WP E5: Noise from Traction Equipment and Running Gear

Transcription

1 Traction Equipment WP E5: Noise from Traction Equipment and Running Gear WP Leader: Ansaldobreda Partners: Lucchini, Trenitalia,, DB, SNCF SubContractor : UniNa S. Ferraiuolo, A. Miele Ansaldobreda SpA N. Rusciano, M. Viscardi,, S. Melchionna University of Naples Federico II Dept. Of Aerospace Engineering (massimo.viscardi@unina.it)

2 Activity Main objectives (TRACTION) to investigate the genesis and characteristics of the noise emission due to electric traction. to assess potential prototype solutions, taking into account integration into railway system constraints, global retrofitting feasibility and good cost/benefit ratio also from the maintenance point of view. To provide prototypes of improved components so that noise reduction achieved can be evaluated in an assembled system.

3 The SIRIO platform Low Floor Tram The objective is to reduce noise in the low speed range of the vehicle. The speed average in the centre of Milan is about 25km/h SIRIO MILANO MAIN CHARACTERISTICS Length ( 7 modules ) mm Length ( 5 modules ) mm Body width 2400 mm Maximum height 3400 mm Floor height from t.o.r 350 mm Seating places ( 7 modules ) 71 Seating places ( 5 modules ) 50 Total passenger capacity ( 7 modules ) 285 Total passenger capacity ( 5 modules ) 191 Wheel arrangement ( 7 modules ) Bo,2,2,Bo Wheel arrangement ( 5 modules ) Bo,2,Bo Rail Gauge 1445 mm Boogie wheelbase 1700 mm Wheel diameter 660 mm Line voltage 600 Vcc/Vdc Maximum speed 70 Km\h Maximum acceleration 1.0 m/s 2 Traction power 4x 106 kw

4 WP E5 objective Noise sources generalities Tram Main Noise Sources External Interior Rolling Noise Squeal Noise Traction Noise Auxiliary equipments Medium-high speeds Low speeds Low speeds Acceleration Deceleration Standstill/low speed Medium-high speeds Low speeds Low speeds Acceleration Deceleration Standstill/low speed

5 State-of of-the-art on Traction Noise 1/3 The noise sources related to electrical traction motor can be divided into two fields: Electromagnetic noise Windage noise Windage noise is generated by the interaction of the moving parts of the rotor when the cooling air passes through the motor. The primary sources of windage noise in an induction motor are the fans and the rotor bars in the vent areas and at the rotor ends.

6 State-of of-the-art on Traction Noise 2/3 Magnetic noise has two principal components: constant-level magnetic noise and load related magnetic noise. The constant-level magnetic noise is a result of the force and vibration that are generated by the interaction of the fundamental magnetic flux wave with the rotating magnetic parts of the rotor. This noise does not change in magnitude with load but can be minimised by the proper slot combination were established. The load-related magnetic noise is generated when current is induced into the rotor bars under an increasing load. The electric current in the bars create a magnetic fields around the bars that applies an attracting force in the stator teeth. These radial and tangential forces, which are applied to the stator teeth, create vibration and noise.

7 State-of of-the-art on Traction Noise 3/3 Force Distribution on the Stator The forces applied to the stator teeth are not evenly distributed to every tooth at any instant in time; they are applied with different magnitudes at different teeth, depending on the relative rotorstator-tooth location. This results in force waves over the stator circumference that will produce flexural modes m of vibration.

8 Vibro-Acoustic characterisation of the traction motor Acoustics Vibrations On-vehicle/pass by noise measurement (close to the engine/ inside the vehicle/ outside on the field) Vibration characterization of the motor on the vehicle Laboratory acoustic characterisation (sound intensity) Vibration characterization of the motor in the lab Operational deflection shape by the use of laser Vibrometry (laboratoty)

9 FEM-BEM numerical modeling Numerical Traction Motor FEM Model Calculation of the magnetic loads (true stator current spectrum) Calculation of the stator vibration field BEM Modeling (LMS-Sysnoise Sysnoise) Numerical-Experimental Acoustic Correlation

10 FIRST STEP: on vehicle traction noise characterization (special interest to electrical engine noise) The first phase of measurements were performed in AnsaldoBreda factory test track in Pistoia, and were addressed to best distinguish the relative contribution of traction equipment and of wheel-rail interaction noise.

