TRANSACTIONS ON ELECTRICAL ENGINEERING

CONTENTS Mašek, Z.: Control System for Hydrostatic Transmission of Railcar M27 . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

56 – 61

Dvořák, P., Fajt, T., Sošková, I. and Hloužek, J.: Water Cooled Induction Traction Motor for 100 % Low Floor Tram Car . . .

62 – 68

Kopecký, M., Švanda, J. and Vlček, M.: Real-time Simulation of 3 Parallel PWM Rectifiers . . . . . . . . . . . . . . . . . . . . . .

69 – 77

Foltyn, D.: Electric Drives in the Modernized Power Plant Tusimice II . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

78 – 81

Dvorský, E., Raková, L.: Modelling of Regulatory Electricity Networks Ability with Rotating Flywheels . . . . . . . . . . . . .

82 – 85

Vol. 3 (2014)

No.

3

ERGO NOMEN

pp.

56 - 85

TRANSACTIONS ON ELECTRICAL ENGINEERING Publisher:

ERGO NOMEN, o.p.s., K13114 FEE CTU in Prague, Technicka 1902/2, 166 27 Praha 6, Czech Republic E-mail: [email protected]

Editorial Office:

PIVONKA Pavel BAUER Jan HAVLICEK Radek MERICKA Jiri NOVA Ivana VONDRICH Jiri ZDENEK Jiri

Periodicity: Language: Scope: On-line version:

Quarterly English International scientific journal of electrical engineering www.transoneleng.org ISSN 1805-3386

Each paper in the journal is evaluated by two reviewers under the supervision of the International Editorial Board. International Editorial Board Editor in Chief: Prof. LETTL Jiri, Czech Technical University in Prague, Czech Republic Members: Prof. BAUER Palo, Delft University of Technology, Netherlands Prof. BRANDSTETTER Pavel, VSB-Technical University of Ostrava, Czech Republic Prof. DOLEZEL Ivo, The Academy of Sciences of the Czech Republic, Czech Republic Prof. DUDRIK Jaroslav, Technical University of Kosice, Slovakia Prof. NAGY Istvan, Budapest University of Technology, Hungary Prof. NOVAK Jaroslav, University of Pardubice, Czech Republic Prof. ORLOWSKA-KOWALSKA Teresa, Wroclaw University of Technology, Poland Prof. PEROUTKA Zdenek, University of West Bohemia, Czech Republic Prof. PONICK Bernd, Leibniz University of Hannover, Germany Prof. RICHTER Ales, Technical University of Liberec, Czech Republic Prof. RYVKIN Sergey, Russian Academy of Sciences, Russia Prof. SKALICKY Jiri, Brno University of Technology, Czech Republic Prof. VITTEK Jan, University of Zilina, Slovakia Prof. WEISS Helmut, University of Leoben, Austria

Responsibility for the contents of all the published papers and technical notes is upon the authors. Template in MS WORD and basic typographic rules to be followed see www.transoneleng.org.

Copyright:

©2014 ERGO NOMEN, o.p.s. All right reserved.

Transactions on Electrical Engineering, Vol. 3 (2014), No. 3

56

Control System for Hydrostatic Transmission of Railcar M27 Zdeněk Mašek University of Pardubice, The Jan Perner Transport Faculty, Department of electrical and electronic engineering and signalling in transport (KEEZ), Czech Republic, e-mail: [email protected]

Abstract — The paper describes functions of a control system for hydrostatic transmission for the reconstructed railcar M27 which is used for transport of passengers on a narrow gauge railway near Jindřichův Hradec, Czech Republic. Software for this control system was developed at the University of Pardubice. Keywords: rail vehicle, control system, hydrostatic transmission, tractive effort, JHMD, M27, MUV 74.1.

I. INTRODUCTION Jindřichohradecké místní dráhy a.s. (JHMD) company provides regular railway service on a narrow gauge railway between Jindřichův Hradec and Obrataň in the Czech Republic. During the year 2012 the JHMD decided to start reconstruction of four railcars of type M27 (805.9). Originally the M27 was manufactured in Romania in the middle of 80’s. TABLE I summarizes technical parameters of the original railcar M27 (805.9). TABLE I. TYPE SIZES FOR CAMERA-READY PAPERS Gauge

750 mm

Length

15 920 mm

Weight

24,5 t

Wheelset arrangement

B’ 2’

Maximum towing capacity

57 kN

Engine type

Raba-MAN D2156HM6U

Engine output

141 kW

Transmission

Hydrodynamic

Maximum speed

60 km/h

The reconstructed railcar M27 has completely new design, engine, new hydrostatic transmission, wheelset arrangement, control system, interior, air conditioning, lights etc. Main frame and bogies are almost the same as on the original M27.

A hydrostatic transmission is used instead of the original hydrodynamic transmission. The control system for the hydrostatic transmission used in the reconstructed vehicle was developed at the University of Pardubice/DFJP-KEEZ and at first it has been succesfully used on special rail vehicles MUV 74.1 by CZ Loko company during the years 2012 and 2013. TABLE II summarizes technical parameters of the M27 after reconstruction. TABLE II. TYPE SIZES FOR CAMERA-READY PAPERS Gauge

750 mm

Length

15 920 mm

Weight

24,5 t

Wheelset arrangement

B’ B’

Maximum towing capacity

24 kN

Engine type

Tedom 242R6VHTA26

Engine output

242 kW

Transmission

Hydrostatic (Parker) 2x hydraulic pump PV270 2x hydraulic motor F12-250

Maximum speed

60 km/h

II. WHEELSET ARRANGEMENT There are two bogies and two independent hydraulic circuits, one for each bogie. It allows to continue driving if one circuit fails. Each circuit consists of an axial piston pump with variable displacement (Fig. 2, pos. 3) and a hydraulic motor with constant displacement (Fig. 2, pos. 4 and 5). A cardan shaft is used to couple wheels with the hydromotor. Both hydraulic pumps are connected directly to the engine output shaft.

Fig. 2. Wheelset arrangement Fig. 1. Reconstructed railcar M27

Transactions on Electrical Engineering, Vol. 3 (2014), No. 3

57

III. HYDROSTATIC TRANSMISSION

M =

A. Overview Hydrostatic transmission is commonly used in off-highway vehicles. In locomotion applications it is commonly used for propeling of auxiliary devices such as fans. Due to its benefits it is also succesfully used in traction drives especially on low power shunting locomotives and special maintenance vehicles like tampers. Benefits of the hydrostatic transmission are high power density (easy installation to vehicle where only a small room is available) and continuous transfer ratio that allows to achieve the prescribed speed-tractive-effort curve and utilizes engine power well. Disadvantages are lower efficiency, possible oil leaks, sensitivity to oil cleanness and increased demands and costs for maintenance compared with modern electric drives. Basic structure of one hydraulic circuit (one bogie) of the M27 vehicle is in Fig. 3.

Vg max 2π

⋅ p⋅β

(3)

where p is the differential pressure across the pump or motor. The theoretic power (input in the case of pump or output in the case of motor): P = M ⋅ω = Q ⋅ p

(4)

All above mentioned equations are theoretic, i.e. efficiency is not included. Maximum pressure used on the M27 is 330 bar, the oil flow rate for maximum velocity 60 km/h is 367 litres per minute. In the first proposals the proportional brake valve should have been used for hydraulic braking but in the final version the hydraulic braking is not implemented. Instead of it the brake valve with its bypass valve are fully opened during normal operation to avoid pressure losses. If the driving direction is not set (the direction control valve is disengaged and is in the central position) the brake valve is activated and its bypass valve deactivated to ensure minimum pressure for a proper operation of the pump. This minimum pressure is about 20 bar. Control law The hydrostatic transmission is controlled according to the ideal tractive effort curve of the vehicle. In Fig. 4 you can see fundamental control characteristic of the hydrostatic transmission equipped with the pump and motor, both with variable displacement.

Fig. 3. Diagram of one hydraulic circuit

The open hydraulic circuit consists of the axial piston pump (HGR) Parker PV270 acting as a source of oil flow, hydraulic motor (HMR) Parker F12-250 with constant displacement, directional control valve (HR) and brake valve (BV). The bypass valve is connected parallel to the brake valve (not shown in Fig. 3) for lowering pressure losses of the brake valve if it is open. The tractive effort depends on the oil pressure on the pump output . Therefore the oil pressure is the main controlled variable. The oil flow rate and vehicle speed are dependent variables. The oil pressure is controlled by the pump displacement, which is represented by the dimensionless quantity β, according to the control law

β=

Vg Vg max

(1)

where Vg max is the nominal displacement [cm3] and Vg is the actual displacement [cm3]. The theoretic oil flow rate in [m3/s] is: Q=

Vg max 2π

⋅ω ⋅ β

(2)

where ω is the angular speed of the pump or motor. The theoretic torque (input in the case of pump or output in the case of motor):

Fig. 4. Control characteristic of HS transmission

The motors used in the M27 have constant displacement (β2 = 1) therefore the part named “secondary control” in Fig. 4 does not exist, the control range is restricted to the pump operation only (primary control). In the region of constant power (from β1 = min to β1 = 1) the engine load torque M is kept constant (5).

M 1T = =

VgmaxHG

VgmaxHG 2π

2π

⋅ pmin =

VgmaxHG 2π

⋅ pmax ⋅ β1min = (5)

⋅ p ⋅ β1 = const.

The pump is commanded with the desired pressure p according to the desired tractive effort. Quantities β1 , flow Q and vehicle speed v result from actual situation

Transactions on Electrical Engineering, Vol. 3 (2014), No. 3

(running resistance). The rngine speed (not shown in Fig. 4) is commanded according to the desired tractive power. Speed for the desired power is a compromise between engine manufacturer requirements and minimum brake specific consumption. In the region of the constant tractive effort, the pressure is set according to the desired tractive effort (330 bar maximum), the engine load is proportional to the actual pump displacement. The vehicle velocity is directly proportional to the actual pump displacement (β1). The transmission ratio is continuously changing (6). V n g max HG β1 ⋅ i = 2T = T β n V 1 g max HM 2

(6)

The subscript T stands for “theoretic”, i.e. efficiencies are not included, n1 is the pump (engine) speed and n2t motor speed (theoretic). In Fig. 5 you can see the tractive effort curve of the reconstructed railcar M27 including grade resistance curves.

58

To achieve the desired vehicle velocity while running resistance is low can be uneconomical because the oil flow and therefore the engine speed is high but the engine load is low due to low running resistance (curves don’t intersect). Fortunately the JHMD’s railway is situated to a hilly country so the aforementioned disadvantage can be avoided. The motors with variable displacement of the used size and for a reasonable price were not available for this vehicle. B. Control structure A simplified control structure is shown in Fig. 7. Driver sets a desired tractive power. The desired power is evenly distributed across both hydraulic circuits (bogies). The traction control is not implemented. In the case of one circuit malfunction the broken circuit is commanded to zero desired power and the circuit is disabled. The desired power is ramped and goes to the tractive power limiter that limits the desired tractive power in all circuits proportionally if the engine load is above a specified value. The actual engine load at current speed is received from the engine ECU via CAN bus. Next a desired load torque M1t* is computed from the desired tractive power and engine speed n1*. The M1t* is then converted to the desired pressure p* with the help of equation (5). Actual β1 is sensed by the LVDT sensor inside of the pump. These computations are done for each hydraulic circuit. The desired speed of the engine is computed according to the desired tractive power including defined constant margin for accessory loads that can switch randomly on and off.

