LOAD IMPACT EVALUATION OF NON-RESIDENTIAL CRITICAL PEAK AND PEAK DAY

LOAD IMPACT EVALUATION OF NON-RESIDENTIAL CRITICAL PEAK AND PEAK DAY PRICING 2019 DRMEC Load Impact Workshop April 26, 2019 Kelly Marrin, Project Director Energy solutions. Delivered.

AGENDA Program Descriptions Methodology Ex Post Impacts Ex Ante Impacts Key Findings Applied Energy Group · www.appliedenergygroup.com 2

Program Description

PROGRAM DESCRIPTION Critical Peak Pricing / Peak Day Pricing Program Basics: Non-Residential customers only Statewide price responsive DR program Customers experience an increase in price (above existing onpeak price) during events Operates year-round Events: Event hours are 2-6 PM Number of events per year varies PG&E 9 to 15 SCE 12 SDG&E maximum of 18 Customers are notified on a day ahead basis Applied Energy Group · www.appliedenergygroup.com 4

PROGRAM DESCRIPTION Program changes PG&E: Defaults are on hold until TOU period transition change is implemented in November 2020. 46,000 customers unenrolled as they transitioned to CCA SCE: No key changes in 2018 2019 begins the default of customers with demands below 200 kW and large Ag and Pump customers Event window changed to 4-9 PM effective March 1, 2019 Capacity Reservation Level (CRL) and CPP-lite options are no longer available SDG&E: New TOU periods established CPP event window moved from 11 – 6 PM to 2 – 6 PM Underlying TOU on-peak period is 4 – 9 PM Commission approved time periods for grandfathered TOU rates Applied Energy Group · www.appliedenergygroup.com 5

PROGRAM DESCRIPTION 2018 Participation, Typical Event Day Participation by Size Small 20 kW Medium 20 x 200 kW Large 200 kW Total Participation by Industry 1. Agriculture, Mining & Construction 2. Manufacturing 3. Wholesale, Transport, Other Utilities 4. Retail Stores 5. Offices, Hotels, Finance, Services 6. Schools 7. Institutional/Government 8. Other/Unknown Total Applied Energy Group · www.appliedenergygroup.com PG&E 119,004 34,014 1,712 154,731 SCE 106 659 2,251 3,016 SDG&E 12,854 1,211 14,065 PG&E 6,031 SCE 76 SDG&E 366 5,043 17,218 694 512 1,053 893 13,018 46,968 269 823 1,987 7,005 3,371 24,757 38,324 154,731 281 261 99 3,016 749 1,885 128 14,065 6

PROGRAM DESCRIPTION Communication Around Events Not all the participants were aware of events SCE and SDG&E provide day ahead notification to customers with contact information PG&E provides day ahead notification, and an enhanced level of support which included post event feedback Applied Energy Group · www.appliedenergygroup.com 7

Methodology

INTRODUCTION TO REGRESSION Regression analysis is about identifying and estimating statistical relationships between variables. Regression analysis studies the dependence of one variable, the dependent variable, on one or more other variables, the explanatory variables, with a goal of estimating and/or predicting the mean of the former in terms of the known values of the latter.1 Yit β0 β1 xit . . . . βn xit I We use regression models to estimate the counter-factual – what would have happened in absence of an event The model uses information from non-event days to predict how much energy customers would have used in absence of an event 1 Gujarati, D., Basic Econometrics, p.18, McGraw-Hill, 2003. Applied Energy Group · www.appliedenergygroup.com 9

ESTIMATING IMPACTS Actual consumption on an event day Consumption on the same day but in absence of an event Calendar Variables Calendar Variables Participati on Variables Weather Variables Actual Load Refere nce Load Applied Energy Group · www.appliedenergygroup.com Participati on Variables Weather Variables Referenc e Load Actual Load Impac ts 10

