rubber hardness increases rolling resistance and that the effect seems to lie in the range of 1-3 % per unit of Shore A. Cr 0,017 0,016 0,015 0,014 0,013 0,012 0,011 y = 0,0002x + 0,003 y = 9E-05x + 0,0066 y = 0,0003x - 0,0096 AAV4 SRTT ES16 ES14 TUG1995 y = 0,0003x - 0,0096 0,010 0,009 y = 0,0003x - 0,0066 0, Rubber hardness in Shore A Figure 8.6: Rolling resistance coefficient C r versus tread rubber hardness, for four tyre types, the latter of which are represented by 2 or 3 tyre samples having different rubber hardness. Measured on the TUG 1.7 m drum, where C r values have been averaged for three different speeds and two different drum surfaces. 48

59 8.3 Tyre load Artesis project - background In there had been a cooperation project between BRRC and Artesis Hogeschool in Antwerp, in which Artesis had access to the BRRC trailer for making rolling resistance measurements [De Bie & Hofmans, 2011]. Earlier, they had also made coastdown measurements of rolling resistance, using texture equipment and test vehicles from BRRC. Most of the activities dealt with issues important for measurement methods; but some addressed road surface influence on rolling resistance, see [Sandberg et al, 2011]. Activities related to road surface influence on rolling resistance are reported below Artesis project effect of tyre load The influence of the vertical load on the rolling resistance force and the rolling resistance coefficient was investigated in the Artesis project. Normally a tyre load of 1939 N is applied on the tyre on the BRRC trailer. To investigate the influence of load the RRC measurements were performed with the following loads: 1285 N, 1416 N, 1547 N, 1678 N, 1809 N and 1939 N. In Figure 8.7 the rolling resistance force as a function of the vertical load is shown. The values are shown for two driving directions and with and without temperature ("T") correction. The latter was in accordance with [Descornet, 1990]. Figure 8.7: Influence of vertical load on rolling resistance force. From [De Bie & Hofmans, 2011]. The influence of the load on rolling resistance force is linear. This means that the rolling resistance coefficient is constant and independent of the load. By dividing the rolling resistance force with the vertical load, the data in Figure 8.7 confirms this. Applying the temperature correction seem to give a little better correlation. 49

60 8.4 Tyre inflation To investigate the influence of tyre inflation, rolling resistance measurements were performed in the Artesis project with different tyre inflations: 1.2, 1.7, 2.2, 2.7, and 3.2 bar. In Figure 8.8 the relation between the RRC and the tyre inflation pressure is shown. The results show that when tyre inflation decreases, RRC increases according to an essentially linear relation. A difference of 1 bar (100 kpa) gives a difference of approximately of the RRC value, which in this case is about %. This means that tyre pressure is a very important factor for the rolling resistance. It shall always be closely monitored. Figure 8.8: Correlation between the average C r (RRC) and tyre inflation pressure. From [De Bie & Hofmans, 2011]. It has been suggested that inflating the tyres with nitrogen instead of air may give more stable inflation and temperature conditions in the tyres [Sandberg & Ejsmont, 2009]. In the MIRIAM RRT project, all regular rolling resistance measurements were made with tyres inflated with nitrogen. However, some special tests were made on a highway with the BRRC trailer to compare the normal air inflation with nitrogen inflation. These tests showed that nitrogen inflation gave some advantages. The main results are shown in Figure 8.9. Temperature seemed to be more stable with nitrogen inside the tyre than with air. The tyre interior temperature dropped less at standstill with nitrogen than with air. A lower interior temperature was reached with nitrogen than with air and temperature increased slower while driving. It should be noted that "air" is not a constant gas, it may e.g. contain more or less water vapour and this would influence the inflation stability; thus different air may give different temperature instability. 50

61 Figure 8.9: Warm-up test with nitrogen inside the ES14/BRRC tyre Continuous measurement of tyre temperatures while driving 15 minutes two times. Tyre inflated with air in the top figure and inflated with nitrogen in the bottom figure. 8.5 Tyre warmup In the MIRIAM RRT the test mentioned in the previous section was also used to test the warmup time, from a condition when the tyre had approximately ambient temperature until a stable temperature was obtained (in this case 215 kpa). It was noted that for the ES14 tyre, 15 minutes of driving at 80 km/h on a highway with the normal test load was sufficient to reach the required stable condition; see Figure Speed Theoretically, rolling resistance should not depend significantly on speed up to "normal" highway speeds. However, the results of [Descornet, 1990] suggested a non-negligible 51