Fig. 5. Tractive effort curve of M27 after reconstruction (new) compared with tractive effort of M27 before reconstruction (old)

Absence of motors with variable displacement reduces the dynamic control range and shortens the constant power region to lower velocities (see Fig. 6).

Fig. 7. Control structure Fig. 6. Partial tractive effort curves of M27 after reconstruction

Transactions on Electrical Engineering, Vol. 3 (2014), No. 3

Due to the used control principle the available engine power at current speed is utilized only in the region of the constant tractive power, not in region of the constant tractive force. This is due to fact that driver commands the desired power not desired force, i.e. the engine speed is set directly according to the throttle position regardless of the fact if the engine would be loaded with this power or not. The brake valve and its bypass valve are fully opened during normal drive to minimize power losses in the system. C. Coasting If coasting is commanded by driver or by external logic then the direction control valves in both circuits are disengaged and oil can freely flow from the output port to the input port of both motors. The motors act in this situation as pumps. The desired power is set to zero but the desired pressure is not zero, instead of it 20 bar is commanded. The minimum pressure 20 bar is needed for a proper pump operation. Without this pressure pump it is not possible to close the pump to almost zero displacement. Almost no oil flow from the pump is generated during coasting, the engine is unloaded. The minimum pressure is maintained by actuation of the brake valve, bypass valve is closed. Engine idles during coasting. Special care must be taken during a return phase from coasting back to pulling when the vehicle is moving. In the beginning of this phase the pumps generate only small flow because the desired power and therefore the engine speed is low but the motors generate flow that depends on the vehicle speed. At the maximum vehicle speed this flow could be twice as flow generated by the pumps. To prevent big flow difference, that motor would have to suck via leakage line from tank, the direction control valve in each circuit is not engaged immediately but after the flow difference decreases under a defined threshold (50 litres per minute). Big flow difference should cause negative pressure in the motor leakage line and cavitation could occur. The same principle is used during “small” coasting while driver commands low power but the vehicle still keeps its velocity. In this situation again a flow difference between the pump and motor exists. If this difference is greater then the defined threshold (150 litres per minute) then the direction valves are disengaged to prevent negative pressure in the motor leakage lines. There are a few differences compared with “big” coasting. During “small” coasting the desired pressure stays at the level according to the desired power, engine speed also the brake valves and bypass valves are fully opened as during normal operation. Return from the“small” coasting back to pulling is automatic and occurs if the vehicle slows enough down causing reaching of the flow rate difference threshold for engaging of the direction valves. IV. HARDWARE COMPONENTS The control system performs only functions closely related to the hydrostatic transmission. All other functions like the engine starting, lighting, information system, doors are controlled by their autonomous control systems. The throttle is an analog type (4-20 mA output) with latch in the bottom position (coasting demand). The

59

external switching logic for blocking traction power is connected to the input named “coasting switch”. Reaction of the hydraulic system on the coasting demand is following. Power is set to zero and the direction valves are disengaged. Oil can flow freely through the hydraulic motors, the vehicle is coasting. The TEDOM engine is controlled in a speed loop, communication runs over the J1939 protocol. The state of engine is displayed on a small color diagnostic display placed on the driver’s desk. State of the hydraulic drive would be also displayed on the same display in a future. An ethernet port on the RRCPU is used for debugging purposes and for firmware and application software download to the RRCPU.

Fig. 8. HW architecture

V. SOFTWARE DEVELOPMENT Main functions of the application SW are following: • Control of the hydrostatic transmission • Control of the engine speed • On board diagnostic of components The application software for the vehicle control unit RRCPU is developed in compliance with legislative norm EN 50128 [2] and with MISRA coding standard. The vehicle computer firmware itself meets requirements for safety integrity level 0. The application SW is written in the C++ language. The application SW uses TROL library by AMiT for access to the CANopen buses, I/O and other

Transactions on Electrical Engineering, Vol. 3 (2014), No. 3

functions in the vehicle control unit RRCPU. Special applications TrolDatGen and TrolView developed by AMiT are used for setting up a project and real-time application debugging including data logging, displaying data in charts, using alarms for detection mechanism of faults and so on. The SAE J939 communication stack for communication with the engine was developed on the KEEZ department because this type of protocol was not supported by the TROL library. Safety functions are secured outside of the application SW using the HW relay logic. It sends a signal to the “coasting switch” input on the RRCPU if it is needed and also mechanically interrupts digital outputs of the RRCAIO module for disengaging of the direction valves in all hydraulic circuits. The result is disengaging of power and vehicle switches to coasting. The safety relay circuit guards following:

60

position, not according to the real power needed from the engine).

• Air pressure in brake system (drop in pressure results in disengaging of power). • If vehicle doors are closed. • Actual position of the hydraulic direction control valves (has to be same as commanded position). • Oil temperature and oil level in tank. Integration testing on the first and second vehicle takes place in these days. After its completion a final validation tests will be performed. The vehicle will then be prepared for the approval procedure. VI. MEASUREMENTS In Fig. 9 there are shown main quantities of the hydrostatic propulsion that were acquired during the vehicle acceleration from zero velocity at 100 % throttle. Measurements were taken on the real vehicle M27.002. At the beginning the vehicle stays at zero velocity, engines idle at 800 rpm, the direction valves are disengaged (outputs of both pumps are connected with an oil tank through the brake valves), the brake valves are activated and create approx. 20 bar pressure (pump 1 and 2 actual pressures are approx. 20 bar in Fig. 9) in order to close both pumps (beta of both pumps is approx. zero in Fig. 9). Driver sets 100 % throttle at the 25th second, i.e. maximum tractive effort. Signal from the throttle is ramped. The direction valves for switching forward direction are activated, oil can flow from now from pumps to motors. Pressure ramps up to maximum value 330 bar. The vehicle begins to move. The pump displacements and flow rates increase which results in a vehicle velocity increase. Pressure is still held on the maximum value. The engine load increases as well. The engine load represents engine torque at a current engine speed. This part of figure corresponds to a constant tractive effort region in the tractive effort curve. The engine speed is set to the value corresponding to 100 % power demand even if the actual power taken from the engine in the constant tractive effort region is just increasing with the vehicle velocity and the commanded engine speed does not correspond to 100 % power demand (engine speed is set according to the throttle

Fig. 9. Vehicle acceleration at 100 % throttle

At the 54th second the velocity reaches approx. 24 km/h. This is the point where the region of constant power begins. It corresponds with the theoretic tractive effort curves shown in Fig. 4 or Fig. 5 quite well. From this point engine further works with an optimal load at optimal speed, i.e. the engine speed is optimal for the demanded power. The pressure demand is computed according to (5) in order to keep the engine load at a constant value approx. 95 %. As the vehicle velocity increases, the pressure and tractive effort decreases to keep the engine load constant. At the 133th second driver commands coasting. The traction power is set to zero, i.e. the pressure demands are zero, the engine goes to idle, the direction valves are disengaged, outputs and inputs of the motors are connected through the direction control valves. It enables free oil flow through the motors. The pumps are closed because the brake valves are activated. The vehicle is coasting. At the 143th second driver commands full throttle again while the vehicle is moving. The drive goes from coasting to pulling in the same way as at the beginning. The only difference is a later activation of the direction valves. The valves are engaged at the moment when difference between the pump flow rate and motor flow rate in each circuit reduces below the defined threshold (50 litres/h). It enables fluent transition without oscillation.

Transactions on Electrical Engineering, Vol. 3 (2014), No. 3

VII. CONCLUSION Main functions of the control system for the hydrostatic transmission of the rail vehicle M27 were described in the paper. In these days integration tests are performed. Results from the first measurements on the real vehicle match theoretical assumptions. It was observed that the hydraulic pumps come from manufacturer with a large deviation in the minimum pressure setting that influences a pump controllability in coasting. Therefore it requires an additional finer setting on each produced vehicle. But overall system behaves as expected. ACKNOWLEDGMENT The research was supported by the TACR grant “Competence Center of Railway Vehicles” No. TE01020038.

61

REFERENCES [1] ČSN EN 50155 – Elektronická zařízení drážních vozidel [2] ČSN EN 50128 - Drážní zařízení - Sdělovací a zabezpečovací systémy a systémy zpracování dat – Software pro drážní, řídící a ochranné systémy. [3] “Systém řízení pohonu na MUV 74 – uživatelská příručka”. Verze 1.03. CZ LOKO a.s. 2013. [4] J. Novák, Z. Mašek, V. Lenoch, L. Mlynařík, “Regulace hydrostatického přenosu trakčního výkonu speciálního kolejového vozidla MUV 74.1 N KSF”, in Mezinárodní konference učitelů elektrotechniky SEKEL 2013. Moravská Třebová, 2013. ISBN 97880-7395-625-7.

Transactions on Electrical Engineering, Vol. 3 (2014), No. 3

62

Water Cooled Induction Traction Motor for 100 % Low Floor Tram Car Pavel Dvořák1), Tomáš Fajt2), Iva Sošková3) and Jakub Hloužek4) 1)

ŠKODA ELECTRIC a.s., Plzeň, Czech Republic, e-mail: [email protected] ŠKODA ELECTRIC a.s., Plzeň, Czech Republic, e-mail: [email protected] 3) TechSoft Engineering, spol. s r.o., Praha, Czech Republic, e-mail: [email protected] 4) TechSoft Engineering, spol. s r.o., Praha, Czech Republic, e-mail: [email protected] 2)

Abstract — The paper deals with the design of the traction motor for 100 % low floor tram car. Within the design it was necessary to deal with many problems which have significant impact on the final product. The most interesting problems were the conception of the torque transmission, mounting of the brake equipment on the motor body, arrangement of the motor inside the bogie, cooling of the motor or protection of the motor against the water and pollution which can enter inside the motor. In this paper there are discussed the electromagnetic model, and special thermal and ventilation model of the motor. The final design was validated within the type testing.

that the environment conditions and position of the motor in the bogie plays a major role. The motor can be exposed in the operation to various pollutions and therefore it needs to be adequately protected.

Keywords: induction traction motor, tram car, electromagnetic field, temperature field, ventilation circuit, equivalent circuit.

I. INTRODUCTION Traction induction motors are the most popular and they are used in many electric vehicles. Most vehicles of the city transport will be low floor for easy entry and exit of passengers. Regarding to this requirement, construction of the vehicle bogie and traction drive must be small and compact.

Fig. 2. View to the traction motor.

Another aspect influencing the design of the traction motor is its size. Necessity of the motor compact dimensions together with torque requirements has an effect on the size of the magnetic circuit, its saturation and cooling. All these requirements lead to a closed compact liquid-cooled traction motor. III. DESIGN OF THE TRACTION MOTOR A. Electrimagnetic model of the traction motor The electromagnetic model can be prepared with analogy between magnetic and electric circuit. In this case the magnetic flux corresponds to electric current and reluctance corresponds to electric resistance. The magnetic circuit can be replaced by an electric circuit which can be solved by standard methods.

Fig. 1. Illustration of the tram-car bogie.

