Available online at ScienceDirect. Procedia Engineering 192 (2017 )

Transcription

1 Available online at ScienceDirect Procedia Engineering 92 (207 ) TRANSCOM 207: International scientific conference on sustainable, modern and safe transport Model railway traction performance measurements and analysis Pavel Sovička a *, Matěj Pácha a, Pavol Rafajdus a a Department of Power Electrical Systems, Faculty of Electrical Engineering, University of Žilina, Univerzitná, Žilina Abstract The purpose of this paper is to introduce measurements and analysis of a :8 scale model railroad in order to provide the possibility to use modern locomotive design methods in this application. Tractive force and vehicle speed are necessary for this analysis and therefore a measuring train car has been developed. On-board strain gage and a magnetic encoder are used to obtain the necessary data. Therefore a part of this paper is devoted to describing the usage of these sensors. Obtained data is processed, analyzed and also compared to a full scale railroad. 207 The The Authors. Published Published by Elsevier by Elsevier Ltd. This Ltd. is an open access article under the CC BY-NC-ND license ( Peer-review under responsibility of the scientific committee of TRANSCOM 207: International scientific conference on Peer-review sustainable, under modern responsibility and safe transport. of the scientific committee of TRANSCOM 207: International scientific conference on sustainable, modern and safe transport Keywords: model railroad; large scale models; :8 scale; railway performance; traction measuring; tractive force calculation. Introduction In the contrary to a popular belief, model trains are actually older as full scale railroad. In the beginnings they were used to experimentally validate physical concepts, for example adhesion. Later they were used for advertising products of locomotive manufacturers and only after that they made their way to hobbyists and children. There are many scales of models available from many vendors, but also from hobbyists around the world. In Czech Republic the most popular scales are :20 known as TT scale and :87 known as H0 scale. However, these models proved somewhat small to some enthusiasts and therefore large scale models were created. One example is the :8 scale, which will be further discussed in this article. These models are much closer to the full scale railroad. Fully functional steam engines hauling * Corresponding author. Tel.: ; address: pavel.sovicka@fel.uniza.sk The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license ( Peer-review under responsibility of the scientific committee of TRANSCOM 207: International scientific conference on sustainable, modern and safe transport doi:0.06/j.proeng

2 Pavel Sovička et al. / Procedia Engineering 92 ( 207 ) carriages with passengers are very popular. The purpose of this paper is to investigate the possibility to use modern train design methods to improve model train designs and to provide necessary input data for the calculations [], [2]... Sičova Drobná Železnice o. s. Equipment of Sičova Drobná Železnice o. s. (known as SDŽ ) was used for the analysis carried in this paper. SDŽ is a hobbyist model railroad named after the builder of the first locomotive, which is based in the city of Ostrava, Czech Republic. Currently SDŽ operates two fully functional steam locomotive models designated and and a variety of carriage types, mostly Pa 3 platform carriages. SDŽ also owns more than 00 m of movable railroad including curved sections and switches. The model railroad uses a 84 mm gauge according to NEM 30. norm [3], which enables interoperability between different model railroads built according to this standard. To acquire necessary data for the analysis a measuring carriage designated MV- was built using the know-how provided by SDŽ members [4]. 2. Measurements As described above a measuring carriage was needed for the necessary measurements. Therefore, the MV- measuring carriage was developed and built. This chapter describes the carriage design as well as methods to process sensor outputs. A strain gage was used to measure tractive effort on the coupling while encoder was used to measure speed. The final part of this chapter is devoted to acquired data processing. All measured data presented in this work were obtained during test runs on a 50 meter long test track. Due to the length of the track and length of the measuring consists, the measuring train travelled only about 35 meters on each run. 2.. MV- measuring carriage The main purpose of this carriage is to measure force on the coupler and speed. For this reason, it was conceived as a measuring device rather than an accurate model. The main dimensions were loosely based on the Ke construction group 7 type carriage operated in the past by Czechoslovak Railways [5]. The MV- is a two-axle carriage with a 84 mm gauge and 25 mm wheel diameter built according to [3]. The frame is a meter long double H letter shape with two crossbars made from steel profiles. The front side of the carriage is the measuring side. It features a Vishay model 64 strain gage mounted into the coupling hook, shortened bumpers to prevent interference caused by bumper touching and RLS RM22IC0009B0 magnetic encoder mounted on the front axle. Sensor outputs are connected to a 56F8322 controller board and transferred to a Freemaster application using a JTAG interface. The rear side of the carriage has a standard coupling and bumper arrangement. Fig. The MV- measuring carriage. Fig 2. 3D development model of the MV- measuring carriage Sensor output processing The strain gage uses a balanced bridge configuration. The bridge is powered by a stable voltage source and the output is connected to an external 24bit ΣΔ (sigma-delta) analog-to-digital converter. This converter is convenient for this application due to its high accuracy. Also the low sampling rate is not a limitation, because high sampling rates are not required during these measurements. The converter is connected by a SPI bus to the controller. No scaling or