11 The Ansaldobreda Pistoia test site Measurement section Total track lenght 800m TEST TRACK

12 Test set-up On- board mesurement set-up Inside the vehicle 2 microphones 2 accelerometers on the floor Motored and trailing boogie area Outside the vehicle 2 microphones in the boogie area 1 accelerometer on the boogie structure 5 accelerometers on the engine 2 microphones in the boogie area 1 accelerometer on the boogie structure Motored boogie Trailing boogie Optical dual trigger system

13 Motored boogie test set-up

14 Trailing boogie test set-up

15 Interior set-up Accelerometer on the floor Microphone: : 90 cm height

16 Pass-by section set-up Test set-up Microphone s s pass by antenna :: 3 microphones (Horizontal/vertical alligned) Beamforming system : 42 channels (wheel array) 1 microphone s s monitor station (7,5 mt from the center line) Optical dual trigger system

17 Pass-by section set-up

18 Pass-by section set-up (2)

19 Monitor station 7,5 m from the center line

20 Pass by measurements results TRACTION - passage at 20 km/h

21 Pass by measurements results TRACTION - passage at 20 km/h

22 s On board measurements results Microphone motored boogie (mic1) Vs Microphone leading boogie (mic4) db Pa s Spectrum Mic1 Bnc WF 301 [ s] Spectrum Mic4 Bnc WF 301 [ s] db Pa Hz Hz

23 On board measurements results Spectrum Mic1 Bnc s Spectrum Mic4 Bnc s Acceleration spectrum 22km/h Spectrum Mic1 Bnc s Spectrum Mic4 Bnc s Mic1 vs Mic4 Pa db db Pa Pa db db Pa Breaking spectrum 20km/h Mic1 vs Mic Octave 1/ Hz AL Octave 1/ Hz AL Engine typical band

24 ENGINE SOUND INTENSITY MEASUREMENTS ISO To simulate different loading condition, power supply was given at various step of current frequency supplying; measurements were performed with 6 frequency alimentation: 15, 20, 25, 30, 35 and 38 Hz. Autopower spectrum 1.00E E E-02 Pa2 1.00E E Hz 20 Hz 25Hz 1.00E E E Hz Autopower spectrum as function of speed

25 LASER VIBROMETER MEASUREMENTS The engine was supplied by the lab inverter with an alimentation frequency from 20 to 40 HZ step 5 Hz, also considering the most critical condition at 38 Hz, and the upper 50 and 60 Hz. The engine was divided into 5 scan area, 4 lateral ( front, rear, rotor and opposite rotor side) and the upper one, on every area the acquisitions were performed for each of the above mentioned frequency supply condition.

26 s Vibro-Acoustic Measurement On board acquisition s Spectrum Mic2 WF 301 [ s] Spectrum Acc6 WF 301 [ s] Vibration db g Acoustic db Pa Hz Hz Detail of accelerometer acquisition (left) and acoustic one (right)

27 Vibro-Acoustic Measurement Laboratory acquisition AutoPow er Sez5:4:S WF 283 [ rpm] rpm db rpm db Pa g AutoPow er Sez4:1:+Z WF 281 [ rpm] Hz Hz Detail of accelerometer acquisition (left) and acoustic one (right)

28 Preliminary conclusions and remark From comparison of field and laboratory tests, it appears a very similar behaviour of the engine both under the vibrational than the acoustic point of view. Both experimental and numerical characterization have pointed out t the strong pure tones character of the noise emitted by the traction motor, mainly in the low speed range. Due also to the high harmonics contents of the current supplied from the inverter in this frequency range. Starting from the results achieved, it could be possible to assess the best technological solution for noise sources reduction in lab, to eventually test it on the vehicle, later on. Taking into account the main genesis of noise, two main solutions s have been preliminary developed: 1.Noise control at source by implementation of Mixed Random Modulation PWM Techniques; 2.Noise control on propagation path by application of a prototype active skin panel. The most promising of these two solutions has been tested at Lab/Vehicle level.

