Optimization of Distributed Energy Resources with Energy Storage and Customer Collaboration

Transcription

1 Optimization of Distributed Energy Resources with Energy Storage and Customer Collaboration NOVEMBER 2014 Jon Hawkins Manager, Advanced Technology and Strategy NOVEMBER 2014

2 PNM SERVICE TERRITORY 2,572 MW generation capacity Peak load approximately 2000 MW Off peak load approximately 1200 MW 14,696 miles transmission and distribution lines Average residential rate approximately $0.11/kWh U.S. SLIDE 2 NOVEMBER 2014

3 PNM S PROJECT APPROACH TO COLLABORATING WITH DISTRIBUTED RESOURCES SLIDE 3 NOVEMBER 2014

4 SOLAR VARIABILITY Clear Day Cloudy Day SLIDE 4 NOVEMBER 2014

5 1:00 AM 2:00 AM 3:00 AM 4:00 AM 5:00 AM 6:00 AM 7:00 AM 8:00 AM 9:00 AM 10:00 AM 11:00 AM 12:00 PM 1:00 PM 2:00 PM 3:00 PM 4:00 PM 5:00 PM 6:00 PM 7:00 PM 8:00 PM 9:00 PM 10:00 PM 11:00 PM 12:00 AM Peak Average Load (MW) SOLAR AVAILABILITY PNM System Peak Average Hourly Load by Month SUMMER PROFILE WINTER PROFILE The best solar production occurs ~ 2 to 8 hours prior to when the most power is needed on the system SOLAR PROFILE SLIDE 5 NOVEMBER 2014

6 PNM PROSPERITY ENERGY STORAGE Project Description First of 16 U.S. DOE Smart Grid Storage Demonstration Projects to go on line Sept 2011 Designed to both smooth PV intermittency and shift PV energy for multiple applications Successfully demonstrating Storage/PV integration to Utility operations Equipment 500 kw PV (fixed C-Si panels) not DOE funded Ecoult/East Penn - Advanced Lead Acid Battery system for shifting 1MWh Ecoult/East Penn - Ultra Battery system for smoothing - 500kW Cyber Secure, High Resolution Data Acquisition and Control System 1 second and 30 samples per second data capture SLIDE 6 NOVEMBER 2014

7 PNM Prosperity Energy Storage PNM PV SMOOTHING Prosperity DEMONSTRATION Energy Storage Smoothing Test Plan Results Clear Day Blue PV Yellow Battery Red Primary Meter Cloudy Day - Altocumulus Sandia National Labs based algorithm Variety of control inputs PV Meter, Irradiance Sensors (average, individual) Variety of gains on input tests different capacities of battery use Optimization target: how much smoothing is enough? Key: Yellow = Battery Output Red=System Output Magnified 5/6/13 with 40 minute magnification Blue=PV Output SLIDE 7 NOVEMBER 2014

8 SMOOTHING ALGORITHM IMPLEMENTATION Developed by Sandia National Laboratories, Implemented by Ecoult Baseline algorithm to respond to the changes in solar output. Dynamic Ability to optimize with different control source inputs. Ability to be tuned by changing input parameter and gains within the equation Experiment Provided the sum of the output of the kw from the fuel cell plus the gas engine (Pge + Pfc) to Aux 1 SLIDE 8 NOVEMBER 2014

9 Moving Average vs Low Pass Filter Smoothing Energy Use Analysis Shows LPF uses 18% more energy use compared to MA Real Energy includes parasitic loads CDF Analysis Shows effective smoothing (quantified) but does not show a big difference between LPF and MA SLIDE 9 NOVEMBER 2014

10 PV SMOOTHING DEMONSTRATION (VOLTAGE) Smoothing Test Plan Results Clear Day Cloudy Day - Altocumulus Corresponding feeder impacts with Prosperity measurement Direct impact on feeder voltage noted Reduction in Load Tap Changer (LTC) operations noted LTC operations measured before and after as one benefit metric Circuit Voltage No Smoothing Prosperity Site Meters - No Smoothing Circuit Voltage with Smoothing Prosperity Site Meter with Smoothing Key: Yellow = Battery Output Red=System Output Blue=PV Output SLIDE 10 NOVEMBER 2014

11 AUTOMATED DISPATCH CAPABILITY - STACKED BENEFITS Internal Optimization Required Firmed PV Allows prioritization of Applications without the need for human intervention Reliability is top Priority - Peak Shaving Further Optimization Determines value of Firming vs Peak Shaving vs Arbitrage Life of Battery and Energy Throughput also a consideration Firm dispatch with weather prediction Peak Shaving Arbitrage SLIDE 11 NOVEMBER 2014

12 COMBINING SMOOTHING AND SHIFTING Simultaneous PV Shifting and Smoothing - 01/14/2013 Entire day of cloudy PV production needed to charge battery for evening peak Firming Key: Blue=PV Output Yellow = Battery Output Red=System Output SLIDE 12 NOVEMBER 2014

