The HIL Based Model Validation Paradigm - Tools, Challenges, and Application Examples

Transcription

1 The HIL Based Model Validation Paradigm - Tools, Challenges, and Application Examples Michael Mischa Steurer Leader Power Systems Research Group at FSU-CAPS steurer@caps.fsu.edu, phone: th Annual Conference of the IEEE Industrial Electronics Society Nov 13, 2013, Vienna, Austria

2 Overview Role of Hardware-inthe-Loop (HIL) FSU-CAPS 5 MW power HIL (PHIL) facility De-risking of PHIL experiments Model Verification and Validation (V&V) PHIL examples FSU-CAPS High Bay PHIL Lab 2

3 Basics of HIL Simulation Approach A device under test (DUT) is interfaced to a simulated environment through HIL interfaces to a real-time simulation model Controller HIL (CHIL) HIL Interfaces use control level (low voltage) signals for I/O Power HIL (PHIL) Power amplifiers and/or actuators are used for interfacing Full power, high fidelity stimulation DUT to be exercised in a wide range of potentially realistic environments Execution of extreme conditions within controlled lab environment DUT to be tested with systems not yet constructed G PHIL Simulation G Amplifier References and Feedback A B C Simulated V abc and I abc Interface Algorithm A B C Real Time Simulation Simulated DUT Control Signals DUT Controller Real Time Simulation DUT Interface Algorithm Simulated T and MRG MRG CHIL Simulation Dynamometer References and Feedback DISTRIBUTION STATEMENT A: Approved for public release. Distribution is unlimited. J. Langston, et.al., Role of Hardware-in-the-Loop (HIL) Simulation Testing in Transitioning New Technology to the Ship, in Proc. of IEEE Electric Ship Technologies Symposium (ESTS), April

4 Role of HIL Simulation throughout Technology Development In situ Modeling and Simulation dominates the entire process Relative Effort Limited Hardware Testing Power HIL HW only Lab Testing CHIL contributes heavily from proof of concept through PHIL testing De-risk early development of Hardware (fast) controller Application (slow) controller De-risking PHIL experiments Control HIL Modeling and Simulation PHIL supports model building and integration phases Experimental data for model construction and validation Stimulation of component through controlled transients Integration testing through emulation of the target environment(s) Proof of Concept Development and Model Building Integration Testing Time 4

5 FSU Center for Advanced Power Systems Established at Florida State University in 2000 under a grant from the Office of Naval Research Focusing on research and education related to application of new technologies to electric power systems Organized under FSU VP for Research Affiliated with FAMU FSU College of Engineering Lead Member of ONR Electric Ship R&D Consortium ESRDC $8 million annual research funding from ONR, DOE, Industry DOD cleared facility at Secret level Research Groups Electric Power Systems Advanced Modeling and Simulation Advanced Control Systems Power Electronics Integration and Controls Thermal management High Temperature Superconductivity Electrical Insulation/Dielectrics Staffing 50 Full time staff of scientists, engineers and technicians, post doc. s and supporting personnel 7 FAMU FSU College of Engineering faculty 45 Students Facility 44,000 square feet, laboratories and offices, located in Innovation Park, Tallahassee; Over $35 million specialized power and energy capabilities funded by ONR, DOE

6 Experimental Capabilities Integrated 5 MW Hardware-in-the-Loop (HIL) testbed 5 MW variable voltage / variable frequency converter: 4.16 kvac, 1.1 kvdc 5 MW dynamometers 225/450, 1,800/3,600; -12,000/24,000 RPM 5 MW MVDC converters: 6/12/24 kv Real-time Digital Simulators (RTDS, OPAL-RT) <2 μs step in real-time Cyber-physical system simulation in RT Superconductivity and crogenics AC Loss and Quench Stability Lab Cryo-dielectrics High Voltage Lab Cryo-cooled systems lab Low power and smart grid labs