29 t Techniques of PWM space vector modulation The current harmonics generated by PWM (Pulse Width Modulation) cause an increase in the magnetic noise of the motor. The switching frequency is relatively low for this kind of application (typically 1 2 KHz) and it is included in the audible band. The use of Random Modulation Techniques allows to distribute the higher tone energy all over the spectrum, getting a noise with a minor annoyance character. It is important to note that any modification to the PWM must not change the total power which is transferred to the motor, therefore the value of fundamental component of voltage and current must not be changed. Therefore we don t expect necessarily a reduction in the total sound power emitted by the traction but a better energy distribution over the spectrum and then a better Sound Quality of the sound perceived. Sirio Tram PWM modulator

30 The following techniques have been considered: 1.RANDOM PULSE POSITION (RPP) 2.RANDOM ZERO VECTOR DISTRIBUTION (RZD) 3.RANDOM CENTER DISPLACEMENT (RCD) 4.RANDOM SWITCHING FREQUENCY (RSF) It has been also considered the family of MIXED RANDOM MODULATION TECHNIQUES (MRMT), obtained by combining Random Pulse Position techniques with Random Switching Frequency technique. Two hybrid techniques have been developed applying RZD and RCD to a modulator which worked with RSF algorithm.

31 FIRST SIMULATION RESULTS FIRST SIMULATION RESULTS The effects of proposed modulation techniques have been evaluated using a Train System Simulator software that allow to compile the control software (the same compiled for the real control board and used on the vehicle) in a C++ environment for PC. Each physical element (control and interface boards, propulsion system, asynchronous motor, main line, wheel-track adhesion,...) is associated to a software object. Such tool is used both for the software unit test and for the system test. The simulation have been performed both in driving and coasting mode at different fixed value of output frequency (31 Hz, 35 Hz, 39 Hz, 44 Hz, 51 Hz, 61 Hz). By using the RSF or RPP techniques some reduction of the higher tones have been carried out at every frequency, but the best results have been carried out by using the MRMT. The figure reported shows a comparison of current spectra between standard PWM and MRMT techniques in driving mode with a fixed output frequency of 31Hz.

32 PRELIMINARY OPTIMIZATION The effects of different changes in the control parameters have preliminarily been tested using a Train System Simulator. Unmodified control Lab experimental noise and vibration measurements performed on the traction motor, have allowed both to evaluate the modification in terms of acoustic results and to extrapolate results at vehicle level through a simplified acoustic model. Modified control

33 Active Noise Control Technique An active-skin may be defined as a continuous covering of the vibrating portions of the system with active, independently controllable actuation (piezoelectric) driven elements and is designed to affect control by altering the continuous structural radiation impedance rather than structural vibration. Active skin panels present three peculiarities: the control system is dynamically separated from the vibrating surface, overcoming the problems related to the vibrating shape complexity and/or to mechanical impedance the shape of the panels may be ad-hoc designed in reference to the excitation frequency spectra the electronic system architecture may be highly simplified. Furthermore, the hybrid behaviour of such device may help to overcome limitation concerned with whole passive or whole active approaches to be effective over a wide frequency range.

34 ACTIVE SKIN PROTOTYPE The developed prototype consisted of a carbon-fibres composite panel (approximate 0,15 m2 area) activated via a set of piezoelectric patches (25x25x1 mm) whose driving signal voltage amplification is realized through a low weight reverse transformer unit. A picture of the skin device is here reported. Active Skin Prototipe First experimental activities were performed over a laboratory specimen, consisting of a rigid wall box, whose top surface was changed with the active skin. The test campaign has been divided into two phases: the first one using a test loudspeaker as noise source and the second (in progress) positioning the motor inside the box.