3 826 Pavel Sovička et al. / Procedia Engineering 92 ( 207 ) calibration is done in the controller itself. However, the ΣΔ converter is commanded for a self-calibration routine via SPI. A laptop is connected to the controller by a Codewarrior USB tap (USB-to-JTAG interface). Freemaster application used on the laptop handles the communication as well as scaling and data recording to output data files, which can be further processed by other software. Because the authors did not have access to an appropriately calibrated force metering unit, the calibration was done using a calibrated weighing-machine. The carriage itself was measured to weigh 8.86 kg. Then it was hanged by the strain gage equipped coupling. Thanks to the known weight of the carriage, the corresponding gravitational force was calculated. The measured value was calibrated to match this gravitational force. The encoder is connected to the controller using a simple interface board and was calibrated according to the methodology described in [6]. p N T s () e f 2 N T r s (2) Maximum expected speed was determined to be 4 m/s (4.4 km/h). To achieve suitable information availability a sample time T s = 0. s was used. The RM22IC0009B0 encoder has N r = 52 impulses per revolution in quadrature mode, or θ p=0.023 radian per one impulse resolution. Using formula () and setting up the counter to a maximum value of ΔN = 52 a maximum measurable speed is rad/s, or 3.94 m/s. As can be seen in Fig 3., the relative speed error above 0.5 m/s is less than 4%, while at lower speeds the relative error e f is increasing rapidly. The absolute error is 0.02 m/s in the entire speed range. Fig 3. Relative speed error as a function of speed given by (2). Fig 4. Example of measured and filtered values of tractive force and speed. During this test, the locomotive was pulling a consist weighing 720 kg Data preprocessing Before analysis was carried out, the acquired data had to be preprocessed due to the high amount of noise present. This noise is caused by two factors: Signal noise in the measurement system. Since the controller board was placed in a temporary container, relatively long cables contributed to the signal noise. Torque ripple from the steam locomotive. The steam engine does not produce a smooth constant torque. Also the output power control is not smooth.