7 FSU-CAPS PHIL Test Facility Fully Integrated with Real-Time Simulator Substation B1 B2 T1 7.5 MVA, 5% T6 C1 B15 SCC kv B5 S10 S4 B3 Future Future Feed Feed SP2 SP4 T7 C kv 4.16 kv B6 S5 B4 S8 B12 Future 4.16 kv Exp. Bus (Port) T9.1 T10.1 B13 B14 T9.2 C4 T10.2 DC Bus VDC 600VDC 1150VDC NEW 4 x MMC converters Delivered : Oct 28, 2013 SP 3.5 / 4.16 kv MVDC Experimental Bus Parallel: 6 kv, 0.8 ka, Series: 24 kv, 0.21 ka 0.8 MW PCM4 4.16kV / 1 kv (DC) M1 M2 5 MW Max PHIL Power Hardware in the Loop SS1 SS3 5 MW VVF AC Bus 4.16 kv Exp. Bus (Starboard) S 1.5 MVA 480 V bus

8 PHIL Challenges Accuracy, Stability, Protection Real-time simulation Fixed time-step with minimum achievable time-step size Limitations on the size and complexity of simulated systems Protection of experiment Amplifiers, Actuators Limited bandwidth Time delays Maximum power, torque, speed, etc. Availability of RT model Interface Algorithms Application specific Ensures stability of PHIL setup Device Under Test DUT Controller Amplifier Amplifier Controls Availability of DUT model for de-risking and tuning of protection Accuracy of models used for surrounding systems (rest-of-system - ROS) Common issue establishing confidence in the models PHIL Interface Flexible Protection of experiment PHIL Interface Controls Interface Component Rest of System Real Time Simulator 4/2013 8

9 De-risking: CHIL Simulation of 5 MW Amplifier Flexible Protection of experiment Device Under Test Amplifier Interface Component Rest of System DUT Controller Amplifier Controls PHIL Interface PHIL Interface Controls Real Time Simulator 06/26/2007 9

10 Simulated PHIL Experiment Device Under Test Amplifier Real Time Simulator DUT Controller Amplifier Controls Flexible Protection of experiment Interface Component Rest of System Hardware in Lab Device Under Test Amplifier PHIL Interface Controls Real Time Simulator DUT Controller Amplifier Controls Transition between modes for every change in the experimental setup 10

11 Model Verification and Validation V r Quantitatively Assess the predictive capability of models. Identify Surroundings Scenarios Observable Quantities Response Quantities In order to be of value, these must be carefully selected By standardizing for common classes of components, improve quality of results V src A B C L src I a I b I c Voltage V max V a V b V c A B C V r P N Active AC/DC Rectif ier T settle DUT V dc I dc R load Time DISTRIBUTION STATEMENT A: Approved for public release. Distribution is unlimited. J. Langston, et.al., Role of Hardware-in-the-Loop (HIL) Simulation Testing in Transitioning New Technology to the Ship, in Proc. of IEEE Electric Ship Technologies Symposium (ESTS), April

12 Synergy Between Verification and Validation and HIL Simulation HIL can facilitate economically carrying out validation experiments on the DUT Definitions of Surroundings, Scenarios, and Response Quantities HIL simulation test plans can be based around scenarios for V&V (including surrounding system models, scenarios, response quantities, etc.) Results can be used for improvement/calibration of DUT models and/or for assessment of prediction error with models Simulation Models of Suroundings Model Predictive HIL simulation experiments can be employed at various stages of the development process (using CHIL for testing controllers, etc.) Model V and V Capability Experimental Data HIL Simulation HIL community needs V&V guidelines and standards DISTRIBUTION STATEMENT A: Approved for public release. Distribution is unlimited. J. Langston, et.al., Role of Hardware-in-the-Loop (HIL) Simulation Testing in Transitioning New Technology to the Ship, in Proc. of IEEE Electric Ship Technologies Symposium (ESTS), April

13 Megawatt Scale High-Speed Generator RTDS Voltage/ Current/ Duty Cycle DAQ Trigger/Synch Voltage - + Open/Close Measured Quantities Voltage Current Speed Rectifier Speed/ Torque VVS 1.6 MW in 400 ms Actual Reference Generator Gearbox Moved from Model to CHIL to full-scale PHIL Offline models used nano-second time step Startup, shutdown procedure Steady-state and dynamic loading (ramping) Dyno Time (s) funded by 4/

