Interpreting Reliability Data Yamille del Valle, Nigel Hampton, Josh - PowerPoint PPT Presentation

Interpreting Reliability Data Yamille del Valle, Nigel Hampton, Josh Perkel, Essay Wen Shu Presentation prepared for the IEEE PES Distribution Reliability Group Joint Technical Committee Meeting (JTCM) January 12-16, 2020 • Jacksonville, FL, USA 1

Notice The information contained herein is, to our knowledge, accurate and reliable at the date of publication. Neither GTRC nor The Georgia Institute of Technology nor NEETRAC shall be responsible for any injury to or death of persons or damage to or destruction of property or for any other loss, damage or injury of any kind whatsoever resulting from the use of the project results and/or data. GTRC, GIT and NEETRAC disclaim any and all warranties, both express and implied, with respect to analysis or research or results contained in this report. It is the user's responsibility to conduct the necessary assessments in order to satisfy themselves as to the suitability of the products or recommendations for the user's particular purpose. No statement herein shall be construed as an endorsement of any product, process or provider. 2

Purpose of the Study • SAIDI, SAIFI are internationally recognized indices used to describe electric service reliability at distribution level • Used to: – Provide Context – Understand correlations – Improve performance – Estimate future performance / establish resource needs • Public Data available (IEEE, EIA, Regulators, CEER, Utility websites etc.) 3

Reliability Growth Model Utility ID = U4 Utility ID = U25 Utility ID = U3 1000 1000 1000 Cumulative IEEE SAIDI Cummulative IEEE SAIDI Cummulative IEEE SAIDI Cummulative IEEE SAIDI 100 100 100 1 1 1 1.5 1.5 1.5 2 2 2 3 3 3 4 4 4 5 5 5 6 6 6 7 7 7 Cummulative Time (years) Cummulative Time (years) Cummulative Time (years) Cumulative Time (years) 4

Perform ance Evaluation Utility ID = 30 Utility ID = 30 Utility ID = 30 200 200 200 150 150 150 Cummulative IEEE SAIDI Cummulative IEEE SAIDI Cummulative IEEE SAIDI Cumulative IEEE SAIDI 100 100 100 90 90 90 80 80 80 70 70 70 60 60 60 50 50 50 40 40 40 30 30 30 20 20 20 1 1 1 2 2 2 3 3 3 4 4 4 5 5 5 6 6 6 7 7 7 Cummulative Time (years) Cummulative Time (years) Cummulative Time (years) Cumulative Time (years) Results include rows where 'ID_4'=301. Results include rows where 'ID_4'=301 Or 'ID_4'=302. Results include rows where 'ID_4'=301. 5

Perform ance Quantification Utility ID = 30 220 Estimated 200 SAIDI 180  = 8% Cummulative IEEE SAIDI Cumulative IEEE SAIDI 160 Actual SAIDI 140 120 100 80 4 5 6 7 Cummulative Time (years) Cumulative Time (years) Results include rows where 'ID_4'=301 Or 'ID_4'=302. 6

Trending / Prognosis Utility ID = U1 Utility ID = U1 Utility ID = U1 6 8 6 8 800 800 800 700 700 700 608.38 600 600 600 500 500 500 428.35 Cumulative IEEE SAIDI Cumulative 400 400 400 CumSAIDI w/o MED CumSAIDI w/o MED CumSAIDI w/o MED 2018 SAIDI 300 300 300 200 200 200 100 100 100 1 1.5 2 3 4 5 6 7 8 9 10 1 1 1.5 1.5 2 2 3 3 4 4 5 5 6 6 7 7 8 8 9 9 10 10 CumTime CumTime CumTime Cumulative Time (years) 7

Perform ance by System 200 EEI_Origin_1 OH Cummulative SAIDI - 2013 to 2015 UG 150 Cumulative IEEE SAIDI 100 90 80 70 60 50 40 30 15 20 30 Cummulative Time (months) Cumulative Time (months) Results exclude rows where 'EEI_Origin_1'="Gen" Or 'CumTime_1'<12. OH contribution to total SAIDI is higher than the contribution of UG However both systems are comparable in size 8

Perform ance by Cause System Level Cause 1 Cummulative SAIDI - 2013 to 2015 BALLOON 20 VEHICLE HIT Cumulative IEEE SAIDI 15 10 9 8 7 6 5 4 10 15 20 25 30 35 Cummulative Time (months) Cumulative Time (months) Vehicle hits have higher contribution to the SAIDI however balloon caused outages are increasing 9

Reliability Index 900 10 800 9 700 8 600 7 6 Cumulative SAIDI 500 Cumulative SAIFI 5 400 4 300 3 200 2 150 1 1.5 2 3 4 5 6 7 1 1.5 2 3 4 5 6 7 Cumulative Time Cumulative Time Compute average yearly Determine second order SAIDI and SAIFI performance change (tip-up or tip-down) Assess SAIDI and SAIFI evolution over time 10 10

Project Status - Benchmarking Average Average SAIFI SAIDI SAIFI SAIDI Utility SAIFI SAIDI Trend Trend Tip‐up/down Tip‐up/down Score Example Green values are good, Amber needs to be monitored, Red is concern Average Average SAIFI SAIDI SAIFI SAIDI SAIFI SAIDI Trend Trend Tip‐up/down Tip‐up/down Score Multivariate Machine Learning Algorithm based on utility data, self adjust with the “information content” of the data Overall score combines all features Outlier rejection included Model builds with experience Influence of individual good / bad years is minimized 11 11

USA Visualization – IEEE Method Algorithm (outlier rejection, level, trend) can also be used for visualization 25 PUDs 220+ Munis 400+ Coops 150+ IOUs 800+ Utilities 12 12

State Cooperatives Example - Median Data Size of bubble relates to number of CoOps reporting Data Scrubbed Outliers Rejected 2103 to 2018 13 13

State Cooperatives Example - Trending Data Scrubbed Outliers Rejected 2103 to 2018 14 14

State Cooperatives Example - Local Context State 1 State 2 State 3 State 4 State 5 15 15

State Cooperatives Example - Benchmarking Average Average SAIFI SAIDI SAIFI SAIDI Reliability Utility SAIFI SAIDI Trend Trend Tip‐up/down Tip‐up/down Score Index Example 1.51 122.42 0.87 0.76 0.23 0.27 4.879 • Multivariate Machine Learning algorithm creates a single reliability index for benchmarking purposes • Index can be used to compare with other state cooperatives • Can be used with peers selected by other criteria (not geographical proximity) • Can be used with any level of granularity State 1 State 2 State 3 State 4 State 5 16 16

For more information please contact: yamille.delvalle@neetrac.gatech.edu nigel.hampton@neetrac.gatech.edu 17