Modeling and Simulation of large scale Power Systems in More Electric Aircraft Y. Ji
M. R. Kuhn
Institue of System Dynamics and Control German Aerospace Center Oberpfaffenhofen-Wessling, Germany
[email protected] [email protected]
Abstract—More electrically powered aircraft reveals some significant advantages such as weight decrease, reduced maintenance requirements and increased reliability and passenger comfort. However, the development of the future more-electric aircraft (MEA) systems is a very challenging task. Model based approach provides significant improvement for total aircraft design process. A key factor of applying model based design approach is dedicated modeling and simulation techniques. Among various available modeling languages and tools in this regard, Modelica - an object-oriented, equation based modeling language is gaining more attention and acceptance in the aircraft society thanks to its strong capability to conveniently model complex physical systems. With the simulation results of component stand-alone tests as well as the tests of an integrated aircraft power network, Modelica has been proved to be suitable for the virtual testing of complex energy systems in the future MEA design process.
I. I NTRODUCTION More electrically powered aircraft reveals some significant advantages such as weight decrease, reduced maintenance requirements and increased reliability and passenger comfort. However, the development of the future more-electric aircraft (MEA) systems is a very challenging task: the respective subsystems are highly integrated to achieve an optimum efficiency and performance at both aircraft and system level. To deal with this task, the traditional document based aircraft design process has to be improved by utilizing high quality mathematical models of components and sub-systems. Model based concept is applicable for every design phase in the aircraft development process, i.e. conception phase, system specification phase, system development phase and system verification phase. Especially in the system verification phase, a virtual integration platform for energy systems allows addressing integration issues prior to their physical integration on the test rigs and also extends test coverage. A key factor of applying model based design approach is dedicated modeling and simulation techniques. Among various available modeling languages and tools in this regard, Modelica - an object-oriented, equation based modeling language is gaining more attention and acceptance in the aircraft society thanks to its strong capability to conveniently model complex physical systems. containing, e.g., mechanical, electrical, hydraulic, thermal, control, electric power or process-oriented sub
components. Modelica was developed as a free, object-oriented and equation-based modeling language. It has significant benefits such as easy re-usability of models and multi-domain modeling capability. In combination with the powerful commercial simulation environment Dymola 1 , a convenient platform is provided to accomplish multidisciplinary system simulation and integration of complete electric networks including all sub-systems. Concerning the outputs delivered by the recent European Commission projects in cooperation with aircraft industry e.g. VIVACE (Value Improvement Through A Virtual Aeronautical Collaborative Enterprise) 2 and MOET (More Open Electrical Technologies) 3 , the level of sophistication of Modelica being a modeling and simulation tool for future aircraft electrical systems has been demonstrated step by step. This paper proposes a Modelica-based solution for the virtual testing process of large scale power systems in MEA. The components and sub-systems in a MEA power network often contain electrical generators, power converts (AC/DC, DC/DC, AC/AC) and various electric power loads such as permanent magnetic synchronous machine and constant power loads. All components have been considered at both functional and behavioral levels to fulfill all modeling requirements of diverse design tasks.With the simulation results of component standalone tests as well as the tests of an integrated aircraft power network, Modelica has been proved to be suitable for the virtual testing of complex energy systems in the future MEA. II. E LECTRIC NETWORK ARCHITECTURES OF MEA In a conventional aircraft system architecture, fuel is converted into power by the engines. Most of this power is expended as thrust to propel the aircraft. The remainder is transmitted via, and converted into, four main forms of nonpropulsive power, i.e. pneumatic, mechanical, hydraulic and electrical power. • Pneumatic power generated by high-pressure compressor to power power the Environmental Control System (ECS) and supply hot air for Wing Ice Protection System (WIPS). 1 www.dynasim.se 2 www.vivaceproject.com 3 www.eurtd.com/moet/
Primary Controls
Primary Controls
Variable frequency generator
GCU Secondary Controls
Commercial Loads HVAC 230V Bus
Engine Systems
Electrical Distribution
Starter / Generator
ATU
ATRU Ice Protection
Environmental Control
HVDC 540V Bus
AC 115V Bus
Landing Gear
115V AC loads ECS
Electric Power
Local Hydraulics
Fig. 1.
Electro Mechanical
DC 28 Bus
•
•
28V DC loads
More electric aircraft Fig. 2.
