Grid + Storage Workshop 9 th February 2016, Athens Developing tools to increase RES penetration in smart grids Grigoris Papagiannis Professor, Director Power Systems Laboratory School of Electrical & Computer Engineering
Outline 1. Challenges on LV Distribution Networks 2. Description of Proposed Control Schemes a. Local Control b. Overlaying Control c. Scheduling Control 3. Evaluation of the Proposed Controls 4. An Integrated Simulation Platform 2
LV Distribution Networks Conventional LV Networks Designed based on a unidirectional power flow scheme from utility to end-users Operational Objective Maintain voltages within permissible levels in accordance to EN 50160 (U Ν ±10 %) Compensation of the voltage drop along the power line Stationary adjustment of the transformer ratio (on-load?) Source: Dr. Bernhard Ernst, EPIA-EDSO for Smart Grids Conference on Grid Management, 24 June 2014, Brussels, Belgium 3
LV Distribution Networks Advent of Distributed Generation (DG) DG units are located close to loads Major Challenges Transition from a unidirectional to a bidirectional power flow scheme Overvoltage and possible violation of EN 50160 Limited DG penetration Source: Dr. Bernhard Ernst, EPIA-EDSO for Smart Grids Conference on Grid Management, 24 June 2014, Brussels, Belgium 4
LV Distribution Networks Other Technical Issues DG not uniformly distributed among phases Considerable voltage unbalances Congestion issues on power lines and/or on transformers Traditionally not monitored and/or controlled Source: Dr. Bernhard Ernst, EPIA-EDSO for Smart Grids Conference on Grid Management, 24 June 2014, Brussels, Belgium 5
LV Distribution Networks Possible Solutions (1/2): Reinforcing the distribution feeders or implementing electric energy storage Reduces network losses Increases feeder capacity Needs investments Use of OLTCs and/or voltage regulators Active voltage regulation No interference with DG units operation Needs investments Reliability issues Operational limitations 6
LV Distribution Networks Possible Solutions (2/2): Employing reactive power control techniques Avoids active power curtailment Limited by inverter rated power Inefficient for LV networks with high R/X ratio Increases network losses Introducing active power curtailment Occurs only at high generation periods Efficient for LV networks Increases DG penetration Revenue loss for DG units owners 7
Proposed Control Schemes Conceptual Design Optimized network operation Scheduling Control (intra- hour day, etc.) Active management of LV distribution networks Overlaying Control (min) Mitigation of overvoltages and of voltage unbalances Local Control (ms) Outcome Increase the penetration of DG units in existing networks Enhance the observability and controllability of LV networks 8
Outline 1. Challenges on LV Distribution Networks 2. Description of Proposed Control Schemes a. Local Control b. Overlaying Control c. Scheduling Control 3. Evaluation of the Proposed Controls 4. The Integrated Simulation Platform 9
MV POWER SYSTEMS LABORATORY Local Control Overview Objective Real-time mitigation of overvoltages and of voltage unbalances Low-level control scheme Main Ascpects Immediate reaction on changes of grid operational state Enhancing the smooth/safe operation of the distribution grid Stand-alone operation, use of local parameters (e.g. voltage at the PCC) MV/LV 1 2 CM ` ` n VM Load Load ~ = AG Load ~ = AG CM: Current Measurement VM: Voltage Measurement AG: Agent DG DG 10
MV POWER SYSTEMS LABORATORY Local Control Overview Actual Implementation Applied to DG units via developed controllable grid-interfaced inverters Two basic control schemes are incorporated The droop control of the injected real power The voltage unbalance mitigation control avoid ON/OFF oscillations MV/LV 1 2 CM ` ` n VM Load Load ~ = AG Load ~ = AG CM: Current Measurement VM: Voltage Measurement AG: Agent DG DG 11
