NREL Experiences on Scheduling Techniques Dr Murali Baggu

NREL Experiences on Scheduling Techniques Dr. Murali Baggu, Dr. Jin Tan, Dr. Ibrahim Krad, Dr. Santosh Veda, Dr. Yingchen Zhang and Dr. Vahan Gevorgian Power Systems Engineering Center National Renewable Energy Laboratory Workshop on “ 99+ Grid Availability for RE Integration” January 23, 2018

U. S. Interconnections 2

Dispatch During “Easy” Week Source: NREL WWIS 2 Report 3

Dispatch During the Worst Week (April) Source: NREL WWIS 2 Report 4

Western Wind and Solar Integration Study 2 Motivation • Thermal generating units that normally run at near constant output would need to cycle more frequently at high penetrations of wind power. • Understand possible impacts on O&M costs and emissions Source: NREL WWIS 2 Report 5

Active power control from wind Can WPPs provide reliable APC response, if we desire it? SYSTEM OPERATOR REGULATOR Why create a new market/policy, if there is currently enough capability and it is given for free? • Interaction of wind APCs • Impact on system reliability and stability Why should we design APC function for wind turbine, if there is no desire? Why should WPP provide APC, if we will lose money ? WPP OWNER /OPERATOR WIND TURBINE MANUFACTOR Gaps between Different Stakeholders 6

Ancillary service Multiple Time-scale Active Power Control Lack of a simulationcapability to show the benefit for different stake holders. Wind Variability Load Following TC AGC IC ms s PFC min hour day week Tools MAFRIT FESTIV WP Dynamic Simulation Tool (PSLF and PSS/E) Generation Dispatch Tool (PLEXOS) O IC: Inertial Control; PFC: Primary frequency control ; TC: tertiary control; FESTIV: Flexible Energy Scheduling Tool for Integration of Variable generation; MAFRIM: Multi-Area Frequency Response Integration Model 7

A Multiple Time-scale Hybrid Simulation Tool Key features: • Bridge dynamic simulation & economic dispatch simulation; • Response to 4 -s AGC and 5 -min dispatch ; • Represent frequency dynamics from seconds to days level considering the variability of the wind. 8

24 -hour Dynamic Simulation • 24 hour simulations • Dynamic responses of conventional generators and the wind plant • Frequency response of the interconnection grid 9

Hawaii Example Bulk System Simulation Goal Investigate, develop, and validate ways that distributed PV and storage can support grid frequency stability on the fastest time scale. Distribution System Grid Frequency Support from Inverter-based Resources in Hawaii • Couple with a distribution feeder model to enable Governor-only (simplified) hybrid transmissionmodel of Oahu distribution simulations HILT • Solvable in real-time using a fixed time step for power hardware-in-the-loop tests 10

Bulk System Simulation N-1 frequency response and output power of renewables Proprietary data and preliminary results. Please do not distribute. 11

Sensitivity Study Proprietary data and preliminary results. Please do not distribute. 12

Subhourly-Reserves – FESTIV Study Objectives: • Analyze nontraditional factors influencing operating reserves • Disseminate Flexible Energy Scheduling Tool for Integration of Variable generation (FESTIV) to industry professionals • Analyze and compare 7 different flexibility reserve requirement methods • Design an improved flexibility reserve requirement method Approach: • Perform a literature review of operating techniques and simulate several different factors in FESTIV to examine the operational implications of these factors • Send an NREL researcher to ERCOT to set up and train ERCOT researchers on how to use FESTIV • Implement 7 different flexibility reserve requirements in FESTIV and analyze the results • Implement the improved flexibility reserve method in FESTIV and compare with the results from FESTIV simulations Key Results and Outcomes: • Nontraditional factors impacting system operations analyzed and disseminated in IEEE Transactions on Power Systems • Performed an apples-to-apples comparison of several different flexibility reserve methods proposed in literature from academia and industry, one of the first comparisons of its kind • Disseminated FESTIV to an industry partner, ERCOT, for their input and validation of the tool; NREL received positive feedback from ERCOT regarding FESTIV • Designed an improved reserve method to help improve reliability metrics with minimal impacts on cost, facilitating the integration of VG under high penetration scenarios • Compared 7 different flexibility reserve requirements within FESTIV and disseminated the results via several different conferences • A new flexibility reserve modification method was designed based on the previous analysis and simulated in FESTIV to compare the results. The new method successfully improves the reliability metrics while minimally impacting cost metrics 13

