Stationary uplift measurement as a diagnostic tool for pantograph monitoring H. Möller, H. Maly, B. Sarnes Abstract: The present paper describes a prototype of an automatic uplift measurement system capable for pantograph diagnostic. It s applications and limits will be discussed. Examples of detected defect pantographs will be given. 1. Introduction Uplift of the contact wire in catenaries caused by a pantograph is a well known safe critical and life cycle determining parameter. Therefore it is normally measured manual [1] [2] one time for admission of new pantographs or new catenaries. To use the knowledge of the uplift it was considered to measure it continuously and use this information to monitor the pantographs. In this paper a brief review of the theoretical aspects are given. Then the technical configuration of the system will be described and the results and experience of the 1 year measurement campaign presented. Future steps for data analysis are proclaimed. 2. Theoretical aspects An increased uplift can be caused by a defect or wrong adjusted pantograph, because uplift is direct proportional to the contact force and can be calculated knowing the elasticity at the mast. This force is the sum of static, dynamic and aerodynamic force, which are functions of correct working parameters of the pantograph. F contact = (F static + F aerodynamic ) v + F dynamic
Fstatic= static, pneumatic force at v train = 0 F aerodynamic = increase of contact force by aerodynamic effects F dynamic = increase of contact force by the interaction pantograph / catenary v = dynamic factor = a function of the train velocity and the acoustic wave speed in the catenary. The uplift (U) can be calculated from the contact force by: uplift (U) = contact force (F) x elasticity (E) = F x E A detailed description of the interaction pantograph / catenary and the mathematical calculations are described elsewhere [3] [4]. Therefore a measured uplift, overstepping (+/-) a threshold value (as a function of train velocity), indicates a defect (F aerodynamic or F dynamic are to large), worn (F aerodynamic or F dynamic are to large), or wrong adjusted (F static is wrong) pantograph. 3. Measurement technique The arrangement of the uplift set up is shown in fig 1,2 and 3.
Fig. 1: Arrangement of the uplift sensor: technical sketch Fig. 2: Photograph of the test point
Fig. 3: Photograph of the mechanical distance sensor Here, uplift of the contact wire is measured by using a mechanical distance sensor. The sensor is mounted on the mast and grounded. It is connected to the contact wire (at a tension of 15kV) using a 1m long insolating fiber. The fiber is connected with a normal wire. This wire is guided over a spool enable change from horizontal to vertical direction - to the contact wire. The electric signal is transmitted with a 50 meter long shielded cable to the amplifier. To achieve a ground free operation optical couplers are used. The signal of the amplifier is directly connected with the AD converter of a PC using the Microsoft Windows NT operating system. The real time aspect of the measurement data will be achieved by using the add on real time system Kithara [5]. That enables to collect data in definite time slots of 20msec, especially necessary to analyse the data in the frequency domain. Fig 4 shows the user interface of the uplift program developed for this application. Using a self calibrating procedure before each measurement an accuracy of 0.5 cm is achieved. This self calibrating is necessary to compensate the thermal expansion of length of the set up and the catenary.
Fig. 4: Uplift program user interface Train speed is measured by two magnetic wheel contacts and the wind speed is detected by a standard wind sensor. Data are transmitted via a standard GSM data modem. For train identification two different methods are tested: 1. Train number identification performed by OCR (optical number recognition) method In this case, passing trains are filmed using a camera with a high shutter speed. The photos are processed by an image processing tool. The output is the number in ASCII format. The correctness can be tested calculating the check sum and compare it with the test digit (the 7 th digit of the train number). Fig. 5 illustrates the train identification by OCR.
Fig. 5: Train identification by OCR This method is capable to detect the number at an arbitrary location on trains, with arbitrary letter sizes and of arbitrary trains (they have not to be prepared in any way). Problem: Illumination at night and bad weather condition (e.g. heavy rain, fog, etc.) can lead to bad or no images which results in no train identification. In this case a time stamp is stored in the data file and the identification must be performed using a timetable. 2. SOFIS [6] Using train identification system SOFIS a TEC transmitter has to be fixed to the trains. Presently more and more trains of DB rolling stock are equipped with this transponders (e.g. ICE 1-3, BR 101). Disadvantages of this system are that the trains of the local transportation as well as trains from other railways than DB have no transponder. They have to be identified using the timestamp in combination with a timetable. The combination of method 1 and 2 will result in almost all cases in a reliable train identification. 4. Results Functionality of the system as an important diagnostic tool is proofed by practical results. The automatic uplift measurement system has already detected several wrong adjusted and defect
pantographs. A selection of these cases will be presented. Furthermore a statistic over all events and the reliability of the system will be given. The pantograph check point is installed on the line Augsburg München. The technical data of that point are: RE 200 catenary max. permitted uplift = 12cm elasticity = 0.077 cm / N max. train speed: 200km/h Fig. 6 shows a typical uplift spectrum of a train with 2 normal operating pantographs. The speed is 200km/h which results in 12.1cm uplift from the first and 8.6cm uplift from the second pantograph. Fig 6: Uplift spectrum of a train using 2 pantographs We found that an uplift larger than 14.5cm on RE200 catenaries indicates a defect pantograph. Therefore we check all trains if they produce uplifts larger than 14.5cm at this specific measurement point.
