Prediction of Contact Wire Wear in High-speed Railways

Transcription

1 Prediction of Contact Wire Wear in High-speed Railways 1 T. USUDA M. IKEDA Y. YAMASHITA 1 Railway Technical Research Institute, Tokyo, JAPAN Abstract Although wear of contact wire has been one of the most serious problems in catenary system maintenance, mechanism of wear progress of contact wire has not been clear yet due to its complexity. The authors have attempted the identification of quantitative effects of a contact force of pantographs and an arc due to contact loss on the contact wire wear from the measurement data obtained on Shinkansen commercial lines. For this purpose, the authors have developed a contact force measurement method by equipping a catenary with sensors. An arc measurement equipment which detects UV ray emitted from an arc at the wayside has also been developed. The contact force of pantographs, the arc due to contact loss and collected current of all the trains which pass through two measurement sections on Shinkansen commercial line have been measured periodically. We propose a wear prediction model of contact wire from data which are measured by these method and equipment. Using this model, the wear of contact wire on the commercial line can be predicted. The predicted contact wire wear by this model agrees well with the measurement data, therefore, this model can be utilized for the efficient maintenance of the overhead catenary system. 1. Introduction Although wear of contact wire is one of the most important problems in the catenary system maintenance, there are many unsolved matters, because the mechanism of the wear progress of contact wire is very complicated. The factors influenced on the contact wire wear depend on pantograph types, amounts of the current collection, contact strip types, velocity of trains, catenary types, erection accuracy, surface conditions of the contact wire, and so on. It seems that these factors mutually have various effects on the contact wire wear. Especially, it is empirically known that the contact wire wear is largely related with the contact force of the pantograph and the arc due to contact loss. Various studies in this field based on the test bench have been performed up to now [1],[]. However, there has been no study to develop a prediction model of contact wire wear, because it has been difficult to simulate wear phenomena in the field including the effect of oxidation of the contact wire surface, etc. by the test equipment. The authors have started to research the quantitative effects of the contact force of the pantograph and the arc due to contact loss on the contact wire wear in order to clarify the mechanism of progress of contact wire wear. We compare data in some sections. However, the contact force of all pantographs, which causes the contact wire wear in a specific section, must be measured, because several types of pantographs pass and trains run with various velocities on the actual commercial lines. Therefore, the authors have developed a contact force measurement method [3] by equipping a catenary with sensors for this study. An arc measurement equipment which detects UV ray emitted from an arc at the wayside has also been developed. To build a wear prediction model of contact wire, the contact force of pantographs, the arc due to contact loss and collected current of all the trains which pass through two measurement sections of Shinkansen commercial line have been measured periodically. The wear of contact wire on these sections has also been measured periodically. From these results, the prediction model of contact wire wear has been built. An overview of this study is shown in Fig. 1. In this paper, these measuring methods, the measurement results and the prediction model of contact wire wear are reported. In addition, evaluation indexes of the current collection performance are discussed.. Measurement theory of contact force on the overhead catenary system In this chapter, a theory of the contact force measurement on the overhead catenary system (OCS) as indicated in Fig. is explained. To derive a equation of motion, the overhead catenary system is modeled as an infinite string supported by droppers force h i (i=1~n). is the Dirac delta function, the pantograph contact force f is assumed as convergent force. The linear density of contact wire, the tension of contact wire, the position of the i-th dropper and the velocity of train are described as, T,

2 i the behavior of contact wire y Fig. 1 Overview of this study x, and v, respectively. At the low frequency where we can neglect bending stiffness of contact wire, x, t is expressed by the following equation: h x x T i i n f t x vt. (1) i1 y t y x The length of the measurement area is described as L. When the pantograph is in the measurement area, integrating Equation (1) over measurement area gives the following equation: n L y y y f t hi T L L L dx i 1 x x x x. () t The first term of the right-hand side of the Equation () is the sum of dropper forces in the measurement area. The second term is the vertical component of the tensile force of contact wire, and calculated by the gradient of the contact wire at both ends of the measurement area. The third term is the inertia force of the contact wire. In calculation of the last term, continuous distribution of the vertical acceleration of the contact wire is necessary. However, generically it is difficult to measure the spatially continuous distribution of the contact wire vertical acceleration. Therefore, in this paper, the inertia force is estimated using the accelerations of the p points, and the Equation () is approximated by the following equation: y f t h T w x x t n p y y j i j L L, (3) i1 x x j 1 where w is inertia force correction coefficient for the j-th acceleration measuring point. We can j identify this value by excitation test results. A method to identify the inertia force correction coefficient is described into the following. Equation (3) can be transformed into frequency domain as follows:. (4) F F F w A Dividing Equation (4) by h t j j j 1 F, we get p w j Aj Fh Ft j 1 1. (5) F F F The real part of Equation (5) is expressed by the following equation, p wj Aj F F h t j 1 Re 1. (6) F F F The imaginary part of Equation (5) is expressed by the following equation, p p wj Aj F F h t j 1 Im. (7) F F F