This paper deals with the design aspects of the traction motor for a tram-car. The described traction motor is induction, squirrel cage motor and it is water cooled. The traction motor is mounted outside of the bogie (see Fig. 1) and it is connected with a gear box. II. DESCRIPTION OF THE TRACTION MOTOR Nowadays, the design of an induction motor is not difficult discipline. On the other hand, requirements which are necessary to be fulfilled are often very rigorous. Due to this the design of the motor is very difficult and problematic. At the beginning it is necessary to calculate, Fig. 3. Flux lines in the motor cross-section .

Transactions on Electrical Engineering, Vol. 3 (2014), No. 3

The model of the motor can be reduced to one pair of poles (see Fig. 3). In this figure there are illustrated flux lines and there can be seen one pair of poles in the magnetic circuit. The calculation method is based on analytical formulas [5] and it uses vast range of simplification aspects and it is suitable in first stage of the design. The simplification is in symmetrical magnetic circuit (only circular outer perimeter is enabled) and/or the field intensity in several parts of the magnetic circuit is considered to be constant, etc... Main advantage of this method is quickness of obtaining first results. The results from first stage of the design can be used for optimization by a suitable numerical method. In present, the most popular is the finite element method (FEM). Basis of FEM is partition of the geometry of the motor to a finite number of the triangles (in 2D) or tetrahedrons (in 3D).

63

testing. Accuracy of the prediction of equivalent circuit parameters is necessary for minimization of further errors with bad mathematical model. C. No-Load operation This operation is characterised by the rotor synchronous speed. The slip is close to zero, therefore the rotor resistance R21 is much bigger than RFe and the equivalent circuit will have a new topology (see Fig. 6).

Fig. 6. Equivalent circuit in no-load operation.

The distribution of the flux density in the motor crosssection gives information about saturation of the magnetic circuit (see Fig. 7).

Fig. 4. Illustration of the 2D triangle mesh.

In Fig. 4 there is shown the 2D triangle mesh of the motor cross section. For acceleration of the solution in the finite element analysis (FEA) geometrical symmetry may be also considered and used. Computation of next characteristic values is relatively simple. B. Equivalent circuit For analysis and control of the motor it is necessary to define values of the equivalent circuit (see Fig. 5). This circuit is basis of the mathematical model of the control algorithm.

Fig. 7. Distribution of the flux density in the motor cross-section .

Difference between analytical and numerical solution of the magnetic flux density is very small (see Table I). TABLE I. COMPARISON OF ANALYTICAL AND FEM RESULTS flux density B (T) in stator

Fig. 5. Full equivalent circuit of the induction motor.

Main circuit parameters are winding resistance of the stator (R1) and rotor (R21), leakage inductance of the stator (L1) and rotor (L21), main inductance (Lh) and core resistance (RFe). All of these values are possible to be obtained from calculation and then to be adjusted after

flux density B (T) in rotor

placement

analytical

FEM

analytical

FEM

yoke

1.5

1.6

1.4

1.4

teeth

1.6

1.7

1.9

1.7

The magnetic flux and terminal voltage dependency on no-load current gives magnetization characteristic. In Fig. 8 it is possible to compare the magnetisation characteristics. No load operation gives information about the value of the stator current, losses, main inductance and core resistance.

Transactions on Electrical Engineering, Vol. 3 (2014), No. 3

64

Fig. 8. Magnetisation characteristic.

Fig. 10. Locked rotor characteristic.

D. Locked rotor operation This operation is characterised by the blocked rotor – the slip is close to 1. The rotor resistance R21 is small compared to RFe and the equivalent circuit will have a new topology (see Fig. 9).

E. Nominal loading The traction motor is loaded by nominal torque (Nm) and speed (rpm). This operation is maintained till stator winding temperature reaches steady state. For steady state temperature it may be considered a condition where the temperature varies no more than 2 K per hour [3]. After that, values of stator resistance, frequency, power losses and power factor may be measured.

Fig. 9. Equivalent circuit in locked rotor operation.

Fig. 11. Losses distribution.

This operation gives information about the leakage inductance (L1 + L21) and resistance (R1 + R21). The slip is equal 1, frequency of the rotor field equals with the stator frequency and the skin-effect in the rotor must be considered. For dividing the leakage inductance to stator and rotor value it is suitable to use the ratio L1 / L21 from the analytical design. For the analysed motor this ratio is about 1.2.

Based on all parameters from the no-load, locked rotor and nominal loading operation it is possible to make distribution of losses in the motor (see Fig. 11). In Table II there are recorded the most important parameters of the motor. Calculation of the electric parameters was referenced to temperatures extracted from the measurement. In Table III there are stored parameters of equivalent circuit (see Fig. 5) of the motor referenced to 20 °C. In both of tables the values named (M) are

Transactions on Electrical Engineering, Vol. 3 (2014), No. 3

65

measured and values named (C) are calculated values. Regarding the values, the model gives quite good results.

TABLE II. COMPARISON OF ANALYTICAL AND MEASUREMENT RESULTS name

unit

calculation (C)

measure. (M)

M vs C (%)

power output

kW

100

100.3

0.3

speed

rpm

2217.5

2218.4

0.04

L-L voltage

V

420

420.6

0.15

phase current

A

178.3

177.8

-0.28

frequency

Hz

75

75

0

Fig. 12. Illustration of the water channel.

total losses

kW

8.6

8.5

-0.5

-

0.84

0.84

0

K

70

68

-2.94

°C

60

60

0

%

1.44

1.42

-1.83

kW

101.5

101.7

0.27

W

87.4

120

27.2

W

2948.4

2562.2

-15.1

W

2166.2

1322.7

-63.8

core losses

W

862

1537

43.9

add. losses

W

2372.2

2965.8

20

IV. VENTILATION AND THERMAL DESIGN Liquid is considered as a most efficient coolant. Liquid – water cooling was chosen for cooling of the analyzed motor. Due to this, the construction of the motor may be very compact instead of self air ventilation. In case of self ventilation the length of the motor depends on dimensions of the ventilator. Also from the noise point of view the water cooled motors are less noisy than self air ventilated motors. Advantage of water cooling is in the better possibility for assembly of the motor with the bogie (see Fig. 1). Regarding to this the cooling via water channel placed on outer perimeter of the stator was chosen (see Fig. 12). Thanks to this, the stator is more actively cooled than rotor and it brings some problems with cooling of other motor parts (rotor winding, bearings, etc.). Due to the stator frame, the next problem was impossibility to use whole area of the stator lamination for cooling (see lower part of the stator frame in Fig. 12). Coolant flow quantity was defined based on calculation to 20 l/min and work temperature of the water on the inlet is 60 °C. The method of the ventilation and thermal equivalent net (VTE net) and method are based on numerical calculations (CFD) in the computer program of ANSYS Fluent. These methods were used for the ventilation and thermal calculation of the analyzed traction motor. Both methods of calculation use as input parameter the distribution of the losses in the motor based on a electromagnetic calculation. In the thermal calculation the heat transfer by conduction and heat transfer by convection were considered between the solids and liquid.

power factor stator winding temperature rise water temperature Slip (hot) power in the air gap mechanical losses Cu losses – stator Cu losses – rotor

TABLE III. COMPARISON OF ANALYTICAL AND MEASUREMENT RESULTS – EQUIVALENT CIRCUIT PARAMETERS

name

unit

calculation (C)

measure. (M)

M vs C (%)

R1

mΩ

20.3

19

-7

R21

mΩ

14

14

0

L1

mH

0.23

0.235

2.2

L21

mH

0.192

0.196

2.2

Lh

mH

5.623

5.483

-2.6

RFe

Ω

46

70.6

34.8

Temperature of the stator and rotor winding has very significant impact on parameters of the motor mathematical model .Good estimation plays important role for the motor control. For this it was created a ventilation and thermal model of the analyzed motor.

A. Ventilation and thermal equivalent net (VTE net) This method is based on the equivalent net which consists of the heat sources, thermal nodes and places of heat dissipation. These elements are connected to each other with thermal resistances. The method is very simple, fast and universal. Therefore it can be used also in case of the motor 3D model not done yet. Results of the analysis by this method are known within several minutes. For the analysis it was used the universal thermal model of the water-cooled motor. This model is a standard model used for such types of the motor of the SKODA ELECTRIC a.s. production. The model of the motor is possible into divide to two parts. First is the ventilation. The ventilation calculation

Transactions on Electrical Engineering, Vol. 3 (2014), No. 3

for water-cooled motor gives information about the pressure drop of the cooling channel and defines average of the water velocity. The pressure drop calculation is based on definition of the ventilation resistance of each parts of the water channel. Result of the calculation is the pressure drop (25 kPa) for the defined water flow quantity (20 l/min). Average water velocity is 0.9 m/s. Second is the thermal calculation. Aim of the solution is to define the temperature of monitored nodes of the thermal net. As input there are losses, boundary conditions (altitude, ambient temperature, inlet water temperature and speed of the rotor). As result of the analysis there are the average temperature rise of the stator winding (66 K), temperature rise of the water (5.9 K) and maximum temperature of bearings (108 °C). B. CFD method The CFD method is possible to be used with a real 3D model of the traction motor (see Fig.13). However, for the CFD analysis it is necessary to modify the real 3D model. Some details in geometry which has no influence on the main problem (water flow) will be deleted or modified. These modifications were done with help of the ANSYS DesignModeller and ANSYS SpaceClaim programs.

66

This analysis was done with help of the TechSoft Engineering Company, which did all of the CFD analysis. Based on the ventilation simulation (see Fig. 15) of water flow inside the water channel it was defined its pressure drop (19.5 kPa) for water flow quantity (20 l/min).

Fig. 15. Pressure field of the water inside water channel.

Together, places were found where the water flow is not good (see Fig. 16). Also places with worst cooling efficiency were found The result of the thermal analysis gives information about temperatures of all parts of the motor and water inside the water channel. The places with high temperatures which were predicted before analysis were confirmed. The highest temperatures about 230 °C were identified near middle of the rotor cage, rotor lamination and shaft – also in the area of losses formation, which is far from the cooling channel. The stator winding is also significantly warm place, especially its front parts. Fig. 13. Computer model of the traction motor geometry.

The mesh for the calculation (see Fig. 14) was created by the ANSYS Meshing. The problem of water flow was defined as an incompressible, viscous and turbulent [4]. For the complete model the inner air flow was considered. This air flow was caused by the rotor rotation.

Fig. 16. Water flow in the water channell.

From the calculation on the 3D geometric model it is possible to see the temperature distribution in all parts of the motor (see Fig 17). Especially it is possible to analyze the stator winding temperature (see Fig. 18). The warmest area of the stator winding is on the bottom, where is no cooling (compare with Fig. 12). Fig. 14. The zoom to mesh.

Transactions on Electrical Engineering, Vol. 3 (2014), No. 3

67

temperature of bearings is more than 120 °C. Therefore the better and more detailed bearings model may be considered. After that, the results of the simulation gives lower values. C. Comparison of methods Each simulation is necessary to proof and compare with the measurement on a real motor prototype. This step is most important in case of problematic solution.

Fig. 17. Temperature distribution in the longitudinal section of the motor.

