1. Frankfurt (Germany), 6-9 June 2011 AN OPTIMISATION MODEL TO INTEGRATE ACTIVE NETWORK MANAGEMENT INTO THE DISTRIBUTION NETWORK INVESTMENT PLANNING TASK Robert MacDonald Graham Ault University of Strathclyde Robert MacDonald, Graham Ault – UK – RIF Session ….. – 1025

2. Frankfurt (Germany), 6-9 June 2011 Active Network Deployment  ANM schemes emerging as alternative to network reinforcement Power-Flow Management via DG curtailment can eliminate thermal constraints Requirement to integrate ANM Deployment into planning stage of Network development  Requirement to model dynamic operational characteristics of ANM schemes Must model dynamic changes in operational states Consider uncertainty in demand, intermittent DG output

3. Frankfurt (Germany), 6-9 June 2011 Network Planning Optimisation Model  Objective is to find lowest-cost investment decisions over planning period  Deployment of ANM may add operational cost as compensation for curtailed energy  Stochastic Programming used to incorporate uncertainty into optimisation model Find optimal investment solution which hedges against future uncertainty Estimated operational cost over planning period calculated using Monte-Carlo method

4. Frankfurt (Germany), 6-9 June 2011 Problem Decomposition  3 quasi-independent sub- problems Master: Make investment decisions Feasibility: Check investment decisions meet security criteria Operation: Calculate expected operational actions and cost over planning period  Sub-problems coupled by Benders cuts Cuts share optimality information between sub-problems in form of constraints Master Investment Problem (Binary Programming) Feasibility Sub-Problem (Linear Programming) Network Operation Sub- Problem (Customised Load Flow) Investment decision variables are fixed and sent to next sub- model If investment results in infeasible operation – infeasibility cuts generated and sent back to Master Problem If no optimality, optimality cuts sent back to Master Problem If feasible – Master decision variables fixed and sent to Operation Sub- Problem Solution Solved once optimality criterion met

5. Frankfurt (Germany), 6-9 June 2011 Basic test-case  Section of rural network  4 Scenarios for new DG connections: 20MW – Wind 20MW – Non-Wind with Full Rated output 30MW – Wind 30MW – Non-Wind with Full Rated output  2-year planning period  Investment decisions: Deploy ANM at DG (CAPEX:100, OPEX:1) Upgrade weak line capacity (CAPEX:500/1000) 100 302303305 306 307 308 309 1103 11041105 1102 Thermal Overload

6. Frankfurt (Germany), 6-9 June 2011 Basic test-case results DG Units Connected Investment Decision DG Output (MWh) Curtailed Energy (MWh) % Energy curtailed 1: 20MW Wind ANM15198727251.8% 1+2: 30MW Wind ANM1940023806619.6% 1: 20MW Non-wind (Rated Output) Line Upgrade350400 (300560) 0 (49840)0 (16%) 1+2: 30MW Non-Wind (Rated Output) Line Upgrade525600 (300560) 0 (225040) 0 (42.8%) 100 302303305 306 307 308 309 1103 1104 1105 1102 DG 1 DG 2 ----- 132kV ----- 33kV ----- 11kv

7. Frankfurt (Germany), 6-9 June 2011 Conclusions  Incorporated deployment of ANM scheme into network planning optimisation model  Stochastic Programming structure considers probabilistic nature of intermittent DG and demand  Basic test cases validate decomposition approach