Local Control Overview An implementation example MV (1) LV (2) 3 4 5 6 7 8 11 13 12 9 10 56 57 61 63 62 79 1 2 3 4 5 6 7 14 30 26 29 31 58 64 16 17 18 19 15 27 28 37 38 39 41 36 32 40 33 34 35 59 60 66 67 65 68 70 69 Load 1 ~ = Load 2 Load 3 ~ = Load 4 ~ = Load 5 20 21 42 43 45 44 74 75 71 72 DG 1 DG 2 DG 3 22 46 76 73 23 48 50 47 77 24 49 51 78 25 52 53 54 55 Examined Control Schemes No control (NC) Voltage unbalance mitigation control (VUMC) Droop control (DC) Combined operation of DC and VUMC (DC & VUMC) 12
Local Control Overview An implementation example No control (NC) Unacceptable overvoltages along the feeder 3 rd DG is switched off Loss of power Positive-sequence Voltage (pu) 1.12 1.11 1.1 1.09 1.08 1.07 1.06 1.05 2 3 NC VUMC Upper threshold (EN 50160) 4 5 6 7 Nodes 13
Local Control Overview An implementation example Phase-to-Neutral Voltages (pu) 1.2 1.15 1.1 VAN NC VBN NC VCN NC VAN VUMC VBN VUMC VCN VUMC 1.05 2 3 4 5 6 7 Nodes 14
Local Control Overview An implementation example Voltage unbalance mitigation control (VUMC) Mitigation of voltage asymmetries Unacceptable overvoltages (at nodes 6 & 7) Disconnection of DG units Loss of power Positive-sequence Voltage (pu) 1.12 1.11 1.1 1.09 1.08 1.07 1.06 1.05 2 3 NC VUMC Upper threshold (EN 50160) 4 5 6 7 Nodes 15
Local Control Overview An implementation example Introduction of Droop Control LV grid distribution lines have a high R/X ratio Efficient voltage control can be achieved by controlling the injected real power of the DG units P inj P ref ʋ min ʋ cpb ʋ max ʋ g Outcomes Unacceptable overvoltages along the feeders can be avoided by curtailing part of the injected real power of the DGs DG unit switch-offs due to overvoltages are avoided 16
Local Control Overview An implementation example Droop control The voltage profile along the feeder is actively controlled Overvoltages are efficiently mitigated No DG unit is switched off Positive-sequence Voltage (pu) 1.12 1.11 1.1 1.09 1.08 1.07 1.06 1.05 2 3 NC DC Upper threshold (EN 50160) 4 5 6 7 Nodes 17
Local Control Overview An implementation example Combined operation of DC and VUMC NC and VUMC control overlap DC and the combination of DC with VUMC control schemes overlap The proposed control reduces overvoltages Positive-sequence Voltage (pu) NC DC VUMC DC & VUMC Upper threshold (EN 50160) 1.12 1.11 1.1 1.09 1.08 1.07 1.06 1.05 2 3 4 Nodes 5 6 7 18
Local Control Overview A field trial example 100 80 Current Magnitude (A) 60 40 20 0 IA IB IC IN 19
Local Control Overview Synopsis Droop Control (DC) Overvoltages are reduced Voltage unbalances are not efficiently mitigated Voltage Unbalance Mitigation Control (VUMC) Voltage asymmetries are tackled It cannot control overvoltages Combined operation of DC and VUMC Unacceptable overvoltages are avoided Higher penetration of DG units is feasible Voltage unbalances are mitigated 20
Outline 1. Challenges on LV Distribution Networks 2. Description of Proposed Control Schemes a. Local Control b. Overlaying Control c. Scheduling Control 3. Evaluation of the Proposed Controls 4. An Integrated Simulation Platform 21
Overlaying Control Overview Uniform active power curtailment among DG units of the same feeder Objectives Use of OLTC to actively control network voltages Congestion management Centralized and middle-level control scheme Main Aspects Moderate response (15-minutes) Increasing the energy capture from DG units Enhancing the safe operation of the distribution grid 22