The average temperature rise of the stator winding is 73 K and the maximum temperature of the front parts of the stator winding is 155 °C. The reason of this value is the nonsymmetric cooling channel and rotation of air heated to high temperature inside the motor, which is in contact with the front parts of the stator winding.

Fig. 20. Specification of points for temperature measurement.

Fig. 18. Temperature distribution of the stator winding.

It was calculated the temperature distribution of the water inside the water channel (see Fig. 19). The total temperature rise of the water is 5.8 K.

During prototype production places for location of Pt cells were specified (see Fig. 20). In these places the temperature was measured and its value recorded. In tables (see Table III and Table IV) there is comparison of calculated values from the VTE net and CFD analysis with values from measurement.

Fig. 19. Temperature distribution of the water.

The results of the simulation gives information, that most problematic parts are bearings. It was confirmed, the

Values are in (K)

ventilation and thermal equivalent net

CFD

measurement

comparison between CFD and measurement

TABLE IV. COMPARISON OF CALCULATIONS

average temperature rise of stator winding

66

73

68

7.4

Transactions on Electrical Engineering, Vol. 3 (2014), No. 3

ventilation and thermal equivalent net

CFD

measurement

comparison between CFD and measurement

V. CONCLUSION The electromagnetic, thermal and ventilation model of the closed, water cooled, squirrel cage induction traction motor was defined. Based on the design procedure the real prototype of the motor was created. On the prototype motor there were done the measurements and the used models were checked. Based on measurement results the design methods were slightly modified and they are used for design of further traction motors for another projects realised in SKODA ELECTRIC a.s.

Placement (see Fig. 20) Values are in (°C)

TABLE V. COMPARISON OF CALCULATIONS

68

1

145

134

139

-3.6

2

145

145

142

2.1

3

144

134

131

2.3

REFERENCES

4

144

145

134

8.2

5

-

136

135

0.7

6

-

144

140

2.9

7

-

136

141

-3.5

8

-

144

144

0.0

9

193

182

181

0.6

[1] I. P. Kopylov, and col., Stavba elektrických strojů, Praha: SNTL, 1988. ISBN-04-532-88. [2] K. Hruška, Určení parametrů náhradního schématu asynchronního stroje v programu FEMM. Plzeň, 2008. [3] ČSN EN 60342-2, ed. 2: 2011 [4] ANSYS Fluent 14.5 – manuals and help [5] J. Pyrhönen T. Jokinen, V. Hrabovcová, Design of rotating electrical machines, John Wiley & Sons, Ltd, 2008. ISBN-978-0470-69516-6

10

103

117

91

28.6

11

108

126

98

28.6

12

66

64

66

-3.0

13

66

89

77

15.6

inlet

60

60

60

0

outlet

65.9

65.8

65.5

0.5

There is good conformity between the measurement and calculation. Both simulation methods used simplified bearing model, therefore the maximum differences of the temperature are in location of bearings.

Transactions on Electrical Engineering, Vol. 3 (2014), No. 3

69

Real-time Simulation of 3 Parallel PWM Rectifiers Michal Kopecký 1), Jan Švanda 2) and Martin Vlček 3) ŠKODA ELECTRIC, a.s./SW2 Prague Department, Plzeň, Czech Republic 1) e-mail: [email protected] 2) e-mail: [email protected] 3) e-mail: [email protected]

Abstract — This paper describes the development of a real-time model up to 3 parallel PWM rectifiers and its implementation on FPGA using LabVIEW development environment. The main benefit of this real-time model is the fact that there is no need for a real device or a test stand for debugging of traction drive control SW. The Hardware-inthe-Loop testing with similar RT model of an induction machine has already brought large financial and time savings. Moreover, destructive states or states difficult to evoke can be tested using such a real-time model. Keywords — PWM rectifier, real-time simulation, Hardwarein-the-Loop (HIL), LabVIEW FPGA, traction drives control SW.

I. INTRODUCTION A. Simulated System The simulated system (see Fig. 1) consists of three parallel branches with one-phase PWM rectifiers connected to the common DC link circuit. Usage of this AC/DC conversion system brings following main advantages: possibility of energy recovery, consumption of sinusoidal current and DC voltage stabilization. Each one-phase PWM rectifier consists of 4 IGBT transistors which form a full bridge. The system is controlled by generation of PWM signals at these transistors [1]. The use of three parallel branches allows reduction of power transmitted by each branch. Moreover, the 3 parallel PWM rectifiers can be controlled with offset to each other. Such a control can damp desired frequency band in the input AC current spectrum in order to fulfil the EMC limits. B. Model Adaptation on Less Parallel PWM Rectifiers A very important thing to be mentioned is that the model of 3 parallel PWM rectifiers could be easily used to simulate a system with only two or one parallel branch. One of the branches is not considered if we set resistance magnitude (e.g. 1010 Ω) and the Ras to a huge transformation ratio p to zero (this causes zero secondary voltage uas) and ensure that the transistors in the disconnected rectifier are switched off. For example, the configuration with two secondary windings was used in the electric multiple unit Škoda 7Ev-RegioPanter. The AFE control SW for this unit was the first application which was tested with the real-time model. C. Connection System Simulation The model provides also the simulation of the system connection in the right order as well as failure states (see

TABLE I). For this purpose there are 4 switches introduced into the model: S0 (Start Voltage), S1 (Charging Contactor), S2 (Line Contactor) and S3 (Load Contactor). It should be mentioned that the switches in parallel branches are coupled, i.e. they can be either on or off in all parallel branches. TABLE I. SWITCHING LOGIC FOR SIMULATION OF SYSTEM CONTROL Switches States System State S0

S1

S2

S3

0

0

0

0

1

0

0

0

1

1

0

0

1

1

1

0

Charging resistance bridged.

1

1

1

1

Current load connected. Operational state.

1

0

1

0

1

1

0

1

Nothing is simulated. Primary voltage connected, all secondary windings open. Connected over charging resistance, Capacitor started charging.

Charging over small resistance. Huge currents. Failure state Charging with connected load. Overcurrents. Failure state.

From the point of view of simulation, all system states with S1 switched on lead to the same model only with different parameters (of resistance or current load). For this reason, we will discuss below only how the operational state with all switches on can be simulated. II. STATE SPACE MODEL A. Transformer The first issue to be considered is the separation of the primary circuit from the rest of the system. The model of the secondary side requires secondary voltages as input. These could be computed from the primary input voltage easily using the transformation ratios:

uas1 = p11uap uas 2 = p12uap .

(1)

uas 3 = p13uap On the other hand, the information about primary current iap has to be provided to the system controller. Once all secondary currents are computed in the state

Transactions on Electrical Engineering, Vol. 3 (2014), No. 3

space model, the primary current can be obtained by formula:

iap = p11ias1 + p12ias 2 + p13ias 3 .

(2)

70

Equations (1) and (2) allow elimination of primary circuit from simulation. Therefore, the only secondary side can be considered below.

Fig. 1. Block diagram of the simulated system

B. State Space Representation The mathematical model of the simulated system relies on the state space representation:

x& = Ax + Bu y = Cx + Du

.

(3)

Because the 3 parallel PWM rectifiers are connected to the same DC link, it is impossible to solve equations for each branch independently, but it is necessary to solve system describing the whole circuit with matrix A of dimension 4x4. There are four state variables – inductor currents and capacitor DC voltage, forming the state vector:

 ias1    i x =  as 2  .  ias 3     ud 

(4)

The input vector (5) contains 3 secondary voltages, DC load current and also diode and transistor forward voltages. These values are constant but the dependence of state and output variables on them varies in different system states.

 uas1     uas 2  u  u =  as 3  .  il   UD     UT 

(5)

The output vector consists of all other quantities in the circuit, i.e.:

 u av1     u R1   u L1     u av 2   uR 2  . y=  uL 2  u   av 3   uR3  u   L3   id 

(6)

Unlike the most common problems, the state matrices are not constant in time in our case. But they change according to the individual states of each parallel circuit. These states are determined by the semiconductor elements which current flows through.

Transactions on Electrical Engineering, Vol. 3 (2014), No. 3

Because solving of the state space system is relatively easy and well known problem, the crucial point of the model development is to invent the switching algorithm which selects the appropriate state matrices in the current simulation step. C. State Matrices Construction Before analysing the possible circuit states, the part-bypart matrices construction (see TABLE II) is introduced. This construction, which substantially simplifies the whole switching algorithm, is based on the independence of individual parallel branches on each other. This is due to the fact that the state of each parallel branch depends only on the quantities in this branch and DC link voltage but this state is independent of the other branch quantities. TABLE II. OVERALL STATE MATRICES COMPOSING OF ROWS CORRESPONDING TO INDIVIDUAL PWM RECTIFIERS Branch

Corresponding variables

Matrices, Rows 1st row of matrix A

1st PWM Rectifier branch

ias1

1st row of matrix B 1st to 3rd row of matrix C

uav1, uR1, uL1

1st to 3rd row of matrix D 2nd row of matrix A 2nd PWM Rectifier branch

ias2

2nd row of matrix B th

4 to 6 row of matrix C uav2, uR2, uL2

3rd row of matrix A 3st PWM Rectifier branch

ias3

3rd row of matrix B 7th to 9th row of matrix C

uav3, uR3, uL3

7th to 9th row of matrix D

Further, the id current is computed as a linear combination of ias currents according to which parallel branches are connected to the DC link and with orientation of which in the current time step: id = k1ias1 + k2ias 2 + k3ias3 , (7) where coefficients k1, k2, k3 are assumed to be 0 or ±1. The last rows of the state matrices can be constructed using these coefficients: k A4,⋅ =  1  C1

k2 C1

k3 C1

 0 

 . 1 B4,⋅ =  0 0 0 − 0 0 C1   C10,⋅ = ( k1 k2 k3 0 )

switching for one PWM rectifier. Once the next states in all parallel branches are determined we can construct the state index vector of 3 elements and according to the TABLE II also the state matrices for the next time step computation. A. Possible States of One PWM Rectifier We will follow the algorithm development history describing also the problems we came across during it. Theoretically, there are 24 = 16 possible states needed to be involved for the full description of one PWM rectifier considering all possible combinations of semiconductor elements conduction. In fact, the number of all possible states can be reduced to 7 by excluding prohibited states (short-circuited branch – causes error of application) and by merging some switching combination into one (in terms of external behavior it does not matter e.g. the sequence of transistor and diode conducting). B. Basic State Switching Model This simplest model was used only for simulation of the uncontrolled single phase rectifier (i.e. diode bridge). In this case there are only 3 system states: Open circuit (No. 1), Positive branch conduction (No. 2) and Negative branch conduction (No. 3). However as we shall see later, it is not sufficient in a more complex situation. The main idea of this model is very simple. Before each step of the state space model a sequence of conditions is evaluated and the first fulfilled condition determines an appropriate state. It is important to mention that here as well as in the whole paper the algorithm depends on the sequence of condition evaluation. More than one condition can lead to the given state, e.g. the state No. 3 (see Fig. 3) can be chosen if one of the following conditions is fulfilled.

th

4th to 6th row of matrix D

71

(8)

D10,⋅ = ( 0 0 0 0 0 0 )

III. STATE SWITCHING ALGORITHM Because of the independence of each parallel branch state on other branches we can consider only the state

if (ias = 2*UD %Zero current in circuit and there is sufficient voltage to diode D3 and D2 begin conducting. C. Correction of Negative Current id The calculated AC current ias can change its sign between two consecutive simulation steps. As a consequence, also the DC current id changes its sign from plus to minus. However, this is not physically possible. The global variable direction used by decision on the next state, avoids simulation to go to a wrong state on the basis of that established configuration only. However, already computed solution at that simulation step would stay wrong (blue curve in Fig. 2). A very simple additional condition was introduced into the model, which returns id up to 0, whenever id < 0 occurs (green curve in Fig. 2).

