Jtam.dvi JOURNAL OF THEORETICAL AND APPLIED MECHANICS 41, 4, pp. 887-901, Warsaw 2003 MODELLING AND IDENTIFICATION OF CATENARY-PANTOGRAPH SYSTEM Anna Kumaniecka Institute of Mathematics, Cracow University of Technology e-mail: pukumani@cyf-kr.edu.pl Michał Prącik Institute of Mechanics and Machine Design, Cracow University of Technology e-mail: mp@sparc2.mech.pk.edu.pl Thepurposeof thispaper is topresenta surveyandanalysis of thedynamic in- teractionbetween the pantographand a catenary system.Theworkpresented in this paper emphasizes the modelling, simulation of co-operating systems and identification done by experiments on a physical model. For analytical and numerical investigation the overhead electrification system is modelled by visco-elastically connected double Bernoulli-Euler beams. For the panto- graph–amodel ofmassesvibrating in thevertical planeat a constantvelocity and having four degrees of freedom (DOF) is adopted. The construction of a simple virtualmodel for computer simulation and some results obtained from investigations carried out on this model are presented. The work contains some numerical examples that illustrate presented theoretical investigations. The analysis of the virtual and two physical experimental models gives the possibility and benefit of their mutual verification and comparison. Key words: dynamics, catenary, pantograph 1. Introduction The collection of the current from the overhead equipment is a problem of primary importance for a high speed railway system, and becomes a challen- ging task for a medium voltage line. There is a great interest in the develop- ment of new pantograph-catenary power transmission systems with improved dynamic performance. Therefore, the dynamic interaction between the panto- graph and catenary has been studied extensively (Poetsch et al., 1997). 888 A.Kumaniecka, M.Prącik The modelling and simulation of the dynamic behaviour of the catenary- pantograph interaction is an important part when assessing the capability of a current collection system for railway traffic. The capability of a current collection system depends on the interaction between the locomotive, panto- graph and the catenary system. The interaction between the pantograph and the overhead wire forms dynamically coupled systems which affect each other through the contact force. To maximize the life span of the pantograph and overhead equipment, and to ensure that the locomotive receives good quality current, the dynamic behaviour of both the pantograph and overhead wire needs to be properly understood and optimized. Therefore, this system has been analyzed by several scientists and engineers in recent years (Drugge et al., 1999). The variety of designs of current collection systems is large. De- pending on the country, railway company and type of traffic, different designs of pantograph and catenary systems are used. Pantograph designs based on a symmetric or asymmetric configuration have been evolved. There are three common types of catenary systems: simple, stitched and compound catenary (Roman, 2001). The large variation in infrastructure characteristics in different countries and railway companies, different designs of the pantograph and catenary, dif- ferent types of railway vehicles running at different speeds makes it almost impossible to develop a final simulation model of the system. The work presented in this paper emphasizes the modelling, simulation of co-operating systems and identification done by experiments on a physical model of the system. For analytical and numerical investigations, the over- head electrification system is modelled by two Bernoulli-Euler beams visco- elastically connected with each other and elastically supported through the upper elements (Kumaniecka et al., 2002). For the pantograph amodel vibra- ting in the vertical plane at constant velocity of four degrees of freedom is adopted (Grzyb and Kumaniecka, 2000). The dynamic state of the investiga- ted system is described by a set of coupled partial equations with a complex boundary conditions. The construction of a simple virtual model for computer simulations and some results obtained from investigations carried out on that model are pre- sented. It is assumed that the supply contact wire gives kinematic excitation to the 4 DOF pantograph model, and that the longitudinal dry friction force is related to the normal load component. The analysis of the virtual and two physical experimental models is themost important task of the present paper because of prediction, verification and comparison, reasons. Modelling and identification of catenary-pantograph system 889 The concept of model investigations and description of the testing facili- ty which were designed and built at Department of Mechanical Engineering, Cracow University of Technology, were introduced. The criteria based on the dimensional analysis were adopted to construct physical models of the panto- graph and catenary system (Prącik and Furmanik, 2000a). In