Real-Time Simulation of Predictive Control of DC Vehicular Microgrids Asal Zabetian-Hosseini and Ali Mehrizi-Sani
2 of 2 Outline Goal Fuel Cell and FC-PU Controller Design Simulation and Experimental Results Conclusion
3 of 2 Motivation Fuel cell technology has promise for the automotive industry due to its high efficiency, cleanliness, and sustainability. Hybridization of a fuel cell with a battery can be used to change load-dependent characteristics of the fuel cell: In a hybrid energy system, the energy management system (EMS) keeps the operating point of the system within permissible limits. The controllers minimize the effect of load changes on the load voltage. https://www.mnn.com/greentech/transportation/blogs/why-hydrogen-fuelcell-cars-taking-off https://gooeyrabinski.com/category/hydro gen-fuel-cell-cars/
4 of 2 Goal The goal of proposed predictive controllers in this study compared to conventional PI-controllers is to provide a control structure with Lower overshoot during transients; Faster transient response; and High robustness against parameter errors The performance of proposed predictive controllers is validated via real-time simulation to show the feasibility of the controller.
5 of 2 Fuel Cell and FC-PU We use a step-up converter to connect the fuel cell to the DC link and a bidirectional converter to connect the battery to the DC link. 2e H 2 à 2H 2e Anode Electrode Membrane H H 2H 2e½ O 2 à H 2 O Cathode Electrode (a) Topology of the FC-PU Utilized FC-PU Source: Asal Zabetian Hosseini, Younes Sangsefidi, and Ali Mehrizi-Sani, Model predictive control of fuel cell power units, in IEEE Ind. Electron. Soc. Annu. Conf. (IECON), Beijing, China, Nov. 217.
6 of 2 Controller Design (1D Bi )I Bi (1D FC )I FC Battery V batt I Bi R Bi L Bi C V DC I L (1D Bi )V DC 1D Bi D Bi I C D FC R FC L FC (1D FC )V DC I FC V FC Fuel Cell The design of the controllers is based on the average model of the system.
Controller Design Control structure has three control loops: 1. I FC tracks its reference value I FC 2. I Bi track its reference value I Bi 3. V DC tracks its reference value V DC by regulating I Bi V EOC V batt K i,batt s I FC I FC L FC f s,i R FC V FC V DC 1 1 V V FC DC 1 D FC 1 R FC L FC s Plant: Fuel cell converter Fuel cell current control loop I FC I L (1D FC )I FC V batt V DC 1 V I DC Cf C s,v I FC 1 V V batt DC 1 D Bi 1 I I Bi Bi L Bi f s,i R Bi L Bi s 1D Bi DC link voltage control loop R Bi Plant: Bidirectional converter Bidirectional current control loop Control structure. Source: Asal Zabetian Hosseini, Younes Sangsefidi, and Ali Mehrizi-Sani, Model predictive control of fuel cell power units, in IEEE Ind. Electron. Soc. Annu. Conf. (IECON), Beijing, China, Nov. 217. I L (1D FC )I FC I C 1 C s 1D Bi Plant: DC link V DC 7 of 2
8 of 2 Proposed MPC Structure Advantages of the proposed MPC structure Faster predictive controller even for the outer voltage control loop by using two separate sampling frequencies Lower overshoot during transients and load changes Robustness against parameter errors (up to 3%)
9 of 2 Proposed MPC Structure Current controllers of the FC-PU (controllers of I FC and I Bi ): The same formulation is used to calculate desired D bi to achieve I Bi in I Bi [k1]. V EOC V batt K i,batt s I FC I FC L FC f s,i R FC V FC V DC 1 1 V V FC DC 1 D FC 1 R FC L FC s Plant: Fuel cell converter Fuel cell current control loop I FC
1 of 2 Proposed MPC Structure Voltage controller of the FC-PU (controller of V DC ): I L (1D FC )I FC V batt V DC 1 V I DC Cf C I FC 1 V V batt DC 1 D Bi 1 I I Bi Bi s,v L Bi f s,i R Bi L Bi s 1D Bi DC link voltage control loop R Bi Plant: Bidirectional converter Bidirectional current control loop I L (1D FC )I FC 1D Bi I C 1 C s V DC
12 of 2 Real-Time Simulation Recorded data DC link (load) Battery Simulation file in RT-LAB Fuel Cell
13 of 2 Real-Time Simulation The performance of the proposed MPC structure is validated via real-time simulation case studies using OPAL-RT. The 3.46 GHz setup employs OP31(hardware), Redhat_v2.6.29.6(operating system), and RT- LAB1.7(software). Simulation sampling time is 2 µs. OPAL-RT used in LIPE. The system is 1 V, 3 kw. The inner control loop sampling frequency is 5 khz and the outer control loop sampling frequency is Hz.
