NERC Load Modeling Activities Ryan D. Quint, PhD, PE Senior Engineer, System Analysis, NERC MRO Fall Reliability Conference November 2016
Introduction Kickoff January 2016 LMTF webpage Chair: Dmitry Kosterev, BPA Focus Areas 2 Coordination with Regional load modeling groups Broad expertise involvement Full utility industry coverage and participation Consolidate and share load modeling practices across industry Tasks Support industry wide advancement of dynamic load modeling Develop guidelines, technical references, industry webinars, etc. Help ensure robust software implementation Share lessons learned and study approaches
History & Background Why is this important? How did we get to where we are today? 3
Why Dynamic Load Modeling? Validation of power voltage oscillations WECC: July 2 and August 10 1996, August 4 2000 Fault Induced Delayed Voltage Recovery (FIDVR) Observed as early as 1980s in Southern California, Florida, Georgia, mid West Related to stalling of residential airconditioners Distributed energy resources Emerging need in early 2010s 560 540 520 500 480 460 440 420 400 10 0 10 20 30 40 4
Air Conditioner Testing Round 1 SCE, EPRI, and BPA tested singlephase residential A/C units Voltage sags, ramps, oscillations, frequency excursions Stall for sudden ΔV to 50 60% of nominal in less than 3 cycles Once stalled, they remain stalled Cannot overcome load torque coolant pressure must equalize Reactive power up to ~ 7x rated Thermal protection trips in 2 30 seconds 5
Air Conditioner Testing Round 1 3.6 0.66 Power (kw) 3.4 3.2 3 2.8 0.64 0.62 0.6 0.58 Stall Voltage (per unit) 2.6 80 85 90 95 100 105 110 115 0.56 Ambient Temperature (F) 6
Development of Initial Performance Model Compressor Real Power [W] 12000 10000 8000 6000 4000 2000 115F 110F 105F 100F 95F 90F 85F 80F STALL Real Power STALL RUN Motor stalls when voltage drops below V stall for duration T stall V stall ~ 0.52 0.6 pu 6 T stall ~ 0.033 sec Real Power Compressor Reactive Power [VAR] 0 0 50 100 150 200 Reactive Voltage Power [V] 12000 115F 110F 10000 105F STALL 100F 95F 8000 90F 85F 80F 6000 4000 STALL Real Power (per unit) 5 4 3 2 1 STALL STALL RUN 2000 RUN 7 0 0 50 100 150 200 Voltage [V] 0 0 0.2 0.4 0.6 0.8 1 1.2 Voltage (per unit) Source: BPA
Electromagnetic Transient Simulations 5 kw 1 ph A/C H = 0.048 sec Speed is pulled down very strongly by negative T elec Negative peak of electrical torque ~8x rated torque Current by stalled motor 5x rated current 8 Source: J. Undrill
Electromagnetic Transient Simulations 100 kva 3 ph Motor H = 0.3 sec Speed minimally pulled down by negative T elec Negative peak of electrical torque ~1 2x rated torque Current returns to near rated current 9 Source: J. Undrill
Point-on Wave Simulations Common mode failure of testing Applied voltage sag at point on wave zero crossing in every test Instantaneous voltage drop to 0.62 pu for 3 cycles Voltage zero crossing Voltage peak Voltage 45 deg point Worst case zero crossing 10 Source: Univ. Wisconsin
Stall Voltage vs. Point on Wave Simulation Source: Univ. Wisconsin 11
Stall Voltage vs. Point on Wave Testing Source: Univ. Wisconsin, BPA 12
Scroll vs. Reciprocating Scroll Compressors Vast majority of new compressors Better fault ride through ability May run backwards after fault ~1 1.25x rated current not locked rotor up to tens of minutes Estimated to be ~ 50% of A/C fleet today (2015 NERC FIDVR Workshop) Reciprocating Compressors Majority of fleet until 2000s Disappearing due to energy efficiency requirements Source: BPA 13
Meeting the Needs of TODAY for Dynamic Load Modeling Addressing issues and practices with the existing dynamic load models 14
Technical Reference Document Documents current state of dynamic load modeling Follow up to the NERC FIDVR Workshop in September 2015 Final NERC PC approval and publication in December 2016 15
Network Boundary Equations & Initialization Standardized procedures for software initialization Familiarize and standardize practices for dealing with current sources in dynamics motor model numerical issues Overcome crashing issues 16 Source: PowerWorld
Common Initialization & Network Boundary Equations Source: PowerWorld 17
Load Model Benchmarking Initial LMTF member and vendor benchmarking NERC LMTF default data set tested using standard test events Results being compiled to identify any discrepancies Fixing any software implementation issues identified Voltage Flow (Measured at From End) Bus V mag [pu] From Bus To Bus Flow MW Flow MVar 1 1.020 102 101 165.0 82 101 1.020 101 1 165.3 90.1 102 0.999 18
Load Model Benchmarking 3.59 3.585 3.58 P-M1 ---- MOTOR A P (PU ON SYSTEM MVA BAS PTI GE 3.575 3.57 3.565 3.56 3.555 3.55 3.545-20 0 20 40 60 80 100 120 Source: PacifiCorp 166 P load [MW] 165 PSLF PSSE PowerWorld TSAT 164 163 P load [MW] 162 161 160 Source: PowerWorld 159 4 6 8 10 12 14 16 18 20 Time [s] 19
Robust Default Data Sets Developed robust default data sets for use across regions as starting point Suitable and reasonable parameters for tripping Can be modified with regional data, composition data 20
Reliability Guideline: Load Composition Guideline on developing representative load composition data for dynamic load models 21 Source: WECC
Reliability Guideline: Load Composition Guideline on developing representative load composition data for dynamic load models 22
Reliability Guideline: DER in Dynamic Load Models Guidance on modeling DER in dynamic load models and powerflow models Coordination with DERTF efforts 23
Reliability Guideline: DER in Dynamic Load Models Utility Scale Distributed Energy Resources (U DER): distributed energy resources directly connected to the distribution bus or connected to the distribution bus through a dedicated, non load serving feeder. These resources are specifically three phase interconnections, and can range in capacity, for example, from 0.5 to 20 MW although facility ratings can differ. Retail Scale Distributed Energy Resources (R DER): distributed energy resources that offset customer load. These DER include residential, commercial, and industrial customers. Typically, the residential units are single phase while the commercial and industrial units can be single or three phase facilities. 24
Reliability Guideline: DER in Dynamic Load Models 25
Reliability Guideline: DER in Dynamic Load Models Source: PowerWorld 26
Industry-Wide Webinar Broader industry engagement Topics to cover Fundamentals of end use load Composite load model Parameters and their meaning Load composition data Distributed energy resource modeling Sensitivity analysis December 9 announcement to be made shortly 27
Utility Forum: Dynamic Load Modeling 28
Meeting the Needs of TOMORROW for Dynamic Load Modeling Addressing issues and practices with future dynamic load models 29
Progressive Stalling and Tripping Source: J. Undrill Source: PowerWorld 30
Efficient Data Format & Model Management Modularized load model structure Plug n play concept Distinction between network and load components Explicit DER representation Data input groups 31
Dynamic Load Modeling in Real-Time Stability Analysis Why do we ignore real time modeling practices? Why do we require induction motor load in planning studies but not in real time studies? What are the limitations in moving towards inclusion of induction motor load in realtime models? How do we proceed cautiously? 32
33