14 Dynamic HIL Testing of Large Inverters Substation B1 B2 T1 B kV S10 Power Grid Simulation 6.3 MVA Variable Voltage Source (VVS) B13 T9.1 T9.2 VVS 1 VVS 2 Real Time Simulator RTDS 466/4160V 3.93MVA Z5.6% AC Bus1: kv I max ka T10.1 T5 B14 B kV AC Bus Real Time Simulator RTDS PV Array Simulation DC Bus: VDC I max +/- 2.5 ka PV Inverter 4160/480V 1.5MVA Z5.86% AC Bus2: kv I max 1.8 ka PHIL testing of MW-scale converters is possible today! Low voltage ride through Fault current contribution Unbalanced voltage Anti-islanding up to 1.5 MW funded by 4/

15 Inverter with 0.8 PF lagging Voltage (kv) Vsim Vm Vr I m Time reactive (min) current Active Power (MW), Reactive Power (MVAR) V required to drive through T5 Psim Pact Qsim Qact Time (min) See J. Langston, et al. Power Hardware-in-the-Loop Testing of a 500 kw Photovoltaic Array Inverter, in Proc. of IECON, Montreal, Canada, 2012 T5 B11 Vm 4 Z Vr Simulated grid impedance X < X T5 4/

16 PHIL testing of SiC converter 4.16 kv AC -1 kv DC B13 AC side DC side 4.16 kv 385 V T9.1 Simulates surrounding system (sources, loads) Provides ultra-fast protection T9.2 RTDS VVS 1 Voltage Ref (Vab (t), Vbc (t)) Voltage & Current Fdbk Voltage & Current Fdbk Current Ref (duty cycle) VVS V B14 T V 438 A max (cont.) 45 Hz (approx.) 100 Hz Small signal bandwidth limit approx. 1 khz Device under test funded by V 2.5 ka max 0 approx. 100 Hz Small signal bandwidth limit approx. 1 khz 16

17 Concluding Remarks PHIL testing is advancing rapidly A tool to address several challenges associated with transitioning technology (de-risking) Emulate a wide range of surroundings and scenarios, simulate yet unrealized systems Impact of PHIL interface more pronounced at MW scale experiments Aim for close coupling between reference and amplifier but acknowledge the limitations in the simulated PHIL setup Develop affordable faster switching amplifiers Improve real time simulation of models Proper model construction and validation is key to success Simulation based preparation of MW scale experiments expected to save time and money Improve development cycle Discover hidden issues early Need to develop common guidelines and standards to accelerate adoption of HIL paradigm Team at work in FSU- CAPS control room 500 kw PV converter in FSU-CAPS lab 4/

18

19 V&V: Quantification and Assessment of Model Predictive Capability Uncertainty Numerical Error P y y X Simulation y P y y x 2 o o o o x o Prediction y x 1 Err x In order to provide confidence bounds on predictions, combine model uncertainty numerical error predicted model form error into prediction at untested operating point DISTRIBUTION STATEMENT A: Approved for public release. Distribution is unlimited. J. Langston, et.al., Role of Hardware-in-the-Loop (HIL) Simulation Testing in Transitioning New Technology to the Ship, in Proc. of IEEE Electric Ship Technologies Symposium (ESTS), April

20 DC-side: Photovoltaic Emulation 1.6 Current (ka) Power (MW) Maximum power point Voltage (kv) Voltage (kv) Ensure that DC-amplifier controls allow PV-emulation in conjunction with PV inverter dynamics. 4/

21 Grid-side PHIL Interface Model/Simulation RTDS Filter PCC + Σ Filter Vmag sim PI Controller Controller Simulated Feeder VVS Voltage Magnitude Reference Choice: Voltage Current, but impacts stability Know your limits: Filters for bandwidth adjustment Equipment Hardware PV Inverter Device under Test Currents Voltage Vmag Id, Iq Trans- T5 former AC VVS Protect: Open loop operation through feedback gain adjustment 4/

22 Cyber-Physical RT Simulation RT simulation of Controls Communications Distributed Controls (DC) computing layer Fast data links between DC and power components: RT simulator specific RT simulation of electric power system