35 Overall Overall Passive control tests: : SPL measur. Punto 5-38Hz 100 Senza Con Punto 5-33Hz 100 Senza Con db db , ,5 Overall , ,5 1/3 octave band 1/3 octave band Punto 6-38Hz 100 Senza Con Punto 6-33Hz 100 Senza Con db db , , , ,5 38Hz Overall 20 33Hz 1/3 octave band 1/3 octave band

36 Harmonic Signal Control alghorithm Noise spectra is strictly correlated to the electrical driving frequency Adaptive feedforward The active portion of the hybrid system relies upon the application of the filtered-x LMS (least mean square) feedforward algorithm in a single-input, single output (SISO) configuration.

37 (33Hz) vibration noise Operative Operative test 100 OFF ON -1,1dB -5,7dB -1,5dB 80-12dB Overall , /3 octave band OFF ON 60 db Hz Overall , /3 octave band Hz -15,2dB 762 Hz -20dBV OFF ON 90 OFF ON db dbv dbv Hz Vibroacoustic analysis control system definition exp. tets

38 Conclusion and future work Passive peculiarities of the skin panel may guarantee a good TL in the whole frequency s range The active portion of the system may improve the TL performance at the main operating frequencies Future development : To increase the critical frequency of the skin panel To improve the algorithm s s performance and to test different ones (feedback or modified feedforward algorithms )

39 CONCLUSION at DR nodal point Solution 1 The simulations performed show that Mixed Random Modulation Techniques (MRMT) reduce the amplitude of the current harmonics generated by the deterministic PWM and are more effective with respect to single Random Modulation Techniques (RMT). A better energy distribution over the spectrum and then a better sound quality of the noise perceived has been found in preliminary lab tests. Both the simulations and the experimental results also show that the amplitude of the fundamental components of motor current and voltage have not been modified, so the dynamic performance of the vehicle shouldn t be affected significantly. Future development The tests on the vehicle should be oriented to the optimization of the algorithm used both for the dynamic performance of the vehicle and for the noise emission. Sound Quality parameters will be evaluated. Solution 2 Passive peculiarities of the skin panel may guarantee a good TL in the whole frequency s range. The active portion of the system may improve the TL performance at the main operating frequencies Future development: To increase the critical frequency of the skin panel To improve the algorithm s performance and to test different ones (feedback or modified feedforward algorithms).

40 Sinthesys of the ACTIVITIES DEVELOPED DURING THE LAST PERIOD The experimental noise and vibration measurements performed in laboratory l on the traction motor, have pointed out that the most promising random modulation technique is the Mixed RSF+RZD. The activities developed in the last period were oriented to the optimisation of the modified control in terms of acoustic results and of control accuracy. Among next slides, main results of the vehicle implemented solution, will be reported.

41 PWM space vector modulation main results 0022.S3A 0020.S3A db 01/07/ db 01/07/ ,50 31, A L 12,50 31, A L Hz LAeq LAeq Cursore: (A) Leq=86,0 db Cursore: (A) Leq=90,9 db Typical spectrum unmodified and modified algorithm at simulated max acceleration phase Hz 0022.S3A 0020.S3A db 01/07/ db 01/07/ ,50 31, A L LAeq Cursore: (A) Leq=82,5 db Hz 40 12,50 31, A L LAeq Cursore: (A) Leq=78,2 db Hz Typical spectrum unmodified and modified algorithm at simulated 20km/h (tractioned)

42 Conclusion and future work An in-depth characterization of the Sirio electrical engine has been realized A driving SW modification has been proposed and experimentally tested; interesting results have been measured in terms of annoying parameters An hybrid (passive/active)control methodology has been proposed and laboratory tested to reduce the radiated noise. Extended test have been planned to be performed.

43 Thank You!!!