13 AUTOMATED OPTIMIZATION OF ALL CAPABILITIES NWS Next day Weather Forecast % Cloud Cover Temperature 228 Available Points from Prosperity site Met Data System Data Meter Data SCADA Data Currently Monitoring 3 Feeders ~ 6 sec poll rate Utilizing set thresholds System optimizes functionality based on priorities to perform: Emergency peak shaving Peak shaving Arbitrage (wind and PV) PV Firming All while simultaneously smoothing PV and optimizing for battery life Market Pricing Currently using CAISO Real time price (SP15) LMP Forecast price (SP15) SLIDE 13 NOVEMBER 2014

14 PRIORITIZED OPERATIONS EXAMPLE USING CONSERVATIVE THRESHOLDS ON INPUT VARIABLES Charge due to RT price Emergency Peak Shaving Charge due to State of Charge Winter Evening Firming SLIDE 14 NOVEMBER 2014

15 FEEDER CONFIGURATION Approximately 1.7 miles (2.73 km) between projects as the crow flies 1.7 miles Approximately 2.5 circuit miles (4.02 km) SLIDE 15 NOVEMBER 2014

16 NEDO BUILDING MICRO GRID 50 kw PV 50 kw Battery Storage 240 kw Gas Engine 80 kw Fuel Cell Thermal Storage micro EMS Thermal Storage Capable of islanding Building Energy management SLIDE 16 NOVEMBER 2014

17 RESEARCH OBJECTIVE Objective: Reduce battery operation in PV-smoothing systems by novel control schemes. Smoothing PV power with a coordinated battery and gas genset reduces the required battery capacity and increases battery life. Simulations demonstrate a reduction in battery operation (SOC range) when the battery is paired with a gas engine-generator (GE). Research Partners: Special Thanks: Abraham Ellis 1, Jay Johnson 1 Atsushi Denda 2, Kimio Morino 2, Jon Hawkins 3, Brian Arellano 3, Takao Ogata 4, Takao Shinji 4, and Masayuki Tadokoro 4 1 Sandia National Laboratories 2 Shimizu Corporation 3 Public Service Company of New Mexico (PNM) 4 Tokyo Gas Co., Ltd. Acknowledgment of support to Dr. Imre Gyuk, Electricity Storage Program Manager, DOE Office of Electricity Slide courtesy of Sandia National Laboratory Jay Johnson SLIDE 17 NOVEMBER 2014

18 COORDINATED, DISTRIBUTED PV SMOOTHING Control was also tested without the feedback loop between the two sites. Microgrid PI Server at the Aperture Center P GE P PV PNM PI Server at Ops Center P PV P GE P GE Building Energy Management System (BEMS) [Commercial Site] BEMS calculates the genset setpoint to achieve PV smoothing. P PV P GE Aperture Center Microgrid P GE- Set PV Genset P PV Battery P Bat-Set Battery Energy Storage System (BESS) [Utility Site] BESS calculates the battery setpoint to achieve PV smoothing incorporating P GE. Slide courtesy of Sandia National Laboratory Jay Johnson SLIDE 18 NOVEMBER 2014

19 COORDINATED VS UNCOORDINATED CONTROLS Experimental reduction in battery operation with coordination. Theoretical (simulated) reduction in battery operation with coordination. Slide courtesy of Sandia National Laboratory Jay Johnson SLIDE 19 NOVEMBER 2014

20 BATTERY SOC RANGE STUDY Higher frequency PV power output leads to SOC drift with the coordinated control. Therefore, in certain cases the coordinated controller does not reduce the SOC range of the battery as originally expected. Two simulations with P PV square waves. GE reaches GE nom each cycle. GE stays below GE nom so P bat is always biased positive. Low frequency PV power allows the GE to reset and the coordinated battery SOC range is smaller. High frequency PV power doesn t allows the GE to reset and the coordinated battery SOC range is larger. Slide courtesy of Sandia National Laboratory Jay Johnson SLIDE 20 NOVEMBER 2014

21 BATTERY THROUGHPUT ANALYSIS Simulation: Total energy throughput reduction from using the coordinated controller is kwh Experiment: Total energy throughput reduction from using the coordinated controller is kwh The blue area is where the coordinated battery is working less than the uncoordinated battery. The red area means the coordinated battery is working harder than the uncoordinated battery. Slide courtesy of Sandia National Laboratory Jay Johnson SLIDE 21 NOVEMBER 2014

22 FUTURE WORK Frequency Response React to external market signal (1-3 sec signal) Desktop analysis of modified battery system (if we were to build it today) Economic analysis based on new design as well as new functionality (ESVT and ESCT) Reactive Power support Continued Battery control system software revisions for battery system health and operation Possible research of feeder as a microgrid Equipment Damage Over Frequency Trip Governor Response Under Frequency Load Shed Under Frequency Trip Contingency Response Equipment Damage Courtesy Ecoult Governor Response Time Correction Normal Range with AGC Corrective Action Time Correction Governor Response SLIDE 22 NOVEMBER 2014

23

24 SLIDE 24 NOVEMBER 2014