14 of 2 Challenges Limitation in simulation time step of OPAL-RT in real-time mode Limitation in the size of the recorded data by RT-LAB Limitation of the switching frequency of converters and the sampling frequency of the inner control loops Limitation in the number of recorded signals Increasing the fluctuation and overshoot in currents.
15 of 2 Case Studies The performance of the control structure is investigated in three case studies: 1) Response of the fuel cell current controller to a step change in reference current value 2) Response of the bidirectional current controller to a step change in reference current value 3) Response of the DC link voltage controller to a step change in refence voltage value Conventional and proposed control structures of FC-PU are simulated in MATLAB/Simulink. Then, the performance of the proposed MPC structure is also validated via real-time simulation case studies using OPAL-RT.
16 of 2 IFC (A) 3 2 1 1 Case 1:Performance of the fuel cell current controller 2 1.19.21.23 Result Referance 2 1 1 2 1.2.22 1 Result Referance 2 1 1 2 1.19.2.21 Result Referance DFC.5.5.5.2.4.6 Time (s) (a).2.4.6 Time (s) (b) Simulation results of I FC and D FC responses to a step change in I FC for the (a) conventional, (b) proposed current controllers, and (c) real-time simulation of proposed controller. PI controller MPC controller Real-time Settling time 2 ms.5 ms.5 ms Overshoot 6.7 A 1.5 A 2.5 A.2.4.6 Time (s) (c)
Case 2: Performance of the bidirectional current controller (battery) IBi (A) 3 1 3.2.22 Result Referance 3 1 3.198.2.22 Result Referance 3 1 3.199.2.21 Result Referance DBi.5.5.5.2.4.6 Time (s) (a).2.4.6 Time (s) (b) Simulation results of I Bi and D Bi responses to a step change in I Bi for the (a) conventional, (b) proposed current controllers, and (c) real-time simulation of proposed controller. PI controller MPC controller Real-time Settling time 25 ms.4 ms.5 ms Overshoot 3.8 A 3.1 A 4.9 A.2.4.6 Time (s) (c) 17 of 2
VDC (V) IBi (A) 1 1 - Case 3: Performance of the DC link voltage controller (tracking the reference value). Result Referance.2.4.6 Time (s) (a) 1 1 Simulation results of V DC and I Bi responses to a step change in V DC for the (a) conventional, (b) proposed voltage controllers, and (c) real-time simulation of proposed voltage controller. PI controller MPC controller Real-time Settling time 75 ms 16 ms 16.5 ms Overshoot 5.4 A.8 A.9 A Result Referance -.2.4.6 Time (s) (b) 1 1 - Result Referance.2.4.6 Time (s) (c) 18 of 2
19 of 2 Conclusion Settling time and overshoot are decreased significantly in proposed MPC structure compared with the conventional PI-based controller. Power electronic converters and controllers are modeled in real-time. Offline and real-time match closely. Real-time simulation validates the performance of the MPC controllers and DC microgrid in real-time.
2 of 2 THANK YOU Real-Time Simulation of Predictive Control of DC Vehicular Microgrids Asal Zabetian-Hosseini and Ali Mehrizi-Sani