•
Actuators DCCU
Electric Pneumatic
Mechanical power from engines to central hydraulic pumps and other mechanically driven subsystems, and to the main electrical generator. Hydraulic power by hydraulic pumps to drive the actuation systems for flight control, landing gear and numerous ancillary systems. Electrical power by generator to supply electric energy to the avionics, to cabin and aircraft lighting, to the galleys, and to other commercial loads
Currently, each system in aircraft has become more and more complex over decades of developments, resulting in an architecture that is far from being optimal. Therefore, an optimized aircraft architecture - More Electric Aircraft depicted in Figue 1 (e.g. no gearbox, reduced engine bleed, local hydraulic source and more electrical power) has been proposed [14]. Using more electric energy could substantially reduce the consumption of non-propulsive power. Although the aircraft industry clearly tends to design future aircraft with more electric power energy, discussions about the optimal energy power network is still ongoing. Comparing with AC power distribution system, high voltage DC power distribution system (HVDC) is gaining more attentions [12] thanks to the benefits like mass/volume saving in the feeders, possible use of regeneration energy from electric actuators and etc. For instance, a possible DC power distribution network configuration is depicted in Figure 2. This architecture contains a 18-pulses auto-transformer rectifier unit (ATRU) which transfers 230V AC voltage to 540V DC voltage supplying DC loads such as electric environmental system (ECS) and electric actuators. By replacing the variable frequency generator in the proposed architecture by a switched reluctance generator which directly generates 540V DC voltage, the ATRU even can be saved [4], [6].
Electric network architectures for More Electric Aircraft
III. M ODELING REQUIREMENTS OF AIRCRAFT POWER SYSTEMS
A key-factor for the success of developing the more electric aircraft is to incorporate high quality system models in the complete aircraft design process [1], which briefly can be divided into 4 major phases: concept phase, system specification phase, system development/validation phase and system verification phase [3]. • Concept phase: During concept phase a two-fold iterative optimization is performed on aircraft manufacturer side. This includes aircraft concept and global energy system architecture optimization. Methods and tools for the tasks during the concept phase have been investigated e.g. ENADOT [13], with which an optimal architecture of electrical energy system can be achieved. • System specification phase: In this phase, a frozen energy system concept is provided by the aircraft manufacturer. Additionally, more detailed aircraft data about structure, cabin, light physics, engine and electrical power generation are available. The selected system suppliers conduct full concept definition where all the requirements and risks are understood. The aircraft manufacturer’s requirements is transformed to the level of equipment suppliers by the system suppliers. Stability studies and failure analysis of aircraft electrical network are typical activities during the system specification phase. • System development and validation phase: In this phase preliminary and detailed design of equipment takes place. Verification and validation for artifacts are done, which are produced during this phase. • System verification phase: The typical tasks of the aircraft manufacturer in this phase are: monitor supplier system development by verification of system performance and functions, integrate systems in physical and functional aircraft, verify integrated systems and validate simulation
Design phase Concept System specification System development System verification
Typical task Architecture optimization Stability studies Control design Virtual testing
Required model Level 1 Level 2/3 Level 2/3 Level 2/3
TABLE I M ODEL REQUIREMENTS IN DIFFERENT AIRCRAFT DESIGN PHASE
models versus test results. The objective in the system verification phase is to demonstrate the maturity of the systems in a realistic integration and verification of moreelectric aircraft systems, capable of covering all phases of the development process. The virtual integration platform for energy systems allows addressing integration issues prior to their physical integration on the test rigs and also extend test coverage. Power quality investigation of the integrated network will be of the interest in this phase. Today, aircraft industry utilizes a multi-level approach for the design of the aircraft system [8]. • Architectural models consist of algebraic equations and are used for steady-state power consumption calculations. • Functional models are derived from behavioral models by time averaging of high frequency periodical switching waveforms. A functional model reflects only low frequency behavior of the original system excluding switching ripples and can be used for stability study [10], [9] and control design. • Behavioral models are based on equations derived from the subsystem structure and electrical circuit. The behavioral models reflect both low and high frequency dynamics including switching effects serving to perform power quality simulation and analysis [7], [15]. The Table I presents an overview of model requirements in all aircraft design phases. In this paper, only functional and behavioral models have been considered. IV. M ODELING OF MEA POWER SYSTEMS High quality models of components and the integrated power network is a precondition for the virtual testing process in the aircraft system verification design phase. An electric power network in the MEA typically covers several physical domains such as electrical systems, magnetic systems, mechanical systems and control. Therefore, a modeling language to be selected for virtual testing process has to be capable for modeling multidomain systems. Furthermore, powerful solvers which can well deal with complex dynamic systems including switching actions are needed. An electric power network often is very complex large scale system, for instance, a proposed aircraft power network ( Figure 5) containing an variable frequency generator (VFG), a high power 18-pulses auto-transformer rectifier unit (ATRU) , a 3 phase auto-transformer, a DC/DC power converter (DCCU), a permanent magnetic synchronous machine (PMSM) and a 115V AC constant power load (CPL) has more than 100 dynamic state variables.