Overlaying Control Overview Proposed Algorithms Fair power sharing control (FPS) Uniform active power curtailment Use of the sensitivity matrix OLTC control Maintain network voltages within limits Reduce the curtailed power of DG units Congestion management Transformer protection from overloading Active power curtailment of DG units 23
Overlaying Control Overview Conceptual design 15 minutes timeslot First 5 minutes: Only Local Control Remaining 10 minutes: Activation of Overlaying Control OLTC OLTC OLTC t t+15 min t+30 min t+45 min Local Control FPS & Congestion FPS & Congestion FPS & Congestion Detection of Voltage Issue Calculate New Tap Setting t+15 min t+30 min Reset Load, MPP Local Control FPS & Congestion Check for OLTC activation OLTC 24
Overlaying Control Overview Distributed Multi-Agent System (MAS) Hierarchical Architecture (Agent-Aggregator) Actual implementation An Agent in each controllable DG unit Aggregator is usually sited at the DSO Wired or Wireless communication 25
Overlaying Control Overview An implementation example 3 4 5 6 7 8 11 13 12 MV LV (1) (2) 9 10 56 57 61 63 62 79 Network characteristics 79 buses 70 unbalanced time series loads 30 PV installations 14 30 26 29 31 58 64 16 17 18 19 20 21 22 23 24 25 15 Nodes Current PV nodes New PV nodes 27 28 37 38 39 41 42 43 45 46 48 50 49 51 52 53 54 55 36 32 40 33 44 47 34 35 59 60 66 67 65 68 70 69 74 75 76 77 78 71 72 73 Examined Control Schemes No control (NC) Local Control (Local) Local & FPS (LF) OLTC & Local (OL) OL & FPS (OLF) OLF & Congestion (OLFC) 26
Overlaying Control Overview An implementation example Comparative results along a feeder (1/2) No Control (NC) No active power curtailment Severe overvoltages at the last nodes Voltage Magnitude (pu) 1.15 1.1 1.05 1 NC Local LF OL OLF Voltage threshold Local Control (Local) Efficient overvoltage mitigation Non-uniform active power curtailment along the feeder Local and FPS Control (LF) Uniform active power curtailment Increased overall active power curtailment compared to Local Control Active Power (%) 0.95 MV LV 11 13 14 18 19 20 Nodes 21 22 23 24 25 110 NC Local LF OL OLF 100 90 80 70 60 50 40 PV 13 PV 14 PV 18 PV 20 PV 22 PV 24 PV 25 PV units 27
Overlaying Control Overview An implementation example Comparative results along a feeder (2/2) OLTC and Local (OL) Less active power curtailment compared to the Local Control Effective overvoltage mitigation Non-uniform active power curtailment among DG units OLTC, Local and FPS (OLF) Effective overvoltage mitigation Uniform active power curtailment Increasing the overall injected active power compared to the LF scheme Voltage Magnitude (pu) Active Power (%) 1.15 1.1 1.05 1 NC Local LF OL OLF Voltage threshold 0.95 MV LV 11 13 14 18 19 20 Nodes 21 22 23 24 25 110 NC Local LF OL OLF 100 90 80 70 60 50 40 PV 13 PV 14 PV 18 PV 20 PV 22 PV 24 PV 25 PV units 28
Overlaying Control Overview An implementation example Time series simulation (1/5) 15 minutes timeslot concept Execution time: less than 2 minutes for a time period of 24 hours Considering the different control schemes, similar conclusions can be drawn about the total injected power For most of the time, OL and OLF overlap 29 Total Injected Active Power (kw) Total Injected Active Power (kw) 300 250 200 150 100 50 Local LF OL OLF 0 0 4 8 12 Time (h) 16 20 24 300 Local LF OL OLF 250 200 150 100 50 10 12 14 16 Time (h)
Overlaying Control Overview An implementation example Total Curtailed Active Power (kw) Network Losses (kw) 100 80 60 40 20 Local LF OL OLF 0 10 12 Time (h) 14 16 12 Local LF OL OLF 10 8 6 4 2 0 10 12 14 16 Time (h) Time series simulation (2/5) The OL control scheme minimizes the curtailed power 30 However, the OLF applies a more uniform active power curtailment among the DG units of the same feeder Superior performance of the OLF in comparison with the other control schemes Network losses proportional to the generation due to reverse power flow Network voltages are actively controlled
