Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
History of numerical weather prediction wikipedia , lookup
Computational fluid dynamics wikipedia , lookup
Perceptual control theory wikipedia , lookup
Molecular dynamics wikipedia , lookup
Agent-based model wikipedia , lookup
Live, virtual, and constructive wikipedia , lookup
Military simulation wikipedia , lookup
Multi-state modeling of biomolecules wikipedia , lookup
NOVEL CASPOC-BASED SOFTWARE FOR MULTILEVEL SIMULATION OF SWITCHED RELUCTANCE DRIVES Alexei Matveev1, Peter J. van Duijsen2 Moscow Power Engineering Institute (Technical University) Krasnokazarmennaja str., 13, 111250, Moscow E-250, Russia phone: +7 095 974 14 47, fax: +7 095 974 15 54, e-mail: [email protected] 2 Simulation Research P. O. Box 397, NL-2400 AJ, Alphen aan den Rijn, The Netherlands phone: +31 172 492353, fax: +31 172 492477, e-mail: [email protected], www.caspoc.com 1 Abstract. In this paper a Rapid Application Development (RAD) tool for Switched Reluctance Machine (SRD) design is presented. The RAD tool is coupled to simulation software capable of simulating and optimizing the entire drive system. I. Introduction SR drives are beneficial and practical because they have a high torque to weight ratio and are very reliable. This makes them ideal for applications in industry, especially in the area of variable speed applications. SR motors can have various configurations [1] (number of phases, number of stator and rotor poles, winding connections), the converters for SR motors can also have various topologies (more then 20 in the classification of Barnes and Pollock [2]). Geometric shape of the motor cores, winding configuration and used materials also strongly influence the drive behaviour. All this makes the task of SRD designer very complicated and time consuming. Many different designs have to be analysed to find the best one. To estimate if some proposed mechanical and electrical design corresponds the desired specification and whether the designed drive fulfills the EMC requirements and can work safely regarding thermal and mechanical aspects the following characteristics have to be obtained for every of several chosen speeds in the working speed range: average torque generated continuously r.m.s. and peak values of phase current torque ripples drive efficiency drive power factor total harmonic distortion winding hot-spot temperature rise critical speeds acoustic noise To find most of the characteristics, the instantaneous values of voltage, current, flux linkage and torque have to be calculated on certain period of time, because the SRD operation is a series of transients in a highly nonlinear magnetic system, with no discernible steady state that can be expressed by simple algebraic formulas. Therefore, for design of SRD, a computer simulation with help of specialized program package is a sine qua non. All elements in the drive system (converter, machine, control unit, load, supply network) influence each other's behavior, and therefore it is impossible to analyze the elements separately, to get the overall behavior. So, simulation of the complete drive system (which is a typical mechatronic system) has to be performed. This means that the used package have to be able to carry out the simulation of the entire drive system. The task of the authors was to develop methods of design of SRD and software implementation of these methods. Short investigation of the existing simulation programs was done (section II). It was concluded that the best way to solve the problem is the development of the new specialized software on the basis of the existing simulation package CASPOC because CASPOC has intrinsic features for simulation of entire drive systems (section III). The created model of SR drive and it’s experimental verification are described in sections IV and V respectively. Some avdantageous features of the new software called CASPOC-SRD are presented in section VI. Conclusions and future plans can be found in section VII. II. Existing simulation programs There exists a large amount of programs for the simulation of electric circuits and drive systems [3]. Well-known programs are SimulinkTM and MatrixXTM for block-diagram simulations and SpiceTM and its derivatives for circuit simulation. Some design departments have their own proprietary software packages for the design of SRD. Alternatively, there are available commercial software packages such as the PC-SRD (Speed™) and the Tesla RAD design suite [4]. The modeling and simulation programs are mainly devoted towards one type of technique. It is either a circuit simulation for power electronics, a block-diagram oriented program for the electrical machine or mechanical load or a high level programming language for the control. The modeling and simulation of the entire system, the power converters, electrical machine, mechanical load