Transcript Slide 1
Optimization Energy Planning here August 2012 About Plexos • Plexos is the software used to produce the Integrated Resource Planning. • PLEXOS is a MIP-based next-generation energy market simulation and optimization software • Co-optimization architecture is based on the Ph.D. work of Glenn Drayton* • Advanced Mixed Integer Programming (MIP) is the core algorithm of the simulation and optimization • Foundation for the mathematical formulation of the New Zealand, Australia, and Singapore energy and spinning reserve markets • PLEXOS licensed in United States, Europe, Asia-Pacific, Russia, and Africa (17 countries, more than 100 sites) PLEXOS 4.0 first released in 2000 2015/07/21 2 Energy Planning Applications LT Plan – Optimal investment New Builds/retirements PASA – Optimal reserve share Maintenance Schedule MT – Resource Allocation Operating Policies ST – Chronological Unit Commitment Detailed by-period results 2015/07/21 3 Plexos Applications – Long-term decision horizon Decision horizon second Operational Planning Operational minute hour day week month year Planning 3 years 20 years – Long-term studies with decision variables impacting across all the horizon, e.g. • Object function minimizes NPV of: – Cost of new builds – Cost of retirements – Fixed operating costs – Variable operating costs 2015/07/21 4 LT-Objective Two types of costs Capital costs C(x): Cost of new generator builds Cost of transmission expansion Cost of generator retirements Production costs P(x): Cost of operating the system with any given set of existing and new builds and transmission network Notional cost of unserved energy 2015/07/21 5 Simplified LT Plan Formulation for Generation Expansion • Sum for all hours and years and generators: • Minimise the following Objective Function: Expansion DFy x BuildCostg x Pmaxg x GensBuiltg,y DFy x FOMCostg x Pmaxg x (GensExistg + Dispatch GensBuiltg,y) DFh x [VoLL x USEh + (SRMC) x GensLoadg,h] • Subject to: Linkage • GenLoadg,h + USEh = Demandh • GenLoadg,h <= Pmaxg x (GensExistg + GensBuiltg,y) • Integrality: GensBuiltg,y is an integer 2015/07/21 6 Problem Reduction • Aggregate the LDC to reduce the number of variables (under LT Plan) • Reduce the number of LDC blocks • Change the period over which the LDC blocks are calculated (from weeks to month, for example) 2015/07/21 7 Stochastic process in PLEXOS • To model input values as stochastic drivers • Define stochastic Variables in Variable class • Specify stochastic characteristics • Specify number of iterations for stochastic sampling and simulation • Properties of the Stochastic object • Assign Variables to stochastic drivers 2015/07/21 8 Stochastic process in PLEXOS • Define a Variable object in Variable class • Define stochastic characteristics • Two methods to define stochastic Variables • Exogenous sampling: user-defined profile samples (with assigned probability for each sample) • Endogenous sampling: user-defined expected profile that will be scaled up and down by random samples with random numbers with user-specified distribution 2015/07/21 9 Example of Stochastic process in IRP • Stochastic modeling for wind generation • Process • Define “Wind Stochastics” in Variables class • Define Stochastic distribution of the Variable as •Hourly sampling •Variable.Profile = Wind Generation Profile (wind.csv) •Error Std Dev = 30% •Max Value = 50, Min Value = 0 • Run a Model with the Variable, report sample details and compare the solution with the base case 2015/07/21 10 Projects • Wind Forecasting Establish a centralized wind forecasting To investigate how the integration of large amount of renewables impact: Lookup Y program. Ancillary reserve requirements to back up intermittent generation. Lookup X The overall system load volatility given large amount of renewables. Generators ramping requirements to circumvent stochastic generation. System emissions: how the system’s emissions profile is impacted if stations are cycled. Cost of cycling: a determined of the cost of cycling. 2015/07/21 11