SUBGROUP LEVEL MODELING APPROACH Utility Each utility and size group is at a different stage in the default schedule Design was selected based on eligible non-participants favoring the development of a control group when feasible PG&E Size Group Ratio Small 1.3 Medium 1.6 Large Medium 3.7 3,904. 3 199.0 Large 5.3 Medium 0.4 Large 0.9 Small SCE SDG&E Analysis Method Within Subjects Within Subjects Matched Control Matched Control Matched Control Matched Control Within Subjects Within Subjects For all subgroups, regardless of design, we developed hourly fixed effect regression models Subgroups include: utility, size, and industry Each model was optimized and validated using our optimization approach Applied Energy Group · www.appliedenergygroup.com 11

SUBGROUP LEVEL REGRESSION APPROACH Baseline 1 Develop a set of candidate models using building blocks set up in logical groups 16-20 Candidate Models Applied Energy Group · www.appliedenergygroup.com Impacts AM Load Calendar Weather Event 12

SUBGROUP LEVEL REGRESSION APPROACH 2 3 Testing and optimization process that minimizes error and bias to select the best model for each subgroup Model the actual load Model the reference load MAPE and In-sample testing Out-of-sample MPE testing Reference Load Actual Load Impacts Calculate the impacts Applied Energy Group · www.appliedenergygroup.com 13

Ex-Post Impacts

EX POST IMPACTS Event Summary Date 6/12/2018 6/13/2018 7/06/2018 7/09/2018 7/10/2018 7/16/2018 7/17/2018 7/18/2018 7/19/2018 7/24/2018 7/25/2018 7/27/2018 8/01/2018 8/02/2018 8/06/2018 8/07/2018 8/09/2018 9/28/2018 10/18/2018 Total Day of Week Tuesday Wednesday Friday Monday Tuesday Monday Tuesday Wednesday Thursday Tuesday Wednesday Friday Wednesday Thursday Monday Tuesday Thursday Friday Thursday Applied Energy Group · www.appliedenergygroup.com PG&E X X X X X SCE SDG&E X X X X X X X X X X 9 X X X X X X X X X 12 X X X 6 15

EX POST IMPACTS – PG&E Average Summer Event, Average Event Hour Utility PG&E # Enroll ed Ref. Load (MW) Load Impa ct (MW) % Load Impa ct Event Temp Large 1,712 445.5 23.9 5.4% 93.1 Mediu m 34,014 750.0 4.9 0.7% 93.2 Small 119,00 4 243.7 (0.1) 0.0% 93.0 154,7 31 1,439 .2 28.8 2.0% 93.1 Size Group ALL PG&E Large customers provide the majority of the impact Small customer impacts are essentially zero – negative impacts result from modeling noise or bias Hottest overall weather of the three IOUs Applied Energy Group · www.appliedenergygroup.com 16

EX POST IMPACTS – SCE Average Summer Event, Average Event Hour Utility SCE ALL SCE # Enroll ed Ref. Load (MW) Load Impa ct (MW) % Load Impa ct Event Temp Large 2,251 583.7 14.2 2.4% 89.8 Mediu m 659 45.9 0.2 0.5% 89.4 Small 105 0.2 0.0 2.7% 88.9 3,016 629.9 14.5 2.3% 89.4 Size Group Again large customers provide the majority of the impact Participation in medium and small classes is opt-in so contributions are low Applied Energy Group · www.appliedenergygroup.com 17

EX POST IMPACTS – SDG&E Average Summer Event, Average Event Hour Utility SDG& E Size Group # Enroll ed Ref. Load (MW) Load Impa ct (MW) % Load Impa ct Event Temp Large 1,211 348.1 6.9 2.0% 88.5 Mediu m 12,854 437.5 1.9 0.4% 88.2 14,065 785.6 8.8 1.1% 88.3 ALL SDG&E Again large customers provide the majority of the impact Small customers are not included in this evaluation Coolest weather of the three IOUs Applied Energy Group · www.appliedenergygroup.com 18

EX-POST IMPACTS Communication Level Aggregate Type No Communication Communication PG&E # Enrolled 26,614 145,197 Applied Energy Group · www.appliedenergygroup.com (MW) % Load Impact Avg. Event Temp. Ref. Load Load Impact 762 4.4 0.6% 90.3 47.7 2.3% 91.0 SDG&E 2,093 SCE 19