Transactions on Electrical Engineering, Vol. 3 (2014), No. 3

72

D. Infinitesimal Occurrence of Open State We observed that the system cannot immediately change its state by the current ias zero crossing – the state No. 1 (open circuit) always has to occur at least for one time step. This treatment works fine for an uncontrolled rectifier (with nonzero diode threshold voltage UD). However, it brings problems in the simulation of PWM controlled rectifiers . We will demonstrate them on our model situation. Let us consider situation with transistor T2 constantly closed (e.g. pwm = [0 1 0 0]) and system being in the state No. 5, when T2 and D4 conduct (see Fig. 4).

uR

ias uas

10

id [A]

5 0 -5 -10

5.02 5.04 Time [s]

5.06 -3

x 10

Fig. 2: Correction of the current id

uL

-uas

without correction with correction

id

C1

uav

ud

il

UD

-ias

D3

UD D2

il

ias

Fig. 3: State No. 3 diagram (diode D3 and D2 conducting)

ias

uR

UT

uL

T2

C1

uav

uas

ud

il

UD D4

il

Fig. 4: State No. 5 diagram (transistor T2 and diode D4 conducting)

The value of short-circuit current ias is continuously decreasing until it changes its sign in one particular simulation step. In this case the system should go to the already known state No. 3 which would also correspond to the physical reality. Unfortunately, the global variable direction=1 does not allow smooth state transition and the intermediate state No. 1 (open circuit) is set for one simulation step. As a consequence, the step change of voltages uav and uL occurs in this step (see Fig. 5). Let us remark that the solution does not bring omission of the global variable direction from the algorithm switching. In this case, smooth transition from the state No. 5 to the state No. 3 occurs without any problems. However, the very next state transition fails. In that case the system in the state No. 3 should not move to the state No. 5 but the state No. 1 should be established. Unfortunately, the switching algorithm does not catch this, because of missing condition with the global variable direction. As a consequence, the wrong state No. 5 is established and this wrong decision results into oscillation of the numerical solution (see Fig. 6). From the description above we can conclude that the key issue of the whole switching algorithm is to determine if the ias current zero crossing is or is not physically

realistic. This reality is done by voltages relationships in the simulated circuit. The final solution of the problem described above has brought infinitesimal occurrence of open state philosophy which relies on the following idea: if the current ias changes its sign between two consequential steps, we can suppose that the open circuit state (No. 1) always occurs for infinitesimally short moment (shorter than the simulation step!). The consequence of this occurrence are settings ias = 0, uR = 0 and uL = 0. While a reason for the zero resistance voltage is obvious, by inductance voltage uL we can imagine that the through-going current is for very short time, but still constantly zero and therefore also its derivative is zero. After this infinitesimal occurrence of open circuit state the voltages conditions for transition to the states with different ias current sign can be evaluated and if some of them is fulfilled, the system could go to the appropriate state. It is important to remark that the values of quantities by infinitesimal open circuit state occurrence do not appear in numerical solution, in contrast to Fig. 5. By the way, this approach has introduced the current state knowledge into our state switching algorithm, which will be necessary in all state switching managers listed below.

Transactions on Electrical Engineering, Vol. 3 (2014), No. 3

Amplitude [V], [A]

200

uas ias uav ul ud

0 -200 -400 -600 -800

0.0118 0.0118 0.0118 0.0118 0.0118 0.0119 Time [s]

Fig. 5: Vault of voltages uav and uL by setting of the intermediate state No. 1

Amplitude [V], [A]

15000

uas ias uav ur ul

10000

73

First, there are evaluated conditions for the state in the same group in the function Groupi_conditions (appropriate function is determined by group index i according to the state in the previous time step). As inputs to this function, the just computed state variables values, the current value of the voltage uas and the values of the voltages uR and uL from the previous time step are sent. If none of these conditions is met (i.e. function Groupi_conditions returns one), the second function belonging to the other group is called as well – but in the inputs there are the values of ias, uR a uL replaced by zeros (because of the infinitesimal open state circuit occurrence). In case the previous state index is equal to 1, there are called both function with the classic inputs and for the next state index maximum of both returned values is chosen. We attach the MATLAB source code of the switching manager described above:

5000

0

-5000

0

0.01

0.02 Time [s]

0.03

0.04

Fig. 6: Oscillation of uav and uL caused by omitting the global variable direction condition

E. Model with Groups According the ias Current Direction According considerations above we introduced the concept of state partition into two groups according to the sign of the current ias. The default state No. 1 with zero ias current stands out of both groups. Whereas a transition between two states in the same group proceeds immediately, a transition between two states from different groups is possible only with infinitesimal occurrence of the open circuit state. For practical implementation of such a switching algorithm, we have left the separate condition principle. Now, the state conditions are evaluated together in two functions: function state_index = Group1_conditions(ias, uas, uR, uL, ud, pwm), function state_index = Group2_conditions(ias, uas, uR, uL, ud, pwm). These functions start with the state No. 1 (i.e. state_index = 1) and process conditions for individual states bottom-up according to the state index. Finally, the function returns the state with the highest index in the appropriate group where the conditions are met. If the function returns the default value 1 it means that no conditions for any state in the group are fulfilled. The next state alone is chosen in

function new_index = choose_state_using_Groups(index, ias, uas, uR_o, uL_o, ud, pwm).

switch index %Decision-making according the previous state index case {2,5,7} %Group1 (ias > 0) new_index = Group1_conditions(ias, uas, uR_o, uL_o, ud, pwm); if new_index == 1 %No state conditions in group 1 fulfilled new_index = Group2_conditions(0, uas, 0, 0, ud, pwm); end case {3,4,6} %Group2 (ias < 0) new_index = Group2_conditions(ias, uas, uR_o, uL_o, ud, pwm); if new_index == 1 %No state conditions in group 2 fulfilled new_index = Group1_conditions(0, uas, 0, 0, ud, pwm); end case 1 %Open circuit new_index = max(Group1_conditions(ias, uas, uR_o, uL_o, ud, pwm), Group2_conditions(ias, uas, uR_o, uL_o, ud, pwm)); end The advantage of this switching manager is not only the solution of both problems described above but also the relatively easy portability on FPGA because the switching algorithm takes the same amount of time in every step. Moreover, the evaluation of both functions Group1_conditions and Group2_conditions can execute in parallel. F. Correction in Case of Open State Circuit Choice in Model with 3 Parallel PWM Rectifiers If the next state chosen in the function choose_state_using_Groups is the open circuit state No. 1 in certain parallel branch, non-zero ias current flowing through this branch is physically impossible.

Transactions on Electrical Engineering, Vol. 3 (2014), No. 3

For this reason, it has to be corrected to zero. As a consequence, we should recalculate also the voltage ud. This is done using the reverse approach to the Euler method for the state space numerical solution, i.e. from the already computed new value of ud we subtract the contribution of the appropriate ias current (the one that should be zero). The use of the coefficient vector k defined by (7) brings advantages to do that. Be aware that values of coefficients k1, k2, k3 do not correspond to the groups according to the sign of ias current. However, their values depend on addition of individual currents charging capacitor in the DC link (see TABLE III). TABLE III. VALUES OF COEFFICIENTS ACCORDING TO THE STATE INDEX Coefficient ki value (i = 1, 2, 3)

State index of the parallel branch i

Semiconductor element conducting

2

D1&D4

6

T4&T1

3

D3&D2

7

T2&T3

1

None

4

D3&T1

5

T2&D4

1

-1

0

The final MATLAB implementation of the switch manager for 3 parallel PWM rectifiers combines both choosing of the next state and correction in the case of open state circuit in one execution of the for loop. The source code is provided below.

k_o = k; %Storing "old" vector value used in state space numerical solution for i=1:PPR %PPR = 3 (#PWM Rectifiers) %Choosing of a new state index of the i-th PWM Rectifier index(i) = choose_state_using_Groups(index(i), ias(n,i),uas(n,i), ur(n-1,i), ul(n1,i), ud(n), pwm(n,i,:)); %Correction in case of state No. 1 – open circuit. if index(i) ==1 ias(n,i) = 0; ud(n) = ud(n)-ko(i)/C1*h*ias(n1,i); end; end; G. Final Implementation of 3 Parallel PWM Rectifiers Model We recapitulate the whole model of 3 parallel PWM rectifiers implementation below. In each simulation steps: 1. Numerical solution of the state equation x& = Ax + Bu using Euler method.

74

2.

3.

Choosing new state using the state switching manager and state matrices updating according to the new state Evaluating the outputs from the equation

y = Cx + Du

The fact that the outputs are computed with already new state matrices brings not only higher accuracy but also no need of additional correction of the current id (see Fig. 2). Because id is computed from already corrected currents ias it is ensured that it will be non-negative. IV. THE REAL-TIME MODEL A. SW and HW resources In order to run the model described above in real-time we need to move on different platforms than PC: real-time system and especially field-programmable gate array (FPGA). TABLE IV. SW AND HW RESOURCES FROM NATIONAL INSTRUMENTS Resource

Usage

NI PXI-7854R Multifunction RIO with Virtex-5 LX110 FPGA NI PXIe-8133 1.73 GHz QuadCore Real-Time Controller NI PXI-4110 Triple-Output Programmable DC Power Supply NI PXIe-1065 18-Slot 3U PXI Express Chassis

Numerical model, Inputs and outputs operation Communication with FPGA model and with PC Interface HW interfaces power supply Chassis, Backplane communication

LabVIEW 2012 SP1 f3 LabVIEW Real-Time 12.0.0

SW used by model implementation

LabVIEW FPGA 12.0.0

We make demand of the highest possible computational repeat on the FPGA model. This should be at least 20 times higher than switching frequency (900 Hz in most applications). However, there is only a limited number of FPGA resources. We will describe our approach how to fulfil these contradictory requirements by reaching the highest model accuracy as possible later. B. Numerical Format For implementation of the computational VI on FPGA in floating point the IP Xilinx CORE Generator Blocks (see Fig. 7) were used. The advantages of this approach are easy portability of the model for systems with (totally) different parameters and high model precision. On the other hand this approach increases significantly the FPGA sources usage. Not many operators in double precision can be placed in our code to avoid FPGA resources overuse. In our case, there were only two multipliers and two adders used. The crucial part of our work is then to invent how to order the input data and the intermediate results as the inputs of individual operators to get the results in the shortest time. This can be done using FPGA high-throughput 2-wire protocol (see TABLE V).