Table 1 shows the results of the test point. In 12 month 36981 pantographs where monitored. In 79 cases we found an uplift larger than 14.5 cm. In 84% defects or wrong operating conditions were found when the system detects an uplift larger than 14.5cm and gives an alarm. False alarm percentage is only 16 % of all cases when we measured an extensive uplift. This shows that the pantograph check point produces highly reliable results. total number percentage Monitored pantographs (uplift) 36981 alarms (uplift > 14.5cm): 79 0.2 % (of all monitored pantographs) defects found on the alarmed pantographs 58 84 % (of all inspected alarms) false alarms 11 16 % (of all inspected alarms) not inspected alarms 10 7.6 % (of all alarms) Table 1: Results of the testpoint 79 36981 Several kinds of defects were found, using that method: broken springs of the pantograph head missing wind spoilers (see fig 7) wrong adjusted fairings wrong adjusted static contact force wrong horns mounted on a pantograph defect contact strip Fig. 7 shows an example of detected defects of a pantograph. In this case wind spoilers were missing. Therefore the aerodynamic behaviour of the pantograph was wrong and results in a high contact force and therefore a high uplift.
Fig. 7: Defect pantograph: missing wind spoiler (left image), correct pantograph: existing wind spoiler (right image) The improvements to the operating conditions of pantographs can be estimated by fig 8. It shows the number of uplifts larger than 14.5cm by 1000 measurements in dependence on time (months of monitoring). The decreasing number shows that a continuous monitoring will increase quality of pantograph settings. The fact that the alarm rate don t decrease to zero shows that a continuous monitoring is necessary to prevent an increasing number of trains with wrong or defect pantographs. Fig. 8 Number of alarms in dependence on the monitoring period
5. Diagnostic future aspects The quite good results of the uplift measurement are discussed above. One disadvantage of this method is that only meaningful defects of pantographs can be found. Another problem is that a large uplift indicates an error but not the type of error. To cope with this, another possibility to use this system as a more sophisticated tool will be the vibration analysis performed by FFT transformation of the uplift spectrum (fig. 9). The meaning of this modes to the pantograph are presently studied. Using vibration analysis methods might be a tool to detect defects of pantographs in a very early stage and make the method also capable for a maintenance tool. Fig. 9 Frequency domain of the uplift spectrum: characteristic vibration modes 6. Conclusion The present paper shows that stationary uplift measurement is an easy and proper diagnostic tool to increase quality of the interaction of pantograph and catenary. We estimate that we need approximately 8 to 10 measurement systems at DB net installed on major high speed tracks to monitor nearly the total high speed fleet and to get a good overview of all vehicles
using electric traction energy. It is possible to combine these pantograph check points with other diagnostic points, e.g. hot box detectors or the wheel flat detection systems. Total sum of pantograph / contact wire faults should decrease using a continuous monitoring network. The live cycle costs of the contact wire system will be reduced due to optimal operating conditions of the pantographs on all vehicles using DB network. References: [1] F. Kießling et al.: Fahrleitungen elektrischer Bahnen. B.G. Teubner, Stuttgart, 1997 [2] W. Harprecht, F. Kießling, R. Seifert Elektrische Bahnen 86 (1988). R. Oldenburgverlag, München 1988 [3] K. Petri, Vergleichende Untersuchungen von Berechnungsmodellen zur Simulation der Dynamik von Fahrleitung Stromabnehmer Systemen, HNI Verlagsschriftenreihe, Paderborn, 1996 [4] G. Poetsch, Untersuchung und Verbesserung numerischer Verfahren zur Simulation von Stromabnehmern Kettenwerk Systemen, Dissertation, Gesamthochschule Paderborn, 2000 [5] Kithara Software GmbH, www.kithara.de [6] SOFIS, Identifizierungssystem, Fa. Siemens AG