Overlaying Control Overview An implementation example POWER SYSTEMS LABORATORY Time series simulation (3/5) b) Local 1.15 1.1 1.1 Voltage (pu) Voltage (pu) a) NC 1.15 1.05 1 0.95 0 4 8 12 Time (h) 16 20 1.05 1 0.95 0 24 4 8 1.1 1.1 Voltage (pu) 1.15 1.05 1 20 24 16 20 24 1.05 1 0.95 4 8 12 Time (h) 16 20 24 0 4 8 16 20 24 12 Time (h) e) OLF 1.15 Voltage (pu) Voltage (pu) 16 d) OL c) LF 1.15 0.95 0 12 Time (h) 1.1 1.05 1 0.95 0 4 8 12 Time (h) 31
Overlaying Control Overview An implementation example Time series simulation (4/5) 2 Only two tap changes are performed during the day OLTC is not stressed 1600 Tap Position 1 0-1 -2 OL OLF 0 4 8 12 Time (h) 16 20 24 Total Energy Production(kWh) 1400 1200 1000 800 The energy capture from DG units is increased due to the OL and OLF control schemes 600 Local LF OL OLF Control Schemes 32
Overlaying Control Overview An implementation example Time series simulation (5/5) - An Extreme Case The installed capacity of PV units is increased to cause congestion at the transformer OL and OLF control schemes are not able to tackle the excessive power flow OLFC relieves the stress from the transformer Results in decreased produced energy compared to OL and OLF schemes Energy Production (kwh) Apparent Power (kva) 300 250 200 150 100 50 0 10 12 OL OLF OLFC Transformer Limit 14 16 Time (h) 2100 2050 2000 OL OLF OLFC Control Scheme 33
Overlaying Control Overview Synopsis (1/2) Local Control Efficiently mitigate overvoltages Unequal active power curtailment Local & FPS Control Redistributes uniformly the curtailed power to the DG units Overall curtailment is increased OLTC & Local Control Results in less active power curtailment compared to Local Control 34
Overlaying Control Overview Synopsis (2/2) OLTC & Local & FPS Control Presents a superior performance compared to LF control scheme, concerning active power curtailment OLTC & Local & FPS & Congestion Control Enhanced OLF control scheme Able to handle transformer congestion issues 35
Outline 1. Challenges on LV Distribution Networks 2. Description of Proposed Control Schemes a. Local Control b. Overlaying Control c. Scheduling Control 3. Evaluation of the Proposed Controls 4. An Integrated Simulation Platform 36
Scheduling Control Overview Objectives Minimize the number of undervoltage/overvoltage violations and current congestion issues by employing demand response (DR) techniques Maximize the profit of DR units Centralized and high-level control scheme Main Aspects Maximizing the green energy injection Slow response (intra- hour, day, day-ahead, etc.) Optimizing the operation of the distribution grid 37
Scheduling Control Overview Conceptual design Top hierarchical layer Use of forecast tools and price signals from wholesale electricity market Responsible for the coordination of the lower level control layers (Local and Overlaying Control) Introduction of a traffic light system approach 4 time layers: Tertiary reserve market (t > 168 h) Day-ahead market (t >24 h) Intraday market (1 < t < 24 h) Balancing market (15min < t < 1 h) 38
Scheduling Control Overview Actual implementation Distributed Multi-Agent System (MAS) Hierarchical Architecture (Agent- Aggregator) Development of Scheduling Control Agent (SCA) aggregator Responsible for the dispatch of DR units Cooperation with the Overlaying Control Agent (OCA) aggregator 39
Scheduling Control Overview Synopsis Optimization criterion includes the maximum profit (including income from the markets, non-performance penalties for DR units) Scheduling Control Results in optimal use of flexible energy available from Demand Response Maximizing the DRES operation (minimal curtailment) No power quality violations Pricing is tied to day-ahead, intraday and balancing markets Helps maximizing the effectiveness of Overlaying Control 40