the paper the dynamic coefficient of dry friction between the supply contact wire and lo- comotive pantograph is determined in the experimental identification process performed on special test stands. The number of published reports (Fujii and Manabe, 1993; Kumaniecka, 1998; Kumaniecka andGrzyb, 2000) proves that these problems are up-to-date. 2. Simulation of catenary-pantograph dynamics With respect to the dynamic behaviour, the catenary and pantograph are vital components in current collection systems for trains. The pantograph is mounted on the roof of the train and its frame assembly raises its head assembly into forced contact with the catenary system. A catenary system consists of a contact wire suspended from a catenary wire via an arrangement of droppers.The catenarywire is linked to the supporting structure, while the contact wire is linked to the support via steady arms. To study the dynamic interaction between the pantograph and catenary it is advantageous to use at first physical and then mathematical simulation models. The research has been curried out to understand and improve the performance of the catenary-pantograph system by mathematical modelling for many years (Bogacz, 1983). During the last forty years, many studies of the catenary, pantograph, and pantograph-catenary systems dynamics have been carried out. Many different, more and more complex analytical models of the pantograph-catenary interaction have been considered and developed. Simplified analytical models of such systems have been used in order to de- termine critical speeds and study the influence of principal systemparameters (Bogacz, 1983; Bogacz and Szolc, 1993). The catenary has beenmodelled using different approaches. One degree of freedom mass spring-damper system models with varying stiffness and mass, and stringmodels with an elastic foundation (Kumaniecka and Nizioł, 2000), and recently two-beam models (Grzyb et al., 2001; Kumaniecka et al., 2002) have been used. Models with the two-dimensional representation of the cate- nary are most common (Kumaniecka et al., 2000). 890 A.Kumaniecka, M.Prącik Somemodels for simulation of the catenary dynamical behaviour, develo- ped by the authors of the presented paper, are illustrated in Fig.1 and Fig.2. Fig. 1. The catenarymodel: two visco-elastically connected Bernoulli-Euler beams with the upper finite elastically supported Fig. 2. The catenarymodel: two visco-elastically connected Bernoulli-Euler beams with concentratedmasses. The upper beam rigidly supported on the foundation In the Fig.3, the physical model of the catenary typeWBL 85 3kV, which is common in use not only inmanyEuropean countries but inAsia andAfrica as well, is presented. This type of pantograph will bemounted in Polish loco- motives EU11 and EU43. This model refers to a real system and it has been examined by engineers from Schunk Wien G.m.b.H. (Prącik and Furmanik, 2000a). In the present paper the systemunder investigation consists of two infinite visco-elastically connectedBernoulli-Euler beams,with one elastically suppor- ted on a rigid foundation and the other suspended via droppers. The model employed in the analysis is shown inFig.1. For the pantographa4DOFmodel Modelling and identification of catenary-pantograph system 891 of vibratingmasses in the vertical plane andmoving at a constant velocity is assumed, see Fig.3. Fig. 3. Reduced physical model of a real pantograph system The analysis of small transverse vibrations of the adopted physical model of the system was presented by Kumaniecka et al. (2002). Several simplifying assumptions have been necessary to limit the scope of the problem under consideration: • considerations are restricted to the OXY plane only, • transverse vibrations of the beams in the vertical plane are small, • the beams are prismatic, homogeneous, vertically loaded, • theupperbeamismulti-spanand is fixedby linear springsof the stiffness k at points which are spaced by the distance l, (catenary wire linked to the supporting structure), • the lower beam is of infinite length and linked to the upper beam via elastic elements of the stiffness c, (contact wire suspended from the catenary wire via an arrangement of droppers), • internal damping effects in the beams are neglected, • masses of the visco-elastic linking elements are neglected, • the interaction between the pantograph and contact wire is limited to a single force excitation in an additive form (constant component, harmo- nic variable and delayed variable of the load acting from the wire onto the pantograph), • the contact between the beam and oscillator exists all the time, • gravitational effects are neglected. 