3 w can be identified from Equation (6) and (7) using the least squares method. In this paper, 15Hz LP j filter was applied to contact force. Y Measurement section L Support point X Dropper force Messenger wire Dropper h 1 h h 3 h 4 h 5 h 6 h 7 h 8 h 9 h 1 Auxiliary wire Contact force Contact wire L Inertia force L y, t y, t x Pantograph x L L y, t y, t Vertical components of tensile force = T( - ) x x Contact force = Inertia force + Dropper force + Vertical components of tensile force Fig. Measurement theory of contact force 3. Arc measurement equipment at the wayside by detecting UV ray In this chapter, measuring principle of the arc measurement equipment by detecting ultraviolet (UV) ray indicated in Fig. 3 is outlined. Ray of arcing light incident on a reflecting mirror in this equipment splits into UV and visible rays by the UV reflecting mirrors. The UV ray is reflected by two pieces of UV reflecting mirrors, and it is converged on a photomultiplier by a lens. The visible ray transmitting to the first UV reflecting mirror reaches a camera for confirming the angle of view. UV reflecting mirrors in this measurement equipment are the dielectric multilayer mirrors. Calculated value and measured those reflection characteristics are shown in Fig. 4, respectively. The mirrors are designed so that the reflecting wave length of the mirror is 1~5nm. Strength of the light is in inverse proportion to the square of the distance between arc point and measurement equipment. In the case of arc measurement on the inspection cars, the measurement equipment is generally close to the arc point. However, in the case of arc measurement at the wayside, the arcing light travelling distance is longer than in the case of the measurements on the inspection car. Therefore, it is necessary to have higher sensitivity for the equipment. For this reason, the equipment uses the photomultiplier with the high sensitivity and uses the lens which converge UV ray on the sensor with the aim of high gain. The valid measurement distance of this equipment is about 6m. The valid angle of horizontal view of the equipment is ±. If arc is measured in place where it is 6m away from the track, valid horizontal measurement span of catenary is 43m. Arc detection result measured 6m away from the track in sunny noon and night in June is shown in Fig. 5. This shows that this equipment is able to measure the arc without the effect of sun. However, it needs attention that the equipment is hard to measure the arc which occurs at the shadow of the pantograph head. Fig. 3 The arc measurement equipment at the wayside by detecting UV ray

4 Reflection ratio (%) Wavelength (nm) Calculated value Experimental value Fig. 4 Characteristic of UV ray reflective mirror Voltage (V). 1. Noon 1 Night 1 Time (sec) Fig. 5 Arc detection result 4. Field test 4.1 Outline of field test From 7 to 9, several line tests to measure contact wire wear, pantograph contact force, arc due to pantograph contact loss, and pantograph collected current were carried out periodically in the two sections of Shinkansen commercial line. The condition of overhead catenary system at measurement section A and measurement section B is shown in Table 1. Arcing frequency is different in these measurement sections. Some arc was observed in measurement section A. On the other hand few arcs were observed in measurement section B. Overhead catenary system of two measurements sections was heavy compound catenary system. Table 1 Catenary condition of measurement sections Contact wire Tension of contact wire (kn) age of service (until 9/1) Number of pantograph (per a month) Length of measurement area(m) Gradient of track Section A GT-Sn17mm year 1month 6, Section B GT-Sn17mm year 3month 6, Measurement result of pantograph collected current By measuring current in the substation and the sub sectioning post which adjoins for two measurement sections, the collected current by pantograph passing through the measurement section was calculated. The schematic overview of the measurement is shown in Fig. 6. The measuring results of the quantity of pantograph collected current are shown in Fig. 7. Average quantity of collected current per a pantograph is 16A at measurement section A, and 9A at measurement section B. These large differences in quantity of collected current mainly depend on gradient of track. Fig. 6 Schematic overview of collected current measurement Collected current by a pantograph (A) 3 1 Measurement section A Measurement section B 1 3 Train velocity (km/h) Fig. 7 Measurement results of collected current

