Next Steps in Load Modeling. Dmitry Kosterev, BPA, NERC LMTF Meeting August 2017

Transcription

1 Next Steps in Load Modeling Dmitry Kosterev, BPA, NERC LMTF Meeting August

2 Need for Flexible Load Model Structure Current composite load model CMPLD is rigid with respect to model components and their models End-uses are evolving - Increasing percentage of electronically connected loads (VFDs, ECMs, chargers) - Distributed energy resources (PV solar, batteries) and changing characteristics from IEEE 1547 leagcy to smart features Our understanding of end-uses is also evolving - Test work done by EPRI, SCE, BPA, others Experience with CMPLD model provides a useful feedback on future improvements needed to make models behave better in grid simulations - Progressive tripping of motors and drives as a function of voltage and time There is a compelling argument for a more flexible model structure 2

3 Present CMPLD CMPLD is likely to stay around for next 3 to 5 years(?), as the next generation CMPLD2 development, implementation and validation will take time and resources Therefore, NERC LMTF needs to continue providing technical support to CMPLD use in planning studies Recently completed model benchmarking among main grid simulators used in North America Several data revisions are planned: - More robust motor protection and control data - Update AC stall feature with lower stall thresholds of 0.4 to 0.45 per unit User feedback is essential to drive model enhancements 3

4 2 nd Generation CMPLD Modular model structure - Need for model flexibility with respect to model components - Consistent rules for model data interpretation - Efficient data management no need to repeat the same data 6,000 times End-Use Component Models - Distributed Energy Resources - Single-phase motor models - Performance model:revise stall and reaccelerate characteristics - MOTORC model: validate MOTORC model, include in load component library - Scroll vsrecip? 4

5 2 nd Generation CMPLD End-Use Component Models - Power Electronic Loads: We may need to differentiate between ECMs, motor drives and electronic/computing/charging loads - Protection and Control Modules: Revise protection and control models to allow for progressive motor tripping Model Benchmarking Studies - Develop a set-up for model benchmarking, certify version releases 5

6 Data Load Composition - Planners need to develop understanding of load composition in typical commercial and residential buildings - Understand changes in load composition - Building surveys - Development and validation of next generation Load Composition Model - Development of default data sets Data Management Tools - Load Model Data Tool for data management need to be aligned with the new model structure - More transparent load composition data 6

7 Validation Load Monitoring - Continue (or begin) deployment of data reorders - We prefer continuous point on wave recordings sorry micro PMUs and DFRs (a) non-intrusive load composition monitoring, (b) load composition, (c) disturbance recordings Model Validation Studies - Have reasonable expectation for default data the model is expected to reproduce the phenomenon, not the detail - Tune model to match the event, identify model data improvements 7

8 Going Beyond Modeling Voltage (kv) Basline 50% Constant Current / 50% Constant Power 100% Constant Power Loads of the future are likely to get much simpler Most loads will become electronically connected chargers, motor drives, solid state lighting Frequency (Hz) Time (sec) Basline 50% Constant Current / 50% Constant Power 100% Constant Power Under-Frequency Load Shedding However, it does not mean the problems get easier We need to work with trade organization, like we did with AHRI, to communicate grid requirements to the equipment manufacturers and standard-making bodies Time (sec) 8

9 Thank You Let s do it! 9