4 Pavel Sovička et al. / Procedia Engineering 92 ( 207 ) Acquired data were filtered using Matlab software by a digital IIR filter using a transfer function shown in equation (3), which was implemented using an equation shown in (4). G Y B B z A z 0 z n A Y n B X n B X n 0 (3) (4) In the above equations the A, B 0 and B are respective filter coefficients, z is the discrete operator, n is the current step, X is the filter input and Y is the filter output. The coefficients were calculated using MATLAB Butterworth filter design function for a low-pass filter with a bandwidth frequency of Hz. The data sampling frequency varied slightly around 65 Hz, which is the sampling time value that was used to calculate the filter coefficients. This variation is caused by the Freemaster application properties in Scope mode. To eliminate possible errors introduced by different phase shifting, all filters used in the preprocessing used the same filter type and settings. An example of measured and filtered tractive effort and speed can be seen in Fig Data analysis After processing, the data was analyzed. The traction characteristic of the measured class model locomotive was determined as well as specific rolling resistance of the Pa 3 carriage. When measuring the locomotive characteristics, the MV- front was coupled to the locomotive. When measuring the rolling resistance, the MV- front was coupled to the load. Obtained results are compared to a full scale railroad to find any similarities in terms of performance between these large scale models and real railroad. 3.. Analysis of obtained data The filtered tractive force F = f(t) and speed V = f(t) of each measuring run were combined into traction characteristic F = f(v). Several measuring runs were conducted on the test track with different loads: Regular operation load These runs consisted of up to 4 carriages (including the MV-) with a random number of passengers, simulating regular operation. Heavy weight testing Similar to regular operation tests but with as much passengers as possible. At one of the runs a total load of.3 tons, including carriages, has been pulled at a speed of 0.5 m/s. High speed testing During these runs, the shortest and lightest load of MV- and one Pa 3 carriage with only the train driver and measuring operator on board was used with the aim of reaching the maximum speed. The resulting tractive effort characteristics from each run were combined. The highest achieved value of tractive force for the given speed was taken out of each characteristic and approximated to form the final traction characteristic of the class model steam locomotive, which can be seen in Fig 5. The power output was calculated by simply multiplying the tractive force with speed. The adhesion coefficient μ was also determined. With a maximum achieved tractive effort of 82.3 N and a weight of 79.4 kg, including the water onboard, the coefficient was determined to be using the formula shown in (5), where F tmax is the maximum developed tractive effort, F adh is the adhesion force, m is the locomotive adhesion mass and g is the gravitational constant. max Ft F max adh Ft max m g (5)

5 828 Pavel Sovička et al. / Procedia Engineering 92 ( 207 ) Fig 5. Traction characteristics and power output of the model steam locomotive. Fig 6. Specific resistance of empty (upper picture) and loaded (lower picture) Pa 3 carriages. Specific rolling resistance of Pa 3 carriages was also measured. The MV- was turned towards the load instead of the locomotive. Two Pa 3 carriages were measured empty and loaded. The obtained data was analyzed in the same way as the locomotive measurement data. But in this case, the force F = f(v) was a resistance force of the load. This force was divided by the mass of the carriages and by gravitational constant and the results can be seen in Fig 6. As described in [7], the resulting graph should be a rising quadratic function. This trend was not found in the results. A possible explanation is that the theory accounts for much higher speeds, while the measurement was conducted at very low speeds, only up to 2 m/s. At such low speeds the value should be approximately constant and any variations can be contributed to other phenomena, such as axle track mutual interaction Comparison to full scale railroad Table compares the steam locomotive model to its full scale counterpart, the third series of class 30 steam locomotive. To facilitate the comparison, the respective ratios of given values are added. Table. Comparison between a model steam locomotive and a full scale class 30 steam locomotive [4, 8, 9]. Parameter Model Full scale Ratio Gauge 84 mm 435 mm : 7.8 Length over bumpers 956 mm mm : 8.3 Wheel arrangement C C - Wheel diameter 6 mm 930 mm : 8 Empty weight 65 kg 30.6 t : Number of steam cylinders 2 2 Diameter of steam cylinders 39 mm 345 mm : 9. Number of pipes in the boiler : 5 Maximum boiler pressure 5 atm atm : 2.2 Approximate maximum power 60 W 230 kw : Traction force at comparable speed* 7 N 20.7 kn : 77 Maximum speed 0 km/h 40 km/h : 4 * 5 km/h for the model and 40 km/h for the full scale locomotive It can be seen that mechanical values are nearly perfectly in the :8 scale of the model, but performance parameters are usually much lower, especially the power output. This is because of the lower pressure, lower number of boiler pipes and also :8 scale means (:8) 2, or :64 scale in cylinder surface and (:8) 3, or :52 scale in boiler volume. Power output of a steam engine depends heavily on pressure and volume. A notable exception is the maximum speed,

