NMR Enabled Sample Design Kent Mathis – JEA Craig Williamson – Energy Insights
JEA (formerly Jacksonville Electric Authority) Originally a municipal electric authority In 1997 JEA took over water and sewage for the city Now we’re just “JEA”. We have about 2300 permanent employees. Customer base: 405,000 electric 300,000 water 223,000 sewer Service Territory 900 square miles Several counties in NE Florida
JEA’s Electric System JEA owns and operates three generating plants and all transmission and distribution facilities. A fourth power plant, the St. Johns River Power Park (SJRPP), is owned jointly by JEA and the Florida Power & Light Company (FPL) and operated by JEA. JEA and FPL are also joint owners of Unit 4 at Georgia Power Company's coal-fired Robert W. Scherer Plant (Plant Scherer), which is located in Macon, Georgia. JEA produces 3.2 MW from a methane-fueled generating facility at the Girvin Road Landfill. JEA's net generating capability is 2,361 MW, gross capacity is just under 3,000 MW
The Ascent of JEA NMR-Enabled Load Research Craig Kent Future Craig Kent Present
Cellnet Nomenclature
Establish Team Corporate Data Integration (CDI) Randy Boswell VP CDI Richard Vento Director CDI Dennis Fleming Adv NMR Engineer Tonya Bell LR Analyst Kent Mathis LR Manager Don Wucker DSM Specialist
Initial Corporate Data Integration Goals Develop Team Knowledge, Skills & Abilities Training (LR, DSM, NMR & Tools) Hire LR & DSM Consulting Experts Meet Customers & Discuss Needs Load Research (LR) Demand Side Management (DSM) Advanced NMR
Develop Team Knowledge, Skills & Abilities Training Joined AEIC and pursuing membership in LR committee LR Workshops, Conferences & Classes NMR Understand how the system works Software SAS AnalyzeIT & VisualizeIT Hire Load Research Expert Surveyed Options Traditional consultant vs subscription services Energy Insights
Meet Prospective Customers and Discuss Their Needs Over a 3 month period (March, April, May) met with customers Rates, System Planning & Forecasting, Distribution Engineering Marketing, Customer Service, Account Executives, DSM Operations, System Reliability, Power Quality Billing & Metering, Collections Needs Centered Around Load Research Demand Side Management Analyze Winter Peak & Develop Programs to reduce peak TOU Advanced NMR Functions Distribution System Loading Optimized Theft Detection Customer Load Information & Notification
Initial Load Research Goals Develop Toolset Get NMR Interval Data Flowing – July 2006 Figure out Process Switching Meters On Initial Sample Design Collect & Analyze Summer Peak – Late September 2006 Completed Load Curves System Load Distribution – Ongoing Estimate Rate Class Load Shapes – Long term Design, Select & Install Sample Collect & Analyze Data
Develop Toolset Load Research Toolset Complete Connectivity between Systems Currently using temporary database (MSPR) SAS Purchased in March AnalyzeIT & VisualizeIT Purchased in April Excel
Meter Data Management System (MDMS) NMR Business Intelligence Load Research System PlanningMarket Research Energy Management Other Data Sources (FFDS, FMS, CAIR, etc.) (FFDS, FMS, CAIR, etc.) Other Data Sources (FFDS, FMS, CAIR, etc.) (FFDS, FMS, CAIR, etc.) ITRONS MV90 Primary Meter Data Other Meter Data Field Force Dispatch DispatchSystem(FFDS) Field Force Dispatch DispatchSystem(FFDS) Data Marts External Customers Web Access to ExternalCustomers ExternalCustomers NMR(Cellnet)NMR(Cellnet) Customer Information System (CAIR) JEA Internal Customers Customers Meter Data Management System Flow Making changes to MDMS and creating Data Marts to allow LR data to pass thru
Initial Sample Design Summary Examined Sample Design Population 124 Report & CAIR Rate Name Report Studied 5 rate classes (Res, GS, GSD, GSLD, Int/Cur) Designed samples for Res, GS, and split the population for GSD into two subgroups Consumption Data Cleaned and organized data (frequency distributions in 50 kWh buckets) Developed 2-D Stratification (reduce variability) Delanius Hodges – Produced Breakpoints For Res and GS, we used a 3x2 design of 6 strata, with 3 winter strata (low, medium, high), and two summer strata (low, high), plus one extreme stratum (very high summer or very high winter or both), producing 7 ‘buckets’ For each subgroup in GSD, we used a 2x2 design, with 2 winter strata (low, high), and two summer strata (low, high), plus one extreme stratum (very high summer or very high winter or both), producing 5 ‘buckets’
Residential Sample Implementation
GS Sample Implementation
GSD Sample Implementation
Sample design moving forward We will use the variances and means from the data for August 4 th to calculate required sample sizes and allocate sample points for the summer peak We will use the means and variances for the MV-90 data to calculate the required sample sizes and allocate sample points for the winter peak for GS and GSD, and use judgment for Residential winter peak accuracy We will carefully consider the implications of the expected percent of meters that can’t deliver interval data Current sample will be randomly reduced or increased where necessary We will include interval data from MV-90 customers that are chosen in the sample
Lessons Learned Expectation Toolset will be completed within 2 to 3 months Reality Time to build toolset is longer than promised
Lessons Learned Expectation It’s easy to keep things straight… Reality No connectivity without toolset Each system uses different keys GIS – Transformer ID MV90 – Recorder ID CAIR – Account ID MSPR – SPIDs Difficult to do what should be an easy task
Lessons Learned (cont) Expectation Over 99% of our NMR meters are enabled Reality Over 99% of NMR meters are deployed Only about 90% of NMR meters are active Of the 90%, even a lower % have acceptable Packet Success Ratios (PSRs) None of the above issues are evenly distributed among rate classes
Lessons Learned (cont) Expectation Defining the population will be easy Reality Defining the population is hard No accurate list of customers and their rates exist
“Ahas” Originally afraid JEA faced a long, slow, expensive learning curve - Purchasing Energy Use Strategic Service continues to be high value for cost; purchase did not require formal bidding process Having an NMR system is enabling our LR group to expand beyond traditional LR applications and uses
Going Forward Complete COS Sample Design Complete LR Toolset Collect & Analyze Winter Peak Develop System Load Snapshot Skills Develop Customer Base & LR Products
Contact Information Craig Williamson Program Director, Energy End Use Research tel Kent Mathis Load Research Manager tel