5 4.3 Measurement results of contact force and arc With the measurement theory of contact force on OCS described in chapter, the contact force measurements were carried out in the Shinkansen commercial line. The setup of the sensors for contact force measurement is shown in Figs. 8 and 9. The arc measurements were carried out at the same time using the measuring equipment described in chapter 3. The schematic overview of the measurement of arc is shown in Fig. 1. The measurement results on thickness of the contact wire, arc and contact force in measurement section A and measurement section B are shown in Figs. 11 and 1, respectively. In each figure, Figs. 11(a) and 1(a) indicate supported points of OCS and dropper points, Figs. 11(b) and 1(b) indicate measurement results of thickness of the contact wire, and Figs. 11(c) and 1(c) indicate arcing probability at each point, and Figs. 11(d) and 1(d) indicate a mean contact force at each point. The arcing probability was defined as a ratio of the number of the pantographs which generated the arc divided by the number of passage pantographs at each point. In these figures, the measurement results of thickness of contact wire were measured in October 9, and the arcing probability and the contact force were mean value of all measurement results for two years. The arcing probability and the mean value of contact force were calculated from the measurement results of about 1 pantographs. In these figures, the method of the sliding mean on base of 1m is applied to the arcing probability. These results and measurement results of collected current described in the section 3. shows that the wear progress of contact wire is fast in the position with high arcing probability in the measurement section A where the large current is collected. On the other hand the wear progress of contact wire is fast in the position with large contact force in the measurement section B where the small current is collected. Relationship among contact force, the arcing probability and contact wire wear in each measurement section is shown in Fig. 13, and these results were evaluated from the data of Figs. 11 and 1. The x- axis of this figure is the value that the arcing probability is divided by the velocity of train, in order to evaluate the energy which input to the contact wire per unit length. In this figure, dark color means that the wear of contact wire progresses. The wear area of contact wire was calculated from wear ratio of contact wire for two years test period. From this result, it is presumed that the arc is a main factor of contact wire wear in measurement section A and large contact force is a main factor of contact wire wear in measurement section B. Figure 14 is extended images of sliding surface of contact wire taken with the microscope on points,1 and, in Fig. 1 and point,3 in Fig. 11. Points,1 and,3 correspond to the position where the wear progress quickly in measurement section A and measurement section B, respectively. Figure 14(a) is the 5 times magnified image of the sliding surface contact wire at point,1. Black and copper-colored irregular areas which are considered as the arc spot can be observed. Figure 14(b) is the 5 times magnified image at point, and Fig. 14(c) is the 15 times magnified image at point,3. There is no arcing spot in these Figs. 14(b) and 14(c); sliding surface is smooth and linear scratches caused by mechanical contact are confirmed. Adhesion wear is a main causes of contact wire wear at Fig. 14(c), because scaly pattern which is made by metallic adhesion and concavity which is made by the spalling of transferred metal are observed. These results indicates that the mechanical sliding by the large contact force mainly causes contact wire wear at point,3. These observation results agree well with the conclusion inferred from Figs. 11, 1 and 13 as stated above. Measurement section 4m Telemeter Support point Measurement hanger Messenger wire AX wire Contact wire Accelerometer Pantograph Train running direction Fig. 8 Contact force measurement method in OCS

6 Fig. 9 Schematic overview of contact force measurement Fig. 1 Arc measurement equipment Fig. 11 Measurement results in section A Fig. 1 Measurement results in section B Contact force (N) (mm /year) (mm /year) Arcing probability x Arcing probability 6 x 1-4 divided by velocity divided by velocity (a) Measurement section A (b) Measurement section B Fig. 13 Wear map of contact wire Contact force (N)