and control, is found only rarely. III. CASPOC advantages in simulation of electric drives The application of mechatronic systems requires more modeling efforts compared to modeling of regular electronic systems. This is caused by the difference in models for the electrical machine, the power converter and the control of the drive. The solution for this problem is a multilevel modeling and simulation program. The combination of the circuit model, block-diagram model and modeling language (Pascal, C) is called the multilevel model. The simulation package CASPOC [5] exploits the multilevel model concept. The dynamics of the different subsystems are connected via the multilevel model and therefore the simulation shows the dynamic behavior of the total system. Influences in, for example, the mechanical load which lead to different waveforms in the power converter are directly visible. A main problem with the simulation of mechatronic systems is the simulation time. The simulation for the power converter requires a time step in the order of microseconds. The dynamics of the drive are mostly in the order of seconds. Therefore a large number of time steps is required to simulate the entire behavior of the total drive system. A requirement for the simulation program is therefore that the simulation has to be very fast. This is achieved by applying special ideal models for semiconductor models. CASPOC provides these ideal models. torque curves T f (i, ) are calculated analytically as well from the magnetization curves and the equation i T The static magnetization and torque curves calculated beforehand are then used in simulation. The simulation of the processes in the machine is carried out with help of block-diagram and modeling language. The model core is the well-know phase voltage equation u i*r u ω T u Power Power converter converter (Circuit (Circuit model) model) i SR SR machine machine (B-D (B-D ++ ML) ML) γ s Control Control (Modeling (Modeling language) language) Fig. 1. Multilevel model of SR drive. IV. Multilevel model of SR drive The first multilevel model of a SR drive was included in CASPOC-2001 as an example [6]. The model used a simplified linear model of the SR machine and did not take into account supply net and rectifier, therefore it could be used for equcational purposes only. The multilevel model presented in this paper (Fig. 1) is more advanced. The improvements are first of all in the machine model. For every proposed design of the machine (chosen configuration, geometry, winding data, active materials) the magnetization curves f (i, ) are calculated analytically using the method similar to that of Miller and McGilp [7]. End-effects are taken into account with help of expressions from [8]. The static are Other drive components are modeled as follows: Mechanical Mechanical load load (Block-diagram) (Block-diagram) i d . dt Parameters changes during the simulation provided by means of ”look-up table”-blocks. Supply Supply net net (Circuit (Circuit model) model) dW ' , where W ' di . 0 d i const The differential equation describing the dynamics of the mechanical part is modeled by a blockdiagram. In the block-diagram a time function for the loading torque can be described, or an equation can be build by blocks which describe the relation between, for example, the torque of the load, inertia, friction and the angular frequency. The power converter and the supply net are modeled in a circuit model. Only the interconnections between the elements in the circuit have to be given. This gives an opportunity to model converters of various topologies. There are available special idealized models for semiconductor switches, which require a minimum of simulation time. A digital control is described in the modeling language. The firing of the switches in the power converter is made dependent on the position of the rotor The models of all the drive components are coupled in one multilevel model. Two numerical methods are used for the numerical integration of the differential equations in the models. The differential equations for the inductors and capacitors in the circuit model are numerically integrated using the trapezoidal method. The integrators in the block-diagram and the modeling language are numerically integrated using the Runge Kutta 4th order method. V. Experimental verification of the proposed mathematical model Acquisition of test results for the verification of the proposed mathematical model was carried out in two ways. First, the tests of several SR motors of various configurations (8/6, 12/8, 6/4) were performed in the lab (Fig. 2). Second, test results reported by another researches in conference papers, thesises etc. were accumulated into special database. Simulation results were compared to experimental data and model tuning was done. Fig. 2. One of tested SR motors (left) and DC-machine (right) as a loading device. Fig. 3. Measured and calculated magnetization curves. Fig. 5. Measured and simulated current waves at 1000 rpm, turn-on 4o, turn-off 19o, PWM duty 50%. Fig. 4. Measured and calculated static torque curves. Fig. 6. Measured and simulated current waves at 1500 rpm, turn-on 0o, turn-off 15o, PWM duty 50%. In the figures 3, 4, 5 and 6 a comparison of measurements and calculated data are shown for one of the SRMs (main parameters can be found in the Appendix). The presented comparison shows good agreement of the experiments and simulations. VI. CASPOC-SRD: new possibilities in simulation of SR drives The novel software called CASPOC-SRD is a standalone application built on the basis of standard CASPOC software. Inside a simulation project, a prototype design of a SR machine is embedded. In the SRD design tool the SR machine is prototyped. From the SRD design tool the parameters for the behavioral model of the SR machine are exported and used in the total drive simulation. Based on the simulation results of the total drive system, the prototype can be optimized in the SRD design tool. Compared to existing packages the coupling to the total drive simulation brings the advantage to optimize the SRM for any possible drive configuration. Not only the SRM can be optimized, but the total drive system can be optimized. MOTOR DATA configuration geometry winding data materials CONVERTER DATA Advantages of a direct coupling between a SR machine prototype and a drive simulation are: Test the prototype with a more realistic voltage waveform instead of an ideal square wave. Harmonics in the power converter are influencing the produced torque. Non-linear loads (for example, periodic or pulsating) influence the SR machine behavior. Delays in the control influence the maximum speed, which can be tested for worst case situations The simulation shows the behavior of the entire SRD system (Fig 9). The Graphical User’s Interface (Fig. 8) is similar to that of standard CASPOC. Instant values of current, voltage, torque etc. can be displayed in graphical or text format. VII. Conclusions and future plans The presented model makes it possible to investigate the electromagnetic behavior of the SR drive. It is important to point out that the thermal, mechanical and vibro-acoustic processes have to be analyzed as well. In the future versions of CASPOC-SRD analytical models for estimation of temperature rise, acoustic noise and critical speeds will be available (Fig. 7). control parameters MULTILEVEL SRD MODEL topology diods & IGBTs resistances phase current torque torque ripples efficiency power factor THD temperature rise critical speed acoustic noise speed (To be available in future versions) Fig. 7. Present capabilities of the software and the future plans. Fig. 8.CASPOC-SRD: Motor data entry and simulation ofa 8/6 SRD. REFERENCES [1] T.J.E. Miller: Optimal Design of Switched Reluctance Motors. - IEEE Transactions on Industrial Electronics, Vol. 49, No. 1, February 2002. [2] M. Barnes, C. Pollock: Power Electronic Converters for Switched Reluctance Drives. - IEEE Transactions on Power Electronics, Vol. 13, No. 6, November 1998. [3] N. Mohan, W.P. Robbins, T.M. Undeland, R. Nilssen: Simulation of Power Electronic and Motion Control Systems - An Overview, Proceedings of the IEEE, Volume: 82 8 , Aug. 1994 , Pages: 128 –130. [4]. Gospodaric D. van Duijsen P.J., Rapid Application Development tool Telsa for Fast Prototyping of Electrical Machines, Proceedings PCIM 2003 [5] P.J. van Duijsen: Multilevel Modeling and Simulation of Power Electronic Converters and Drive Systems. Proceedings Power Conversion (PCIM), 1994. [6] P.J. van Duijsen: Multilevel Modeling and Simulation of a Switched Reluctance Machine. Proceedings Power Conversion (PCIM), 1995. [7] T.J.E. Miller, M. McGilp: Nonlinear Theory of the SRM for Rapid Computer-Aided Design. - IEE Proc. Vol. 137, Pt. B. ¹6, November 1990. [8] A. Matveev, V. Kuzmichev, E. Lomonova: A New Comprehensive Approach to Estimation of End-Effects in Switched Reluctance Motors. – ICEM-2002, Brugge, Belgium, 26-30 August, 2002. [9] Caspoc User Manual, Simulation Research, 2003, www.caspoc.com APPENDIX The steel curve: B, T H, A/m B, T H, A/m 0 2 1.51 1300 0.7 500 1.57 2000 1.27 900 1.7 5000 1.44 1100 1.9 14000 The SR motor data: General data Rotor Tooth Height 16.1 mm Rated Power 5 kW Stator Yoke 12 mm Rated Speed 1500 rpm Rotor Yoke 17 mm Shaft Diameter 50 mm Configuration Stator Poles 8 Winding data Rotor Poles 6 Number of Turns in a Coil 70 Coils in Parallel 2 1.18 mm Geometry Stator Diameter 206 mm Single Strand Diameter Rotor Diameter 116.2 mm Supply Air Gap 0.4 mm DC voltage Stack Length 150 mm Converter type – Stator Tooth Width 20.6 mm asymmetrical half-bridge Rotor Tooth Width 23.8 mm Stator Tooth Height 32.5 mm Fig. 9. Simulation of a SR machine with experimental results 540 V