Fig. 3.
Scheme diagram of an 18 pulses auto-transformer rectifier unit
The variable frequency generator consists of a synchronous machine, an exciter and a generator control unit. The VFG model has been completely modeled with Modelica. Both behavioral model with 3 phase AC power connector and functional model using 3 phase RMS (Root Mean Square) power connector are implemented. Using the diode model in the Modelica standard library, the topological model of 6/18-pulse bridge rectifier (RU) can be made very easily. Detailed parameter in diodes such as forward state-on resistance and forward threshold voltage can be freely tuned to best match the real performance of physical components. When modeling the high power 18-pulse auto-transformer rectifier unit (ATRU) model for the MEA electric power network depicted in Figure 3, the inrush current effect at switchingon has to be considered. The magnetic core of auto-transformer is built with Modelica magnetic library including magnetic hysteresis models in order to present the inrush current effect. The functional model ignoring switching actions of diodes is also available. The 3-phase auto-transformer unit (ATU) with input 230V AC voltage and output 115V AC voltage is also modeled by Modelica magnetic library [16]. The geometry parameters of magnetic circuit such as area and length of magnet core as well as the characteristics of magnetic hysteresis loop e.g. coercive field and permanent magnetic field can be conveniently set by users. The full bridge isolated DC/DC buck converter unit (DCCU) contains a full bridge inverter, magnetic transformer, input filter and phase-shifted control. The DCCU typically supplies 28V DC low voltage loads in the electric power network. The scheme of the modeled DC/DC buck converter is depicted in Figure 4. Modeling and simulation of DC/DC converter units could be very challenging job, since this component inherently works with very high frequency PWM modulation. This high frequency switching operation could lead great amount of event actions which will significantly slow down or break the simulation process. The synchronous feature [11] provided by Modelica can efficiently solve this issue by transferring the state events into time events. Classical averaging technique has been utilized to implement the functional DC/DC buck converter. A permanent magnetic synchronous motor (PMSM) can be a very interesting candidate as electric actuator [2] to directly drive flight control systems in the MEA. The Modelica PMSM model addresses all important components in an electric actua-
i1(t)
1 : n
D5
L
iD5(t)
Vg
+ –
vs(t)
–
Q2
Fig. 4.
Q4
C
R
– : n
D4
+
YBD
YBE
YF
Z D
vT(t)
D2
i(t)
+
+
Z E
Z F
v –
YBD
YBE
YBF
D6 Z D
Z E
Scheme diagram of an DC/DC converter unit HB*1'
Z F
/HDNBDLU«
D3
/HDNBDLU«
Q3
D1
/HDNBDLUBD
Q1
&RUH&
/HDNF
&RUH%
/HDNE
R
&RUH$
/HDND
B
DCCU JURXQG
ATRU induc torLine
res is torLine
l_phas e s witch1
PMSM2
Fig. 6.
Modelica model of AC/AC auto-transformer
PMSM
AT U1 p1 p2 p3
VFG
ATU
Fig. 5.
115V AC CPL
An integrated electric power network for MEA
tor unit such as PMSM, input filter, power inverters with deadtime, protections for over/under voltage, pre-charger and motor control unit. Friction in bearing system is also considered. The functional PMSM model has been derived by replacing power inverters by continuous transfer functions and transforming 3phase values into RMS values. An integrated electric power network model including behavioral models of a VFG, an ATRU, an ATU, a DCCU, a PMSM and a 115V AC constant power load (CPL) is depicted in Figure 5. V. V IRTUAL TESTING As previously mentioned, a virtual integration platform for energy systems allows addressing integration issues prior to their physical integration on the test rigs and also extends test coverage. Today, virtual testing of the integrated aircraft energy system is becoming an indispensable task in the system verification design phase. A. General process of virtual testing The total virtual testing process can be briefly divided into two steps. Before performing simulations and tests of the total integrated system model, each subsystem or component model shall be firstly evaluated by so-called component standalone test. A standalone test usually consists of a bunch of tests such as power connection, power disconnection, power consumption at steady state, current harmonic analysis and so on, for one component. Standalone tests are required for both functional and behavioral models. Virtual tests can begin once standalone
tests for all components and subsystems are successful finished, simulations of integrated system model can be done. Finally, specific analysis and post-treatment tasks can be performed based on the simulation results of the integrated models. B. Requirements of virtual testing for More Electric Aircraft Virtual testing covering total electric power systems in MEA is very challenging. To deal with this job, special requirements in terms of methods and tools have to be fulfilled. First of all, component and system models shall be build at both functional and behavioral levels. Furthermore, to simulate complete aircraft power systems which are usually very stiff and suffering from huge amount of event handling actions due to switching components, the performances and robustness of solver have to be ensured. Since various analysis tasks such as harmonic analysis in frequency domain and stability analysis in time domain, tools for analysis and post-processing [5] are required. Finally, it should be possible to customize different test scenarios by scripting tools. C. Standalone tests for electric components A 230VAC/115VAC auto-transformer model depicted in Figure 6 is used to demonstrate standalone tests for components. The tests considered in this paper are harmonic current test, inrush current test, power connection and power disconnection test. The harmonic current analysis aims to determine pollution due to the equipment on different frequency levels. Fast Fourier transformation is performed when the simulation of ATU model reaches steady state. The results for harmonic current analysis with system setup of 230V and 720Hz input voltage is depicted in Figure 7. The result of inrush current test which studies if the ATU may cause inadvertent trip at power up is presented in Figure 8. Power load connection and disconnection tests whose results are depicted in Figure 9 and 10 aim to check if the ATU behavior at load connection and disconnection is compliant with the associated requirement.