Transactions on Electrical Engineering, Vol. 3 (2014), No. 3

75

TABLE V. LOGICAL SIGNALS OF 2-WIRE HANDSHAKING PROTOCOL LabVIEW standard name

Xilinx name [2]

input valid

operation_nd

output valid

rdy

4 1 1 4 4 2 2 4 4 3 3 4 1 2 3 1 1 1 3 3 3 4 4 4 6 6 6 7 7 7 9 9 9

Description

D

C. FPGA Model Performace Considerations Because of the requirements to reduce the model space and timing, we introduced some more or less substantial simplifications. First, we rewrite Euler method (9).

xn+1 = xn + h( Axn + Bun ) .

(9)

We can save 4 multiplications for evaluating xn+1, when we use the multiplied matrices hA and hB instead of the matrices A and B on FPGA The equation (9) will be transformed into shape: (10) x = x + hAx + hBu . n +1

n

n

18 1 5 6 1 5 6 2 5 6 2 5 6 3 5 6 3 5 6

Determinates if there is new valid input data to process. Determinates if the received result of operation is valid.

n

The numerical solver virtual instrument (VI) solving one simulation step contains 3 parallel computational threads, switch state manager VI and building matrices VI placed in single cycle timed loop (SCTL). The main FPGA VI consists of the solver loop, digital inputs (PWM) processing loop and ten times slower analog output generation loop and loop which sends data to the RT controller (from where they are forwarded to PC). The compiled code of this main VI consumes almost all FPGA resources (see TABLE VII). TABLE VII. DEVICE UTILIZATION OF COMPILED MAIN FPGA VI Total Slices

91.9 % (15880 out of 17280)

Slice Registers

68.2 % (47166 out of 69120)

Slice LUTs

66.5 % (45937 out of 69120)

DSP48s

71.9 % (46 out of 64)

Block RAMs

7.8 % (10 out of 128)

Sparse matrices usage, involving multiplication with simulation step h and parallel executions decrease the number of 40 MHz SCTL executions down to 182. By code benchmarking we discovered that due to the overhead whole loop in main FPGA VI is running with tact of 208 ticks of the 40 MHz clock. To ensure that the simulation will be actually running in real-time we decided to choose simulation step h = 6.25 µs, corresponding to 250 ticks of the 40 MHz clock. Fig. 7: Using of Xilinx Floating-Point Operator to multiply 2 numbers in double precision on FPGA

Substantially significant resources and time savings can be reached using the sparse matrices concept. In general, all matrices hA, hB, C and D have together 16 + 24 + 40 + 60 = 140 elements. However, there are only 52 elements which can be non-zero in some combination of individual states (see TABLE VI). Only these non-zero elements are involved in the computation. We only have to keep in mind the sparse matrices structure (see TABLE VI) by model implementation. TABLE VI. COORDINATE-WISE REPRESENTATION OF SPARSE MATRICES Row index of non-zero elements #

Matrix Column indexes of non-zero elements 1 1 2 2 3 3 4 4 4

9

hA 1 4 2 4 3 4 1 2 3 1 1 1 2 2 2 3 3 3 4 hB

10 1 5 6 2 5 6 3 5 6 4

C

1 2 3 3 4 5 6 6 7 8 9 9 10 10 10

15

D. Communication The model receives the PWM pulses from the electronic control unit (ECU) via optical-electrical converter and digital inputs of FPGA. The output analog data are adjusted and generated in slower output generation loop and provided using SW and HW interface to ECU (see Fig. 8). Besides, the FPGA model communicates with an application on PXIe-8133 RT controller. This communication is provided by two DMA FIFOs. Host-ToTarget FIFO sends input quantities, i.e. secondary voltages uas and current consumption il. All computed quantities are sent back using Target-To-Host FIFO to RT target application. However, this application only sends data forward via Ethernet to PC. In PC, the received data are presented and also could be logged. V. MODEL RESULTS A. Model validation We made a simple test to validate the model of one PWM rectifier. The input data for this testing was sinusoidal secondary voltage uas (of the frequency 50 Hz and amplitude 250 V) and approximately constant load current il of 6.65 A. These quantities were measured by an experiment with the laboratory pulse rectifier.

Transactions on Electrical Engineering, Vol. 3 (2014), No. 3

76

The results of the RT model of one pulse rectifier with input data described above are shown in Fig. 10 and Fig. 12. This simulation results were compared with the real measured data to prove that the model behavior corresponds to the behavior of a real device. B. Rectifiers offset testing The data sent from model into PC are sampled with 16 kHz. Therefore, we can make the spectral analysis theoretically up to 8 kHz and use the model for testing electromagnetic compatibility of device by different configurations. Primary current spectra are shown in Fig. 11 and Fig. 13. We can see that the cluster of spectral lines nearby frequency 1600 Hz can be eliminated by the offset control of 2 parallel PWM rectifiers. Fig. 8: Communication diagram of real-time model

Fig. 9: Off-line graph of transition into energy recovery state from the diagnostic environment DISMON. Primary voltage uap in blue, primary current ias in red and DC link voltage ud in black (all in computer units).

Fig. 10: Secondary current ias

Fig. 12: Bridge input voltage uav

Fig. 11: Spectrum of primary current iap by two PWM rectifiers running

Fig. 13: Spectrum of primary current iap by only one PWM rectifier running

Transactions on Electrical Engineering, Vol. 3 (2014), No. 3

VI. CONCLUSION This paper has presented the main issues we encountered by the development of the real-time model of 3 parallel PWM rectifiers. This model allows us to perform a substantial part of the control SW testing without existence of a real device [3]. Moreover, such tests with model can be easily rerun. Therefore, it is planned to use this model to test the control SW in all projects in our company in the future. ACKNOWLEDGMENT The whole research was made under the research project Extension of Traction Drive Real-time Model (No. 65Z6210) in ŠKODA ELECTRIC, a.s. The publications were financially supported by the Technology Agency of Czech Republic (TACR) under the grant Competence Center of Railway Vehicles (No. TE01020038).

77

REFERENCES [1] J. Bauer, “Single-Phase Pulse Width Modulated Rectifier,” Acta Polytechnica, vol. 48, no. 3/2008, Czech Technical University in Prague, pp. 84-87. [2] LogiCORE IP Floating-Point Operator v5.0. XILINX, 2011. [3] M. Kopecký, J. Švanda, and M. Vlček, “Využití real-time simulací při návrhu řízení trakčních pohonů,” XXXIII. konference o elektrických pohonech, Pilsen, pp. 84-89, June 2013. [4] M. Kopecký, V. Buba, “Příspěvek k řízení pulzního usměrňovače lokomotivy 109E,” XXX. konference o elektrických pohonech, Pilsen,, June 2007. [5] M. Kopecký, M. Bednář, “Dosavadní zkušenosti s algoritmy řízení 4Q na lokomotivě 109E a jednotce 5Ev Litva,” XXXI. konference o elektrických pohonech, Pilsen,, June 2007. [6] J. Javůrek, Regulace moderních elektrických pohonů. Grada, 2003. [7] E. Vitásek, Základy teorie numerických metod pro řešení diferenciálních rovnic. Academica, 1994. [8] J. Švanda, Using NI PXI Express and CompactRIO to Develop a Hardware-in-the-Loop Tester for Electric Driver ECUs of Locomotive, National Instrument, Solutions, Case Studies. [Online]. Available: http://sine.ni.com/cs/app/doc/p/id/cs-15423 [Accessed: 8 April. 2014]. [9] High-Throughput LabVIEW FPGA Exercises. National Instruments, 2012.

Transactions on Electrical Engineering, Vol. 3 (2014), No. 3

78

Electric Drives in the Modernized Power Plant Tusimice II David Foltyn1) 1)

ŠKODA PRAHA Invest /Electrical engineering department, Prague, Czech Republic, e-mail: [email protected]

Abstract — In the paper it is presented a brief description of the technology of electricity production in the condensing power plant Tusimice II and mentioned replacement of main electric drives implemented by ŠKODA PRAHA Invest. The focus is placed on the modernization of electric drives as motors for fan mills, air and smoke fans in the boiler room, circulation pumps on absorber of desulfurization, generator and supply pumps in the machine hall, drives for the coal handling conveyors and drives in water management. It focuses also on backup resources – diesel-generators. There is also included practical experience of the implementation. Keywords — the power plant, drive, mill, fan, pump, conveyor.

I. INTRODUCTION A. General destcription of power plant Modernization of four units with installed capacity of 4 x 200 MW ensures the future operation under current European standards until around 2035, when the excavations of neighbouring mine Libous is expected. Power plant efficiency was increased by 6 % to 39 %, leading to 14 % fuel savings per produced MWh. Emissions (NOx, SO2, dust) levels were reduced by an average of 79 %. Own consumption unit was reduced from about 18 MW to 12,7MW.

Fig. 1. View on Tusimice II Beginning of construction Completion of modernization in Fuel Power-production unit The installed capacity of Desulphurized since

2010 2012 brown coal ETU II 4 x 200 MW 1997

II. BOILER For preparation of the coal powder 6 pieces of fan mills with powder burners are symmetrically installed. The ignition of the powder is determined by 6 stabile natural gas burners. Used fan mills provide a reliable boiler operation at all power levels within the control range 50– 105 % of nominal power. The rated boiler output is dependent on the quality of fuel, burning of which ensures the operation of four, respectively five mills; one resp. two mills are in the reserve backup. The transport of the combustion air serves for each boiler, only one air axial fan blows the air flow through the air heater type Ljungström into the air channels of the boiler. A. Mills and separators A new mill is equipped with an electric motor with high performance gearbox with increased rated speed at output and with variable speed by hydraulic clutch. Each fan mill set consists among others of fleeting founded grinding wheel on the main shaft of the gearbox, a new frontal single-end gear boxes, control hydraulic clutch and electric motor. 1) Hydraulic clutch: The mill drive is newly equipped with a hydraulic clutch type 750 SVTL spec with speed control. The input shaft of the hydrodynamic coupling is connected to a fixed coupling with the shaft of the electric motor; the output shaft is connected to a fixed coupling with the shaft gear. For the torque transfer the clutch utilizes the kinetic energy of fluid flowing between the drive and the driven bladed wheel, the flow quantity of oil regulates the speed of the driven parts. Controlled speed range is 76 to 100 %, corresponding to an operating range of the output speed 1106 -1455 rev / min (i.e. the speed of the milling wheel 438 -576 rpm/ min). 2) Electric motor: It used a new induction squirrel-cage motor type 1LA4404-4AN60: • Pn = 800 kW; Un = 6000 V; f = 50 Hz, nn = 1488 ot/min, In = 93 A, Ik = 5.5; ETA = 96.6 %; cos phi = 0.86; Mn = 5134 Nm; protection class IP55; cooling own IC411; insulation class F / B thermal utilization; • Max. permissible motor starting time is 15 s; number of starts allowed per 1 hour – 3 starts from cold / 2 starts from warm state, start-up equipment is controlled through hydraulic couplings, roller bearings with grease lubrication. B. Air and smoke blowers fan 1) Air fan: The fan is designed as a single-stage twospeed axial fan with variable pressure relief performance by turning vane impeller during the fan operation. It is of

Transactions on Electrical Engineering, Vol. 3 (2014), No. 3

horizontal configuration, the suction chamber at 90 °. The fan is designed for conveying air and to overcome the pressure losses related with technologies up to a maximum achievable static pressure. It used a new three-phase induction motor with squirrelcage deep-bar type 5V 227-06/10HV: • Rated power Pn = 1800/500 kW; nominal voltage Un = 6000 V AC; speed 995/595 1/min; roller bearings, shield, bearing with thermometers. • The electric motor is equipped with a cooling air-toair aluminium tube, it is provided with further heating of the stator winding (during disconnection of the device).