6 Pavel Sovička et al. / Procedia Engineering 92 ( 207 ) which is double of what it should be after scaling. This is usual with model railroads, especially with the smaller scales. Table 2 compares obtained specific resistances of the model with specific resistances determined for full scale 4 axle cargo carriages. Because of the low speeds, the value used corresponds to zero speed. As expected, the specific resistances are much higher than on the full scale railroad. It can also be noted, that the ratio between the loaded and empty model specific resistance is 0.6, while the same ratio for the full scale railroad is 0.7. Table 2. Comparison of specific resistance for zero speed between Pa 3 model carriages and full scale 4 axle cargo carriages [0]. Specific resistance Model Full scale Ratio Empty 25 N/kN 2 N/kN 2.5 : Loaded 5 N/kN.4 N/kN 0.7 : 4. Conclusions Locomotive design methods require input data, most notably locomotive performance, specific load resistance and adhesion parameters. While this work provides these information s, the measurements were carried out using only one model locomotive and one type of carriages on a simple 50 m long track, of which only 35 meters were used. Because of this, these data can be considered only as a basic guidance. More measurement runs with different locomotives and loads will have to be made before the locomotive design methods potential can be fully utilized. A notable and partly unexpected outcome is that a 73 kg weighing steam locomotive model with power output around mere 60 W was able to haul.3 tons of load, an equivalent to a small car, although at a very low speed. A more significant impact of this work can be seen in the developed methodology. This paper describes the design of a simple measuring carriage and the algorithms used to process raw data into useful information. Therefore, it can serve as a cornerstone for further research in this area and is already being used to develop a fully electric battery powered model of a Slovak railways class 99 locomotive. Acknowledgements The authors of this article would like to thank Sičova Drobná Železnice, o. s. for the provided support in design of MV- and for providing a complete model railroad set, needed for measurements. This work was supported by the Slovak Scientific Grant Agency (VEGA /0957/6) and by the Department of Power Electrical Systems of University of Žilina. References [] L. Losos, Vlaky dětských snů (Childhood dream trains), Prague: Nakladatelství dopravy a spojů, ISBN: , 989. [2] V. Zuska, Modely železnice v terénu (Railroad models in terrain), Prague: Nakladatelství dopravy a spojů, ISBN: , 988. [3] Normen Europäischer Modellbahnen 30.. [4] Sičova drobná železnice, o.s., [Online]. Available: [Přístup získán 9 207]. [5] Z. Maruna, Československá železniční vozidla - Soubor monografií pro přátele železnic a železniční modeláře (Czechoslovak railroad vehicles - A set of monographies for railroad friends and railway modelers), Prague: Nakladatelství dopravy a spojů, 97. [6] I. I. Incze, A. Negrea, M. Imecs a C. Szabó, Incremental Encoder Based Position and Speed, Acta Universitatis Sapientiae Electrical and Mechanical Engineering, ISSN: , 200. [7] J. Drábek, Energetické vlastnosti železničních dopravních systémů (Energetics properties of railway transport systems), Katedra výkonových elektrotechnických systémov ŽU v Žiline, Zilina, 200. [8] J. Bek a Z. Bek, Encyklopedie Železnice - Parní lokomotivy ČSD [] (Encyclopedia of Railroad - CSD steam locomotives []), Nakladatelství Corona spol. s.r.o., ISBN: , 999. [9] J. Koutný a M. Leštinský, Parní lokomotivy řady 30.0 (Class 30.0 steam locomotives), Prague: Nakladatelství Corona spol. s.r.o., ISBN: X, [0] F. Jansa, Dynamika a energetika elektrické trakce (Dynamics and energetics of electric traction), Prague: Nakladatelství dopravy a spojů, 980.