3RZHUGLVFRQQHFWLRQZLWK+]/RDG 3)
+DUPLQLFDQDO\VLVDW9+]
>$@
VZLWFK
LQSXWFXUUHQWDWQRGH$ LQSXWFXUUHQWDWQRGH% LQSXWFXUUHQWDWQRGH&
2XWSXWFXUUHQWIRUSKDVH$ 2XWSXWFXUUHQWIRUSKDVH% 2XWSXWFXUUHQWIRUSKDVH&
>$@
))7DWQRGH$ ))7DWQRGH% ))7DWQRGH&
$PSOLWXGH
325YROWDJHIRUSKDVH$ 325YROWDJHIRUSKDVH% 325YROWDJHIRUSKDVH&
(
Fig. 7.
(
( )UHTXHQF\+]
(
>9@
(
Current harmonic analysis of ATU at 230V and 720Hz input voltage
Fig. 10. Power disconnection test of ATU with 75% load and 0.8 power factor
,QUXVK&XUUHQW$QDO\VLVDW9+]
,QSXW9ROWDJHIRU3KDVH$ ,QSXW9ROWDJHIRU3KDVH% ,QSXW9ROWDJHIRU3KDVH&
>9@
6LPXODWLRQUHVXOWRILQWHJUDWHGHOHFWULFSRZHUQHWZRUN &XUUHQW9)*SKDVH$
&XUUHQW9)*SKDVH%
&XUUHQW9)*SKDVH&
,QSXW&XUUHQWIRU3KDVH$ ,QSXW&XUUHQWIRU3KDVH% ,QSXW&XUUHQWIRU3KDVH&
>$@
+9'&YROWDJH
Fig. 8.
Inrush current analysis at 205.V and 360Hz input votlage
D. Testing of the integrated aircraft power system
After successful stand-alone tests for all components, various scenarios for testing the complete electric power network can be started. To demonstrate the capability of Modelica/Dymola to deal with large scale power systems, the proposed electric 3RZHUORDGFRQQHFWLRQZLWK+]/RDG 3) VZLWFK
2XWSXWFXUUHQWIRUSKDVH$ 2XWSXWFXUUHQWIRUSKDVH% 2XWSXWFXUUHQWIRUSKDVH&
>$@
325YROWDJHIRUSKDVH$ 325YROWDJHIRUSKDVH% 325YROWDJHIRUSKDVH&
>9@
Fig. 9. factor
Power load connection test of ATU with 75% load and 0.8 power
Fig. 11. Simulation result of integrated electric power network: VFG current and HVDC voltage (behavioral level)
power network depicted in Figure 5 has been simulated in Dymola at both behavioral and functional levels. In this electric power network, the ATRU is connect to grid at 0.0025 second. After the pre-charging ATRU with 25e-3 second, the DC output of the ATRU is connected with the HVDC network. The PMSM which has a 20e-3 second pre-charging time is connected with the HVDC network at 0.055 second. After the power inverter in the PMSM is activated, a constant speed command is given for the PMSM under a constant load. The speed-up process of the PMSM presented by the input currents of PMSM, the output voltage from ATRU, the output AC currents and voltages of the VFG is recorded in the Figure 11 and Figure 12 for behavioral model. The results of same test for functional model are depicted in Figure 13 and Figure 14. In the simulation results, it is clear to see the inrush currents at the moment of switching on ATRU and DC ripple at the ATRU output. These values are very important indicators for the stability study for the electric power network in MEA. [10]
6LPXODWLRQUHVXOWRILQWHJUDWHGHOHFWULFSRZHUQHWZRUN
6LPXODWLRQUHVXOWRILQWHJUDWHGHOHFWULFSRZHUQHWZRUN
&XUUHQW3060
&XUUHQW3060
>$@
&XUUHQW'&&8
&XUUHQW'&&8
>$@
Fig. 12. Simulation result of integrated electric power network: current of PMSM and DCCU (behavioral level)
&XUUHQW9)*506
>$@
+9'&YROWDJH
Fig. 14. Simulation result of integrated electric power network: current of PMSM and DCCU (functional level)
6LPXODWLRQUHVXOWRILQWHJUDWHGHOHFWULFSRZHUQHWZRUN
Fig. 13. Simulation result of integrated electric power network: VFG current and HVDC voltage (functional level)
