Control Strategies for Microgrids Ali Mehrizi-Sani Assistant Professor School of Electrical Engineering and Computer Science Washington State University Graz University of Technology Thursday, November 14, 2013 Graz, Austria
Outline Current and Envisioned Status of the Power System Overview of control requirements Proposed Control Strategy Online set point modulation Results of Evaluation of the Strategy Offline simulation Real-time implementation Fine tuning the parameters of the strategy Applications 2 of 43
Motivation One of the U.S. grand energy challenges is to enable integration of at least 80% renewable energy resources at a competitive cost in the power grid by 2050. While it is technically feasible to run the U.S. economy on renewable technologies available today, what is missing is a flexible power system that accommodates the unique characteristics of renewable resources: Intermittency Lack of inertia Susceptibility to violation of operational limits This work addresses the latter susceptibility to violation of limits. 3 of 43
Global Need The need to address this challenge is confirmed by Department of Energy 2012 microgrid workshop (and 2011/2010) 2013 White House 21st century grid report (and 2012) National Academy of Engineering (2013 grand challenges) Our proposed strategy addresses this challenge: It empowers controllers to closely track their set points even when the host system changes significantly. The existing work does not address this gap: The common designs assume the host system does not experience significant changes. Significance of this work is that it reduces the need for overdesign and subsequently increases asset utilization. 4 of 43
Goal Our goal is to significantly improve the performance of controllers in a system that Is time varying Has limited reserve A prominent example of such systems is a microgrid. Enabling concept for the modern and smart power system as a building block. 5 of 43
Microgrids Definition An aggregate of collocated resources (loads, generation units, and storage units) that are interfaced to the main grid at the distribution level and is capable of operating in the gridconnected mode, islanded mode, and the transition between these two modes. DG DG Primary Controller Grid DG Primary Controller Primary Controller Load Load Load Microgrid 6 of 43
Microgrid Challenges (1/3) Microgrids offer scalability, modularity, and security, but they may experience Frequent changes in the topology; Units susceptible to overcurrents and overvoltages; Operation close to the limits to increase asset utilization; and Limited total capacity. Therefore, changes in the microgrid may have a detrimental effect on the performance of controllers. Controllers are designed for a prespecified configuration. It is imperative to ensure controllers retain their tracking capability under various operating conditions, including those very different from the original design. 7 of 43
Microgrid Challenges (2/3) Existing control design approaches include Model-based automated tuning (Astrom s work) Optimization-based (Gole s work) However, these approaches Require access to updated system models; Need availability of a computational infrastructure to allow redesign; Have limited robustness to topology, operating point, and system parameters; Are difficult to retune; or Are intrusive. 8 of 43
Microgrid Challenges (3/3) Example Effect of large load change on controller performance. Original System Overshoot: 15% Settling time: 32 ms Load Disconnected Overshoot: 26% Settling time: 67 ms 9 of 43
Objective Our objective is to design stringent control strategies that offer close set point tracking while being Robust to topological and operational changes; Independent of the system model; and Operating with little information about the unit to which it is associated. 10 of 43
11 of 43 Shaping of the Response Trajectory Consideration of Dynamic Limits of Devices V (pu) 1.4 1.2 1.0 0.8 No Interruption Prohibited Region No Damage x2 dynamics steadystate 1 ms 3 ms Challenges 20 ms 0.5 s Avoid violating dynamic limits With a small overshoot Achieve a fast response Without changing the existing controller 10 s Δt (a) Δt = 0.5 s Δt = 0 0.7 0.8 0.9 1 1.1 1.2 V (pu) (b) x1