Fig.2 The arrangement of the air fan

Fig.3 The characteristic of the smoke fan

Legend of air fan main parts of: 1. The impeller, 2. suction chamber, 3. input box, 4. switchbox, 5. diffuser, 6. core diffuser, 7. cases of impellers, 8. fan shaft, 10. cooling tube, 11. female coupling, 12. male coupling, 13. clutch cover, 14. drive (electric motor) 17. lubricating stations 18. main bearing 20. fan base, 22. heater of air 24. electrical cabinet 2) The smoke fan: The boiler will be after comprehensive renewal equipped with one smoke axial fan. The fan is designed as a two-stage pressure axial fan with power control by turning vane impeller during the fan running. It will be of horizontal configuration, the suction chamber at 90 °. The fan is designed for conveying the flue gas with a maximum temperature of 200°C and to overcome the pressure drop related with technology up to the maximum achievable static pressure. It is used a three-phase induction motor with squirrelcage deep-bar type 5V 255-0HV:

79

• Rated power Pn = 4000 kW; nominal voltage Un = 6000 V AC; speed 595 1/min; roller bearings, shield, bearing with thermometers • The electric motor is equipped with a cooling air-toair from aluminium tubes, is provided with further heating of the stator winding (during disconnection of the device). Also these types of drives were solved as: • Fan of recirculated flue gas 0,4 kV 180kW • HV source of electrical precipitator L1 0,4kV 115kW • MST an escalator of slag 2 0.4 kV 15kW • MST crusher of slag 2 0.4 kV 30kW • MST Mixer of ash A1 0.4 kV 55kW III. DESUPHURIZATION A. Flue system Each installed desulphurization system consists of a heat exchanger of the raw gas, emergency cooling system, absorber (J1HTF01 BB001, BB001 K1HTF01), duct of pure gas including connection to the cooling tower system of oxidation, air circulators of absorber gypsum slurry pumps, supply of limestone slurry system process water and drain emergency suspension system as the main part of the operation of desulfurization. The flue gas from the boilers 21, 22, boiler 23 and 24 passes through a separate boiler flue fan, which increases the pressure to overcome the pressure system of the boiler, the exhaust gas duct of the raw gas, absorber and exhaust gas duct of clean gas. In our proposal FGD the lime is needed as alkaline reagent and process water to supply the water loss, and leads to the production of gypsum (CaSO4 * 2H2O) in the absorber unit. B. Absorber Each installed desulfurization system consists of the exhaust gas duct of raw gas, emergency cooling system, and absorber. Reconstructed electrostatic precipitators provide concentrations of solid pollutants up to 100 mg/Nm3 in the flue gas at the outlet. Dust extraction is loaded into a new desulfurization equipment in the outdoor configuration that was designed as a two-block, i.e. one desulfurization unit is used for two blocks. Related operational files limestone slurry preparation and dewatering gypsum are original. In the absorber, the flue gases are cleaned by back flow of the limestone-gypsum slurry. The circulation pumps of the absorber (J1, K1HTF10, AP001 20,30,40,50) are installed in a building of desulfurization and used for circulation of the gypsum slurry in the absorber. Gypsum slurry is sucked from the absorber and is transported through a sieve to the injection plane using a pressure pipe. • 10x motor of the circulated pump of the absorber 6 kV 1250 kW • compressor of the oxidation air

Transactions on Electrical Engineering, Vol. 3 (2014), No. 3

•

3x motor of oxidation air compressor 6 kV 1000 kW Also these types of drives were solved as: • The limestone suspension system – pumps of limestone suspension. • Drainage of gypsum – pumps of gypsum suspension. • The process water system – pumps of process water. • Drainage system – mixers of emergency drain tank of suspension. IV. MACHINERY HALL A. Turbine 200 MW steam turbine type KT – 200 – 17,5 is the three-body, uniform pressure condensing steam turbine with reheating for high-pressure part, with one single current high-pressure part, one with a medium current one part and one turbofan low pressure part. All turbine housings are double-walled. The output from the lowpressure part is led into the condenser. Drive solution for positioner with a three-phase induction electric motor controlled by a frequency converter and motor for emergency, aid, lifting oil pump are used. Furthermore, the motors: • condensate pumps • motors for pumps embedded cooling circuit 45 kW, 400 V / 50 Hz • motors for water ring vacuum pump, motor power 160 kW, power supply 6 kV / 50 Hz • motors for the supply pump, motor power 4250 kW, power supply: 6 kV / 50 Hz 1) Motors for the supply pumps 2p general repair of induction motor with squirrel cage with deep-bar type 2V 206 – 02H. Rated power: 4250 kW; nominal voltage: 6000 V; rated speed of 2980 rev / min, frequency: 50 Hz; shape of the motor: horizontal.

80

Direction of rotation: right, looking at the main shaft end. Motor design: closed, IP/44, own ventilation, ring type with the circulation of cold air through a water cooler located on the engine. Motor design: welded, coil stator winding, double-layer insulation in the class "F". Terminations: right. Circulated bearing lubrication from the system of hydro clutch, type designation R 17 K, the efficiency of 94 %. B. Generator • 4 pieces – 2-pole generator design to the end frame type according to the technical data: • 235.3 MVA / 200 MW; 15.75 kV (± 7.5) %, PF = 0.85; 50Hz +0.5 / -1.5 %; In = 8.63 kA, 3000 rpm. - Design according to IEC 34. • Cooling water 33°C inlet temperature (for warranty item). • Degree of protection IP 44. • Direction of rotation clockwise when viewed the turbine. • Location of cases: bottom. • Wake-up machine and auxiliary exciter with permanent magnets for brushless excitation system. • Turbine-generator clutch. • Complete excitation system consisting of two independent channel regulators. Each channel contains AVR (automatic / manual) stabilizer system and related facilities, two power units located beside the generator. The second channel is a complete replica of the first channel. Properties of the excitation system: design according to IEC: 60146-1-1 (rectifiers, inverters), 61000-4-1÷5 (EMI), 60439-1 (ballast).

Fig.4 Diagram of the generator

Transactions on Electrical Engineering, Vol. 3 (2014), No. 3

81

Fig.6 Coal handling

• Fig.5 P-Q diagram of the generator

V. COAL HANDLING Belt drive, traction and stretching for coal handling links are outdated and will be replaced by a new outside the weighing conveyor drive of the traffic routes that due to minimum wear will be existing. On the weighing transport route it will be exchanged only drives of screw conveyors under the weighing tanks and also the drive for conveyor belt T41. At the request of the operator, it was agreed that the new machines will be covered IP55 and will not be equipped with temperature monitoring bearings or winding temperature, the surface temperature during normal operation may not exceed 145 °C. 6000 V machine will be equipped with heating. It was also agreed that the new gearboxes will be designed with a service factor min. 1,8, gearboxes of inclined conveyors are equipped with reverse latch. The concept of the drive belts has been designed to maintain the existing: drive + clutch + driving drum, for conveyors T5 and T21 drive in plug-in design + driving drum, for adjusting conveyors T2A, T2B, T4 drive + drive chain + transfer drum. Conveyors with higher performance for power requirements have been fitted with hydrodynamic starting clutch with one pre-chamber, which will be equipped with a safety switch contact and fusible plug. For conveying where calculation of free run time came 5 s and longer drives are equipped with a brake system (disk + electromagnetic brake), which prevents to smothering of chutes by transported material in a sudden loss of service. Drives 9x belt drive of conveyor belt T3, 6000 V, 50 Hz, 160 kW, driving station. VI. WATER MANAGEMENT Raw water for ETU II is transported from the raw water pumping station on the river Ohře reservoir through a pair of gravity mains pipe to the plant. This water, which is used mainly for filling tower cooling water circuit, is due to the low temperature also used for cooling of some appliances in the machine room.

•

• • •

•

Cooling water pumping station is used for cooling water transportation for condensers TG and Tfeeding pumps. Raw water pumping station, where three raw water pumps pump the water into reservoirs ETU II and ETU I, which are interconnected. From there the water using gravity mains is distributed over the power plant. Water treatment used to treat water to the required parameters for other technological units. Chemical water treatment. Block condensate management removes undesirable impurities (– ion-solute substances and released oxides of metals and alloying elements) and adjust the raw condensate to the quality suitable for its re-introduction into the pipeline of boiler feed water. Water used to ensure refilling water to the cooling tower circuit and collecting, treatment and reuse of liquid waste from the plant operation.

VII. DIESELGENERATOR 2 pieces of diesel engines with an alternator form mono-block attached by flanges, firmly mounted on a rigid frame on which are also mounted cooler and the management and control panels. To prevent transfer of vibrations to the surroundings the whole aggregate (frame) is stored on silent blocks. Aggregate type P1650E – Standy Rated power – stand by 1650 kVA (1320 kW) Dimensions L x W x H (mm) 5294x2039x2307 mm The heat dissipated by coolant 452 kW The amount of cooling air 1410 m3.min-1 Maximum start time DG to the desired power in 15 s Back-up time 8 hours ACKNOWLEDGMENT Thanks to the authors of the detail design we could realise this project. REFERENCES [1] The detail design project of Tusimice II power plant, electrical part

Transactions on Electrical Engineering, Vol. 3 (2014), No. 3

82

Modelling of Regulatory Electricity Networks Ability with Rotating Flywheels Emil Dvorský1), Lenka Raková 2) 1)

West Bohemia University, Department of Power Engineering and Ecology, Pilsen, Czech Republic, e-mail: [email protected] 2) West Bohemia University, Department of Power Engineering and Ecology, Pilsen, Czech Republic, e-mail: [email protected]

Abstract — Electricity as one of energy forms has still significant disadvantage; it cannot be effectively stored. Mostly, there is necessary, for a regulation process, to transform it into another energy form, which is loaded with solutions of technical and economic issues of energy transformation. One of the prospective technology from total transformation ones can be the Flywheel Energy Storage – FES, which is possible to meet the criteria of Uninterruptible Power Supply - UPS. FES has the potential to become a viable option compensation for traditional electric storage technologies or regulatory power source.

I. INTRODUCTION Rotary accumulators, at the current technological level, appear to be a competitive alternative to other types of storage technologies or regulated electrical power sources to ensure the supply of electricity to the consumers in order to cover supply of the required load with the electric quality parameters – a high performance modular variant is shown in Fig. 1.

necessary covered load can be defined as Critical Load – CL, for which the degree of supply security is "1" – uninterruptable load. The current most frequently used method of providing backup power is primarily by batteries and diesel generators. In terms of time the battery is used for uninterrupted power to the load until it is secured by diesel-generators. Maintaining energy parameters, although it is the main task of such sources, must be also provided at delivery quality parameters. For AC appliances these parameters are represented by frequency or voltage compliance within the required limits. Fig. 2 shows the case of FES use for the frequency regulation in a electricity distribution network of NYISO Company, which ensures the necessary regulatory power between sources and load in New York – Fig. 3.