892 A.Kumaniecka, M.Prącik All considerations are done in a reference system OXY . The x co-ordinate is measured along the upper beam from the point x = 0. The functions w1(x,t) and w2(x,t) describe the lower and upper beam transverse displa- cements, respectively. For the pantograph a 4 DOF model of masses vibrating in the vertical plane and moving at a constant velocity is adopted (Prącik and Furmanik, 2000a). This constitutive a discrete non-linear model is presented in Fig.3. It is assumed that the supply contact wire continuously gives a kinematic excitation to the 4 DOF pantograph model, and that the longitudinal dry friction force is related to the normal load component. 3. Analysis of equations of motion of the catenary-pantograph system The mathematical model of the catenary system given in Section 2 was obtained for small transversal vibration of the beams. Motion of the catenary in the vertical plane is governed by equations E1J1 ∂4w1 ∂x4 −N1 ∂2w1 ∂x2 +̺1 ∂2w1 ∂t2 −p+pF =0 (3.1) E2J2 ∂4w2 ∂x4 −N2 ∂2w2 ∂x2 +̺2 ∂2w2 ∂t2 +p+pp =0 where the following notation is used E1,E2 – Young’smodulusof the lower andupperbeam, respectively Ji – cross-sectional moment of inertia, (i=1,2) Ni – tensile force in the beam ̺i – mass density wi(x,t) – transverse displacements x – spatial coordinatemeasured along the nondeformed axis of the beam t – time. The loads p(x,t) acting on the beams and caused by internal forces in springs and damping elements are treated as continuous and can be expressed in the form p(x,t)= ∑ (n) {c[w2(x,t)−w1(x,t)]+ b[ẇ2(x,t)− ẇ1(x,t)]}δ(x−xn) (3.2) Modelling and identification of catenary-pantograph system 893 where c – coefficient of the spring elasticity b – damping coefficient xn – coordinates of the droppers spacing, xn =(2s−1)lw, s∈N lw – distance between the droppers δ – Dirac’s function. The interaction force between the pantograph and contact wire pF can be described by the term pF(x,t)=F(t)δ(x−vt) (3.3) The reaction force pp that comes from supports acting at points xj on the upper beam is of discrete character, but can be treated as disributed and written in the form pp(x,t)= ∑ j kw2(x,t)δ(x−xj) (3.4) where k – coefficient of the springs elasticity xj – coordinates of the supports spacing. Theboundaryand initial conditions adopted for numerical simulation have been based on the assumed vibration mode of the linear system (data from the identification research). Themathematical model for a physical model of the pantograph, adopted in Section 2, was discussed in detail and presented by Prącik and Furmanik (2000b). The structure of the simulation model has been based on a formal notation of motion in the form of ordinary differential equations, and the simulation software package VisSim has been applied. Analytical approach and some results obtained from the numerical investigations carried out on the simulation model were already presented by Kumaniecka et al. (2002). On the basis of the given set of equations a simulating program, applying the package VisSim Analyze ver 2.0, has been built. The block scheme is presented in Fig.4. In Fig.5 an example of sub-blocks (up to the 2nd level depth) integrating the equation of motion of the mass m1P is shown as well. Numerical simulations have been carried out for different data sets. The calculations (results of which are presented in Fig.4) were performed for the following parameters of the system: (construction of the pantograph WBL 85-3kV/PKP): m1L = m2P = 7.93kg, m2 = 8.73kg, m3 = 10.15kg, F1 = 2.0N, Faer = 0, F2 = F3 = 2.5N, PP(t) = 10+50sin[14π(t−0.02)], PL(t)=10+50sin(14πt), c3 =60Ns/m, Fstat =110N. Some results were presented in the paper by Kumaniecka et al. (2002). 894 A.Kumaniecka, M.Prącik Fig. 4. Scheme of the block for simulation of motion Fig. 5. Example of simulation results; displacement of a particular mass Modelling and identification of catenary-pantograph system 895 4. Experiments on the physical model The criteria which are determined on the base of a dimensional analysis method make it possible to construct physical models of the pantograph and catenary system(Prącik andFurmanik, 2000b).Thepurposeof suchmodelling is to preserve mechanical similarity to the real structure of the considered system. Dimensional analysis allows the estimation of the influence of chosen parameters on the investigated quantity and the determination of mechanical similarity criteria. The testing facilities for laboratory experiments have been designed and built at the Department of Mechanical Engineering, Cracow University of Technology. The stand is situated in a long corridor (about 50m), see Fig.6. Fig. 6. Facility for testing natural frequencies andmodes of vibrations of the catenary system of a physical model The catenary interval (30m long) is divided on sub-intervals 5m each – where suspension elements are embedded. The suspension consists of a canti- lever, at the end of which double-arm rods connected with articulated joints are mounted. The main carrying rope is stretched at the top of the whole catenary system. The supply contact wire is kept by lifting grips assembled to the edges of rocking arms of suspensions.This flexible construction is a 3DOF system, i.e. displacements are possible in 3 directions. There are also droppers 896 A.Kumaniecka, M.Prącik (connecting