Jtam.dvi JOURNAL OF THEORETICAL AND APPLIED MECHANICS 46, 2, pp. 435-442, Warsaw 2008 GENERATION OF VIRTUAL TRACK PROFILES USING EXPERIMENTS AND COMPUTER SIMULATIONS Marek Hejman Vladiḿır Lukeš ŠKODA VÝZKUM s.r.o., Section of Materials and Mechanical Engineering Research, Pilsen, Czech Republic e-mail: marek.hejman@skoda.cz; vladimir.lukes@skoda.cz An advantage of computer simulations of structures is the prediction of dynamic properties, loading, strength, fatigue life stability, etc. A new goal of computer simulations is the fatigue life analysis of vehicle struc- tures or that parts. A ”virtual track profile” (mathematical description of the track profile) can solve the problem of proper excitation. The pa- per describes one possible approach of virtual track profile generation. Experimental data and computer simulations are used for the profile generation. Key words: track profile, fatigue analysis, rainflowmethod 1. Introduction An advantage of computer simulations is the prediction of strength, loading dynamic properties, stability, etc. A new goal of computer simulations is the fatigue life assessment of structures or their parts. The problem of such an analysis is a definition of kinematic excitation of numerical models. It can be realized by the ”virtual track profile”. The virtual track profile means a mathematical description of a track surface. The virtual track can serve for the comparison of the aggressiveness of different operational tracks, see e.g. Hejman andLukeš (2005) orHejman (2005). This paper dealswith the virtual track profile generation. The tracks vompound of the described method can be utilized for the strength assessment of structural parts close to the suspen- sion. This restriction is due to the applied criterion of the track generation. The criterion is based on the agreement of the rainflow histogram of relative 436 M. Hejman, V. Lukeš displacements and velocities between the structure and the axles of vehicles. The virtual track can by used in computer simulations or experimental tests. 2. Basic principles The primary cause of fatigue damage Di of a vehicle structure is the kine- matic excitation of a springing. The excitation is transferred by the springing to the structure. Stress responses depend on the character of excitation and properties of the structure (modal shapes, frequencies, damping ratio, etc.) in places of the fatigue life assessment. Stress time-history courses are evaluated using the standard rainflowcountingmethod (RF).The fatigue life assessment is performed by the application of the damage accumulation rule according to Miner. The process is described in Fig.1, where ni is the number of cycles on the ith level and Ni is the number of cycles to failure on the ith level according to the given S-N curve. Fig. 1. A scheme of the inductive process of the vehicle fatigue damage 3. Classification of road surfaces The proper path to the track profile generation is the measurement of a real track profile. But this approach is very complicated and for the purpose of multibody and finite element numerical simulations also unsuitable. There are several reasons why to generate track profiles in some other way. There are no appropriate mathematical models implemented in the MBS software considering full three-dimensional contact of a tire with road une- Generation of virtual track profiles... 437 vennesses. There are problems with numerical simulations and also with real measurements in three dimensions. The examined vehicle drives along the same road in different tracks and it does not hit the unevennesses in the same way. Every exact method of the track profile measurement includes some inaccuracy. For the elimination of this inaccuracy it would be necessary to includemany passages of the track. Relative displacements of the axles and bodywork do not include ”real information” of the track, because the relative displacements are the results of kinematic excitations and dynamic properties of the vehicle. But this quantity is suitable for our purpose because it includes the effects of the variable speed and weight of the vehicle. 4. Virtual track generation Themethod based on the combination of predefined artificial unevennesses is used for the generation of a virtual track profile. It is necessary to make a set of artificial obstacles with different lengths and heights, see Fig.2. The set must be large enough to be able to generate a sufficient virtual track profile. If it is necessary, other shapes of the obstacles can be proposed. Fig. 2. The shape and size of the artificial obstacles NumericalMBS simulations were performed for running over all these ob- stacles. The parameters of the usedmultibodymodel were identified from the experimental data (Polach andHajžman, 2005). The responses, i.e. the relati- ve axle displacements, are identified and then processed by using the rainflow method into one-dimensional histogram. The histogram for each obstacle re- presents a ”basis function” in the track generation process. The least square method is used to find such a combination and number of the artificial ob- stacles that the resulting histogram of the artificial track approximates the histogram of the experimental measurement. 438 M. Hejman, V. Lukeš It is necessary to find such a vector of parameters a that the sum N ∑ i=1 r 2 iwi is minimized, where the residuum vector r is defined as r= [ D d m D v m ] − [ D d p1 D d p2 · · · D d pN D v p1 D v p2 · · · D v pN ] ·       a1 a2 ... aN       Thevector w represents theweight vector, Dpp and D v p are thehistogramvec- tors of relative displacements and their first time derivatives for each artificial obstacle, and Dpm and D v m are the histograms of the experimental data. The shape of the artificial obstacle is shown in Fig.2 and the set of the obstacles is evident fromthe table.The table is not full, because someobstacles in the table corners have improper rations of length and height, and do not represent the existing characteristic obstacles. A program in the Matlab system was created to generate virtual track profiles, see Fig.3. Fig. 3. The program for virtual track profile generation The approach of the track generation is used taking into account the fact that the rainflowhistograms of the selected segments can not be simply added. Generation of virtual track profiles... 