Subhourly Reserves – Key Results Implications of Flexibility Reserves Direction of Control Performance Standard 2 (CPS 2) violations shows need for upward reserves in July and downward reserves outside July Increased VG curtailments results with significant intervals with LMP of 0 $/MWh Flexibility reserves increase thermal generator commitments as requirement becomes more conservative Increased commitments results with an increase in unused thermal capacity that can be used to help system operations 14

Subhourly Reserves – Impacts Ø Shared FESTIV, an NREL developed power system operations tool, with the power system operator in Texas who showed excitement and support for the design and features of the tool Ø A direct analysis of different flexibility reserve methods was completed and their operational implications were examined, one of the first comparisons of its kind Ø With this understanding, a new reserve requirement methodology was designed that improves reliability metrics while minimally impacting cost Ø More wind and solar can be integrated into the system with minimal impacts on operation Ø It would be interesting to investigate the interaction between flexibility reserves and other ancillary service products and compare the use of these explicit reserve methodologies in a stochastic modelling environment 15

Impact of Solar Eclipse on Grid Operations – PLEXOS and PSLF Study Objective: • Learn from the past grid operational experience to prepare for the solar eclipse event on Aug 21, 2017 • Develop tools to investigate the impact on solar eclipse on grid operations • Disseminate the lessons to industry stakeholders and public • Extend the framework to perform extreme event analysis Approach: • Perform spatio-temporal analysis • Establish baseline loading & dispatch • Estimate impact of the eclipse • Develop mitigation measures • Perform field evaluation • Perform post-event analysis and reporting • Extend the framework for extreme event analysis 16

Distributed PV Capacity Utility Area Estimated Distributed PV Capacity (MW) Utility Area APS 2160 PACE Estimated Distributed PV Capacity (MW) 57 BPA 109 PG&E 1513 El Paso 261 PSCO 101 IID 201 SCE 2016 IPCO 453 SDGE 522 LADWP 1077 Sierra 97 NM 449 WAPA 81 NVP 169 9. 2 GW Distributed PV across WECC 17

Perform spatio-temporal analysis • Estimated the capacity and location of utility and distributed PV sites • Estimated the impact on PV output Insolation profiles for a prototypical August 21 were calculated at each generator location using SAMPA o Weather data was supplied from NSRDB TMY (NSRDB) and supplemented with atmospheric data from MODIS (NASA) o SAM v 2016. 08. 01 with PVWatts v 5 was used to calculate generation profiles for each characteristic u. PV and d. PV system at each generator location o Profile for both eclipse and non-eclipse days were generated using the SAMPA-derived insolation profiles o 18

Spatio-temporal analysis 19

Establish baseline loading & dispatch • Performed production simulation using PLEXOS First scenario - standard operation model o Second scenario standard operation model but with the solar eclipse limiting the output of solar generators, both distributed and utility scale solar generation o • The PLEXOS model simulates the real-time dispatch and operations, consisting of a real-time security constrained unit commitment and economic dispatch problem (RTSCUC, RTSCED). This model is used in order to determine the operating schedules (commitments and dispatch) of all generation plants in the model at a 5 -minute resolution 20

Estimate the impact of the eclipse Performed production simulation for modified PV profiles 21

Estimate the impact of the eclipse • Performed transient simulation studies using GE PSLF • Two-part models – powerflow (algebraic) and dynamic models (differential) • Transient simulations to study electro-mechanical transients – Voltage stability – Frequency stability – Angular stability Case 1 2 List of credible contingencies 3 4 5 Contingency Description Three-phase fault at Palo Verde (Bus 194633), followed by loss of the Palo Verde generating units Three-phase fault near Celilo, followed by loss of PDCI Three-phase fault near Intermountain Plant, followed by loss of IPP Three-phase fault at Malin, followed by loss of MALIN to Round Mountain Three-phase fault at Capt. Jack, followed by loss of Captain Jack to Olinda 22

Perform post-event analysis • Peak Reliability collected high resolution data from across the footprint – Real-time Measurements • • Output from renewable plants Flows along transmission lines Inter-BA exchange flows Area Control Error, Schedules, reserves – EMS-solved cases • System State across the footprint as output of state estimation • Selected Four Representative Days – Aug 14, Aug 18, Aug 21, Aug 22 23

Total Generation and System Loading 24

Extend framework for extreme event analysis 25

Thank you! Murali. Baggu@nrel. gov