DOE GMLC Project Computational Science for Grid Management Computational Framework for Grid Applications

Transcription

1 5 th GridOPTICS Workshop November 9-10, 2016 Richland, WA DOE GMLC Project Computational Science for Grid Management Computational Framework for Grid Applications HENRY HUANG Chief Engineer/Technical Group Manager Pacific Northwest National Laboratory (PNNL) PNNL-SA

2 The future grid needs holistic highperformance analytics Renewable (Renewable energy technology, interconnection standards) Smart Grid (Deployment of smart devices, empowering customer involvement and innovation) Renewable Integration (Renewable Integration Modeling) Transmission Grid Holistic High Performance Analytics With end-to-end grid in mind, address questions: What can we use the data for (what data network is required)? How will we address the complexity in order to understand the grid? How will we run and control such a complex grid? Storage and Storage Integration (Deployment of costeffective storage) Transmission Reliability (Advanced measurement and control technology, e.g., synchrophasors) 2

3 Math and computing challenges in modeling and simulation of the future grid Multi-scale spatio-temporal modeling and simulation with stochasticity From micro-second to decades From 10 3 generators nodes to 10 9 end-use devices Large-scale data assimilation for state and parameter calibration Petabyte data/year from high-speed sensors and smart meters. Modeling of multi-system dynamics and dependency Grid, buildings, communication, gas pipelines, weather/wind/solar, water Spatio Temporal 15 years 10 years 5 years 1 year 1 month 1 week 1 day 5 minutes μ-seconds Generation Transmission Substation 3 Transformer Site Load 3

4 Math and computing challenges in optimization and control of the future grid Dynamic security assessment with uncertainty from renewable and DERs Decision support with actionable information from big data from both large-volume measurements and large-scale simulation Optimal planning and operation with uncertainty and dynamics Short time horizon for economic dispatch and unit commitment (5 minutes) Long time horizon for transmission planning (>10 years) Optimization and control of million devices in a hierarchical architecture 33% Sources: EIA, EPA, DOE, FERC, Carnegie Mellon, GlobalData 4

5 Need for a computational framework and efficient solvers A computational framework that enables linking data to computation and software compatibility Efficient solvers that handle large-scale diverse problems Conceptual Design GridOPTICS TM Applications Run-Time Data Communication and Management Data Computation Visualization Actionable 5

6 Reference implementation with GOSS and GridPACK Requirements: scalable, portable, extensible, fast, compact GOSS hide implementation details from application developers a middleware using publication/subscription mechanism GridPACK expose math to solver developers a software framework and library designed for grid applications on highperformance computing platform Applications GridOPTICS Software System (GOSS) APIs APIs APIs Data Computation Visualization Actionable GridPACK Math and Computing Libraries (PETSc, SUNDIALS, Hypre, Trilinos, Minotaur, New Solvers) 6

7 GridOPTICS Software System (GOSS) 7

8 GridOPTICS GridPACK TM GridPACK Applications Base Factory Network-wide Operations Base Network Components Neighbor Lists Matrix Elements Application Driver GridPACK Framework Application Factory Application Components Import Module PTI Formats Dictionary Network Module Exchanges Partitioning Configure Module XML Export Module Serial IO Task Manager Math and Solver Module PETSc Mapper Utilities Errors Profiling Core Data Objects Matrices and Vectors Power Grid Network 8

9 Example use cases to drive the development of the framework and solvers Domain drivers: Uncertainty: Renewable, DERs, EVs Dynamics: physical behaviors (e.g. power electronics), market dynamics Optimization: transmission + distribution + Use Case 1: Security-Constrained AC OPF under Uncertainty Use Case 2: Dynamic Security Assessment with Uncertainties Use Case 3: Optimization under Uncertainty with Transient Security Constraints 9

10 Example: Use Case 2 Dynamic Security Assessment with Uncertainties Impact of increasing penetration of renewable generation on stability Impact of DERs & consumer participation on system operation & planning Optimization of DERs for multiple value streams for the power system Smart Sampling (R) GridPACK TM (Task manager) PF/DSA PF/DSA Post-processing of PF/DSA outputs (R) Web-based visualization Forecast/Actual Values Contingency List Smart sampling realizations with reduced contingencies PF/DSA Output PDFs Smart Sampling Realizations Reduced Contingencies PF/DSA Outputs PDF Commands (If any) GOSS Middleware Actual and Forecast values Contingency List Smart Sampling Outputs Power System Model PF/DSA inputs PF/DSA Outputs PDFs GOSS database 10

11 Questions? 11