themain carrying rope with the contact wire) distributed betwe- en the suspensions. A carefully designed loading system has been adapted to make control possible and stabilize the tensile load for the carrying rope of the catenary system. An auxiliary system has been applied to achieve requ- ired levels of the supply wire tensions. An instrumentation for non-contact (laser) measurements of displacements of the vibrating supply wire and other elements of the catenary system (in the vertical and horizontal directions) has been used. The components of contact forces and characteristic vibration qu- antities can bemeasuredon themovingpantographmodel using strain gauges, accelerometers and other parts of the mobile instrumentation. The pantograph model is assembled on a rail car, which is driven by a cable system with an electric engine powered and controlled via microverter (Prącik, 2001), see Fig.7. Fig. 7. A rail-car with the model of pantograph; in the background the elements of the driving system: guiding wheels, cable rewinding, electric engine can be seen The plan of experiments included: • modal analysis of the experimental model of the catenary system (me- asurements of the natural frequencies and modes of the contact wire vibration for different excitations; impulse response functions using an impact hammer), • analysis of dynamical loads acting on parts of the pantograph model co-operating with the model of a catenary system (measurements of Modelling and identification of catenary-pantograph system 897 forces between the slider-bar of the pantograph and the contact wire for different conditions on the stand;measurements of vibration parameters of particular pantograph elements), • identification of friction (investigation of friction function treated as a function of load and relative slip velocity for different pairs of materials; dampingmeasurements of chosen elements of the catenary system). The main advantage of the physical model is the possibility of taking into account real effects, for instance the fact that the supply wire forms a zig-zag pattern and that the motion can be analysed in 3D space, not only in plane, • application of various stiffnesses of the physical catenarymodel (however particular results are omitted in the paper). Some selected results of the experiments can be illustrated by diagrams in Fig.8-Fig.10. Fig. 8. Spectral density of the free horizontal vibration acceleration of the pantographmodel beam (impact horizontal excitation, piezoelectric accelerometer) The experimental investigations allow one to identify the parameters of the simulation model built accordingly to the physical model. The results of the experimental research carried out for more general pur- poses, not just for the parametric identification, allow one to verify also the assumptionsadopted in the simulationmodel.These assumptions refer, among others, to the effect of the transverse vibrations assosiatedwith the zig-zagmo- tion – conditions bounded for the plane model. On the other hand, the effect of the zig-zag motion could be investigated in measurements carried out on the physical model. 898 A.Kumaniecka, M.Prącik Fig. 9. Exemplary results of investigations on dynamic friction coefficient (in function to relative slip velocity) Fig. 10. Exemplary results of the contact force measurements (mean distribution along the wire) Modelling and identification of catenary-pantograph system 899 5. Final conclusions On the basis of theoretical investigations, numerical simulations and expe- riments, the following conclusions can be drawn. • Parallel analysis of the resultsof investigation carriedoutonthephysical, experimental and simulation models enable their mutual verification. • Dimensional analysis allows the estimation of the influence of chosen parameters on the investigated quantityanddeterminationofmechanical similarity criteria. • During the investigations carried out on the simulation model, the co- nvergence of numerical results by changing the integration step and the order of the applied Runge-Kutta method has been tested. • Analysis of the simulation results indicates thatwhenthe force excitation with a dominant harmonic component acts on the pantograph, then the response of the systemmodel is not harmonic. • Analysis of the measurement results and calculations referring to the fexibility of the catenary linearmodel points out cyclic variability of the local flexibilitymeasured along thewire length, and indicates nonlineari- ty when the level of initial loading changes (vertical interaction forces). This cyclic variability of the flexibility can be explained if the periodic character of the suspension structure is taken into account. The maxi- mum of the nonlinear flexibility as a function of loading appears at the support points of the catenary wire. • Modal analysis of the catenary wire suspension system proves the exi- stence of a wide spectrum of free transverse vibration frequencies of the contact wire, in a range up to 40Hz. • The measurements of dynamic vibration and interaction force between the pantograph and contact wire were taken for different initial loads and velocity of the rail car (∆v0 =0-2m/s within time of 0.8s). As a result, we can state that the loss of contact between the pantograph and catenary is possible. • The identification investigations of the frictionmodel of cooperatingma- terials: pantograph – contact wire, were carried out on a special expe- rimental stand.Measurements for the relative slip velocity of 0-20m/s were possible. As a result, characteristics of dynamic friction coefficient versus relative velocity were obtained for different levels of the contact force. 