7 (a) Sliding surface at point,1 ( 5) (b) Sliding surface at point, ( 5) (c) Sliding surface at point,3 ( 1) Fig. 14 Images of contact wire taken with a microscope 5. Prediction model of contact wire wear In this chapter, we have proposed a new contact wire wear prediction expression based on the data shown in the section 4. and the section 4.3. This wear prediction model has been constructed based on the heat model [1] of contact wire which H. Nagasawa used for the wear map of contact wire. At the beginning, it is assumed that contact wire wear was evaluated by heat quantities which are input to contact wire by the factors of contact force, arc, and electrical resistance between contact wire and contact strip. The model which indicates the area of contact wire wear due to the heat generated by each factor mentioned above after 1, pantographs passing was constructed. The wear area which progresses by the heat input to contact wire per unit length by the mechanical sliding, is defined by the following equation: I wf k1 1 F, (8) I where k 1 is a constant which converts the heat due to the contact force into the contact wire wear area; µ is a dynamic friction coefficient; α is a constant of the current lubrication; I is the collected currents; I is the constant of collected currents; and F is the contact force of pantograph. From the measurement results, α and I were set at. and 5 (A) respectively. In this equation, (1+I/I ) -α is the current lubrication effect [] which has been used against the wear due to contact force. The wear area which progresses by the heat input to contact wire per unit length by the arc is defined by the following equation: * I w k V, (9) v a where k is a constant which converts the heat of the arc into the contact wire wear area. V*, the electrical potential between the contact strip and the contact wire when arc is generated, is assumed to be a constant in this paper. ε is the arcing probability; v is the train velocity. The wear area which progresses by the heat input to the contact wire per unit length by the contact resistance between the contact wire and the contact strip is defined by the following equation:

8 I w, (1) r k3r v where k 3 is a constant which converts the heat of the contact resistance into the contact wire wear area. R, the electrical contact resistance between the contact strip and the contact wire, is assumed to be a constant in this paper. From the equations (8)~(1), the wear area which progress after passing of 1, pantographs is modeled by the following equation: I * I I w wf wa wr k1 1 F kv k3r I v v I I I Af 1 F Ba C, (11) R I v v * where Af k1, Ba kv, CR k3r. Based on this model, A f, B a and C R are identified as (mm /N 1 4 pantographs), (mm /N 1 4 pantographs) and (mm /N 1 4 pantographs) respectively by the least squares method in using data measured in measurement area A and measurement area B. The prediction result of the wear area of contact wire by the equation (11) is shown in Figs. 15 and 16. Figures 15(a) and 16(a) show the measurement and the prediction results of the wear area of contact wire per year. The measurement result is calculated from the wear progress for two years. The calculated values from equation (8) (9) (1) is shown in Figs. 15(b) and 16(b), respectively. These results are plotted in X-Y coordinates as shown in Fig. 17. This indicates that the prediction value generally agrees with the measurement one, therefore, the validity of the wear prediction model proposed in this paper can be confirmed. The contact loss ratio was measured on a specific test train, and contact loss ratio, namely time of contact loss divided by the whole time power running was calculated in the past. The contact loss ratio is average index in each drum. However, Fig. 15(b) shows that the arcing probability and contact wire wear correlate to each other depend on the collected current according to conditions. Therefore, it is supposed that the arcing probability is also important to manage the wear of contact wire in specific point. On this matter, it is necessary to verify further. With respect to the above discussion, it is difficult that all arcs are measured from the ground, since small arc is missed. As a result, this prediction model has a tendency to overestimate the effect of arcing on contact wire wear. Further, there is a possibility that the error of the wear area in the proposed wear prediction model have been included, since wear progress for two years is comparatively small. In addition, when plural pantographs in a train set pass continuously, the sliding surface of contact wire will be softened and the surface roughness will increase by the arcing of the 1st pantograph. As a result, the following pantograph will pass such sliding surface, and cause mechanical wear. This effect is not considered in the proposed wear prediction model. (mm /year) Measurement value Prediction value Distance (m) (mm /year) st term 3rd term nd term Distance (m) (a) Prediction result (b) Each term of prediction equation Fig. 15 Prediction results of the area of contact wire wear in Section A (mm /year).6.4. Mesurement value Prediciton value 1 3 Distance (m) (mm /year) st term 3rd term nd term 1 3 Distance (m) (a) Prediction result (b) Each term of prediction equation Fig. 16 Prediction results of the area of contact wire wear in Section B