62 speed dependence. Measurements by BASt showed a significant speed effect too; thus the BASt trailer's test tyre was equipped with a wind-protecting cover early in the development. The TUG trailer was used the first years in Poland and Sweden without a wind-protecting cover, and large influences of speed were noted. After that, also the TUG trailer has been equipped with a cover the latest years. The BRRC trailer, however, did not yet in the MIRIAM RRT use a wind-protecting cover. Both the measurements reported by Descornet in 1990 and the Artesis project showed that already an increase from 30 to 50 km/h influenced the C r significantly. In the MIRIAM RRT the speeds 50 and 80 km/h were used consistently. It is shown very clearly in the RRT report that the uncovered BRRC trailer gave substantially higher C r at 80 than at 50 km/h [Bergiers et al, 2011]; see also Figure Thus, test tyres must be protected from the ambient air flow by some protecting cover, and this is needed already at 50 km/h of test speed. At 80 and higher speeds testing without a cover is meaningless as the results will depend much on the aerodynamics of the test vehicles in combination with ambient wind. The TUG measurements in the MIRIAM RRT indicated no significant speed effect (Figure 7.13 left) which would suggest that the TUG cover which has a ground clearance of 5-10 cm is effective. The BASt has a cover with higher clearance to the ground, approx. 20 cm, and the effect of speed seemed to be a little uncertain in some cases for the BASt results (Figure 7.13 right). One must be aware of a potential problem with the cover. The test tyre needs cooling by surrounding air and if this surrounding air is flowing slower and will be contained within the cover, the tyre temperature might increase substantially due to the poor cooling while the tyre is running and stable running conditions may be more difficult to achieve. Ambient wind may, depending on wind direction and speed, influence the air near the uncovered tyre/road interface, thus affecting both cooling and tyre air drag. Consequently, construction of the protecting cover is probably rather critical. The main author thinks that the cover should not be too tight over the tyre; instead it should allow for air moving slowly inside the cover to cool the tyre reasonably well. One may even consider having the upper part of the cover partly open or ventilated. It is suggested that the TUG trailer test tyre is tested with and without its cover to see how tyre temperature develops during a complete warmup process. In summary, the research so far shows that test tyres shall be covered effectively from air drag and ambient wind, and when this is done, test speed is not critical. One can thus allow quite wide tolerances on test speed, at least ±10 km/h would constitute no problem. The choice of test speed may be mostly a practical matter and considering how well one can drive over the test section. 8.7 Temperature In the research carried out by Descornet in the 1980's, he found a very substantial effect of temperature on the rolling resistance coefficient [Descornet, 1990]. The results are shown in Figure The black curve shows his results measured on-road in 1986 for one tyre, based on tyre shoulder temperature measurements (this includes effect of inflation pressure being influenced by temperature). In the same figure, the red curve shows the temperature correction required according to ISO for drum measurements (for car tyres). However, this one uses ambient air temperature rather than tyre temperature. If one would use the same temperature in both cases the very large difference in slope would become much smaller. Nevertheless, even the small slope of the red curve implies a significant temperature effect, since a difference in air temperature of 30 o C means a difference in C r of approx. 20 %. Therefore, temperature corrections seem to be necessary, unless the temperatures change just a little over the test 52

63 period. In the MIRIAM RRT, temperature variations were fortunately rather small [Bergiers et al, 2011]. Figure 8.10: Effect of temperature on Cr; measured by Descornet in 1986 (black curve) and correction required in ISO (red curve). Note, however, that Descornet used tyre temperatures while ISO refers to ambient air temperatures. 8.8 Road curvature, ruts, slope and cross-slope When driving in a curve, the tyre reacts to the steering by developing a slip angle; see Fel! Hittar inte referenskälla. in [Bergiers et al, 2011]. Even half a degree of slip angle may change C r by a significant amount [Sandberg (ed), 2011]. Measurements were made in the MIRIAM RRT of the effect of side forces. However, the results were judged to be impossible to use as the measurements were made by the BRRC trailer and it was found that the lack of air-flow-protecting cover could have been responsible for the changes measured due to the side forces. Driving in ruts also creates side-forces, unless the tyre is able to avoid running on the side slopes of the rut, which in practice is difficult; however, in certain trailer systems tyre lateral position may be automatically adjusted to run in this way. A cross-slope of the road will also create side forces. As roads are designed to have a certain cross-slope (often 2 % on flat areas) this is an issue which is important to get knowledge of. Longitudinal slope (gradient) affects rolling resistance significantly, but the effect may be compensated for by running in both directions and calculate the average Cr. Compensation works only as long as the slope is sufficiently low to affect rolling resistance in a linear fashion. Unfortunately, the magnitude of all these effects is unknown. 8.9 The influence of wind shielding The results of the MIRIAM RRT, as well as earlier experience, are discussed in Section 8.6 about speed influence. In summary, the research so far shows that test tyres shall be covered effectively from air drag and ambient wind. 53

64 9 REFERENCE TYRES 9.1 Introduction As it is very impractical to measure rolling resistance using a great number of tyres representing the tyre market, when classifying or ranking pavement properties for rolling resistance, it is practical if not necessary to use reference tyres. The purpose of these is to be representtative of the category of tyres that they are intended to represent and to provide stable and repeatable conditions. A common reference tyre concept is that one tyre shall represent the fleet of automobile tyres on the roads (tyre category C1), and another tyre shall represent the fleet of heavy truck tyres (C3). One might also want to have a tyre representing the middle range; van tyres (C2). Reference tyres must be available for a long time. 9.2 Reference tyres for noise measurements This concept is already implemented in the drafts ISO/DIS and ISO/TS which are two documents specifying the so-called CPX method for classification of noise properties of pavements. The tyres used in the CPX method by ISO are shown on the left (SRTT) and in the middle (AAV4). A draft for an ASTM method for a "Standard Test Method for Measurement of Tire-Pavement Noise Using the On-Board Sound Intensity (OBSI) Method" specifies the use of one reference tyre (the SRTT). 9.3 SRTT and AAV4 The SRTT ("Standard Reference Test Tire") is a tyre specified in ASTM F2393 as a reference tyre for various purposes. The Avon AV4 tyre (designated "AAV4") is a tyre tested and found to classify pavements (for noise) in roughly the same way as a selection of regular heavy truck tyres do. It is in fact a light truck tyre, but as the smallest dimension for this series of tyres is used, the AAV4 fits on large passenger cars, as does the SRTT. Figure 9.1: Reference tyres used in the tests reported in this article. Refer to the text for more information. As it is a reference tyre specified by ASTM, the SRTT is likely to be available for several decades in the future. The AAV4 tyre will not be manufactured in the future unless the users of CPX tyres orders a full batch of 100 or more tyres simultaneously, and to keep them stored 54