900 A.Kumaniecka, M.Prącik References 1. Bielińska E., 1997, Identification of Processes, Silesian Technical University, Gliwice, (in Polish) 2. Bogacz R., 1983, On dynamics and stability of continuous system subjected to a distributed moving load, Ingenieur Archive, 243-255 3. Bogacz R., Szolc, T., 1993, Analysis of dynamic interaction between the continuous string andmoving oscillator,Engng. Trans., 41, 3-4, 361-380 4. Drugge L., Larsson T., Stensson A., 1999, Modelling and simulation of catenary pantograph interaction, Vehicle System Dynamics Supplement, 33, 490-501 5. Fujii Y., Manabe K., 1993, Computer simulation for dynamics of overhead catenary-pantograph system,Computer in Railways, 51-60 6. Grzyb A., Kumaniecka A., 2000, Non-linear dynamical pantograph vibra- tion,Proc. XIV-th Conference ”Rail vehicle 2000”, Kraków-Arłamów,121-128, (in Polish) 7. Grzyb A., Kumaniecka A., Nizioł J., 2001, Analysis of catenary vibration modelled as a set of two beams with concentrated masses, Scientific Issues of Cracow University of Technology, Mechanics, 83, 87-94, (in Polish) 8. KumanieckaA., 1998,Longitudinal and transversal vibration of railwayover- head contact system,Machine Dynamics Problems, 20, 181-192 9. KumanieckaA., BogaczR., Nizioł J., 2000,Dynamical analysis of the ca- tenaryandpantographsystem,Proc. of 7thMini Conference onVehicle System Dynamics, Identification and Anomalies, Budapest University of Technology and Economics, 129-136 10. Kumaniecka A., Grzyb A., 2000,On dynamics of catenary-pantographmo- delled as discrete-continuous system,Machine Dynamics Problems, 24, 1, 105- 119 11. Kumaniecka A., Nizioł J., 2000, Dynamical phenomena in catenary- pantograph system, Scientific Issues of Bydgoszcz Technical and Agriculture University, Mechanics, 47, 228, 15-26 (in Polish) 12. Kumaniecka A., Nizioł J., Prącik M., 2002, Dynamical cooperation of pantograph and catenary,Problems of Exploitation, 1, 44, 145-154 (in Polish) 13. Lasser, M., Karlsson L., Drugge, L., 1996, An interactive model of pantograph-catenary system, Vehicle System Dynamics Supplement, 25, 397- 412 14. NiederlińskiA.,Kasprzyk J., Figwer J., 1997,MULTI-EDIPAnalyser of Multi-Dimensional Signals and Subjects, SilesianTechnicalUniversity, Gliwice, (in Polish) Modelling and identification of catenary-pantograph system 901 15. Poetsch G., Evans J., Meisinger R., Kortum W., Baldauf W., Veitl A.,WallaschekJ., 1997,Pantograph-catenarydynamics andcontrol,Vehicle System Dynamics, 28, 159-195 16. Prącik M., 2001, Modelling and simulation of catenary-pantograph system, Scientific Issues of Cracow University of Technology, Mechanics, 83, 227-234 (in Polish) 17. Prącik M., Furmanik K., 2000a, Problems of modelling and simulation of catenary and pantograph cooperation,Proc. VII-thWorkshop PTSK,Warsaw, 7, 242-250 (in Polish) 18. PrącikM., FurmanikK., 2000b, Some problems ofmodelling and computer simulation of the pantographand supply contactwire co-operation,Proc. of 7th Mini Conference on Vehicle System Dynamics, Identification and Anomalies, Budapest University of Technology and Economics, 137-145 19. PrącikM., FurmanikK., 2001, Simulation and identification of frictionmo- del in the catenary-pantographsystem,Proc.VIII-thWorkshopPTSK,Gdańsk, 8, 305-312 (in Polish) 20. RomanZ., 2001, Simulation of the catenary andpantograph interaction,Tech- nics of Railway Transport, WarsawCNTK, 7-8 (in Polish) Modelowanie i identyfikacja układu lina trakcyjna-pantograf Streszczenie Celem pracy jest analiza dynamicznej współpracy pantografu z liną trakcyjną. Rozważane są zagadnienia związane z modelowaniem i symulacją zjawisk zachodzą- cych w układach zasilania kolejowej trakcji elektrycznej, a także identyfikacją na fi- zycznym modelu doświadczalnym. W pracy przyjęto model sieci trakcyjnej złożony z dwóch równolegle umieszczonych belek Bernoulliego-Eulera.Odbierak prądu, poru- szający się ze stałą prędkością, modelowany jest nieliniowym cztero-masowym ukła- demwykonującym ruch w płaszczyżnie pionowej. Przedstawiono konstrukcję modelu wirtualnego i wyniki symulacji przeprowadzonych na tym modelu. Analiza wyników eksperymentów prowadzonych na fizycznym modelu doświadczalnym i na modelach symulacyjnych pozwala na ich wzajemną weryfikację. Manuscript received December 30, 2002; accepted for print April 15, 2003