Outline 1. Challenges on LV Distribution Networks 2. Description of Proposed Control Schemes a. Local Control b. Overlaying Control c. Scheduling Control 3. Evaluation of the Proposed Controls 4. An Integrated Simulation Platform 41
Evaluation of the Proposed Controls Nodes 3 4 Line 3 5 Line 4 6 Line 5 7 Line 6 8 Line 7 9 Line 8 10 Line 9 11 Line 10 12 Line 11 13 Line 1 Line 2 MV LV (1) (2) Business as Usual (BAU) Case 5 PVs No Control is applied Voltage rise and thermal limit of lines are considered Typical yearly load and generation profiles are used Considering minimum load the maximum permissible rated power of the PV units is determined Voltage rise is the limiting factor in this case BAU PV Nodes New PV Nodes 42
Evaluation of the Proposed Controls MV (1) LV (2) Nodes 3 4 Line 3 5 Line 4 6 Line 5 7 Line 6 8 Line 7 9 Line 8 10 Line 9 11 Line 10 12 Line 11 13 Line 1 Line 2 PV name Nominal Power (kw) PV 5 11,65 PV 7 9,31 PV 9 13,98 PV 11 18,65 PV 13 16,3 Total 69,89 BAU PV Nodes New PV Nodes 43
Evaluation of the Proposed Controls Nodes BAU PV Nodes 3 4 Line 3 5 Line 4 6 Line 5 7 Line 6 8 Line 7 9 Line 8 10 Line 9 11 Line 10 12 Line 11 13 Line 1 Line 2 MV LV (1) (2) Local and FPS Controls BAU PVs are still connected 4 new PVs are installed Local and FPS Control strategies are applied to the PV units. Three different implementations I. V cpb =1.06 pu, for all PVs II. V cpb =1.08 pu, for all PVs III. V cpb =1.08 pu, for old PVs, while for new PVs V cpb is considered equal to 1.06 pu (mixed case) Maximum permissible rated power of the PV units is determined In this case, line thermal limit is the limiting factor New PV Nodes 44
Evaluation of the Proposed Controls MV LV (1) (2) Local Control with V cpb =1.06 pu Nodes 3 4 Line 3 5 Line 4 6 Line 5 7 Line 6 8 Line 7 9 Line 8 10 Line 9 11 Line 10 12 Line 11 13 Line 1 Line 2 PV name Nominal Power (kw) PV 5 11,65 PV 6 32,97 PV 7 9,31 PV 8 57,7 PV 9 13,98 PV 10 65,95 PV 11 18,65 PV 12 49,46 PV 13 16,3 Total 275,97 BAU PV Nodes New PV Nodes 45
Evaluation of the Proposed Controls MV LV (1) (2) Local Control with V cpb =1.08 pu Nodes 3 4 Line 3 5 Line 4 6 Line 5 7 Line 6 8 Line 7 9 Line 8 10 Line 9 11 Line 10 12 Line 11 13 Line 1 Line 2 PV name Nominal Power (kw) PV 5 11,65 PV 6 20,87 PV 7 9,31 PV 8 36,21 PV 9 13,98 PV 10 41,73 PV 11 18,65 PV 12 31,3 PV 13 16,3 Total 200 BAU PV Nodes New PV Nodes 46
Evaluation of the Proposed Controls MV (1) LV (2) Local Control with mixed V cpb Nodes 3 4 Line 3 5 Line 4 6 Line 5 7 Line 6 8 Line 7 9 Line 8 10 Line 9 11 Line 10 12 Line 11 13 Line 1 Line 2 PV name Nominal Power (kw) PV 5 11,65 PV 6 29,85 PV 7 9,31 PV 8 52,24 PV 9 13,98 PV 10 59,7 PV 11 18,65 PV 12 44,77 PV 13 16,3 Total 256,45 BAU PV Nodes New PV Nodes 47
Evaluation of the Proposed Controls 200 180 Old PV units New PV units 160 Total Injected Energy (MWh) 140 120 100 80 60 40 20 0 BAU Local (1.06) Local+FPS (1.06) Local (1.08) Local+FPS (1.08) Local (mixed) Local+FPS (mixed) 48
Outline 1. Challenges on LV Distribution Networks 2. Description of Proposed Control Schemes a. Local Control b. Overlaying Control c. Scheduling Control 3. Evaluation of the Proposed Controls 4. An Integrated Simulation Platform 49
The Integrated Simulation Platform Objectives Evaluate the effectiveness of the proposed control schemes Simulate extended networks fast and reliably Draw Simulink Designing Tool JADE Agents Environment 1-way 2-way MATLAB 1-way 2-way OMNeT++ LAN Simulation Framework OpenDSS Distribution System Simulator 50
Thank you very much for your attention! Questions? 51