Fig. 2. FES for providing frequency regulation "http://beaconpower.com

Fig. 1. UPS system working with FES„http://www.power-thru.com/

The flywheels are accumulating electricity in form of the kinetic energy of rotating masses. The rotating mass can serve as a short-term backup power source of energy, in the event of failure of the main power source. Currently, they founded utilisation as backup resources as UPS systems, which are particularly important for telecommunications systems, data centres, etc. The

Fig. 3. New York distribution system (DS) and its connection with neighbouring systems - „http://www.nyiso.com“/

For flywheels use speaks their better operating characteristics with regard to higher efficiency, compact device functionality at a higher temperature range guaranteeing reliable operation.

Transactions on Electrical Engineering, Vol. 3 (2014), No. 3

The issue of assessing the suitability of FES use in the DS supply is shown on the basis of FES utilisation model in micro network when platform MATLAB / Simulink is used. The aim was to eliminate the existing lack of real data from the FES operation. II. THE TRANSFORMATION PROCESS OF FES FES is composed of a rotating flywheel and induction machine, which works simultaneously as a motor and generator (M-G). During normal operation, it converts electricity from the grid into kinetic energy. If necessary, the stored kinetic energy is converted back into electricity using a generator – Fig. 4.

83

manufacturer of the flywheel. However, it can be also calculated the ultimate tensile strength σ of relevant material from which it is produced:

ωmax = k .

l σ r ρ

In (7) k is the safety factor. The flywheel may be in the operating period T charged or discharged at a constant speed or at a speed increase or decrease. Speed course changes [2] are exponential and it can be expressed for discharge in the given period by: ωmax  K ω( t ) =  − 0 .t ωmax .e J

Fig. 4. The principle function of FES/

The mathematical model of transformation of electricity into kinetic energy and back is derived on base of the equation for kinetic energy in the rotating mass [1]. The value of the kinetic energy of the container body is determined by its mass and velocity: E=

1 2 mv 2

(1)

In case of the flywheel the kinetic energy stored in the rotating mass and the velocity of the mass is determined by the angular velocity ω of the radius r, whereby (1) goes into the shape: E=

1 ( mr .ω )2 2

(2)

Flywheel body mass is spread with different radii at the same angular velocity. The total kinetic energy is then proportional, besides to the rotational speed, also to the moment of inertia J of rotating masses, which in case of a cylinder is:

∫

J = r 2 .dm = m.r 2

(3)

While mass of the cylinder is dependent on the distance l and given specific mass ρ : m = π .ρ .l .r 2

(4)

Moment of inertia is then: J = π .ρ .l .r 4

(5)

The total stored energy in the flywheel expressed by the inertia moment is:

Eakum =

1 2 J .ωmax 2

(6)

Maximum speed is determined by the material properties of the flywheel and is always guaranteed by the

(7)

t ≤T

(8) t ≥T

K0 is the conversion factor of the mechanical torque to electricity. For commercially available flywheels the period T, during which the flywheels operates at ωmax is usually always known. If not available, it can be approximately determined by the rules of dependence between the moment and inertia of the flywheel power output: Tmax =

2 J .ωmax 2.Pmax

(9)

The flywheel power, with exponential speed change, can be derived from (10), in respect of each time:  Pmax  K P= − 0 .t 2 η setrωmax .e J

t ≤T

(10) t ≥T

Another important parameter is the efficiency of the flywheel. The efficiency can be expressed by power losses, which are needed to maintain the speed of the flywheel at the nominal speed or in the standby mode. The power losses, which are necessary to maintain the flywheel in standby mode, are ranging from 0.2 % to 2 % of the overall flywheel power output. For comparison it may be noted effectiveness of conventional UPS systems (batteries), which typically ranges from 95 % to 98 % [3]. III. FES modeling For computation of FES it can be used FES models that are available on the network. One of the examples is presented in [5]. It is also possible to use the published models – [4]. The presented model consists of two basic parts. The first one is a flywheel, which evaluates the voltage and current of the bus-bar which is connected to the load, and controls the power output from the flywheel. In to the model, there is possible to enter the parameters, which are represented by the maximum and minimum angular velocity (rad/s), power (kW), energy capacity (kW.s) efficiency (–), time charge (s), time of maximum power discharge (s) and the time of discharge from the halfpower (s). The output of the model is its flywheel power output (kW), power work output (kW.s) and angular velocity (rad/s). The typical discharge time of the

84

Transactions on Electrical Engineering, Vol. 3 (2014), No. 3

flywheel, according to the manufacturers, is between 10 to 30 seconds, whereby the charge of the flywheel is required from 1 to 10 minutes, depending on the technology used by the flywheel. Control of the flywheel is based on energy bus-bars balance. If the value of the energy bus-bar is greater than it is necessary for load, flywheel begins to charge excess energy. In the absence power the flywheel is discharged so that a critical load can be maintained. When the backup source (diesel) is connected, through which the frequency and voltage of the bus-bar is stabilized, the flywheel goes into charging mode. Solution of discharge model mode of the flywheel based on (10) is shown in Fig. 5, and the solution of charging model mode is shown in Fig. 6.

Fig. 7. The M-G model

For the purpose of the FES functionality verifying, the model was applied on four different types of flywheels with performance 120, 150, 160 and 250 kW. Modelled output discharge characteristics were compared with manufacturers presented ones, which are based on test measurements. Results are shown in Fig. 8.

Fig. 8. The discharge characteristics of flywheels

Fig. 5. The discharge flywheel model

Fig. 6. The charge flywheel model

Fig. 9. FES Parallel connection

The speed of the flywheel rotation is calculated on the base of (6):

In the event when we need available power for an extended period, the units should be connect in parallel. The model is then possible to answer questions like – for the time required to establish the necessary power output of parallel cooperating flywheels. Fig. 9 shows solved number of the parallel units with single output of 250 kW which must be connected to the total power output of 800 kW, to be available for the required time.

ω=

2E J

(11)

The second part of the model is the motor-generator system which performs transformation between the mechanical and electrical systems. Most FES use induction motors or synchronous machines with permanent magnets. The presented motor-generator model uses a synchronous machine (Fig. 7).

85

Transactions on Electrical Engineering, Vol. 3 (2014), No. 3

Fig. 10. Synchronization diesel-generator on the bus

The proposed FES model is connected via a bus with the backup power source which is represented by a diesel-generator that takes, over a specified time, the critical load supply. In this presented case, the modelled critical load was set to 750 kW. Function of the dieselgenerator is evident from Fig. 10. At the time of power failure (t = 0 s), the flywheels begin to deliver the bus load required for covering the critical load. Because the failure is longer than 5 seconds, the diesel generator is after the start synchronized with the network in 17 seconds. The entire network is after this time powered by the diesel generator and the flywheels go into the charging mode. From the speed course of the system – Fig. 11, we can see that the flywheel will be charged back after 190 seconds.

Fig. 11. Speed course of the system

As a result of the continuous power supply, using the FES system, the frequency is also stabilized – Fig. 12 – red curve. +0,2 +0,1

-0,1 -0,2 -0,3

Fig. 12. Speed course of the system

IV. CONCLUSION The created model demonstrated the use of the FES in conjunction with other supplying resources the relevant network is possible. Model example was applied to provide power for critical loads in a micro-network to avoid voltage loss. The use of storage elements in micro-networks may be more effective due to the economic advantages compared with conventional accumulation elements, because their life is about four times higher with increased efficiency. Using of the FES in DS is considerably limited by their discharge time. In conjunction with other regulatory means they can find their place.

ACKNOWLEDGMENT This article was created under the student scientific project SGS 2012-047 foundation.

REFERENCES [1] Understanding Flywheel Energy Storage: Does High Speed Really Imply a Better Design?”, Technical Report, Active power Inc., Austin, TX, white papers no 112, 2008. [2] T. T. Leung, “Concept of a modified flywheel for megajoule storage and pulse conditioning ”, IEEE Transactions on Magnetics, Vol. 27, Issue 1, pp. 403 – 408, Jan 1991 [3] S. Eckroad, “Flywheels for electric utility energy storage,” Technical Report, Electric Power Research Institute, Palo Alto, CA, EPRI report TR-108889, Dec 1999. [4] M. Ahrens, L. Kucera, R. Larsonneur, “Performance of a magnetically suspended flywheel energy storage device”, IEEE Transactions on Control Systems Technology, Vol. 4, Issue 5, pp: 494 – 502, Sept. 1996. [5] ]http://www.calculatoredge.com.

Transactions on Electrical Engineering, Vol. 3 (2014), No. 3

Mašek, Z.: Control System for Hydrostatic Transmission of Railcar M27 The paper describes functions of a control system for hydrostatic transmission for the reconstructed railcar M27 which is used for transport of passengers on a narrow gauge railway near Jindřichův Hradec, Czech Republic. Software for this control system was developed at the University of Pardubice. Dvořák, P., Fajt, T., Sošková, I. and Hloužek, J.: Water Cooled Induction Traction Motor for 100 % Low Floor Tram Car The paper deals with the design of the traction motor for 100 % low floor tram car. Within the design it was necessary to deal with many problems which have significant impact on the final product. The most interesting problems were the conception of the torque transmission, mounting of the brake equipment on the motor body, arrangement of the motor inside the bogie, cooling of the motor or protection of the motor against the water and pollution which can enter inside the motor. In this paper there are discussed the electromagnetic model, and special thermal and ventilation model of the motor. The final design was validated within the type testing. Kopecký, M., Švanda, J. and Vlček, M.: Real-time Simulation of 3 Parallel PWM Rectifiers This paper describes the development of a real-time model up to 3 parallel PWM rectifiers and its implementation on FPGA using LabVIEW development environment. The main benefit of this real-time model is the fact that there is no need for a real device or a test stand for debugging of traction drive control SW. The Hardware-in-the-Loop testing with similar RT model of an induction machine has already brought large financial and time savings. Moreover, destructive states or states difficult to evoke can be tested using such a real-time model. Foltyn, D.: Electric Drives in the Modernized Power Plant Tusimice II In the paper it is presented a brief description of the technology of electricity production in the condensing power plant Tusimice II and mentioned replacement of main electric drives implemented by ŠKODA PRAHA Invest. The focus is placed on the modernization of electric drives as motors for fan mills, air and smoke fans in the boiler room, circulation pumps on absorber of desulfurization, generator and supply pumps in the machine hall, drives for the coal handling conveyors and drives in water management. It focuses also on backup resources diesel-generators. There is also included practical experience of the implementation. Dvorský, E., Raková, L.: Modelling of Regulatory Electricity Networks Ability with Rotating Flywheels Electricity as one of energy forms has still significant disadvantage; it cannot be effectively stored. Mostly, there is necessary, for a regulation process, to transform it into another energy form, which is loaded with solutions of technical and economic issues of energy transformation. One of the prospective technology from total transformation ones can be the Flywheel Energy Storage – FES, which is possible to meet the criteria of Uninterruptible Power Supply - UPS. FES has the potential to become a viable option compensation for traditional electric storage technologies or regulatory power source.

_____________________________________________________________________________________________ TRANSACTIONS ON ELECTRICAL ENGINEERING VOL. 3, NO. 3 HAS BEEN PUBLISHED ON SEPTEMBER 30, 2014