439 The RF histogram, which is the sum of the RF histograms of the separated segments, is not necessary as well as the histogram of the whole signal. It is the consequence of the RFmethod principles (ASTME 1049-85, 1999). As a result of the signal segmentation, some cycles are not closed or are closed at different class of the RF histogram. The dynamic response of the springing depends on the obstacle characte- ristics and on the initial state of the system. To avoid interaction between the responses of different obstacles, the distance between two obstacles must be sufficient. The stress responses in the elements of the suspension or the parts close to the suspensionare dependenton both spring forces and forces in shock absorbers. It is the reason why the RF histograms of relative displacements are used together with the RF histograms of velocities. Fig. 4. Histograms of the stress history The roadways are generally made of materials of different types and com- position. Generally, it is possible to divide road surfaces into two groups. The first group involves homogeneous parts of the roadway. The second group in- 440 M. Hejman, V. Lukeš Fig. 5. Resulting fatigue damage volves significant unevennesses of individual shapes like drains, retarders, pot holes, railway crossing, etc. This simple classification is proposed on the basis of the analysis of stress history obtained from the experimentalmeasurements on a real track. In Fig.4 and Fig.5 some results of the rainflow processing and fatigue damage calculation for one stress history are illustrated. From the signal of stress history, segments corresponding to the significant unevenness were extracted and processed by the RF method and the damage algorithm. This also was done for the signal of thewhole track and for the signal without the significant unevenness. It is evident from the graphs that the significant unevenness contributes to the RF histogram with a small number of cycles but with higher amplitudes and, therefore, the corresponding fatigue damage is bigger than the fatigue damage of the rest of the track. The segmentation of the experimental data and processing by the RF method is influenced by the same error as it was mentioned in the previous Section. Generation of virtual track profiles... 441 In the set of the artificial obstacles, theremust be some obstacles with the characteristic response corresponding to distinctive obstacles in a real track. One possible way to include these obstacles is to identify their profile. An optical method was developed for this purpose. The optical method is based on takingpictures of themeasuredunevenness.Asizing rod is placedabove the unevenness and the unevenness ismarked by a bright adhesive tape, seeFig.6. The acquired pictures are processed in ownMatlab program.As a result, data sets describing the profile in the measured line are obtained. The accuracy of thismethod is not too high, but it is sufficient for the purposeof the numerical simulations. Advantages of the method are the simplicity and efficiency. Fig. 6. The optical method for the measurement of significant unevenness 5. Conclusions Themethod of the generation of the virtual track profilewas presented in this paper. Virtual tracks are used in the multibody numerical simulations and can also be used for the fatigue tests. The virtual track profile is compared of a number of obstacles of different shapes. The selection of artificial obstacles and their number is determined by the least squaremethod.The final rainflow histogram obtained from the computer simulations performed on the virtual track approximates the experimental data. This approach will be further im- proved to eliminate the effect of fragmentation of the rainflowprocedure. This task can be solved by using some optimization methods. Acknowledgment Thisworkwas supported by theGrantAgency of theCzechRepublic, the project ”Generation of the Virtual Track and Their Using for Vehicle Fatigue Assessment”, No. 101/03/1497. 442 M. Hejman, V. Lukeš References 1. ASTM E 1049-85, 1999, Standard practices for cycle counting in fatigue ana- lysis, In:Annual Book of ASTM Standards, Vol. 03.01, Philadelphia, 710-718 2. Hejman M., 2005, Draft of virtual roadway generation bymeans of computer simulations,Proc. Engineering Mechanics 2005, Svratka 3. Hejman M., Lukeš V., 2005, Generation of virtual tracks using tests and computer simulations,Proc. 8th Conference onDynamical SystemsTheory and Applications, Łódź, Poland 4. Polach P., Hajžman M., 2005, Influence of the hydraulic shock absorbers model in trolleybusmultibody simulations on the suspension deformations and comparison with the experimental results, Proc. Engineering Mechanics 2005, Svratka Generowanie wirtualnego profilu drogi na podstawie symulacji komputerowych i danych eksperymentalnych Streszczenie Jedną z korzyści płynących z symulacji komputerowych jest możliwość przewidy- wania właściwości dynamicznych, obciążeń, wytrzymałości, stabilności zmęczeniowej itp. różnych konstrukcji.Wpracy przedstawiono nowy cel analizywspomaganej kom- puterowo – badaniawytrzymałości zmęczeniowej pojazdów i ich podzespołów. Zapre- zentowanometodę „wirtualnego profilu drogi” (matematycznego opisu profilu drogi) do rozwiązania problemu wymuszenia kinematycznego pochodzącego od nierówności jezdni.Wpracy zawartoopis jednego zmożliwych sposobówgenerowaniawirtualnego profilu drogi. W rozważaniach zastosowano symulacje komputerowe oraz oparto się na pomiarach doświadczalnych. Manuscript received February 21, 2007; accepted for print April 4, 2007