9 Prediction value of wear area (mm /year) Section A Section B Measurment value of wear area (mm /year) Fig. 17 Prediction accuracy of contact wire wear area by the proposed model 6. Evaluation index for current collection Using the wear prediction model of contact wire mentioned above, case studies on an evaluation index of current collection performance were carried out. In this chapter, the condition in which the wear area of contact wire does not reach the following threshold value in a time span of a decade were examined in the section where number of passing pantograph is about 8/year. It is assumed that the type of catenary wire is heavy compound overhead catenary system, the type of contact wire is GT-Sn17mm, contact wire tension is 19.6kN and the permissible limit of crosssectional area of worn contact wire is set to 44.7mm. The result of survey is shown in Fig. 18. Figure 18(a) is the permissible value of the arcing probability against the collected current each velocity of the pantograph. It was assumed that the contact force of the pantograph is small enough to occur arcing frequently. In the area under the curve in this figure, life time of contact wire due to wear should be more than ten years. The effect of first term of equation (11) is not considered in order to discuss the area where the contact force is small and arc occurs frequently. This result shows that permissible arcing probability shall be lower, as the collected current increases or train speed decreases. The cause of this result is the effect of speed v which is included in the second term and third term of equation (11). Figure 18(b) is the characteristic of permissible value in the arcing probability against velocity for each collected current per a pantograph. In the area under the line in this figure, life time of contact wire due to wear should be more than ten years. However, the figure data in the low speed require careful attention, since most train passed over km/h in the sections where we measured data and the error of wear prediction model may increase by such as differences of wear mechanism in the low speed. The characteristic of collected current against the velocity is shown in Fig. 18(c). The permissible collected current per a pantograph can be derived under the condition that the contact force of the pantograph is 1N and that no arc occurred. In the area under the curve in this figure, life time of contact wire due to wear should be more than ten years. The second term of equation (11) is not considered, since the pantograph is premised on keeping in contact with the contact wire without arc. We will examine the low speed data, because the low speed data is not included in identifying the wear prediction model. The permissible collected current is small at the low speed. This reason depends on the speed v of denominator of third term in equation (11). In this wear prediction model, the effect of heat conduction and heat radiation of contact wire model are ignored, but it is predicted that this effect increase and should be not negligible at the low speed. These error causes are challenges for the future. 7. Conclusions In this paper, the measurement results of the contact force of pantographs, the arc and the collected current of all the trains which pass through two measurement sections of Shinkansen commercial line have been reported. Relationship among contact force, the arcing probability and contact wire wear was shown from these measurement data. In the measurement section with the large collected current, the wear progress of contact wire is greatly affected by high arcing probability. On the other hand, in the measurement section with the small collected current, the wear progress of contact wire is greatly affected by large contact force. Finally, this paper has proposed the prediction model of

10 Permissible arcing probability (%) km/h 3km/h 35km/h Permissible arcing probability (%) 6 4 1A 3A 5A 8A Collected Current(A) (a) Permissible arcing probability vs. collected current 1 3 Train velocity (km/h) (b) Velocity characteristics of permissible arcing probability 1 Permissible collcted currnet (A) Train velocity (km/h) (c) Velocity characteristics of permissible collected current Fig. 18 Criteria values to secure more than 1 years of contact wire life contact wire wear which has been developed based on these results and the case study results on evaluation index of the current collection performance. However, since our measurement data were obtained in high speed sections, the wear prediction model was not examined at the low speed. We will examine the low speed data. Problem raised by the case when plural pantographs in a train set pass continuously have been left unsolved. The solution of these problems should be challenged in the future. We are going to verify the effectiveness of the proposed wear prediction model from now on, using the measurement data and the experimental data on the test bench. Acknowledgments The authors would like to express our sincere thanks to the East Japan Railway Company for their cooperation with this study and permission to publish this paper. References [1] H. Nagasawa, Koji Kato. Wear mechanism of copper alloy wire sliding against iron-base strip under electric current, Wear, 16, pp , (1998). [] S. Bruni, G. Bucca and A. Collina, Pantograph catenary dynamic interaction in the medium high frequency range, Vehicle System Dynamics Supplement Vol.41, pp , (4). [3] T. USUDA, The pantograph contact force measurement method in overhead catenary system, WCRR, (8). [4] T. USUDA, Estimation of wear and strain of contact wire using contact force of pantograph, Quarterly Report of RTRI, Vol.48, No.3, pp17-175, (7).