12 of 43 Proposed Solution Improving the response by temporarily manipulating the set point Secondary Controller xsetpoint xsetpoint Set Point Modulation Primary Controller Unit x(t)
Set Point Modulation Best Strategy Choose T 1 so that the peak of the response equals the reference Choose T 2 to be the time of this peak Not Implementable Faster-than-real-time simulator Closed-form solution System parameters x(t) 0 T1 tp T2 T1 tp t 13 of 43
14 of 43 Finite-State Machine SPAACE /speɪs/: Set Point Automatic Adjustment with Correction Enabled xpred > xmax Δt < Tmax Salient Features: Based on local signals Independent of model S000 Robust to changes in parameters Independent of time scale x(t) > xmax Δt > Tmax S101 violation wait w4 w1 Δt < Tmax S001 violation xpred > xmax S110 S010 Δt > Tmax wait w3 x(t) > xmax S111 violation w2 xpred < xmax wait w5 x(t) < xmax wait xpred < xmax x(t) < xmax S100
15 of 43 Case Study I: Set Point Change 800 816 DG1 824 858 832 848 846 844 842 828 830 854 852 DG2 834 860 836 840 888 890 IEEE 34-Bus System Added 3 DG units and a load Operates in grid-connected mode DG3 I (pu) I (pu) 1.1 1 0.9 1.1 1 0.9 0 1 2 3 4 5 (a) 0 1 2 3 4 5 (b) Time (ms) System Response DG2 step change from 0.91 pu to 1.09 pu DG1 and DG3 unchanged (40% overshoot)
16 of 43 Case Study II: Simultaneous Change 848 846 I (pu) 1.1 1 0.9 0 1 2 3 4 5 (a) 800 816 DG1 844 DG2 842 824 858 834 860 836 840 832 888 890 828 830 854 852 DG3 I (pu) I (pu) 1.1 1 0.9 1.1 1 0.9 0 1 2 3 4 5 (b) 0 1 2 3 4 5 (c) Time (ms) System Response Simultaneous step change
17 of 43 Case Study III: Load Disconnection 848 846 800 816 DG1 844 DG2 842 824 858 834 860 836 840 832 888 890 828 830 854 852 DG3 I (pu) 1.1 1 0.9 with SPAACE 0 1 2 3 4 5 6 7 8 9 10 Time (ms) IEEE 34-Bus System Added 3 DG units and a load Operates in grid-connected mode System Response Resistive 0.5 pu load change (15% overshoot)
18 of 43 Case Study IV: Unbalanced System 650 646 645 632 633 634 V (pu) 1.5 1 611 684 671 692 675 652 680 Test load DG1 0.5 0 0.2 0.4 0.6 0.8 1 Time (s) IEEE 13-Bus Unbalanced System Added a DG unit and a test load Operates in islanded mode System Response Resistive 1 pu load switched off Unstable system to stable system
19 of 43 Metric A Metric to Assess Improvement in Tracking
20 of 43 HVDC Study System 12 pulse, 500 kv, 1000 MW 345 kv 50 Hz 230 kv 50 Hz CIGRE HVDC Monopolar First Benchmark System Rectifier is current controlled Inverter is gamma controlled
21 of 43 Case I: Rectifier Current Step (0 to 0.55 pu) I (pu) 0.8 0.6 0.4 0.2 0 0 0.02 0.04 0.06 0.08 0.1 (a) I (pu) 0.8 0.6 0.4 0.2 0 0 0.02 0.04 0.06 0.08 0.1 (b) Time (s) Peak (pu) Overshoot Error (S e ) Settling (ms) No SPAACE 0.878 59.6% 1.0 95 With SPAACE 0.708 28.7% 0.56 (Δ=44%) 75 (Δ=26.3%)
22 of 43 Case II: Faulted (I-Side) DC Current, 50 ms I (pu) I (pu) 1.8 1.4 1 0.4 0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 (a) 1.8 1.4 1 0.4 0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 (b) Peak (pu) Error (S e ) Settling (ms) Time (s) No SPAACE 1.744 1.0 800 With SPAACE 1.540 0.45 (Δ=55%) 260 (Δ=67.5%)
23 of 43 Prediction Methods DER 2 1.25 1.2 12 13 14 DER 1 1.15 1.1 3 8 7 6 V (pu) 1.05 1 0.95 0.9 0.85 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 Time (s) No Prediction G1 1 2 11 5 DER 3 10 9 V (pu) 1.25 1.2 1.15 1.1 1.05 1 0.95 0. 9 0.85 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 Time (s) Linear Prediction V (pu) 1.25 1.2 1.15 1.1 1.05 1 0.95 0.9 0.85 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 Time (s) Quadratic Prediction
Prediction Algorithms: Step Change Linear Prediction Quadratic Prediction No SPAACE 0.8 I (pu) 0.6 0.4 0.2 0 Quadratic Pred Linear Pred 0 0.02 0.04 0.06 0.08 0.1 Time (s) Peak (pu) Overshoot Error (S e ) Settling (ms) No SPAACE 0.878 59.6% 1.0 95 With SPAACE (L) 0.708 28.7% 0.560 (Δ=40%) 75 (Δ=26.3%) With SPAACE (Q) 0.777 41.2% 0.668 (Δ=33%) 55 (Δ=42%) 24 of 43