VI. C ONCLUSION An overview of virtual testing process for the electric power network of MEA is addressed in this paper. It has been shown, that Modelica is very suitable for modeling and simulation of very complex large scale power systems. With additional powerful post-processing and scripting features provided by the Modelica platform-Dymola, a successful virtual testing process can be guaranteed. ACKNOWLEDGMENT The research leading to these results has received funding from the European Union’s Seventh Framework Programme (FP7/2007-2013) for the Clean Sky Joint Technology Initiative under grant agreement no. CSJU-GAM-SGO-2008-001 R EFERENCES [1] J. Bals, Y. Ji, M. Kuhn, and C. Schallert. Model based design and integration of more electric aircraft system using modelica. Moet Forum at European Power Electronics Conference and Exhibition, 2009.
[2] Wenping Cao, B.C. Mecrow, G.J. Atkinson, J.W. Bennett, and D.J. Atkinson. Overview of electric motor technologies used for more electric aircraft (mea). Industrial Electronics, IEEE Transactions on, 59(9):3523– 3531, 2012. [3] T. Giese, D. Schlabe, R. Slate, M. Crespo, F. Tichy, and C. Baumann. Extended design office concept definition. Cleansky WP2.1.1 deliverable, 2010. [4] Liqiu Han, Jiabin Wang, and D. Howe. Small-signal stability studies of a 270v dc more-electric aircraft power system. In Power Electronics, Machines and Drives, 2006. The 3rd IET International Conference on, pages 162–166, 2006. [5] Y. Ji and J. Bals. A novel modelica signal analysis tool towards design of more electric aircraft. In Computer Science and Information Technology (ICCSIT), 2010 3rd IEEE International Conference on, volume 5, pages 152–156, 2010. [6] Y. Ji and J. Bals. Modeling and control design of switched reluctance machine using modelica. The 19th IASTED International Conference on Applied Simulation and Modeling, 2011. [7] Y. Ji, A. Pfeiffer, and J. Bals. Optimization based steady-state analysis of switched power electronic systems. In Control and Modeling for Power Electronics (COMPEL), 2010 IEEE 12th Workshop on, pages 1–6, 2010. [8] M. Kuhn and M. Otter. A multi level approach for aircraft electrical systems design. 6th International Modelica Conference, 2008. [9] M.R. Kuhn, Y. Ji, H. D Joos, and J. Bals. An approach for stability analysis of nonlinear electrical network using antioptimization. In Power Electronics Specialists Conference, 2008. PESC 2008. IEEE, pages 3873– 3879, 2008. [10] M.R. Kuhn, Y. Ji, and D. Schrder. Stability studies of critical dc power system component for more electric aircraft using μ sensitivity. In Control Automation, 2007. MED ’07. Mediterranean Conference on, pages 1–6, 2007. [11] Martin Otter, Bernhard Thiele, and Hilding Elmquvist. A library for synchronous control systems in modelica. 9th International Modelica Conference, 2012. [12] X. Roboam. New trends and challenges of electrical networks embedded in more electrical aircraft. In 2011 IEEE International Symposium on Industrial Electronics, pages 26–31, 2011. [13] C. Schallert. Inclusion of reliability and safety analysis methods in modelica. 8th International Modelica Conference, 2011. [14] C. Schallert, A. Pfeiffer, and J. Bals. Generator power optimisation for a more-electric aircraft by use of a virtual iron bird. 25th International Congress of the Aeronautical Sciences, 2006. [15] Ji. Y, J. Bals, and A. Pfeiffer. Multi-level power quality assessment towards virtual testing of more electric aircraft. In IPEC, 2010 Conference Proceedings, pages 28–33, 2010. [16] J. Ziske and T Boedrich. Magnetic hysteresis models for modelica. 9th International Modelica Conference, 2012.