25 of 43 Prediction Algorithms: Fault 1.8 No SPAACE 1.4 I (pu) 1 0.4 0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 Linear Pred Quadratic Pred Peak (pu) Error (S e ) Settling (ms) No SPAACE 1.744 1.0 800 With SPAACE (L) 1.540 0.45 (Δ=55%) 260 (Δ=67.5%) With SPAACE (Q) 1.206 0.21 (Δ=79%) 500 (Δ=37.5%)
Effect of Scaling Factor m Adaptive Nature of SPAACE x(t) u2 u(t) Response without SPAACE u1 T1 T2 Response with SPAACE (1-m)u2 t 26 of 43
27 of 43 Scaling Factor m 0.15 0.15 iq (pu) 0.10 0.05 0.15 0 0 0.02 0.04 0.06 0.08 0.10 0.12 0.14 Time (s) m=0.00 iq (pu) 0.10 0.05 0.15 0 0 0.02 0.04 0.06 0.08 0.10 0.12 0.14 Time (s) m=0.20 iq (pu) 0.10 0.05 iq (pu) 0.10 0.05 0 0 0.02 0.04 0.06 0.08 0.10 0.12 0.14 Time (s) m=0.40 m Peak (A) Error (S e ) Settling (ms) 0 0.149 1.0 80 0.20 0.136 0.806 60 0.55 0.125 0.727 55 0 0 0.02 0.04 0.06 0.08 0.10 0.12 0.14 Time (s) m=0.60 G1 12 DER 3 13 14 3 DER 2 1 2 11 5 8 10 7 9 DER 1 6
28 of 43 Upper Bound of m m 0.5 0.4 0.3 0.2 0.1 0 0 0.2 0.4 0.6 0.8 1 ζ
29 of 43 Physical Analogy x2 x1 x(t) B k B k m u(t) = F + f (t) m F + f (t)
30 of 43 Alternative Methods to SPAACE Model- Free Non- Intrusive No Access to Controller Comments SCALING PI X Limited to performance of the original design RAMP Unnecessary intervention, DC tracking MPC X X Computationally intensive PID X X D as linear extrapolation ES / IFL X X Sinusoidal perturbation o input POSICAST X Essentially open-loop, 2nd-order SPAACE
SPAA If a priori knowledge of overshoot is available SPAA /spɑː/: Set point automatic adjustment x(t) x2 x(t) x2 (1) x(t) with SPAA (2) x2 x1 Time 31 of 43
SPAA Case Study 848 Start-Up Current Control IEEE 34-bus system with 3 DERs DER1 and 3: i d = 1.0 pu, i q = 0 800 816 DG1 846 844 DG2 842 824 858 834 860 836 840 832 888 890 828 830 854 852 DG3 DER2: off to i d = 1.08 pu SPAA assumes ζ=0.361 and ω=8450 rad/s 33 of 43
34 of 43 SPAA vs. SPAACE SPAA SPAACE RATE OF UPDATE After steady state Continuously NEED TO MODEL Yes (approximate) No EFFECTIVENESS Large changes Moderate changes APPROACH Open loop Closed loop RESPONSIVENESS Set point change Any difference in the set point and response (set point change, load switching, faults)
35 of 43 Experimental Implementation RSCAD DC Power Supply NI crio: SPAACE Algorithm NI cdaq: Collect Data Measurement and Control Signals Real-Time Digital Simulator
Experimental Implementation NI cdaq: Data Acquisition measure x(t), u(t), u*(t) x(t), u(t), u*(t) NI crio: SPAACE Algorithm u*(t) Real-Time Digital Simulator (RTDS): Power system Data Monitoring Laptop: RSCAD and LabVIEW status monitoring x(t), u(t) Interactive Execution Data, Data Monitoring x(t): output signal, u(t): set point for output, u*(t): adjusted set point. 36 of 43
37 of 43 3 4 5 2 1 Laptop running RSCAD and LabVIEW 2 Real-Time Digital Simulator 3 DC Power Source 4 NI Compact RIO Controller 5 NI cdaq (Data Acquisition) 1
Test System 38 of 43
39 of 43 Case I: Load Energization (1.2 pu) Without SPAACE V (pu) 1.20 1.15 1.10 1.05 1.00 0.95 0.90 1.78 1.80 1.82 1.84 1.86 1.88 1.90 1.92 Time (s) With SPAACE V (pu) 1.20 1.15 1.10 1.05 1.00 0.95 1.78 1.80 1.82 1.84 1.86 1.88 1.90 1.92 Time (s)
Case II: Step Change in i q Without SPAACE iq (pu) 1.2 1.0 0.8 0.6 0.4 2.65 2.70 2.75 2.80 2.85 2.90 2.95 Time (s) With SPAACE iq (pu) 1.2 1.0 0.8 0.6 0.4 2.65 2.70 2.75 2.80 Time (s) 2.85 2.90 2.95 40 of 43
Conclusions By appropriately designing the trajectory to reduce overshoots, it is possible and safe for a system to operate closer to its limits. Offline (PSCAD) and real-time (RTDS) simulation studies show that SPAACE is effective in mitigating transients: Step change: Mitigating overshoots (37%) Fault: Closer set point following Load energization: Eliminating a peak of 1.15 pu Load disconnection in a unbalanced system: Stabilizing oscillatory behavior of voltage 41 of 43
Applications Systems with Limited Resources Transients may exceed the capacity of the system DER n PCn Microgrid PCC PC3 PC1 PC2 S Main Grid DER 3 DER 2 DER 1 Large AC/DC Systems Segmented Power Systems Emerging Small AC Systems All-Electric Ships and Military Systems Emerging Application: HSIL transmission lines 42 of 43
Thank You Control Strategies for Microgrids Ali Mehrizi-Sani mehrizi@eecs.wsu.edu http://eecs.wsu.edu/~mehrizi 43 of 43