EX POST IMPACTS Utility System Peak Hour # Enrolle d Utility PG&E PDP SCE CPP SDG&E CPP 7/25/2018 145,372 (HE19) 7/6/2018 3,082 (HE16) 8/09/2018 14,109 (HE17) % Load Event Impac Temp t Ref. Load (MW) Load Impact (MW) 1,310.1 28.5 2.1% 97 663.7 16.9 2.5% 107 792.5 4.8 0.6% 87 PG&E had the latest system peak at HE 7 PM, they also had the largest impact with nearly 29 MW SCE had the hottest system peak with a temperature of 107 and an impact of 17 MW SDG&E’s system peak was at HE 5PM, and their impact was nearly 5 MW Applied Energy Group · www.appliedenergygroup.com 20

EX POST IMPACTS Statewide System Peak Hour, 7/25/2018 - HE18 Utility PG&E PDP SCE - CPP SDG&E CPP Statewide # Ref. Load Enrolled (MW) Load Impact (MW) % Load Impact Event Temp 145,372 1,410.9 30.4 2.1% 97 - - - - - 14,043 711.7 5.5 0.6% 80 159,415 2,122.6 35.9 1.7% 96 The total load reduction across all three programs on the statewide system peak was 36 MW SCE did not call an event on the statewide peak day Applied Energy Group · www.appliedenergygroup.com 21

Ex-Ante Impacts

EX ANTE IMPACTS Methodology Use subgroup level regression models from ex post analysis Predict per-customer weather-adjusted impacts for all subgroups Apply Utility and CAISO weather scenarios Use enrollment forecasts from IOUs to forecast aggregate impacts Enrollment was derived based on Default schedules Population growth Historical trends IMPORTANT - RA Window Change 2017 evaluation 2-6 PM (coincident with operating hours) 2018 evaluation 4-6 PM SCE coincident with operating hours Applied Energy Group · www.appliedenergygroup.com 23

EX ANTE IMPACTS Comparison of current and previous ex-ante forecast Previous Forecast, 2018 Aggregate Accounts Impact (MW) 238,238 46.6 103,300 59.6 13,282 15.3 Current Forecast, 2019 Aggregate Accounts Impact (MW) 137,077 9.6 300,243 26.7 14,074 3.7 Utility PG&E SCE SDG&E Statewi 354,820 121.5 451,394 de 40.0 Results are average event-hour impacts for August peak day; Utility Peak 1-in-2 weather conditions. PG&E Decrease in enrollment due to change in default schedule Changes in RA window mean that only 2 of the 5 RA hours are program hours, with three of those hours occurring directly after the event when some customers might be increasing load. SCE Increase in enrollment due to change in default schedule. Decrease in impacts due to more realistic assumptions about impacts for small and medium customers SDG&E Decrease in impacts almost entirely due to changes in the RA window – similar to PG&E above Applied Energy Group · www.appliedenergygroup.com 24

EX ANTE IMPACTS Enrollment and Impacts, Typical Event Day, Utility 1-in-2 Utility PG&E- PDP SCE - CPP SDG&E - CPP Statewide PY 2019 Enrollment 137,077 300,243 14,074 451,394 PY 2019 Load PY 2029 Impact Enrollment (MW) 9.7 222,272 26.8 370,542 3.3 13,281 39.8 606,094 PY 2029 Load Impact (MW) 21.4 29.2 4.5 55.1 Drivers PG&E forecasts increased participation and impacts as default schedule resumes in 2020. SCE enrollments and impacts make an initial jump in 2019 with default, then grow steadily over time with population SDG&E enrollments actually decrease over time as medium customers opt out of the program. Impacts on the other hand increase slightly as large customers join the program. Applied Energy Group · www.appliedenergygroup.com 25

KEY FINDINGS Ex Post Analysis – Typical Event Day Impacts by Utility Utility PG&E SCE SDG& E State wide Size Large Medium Small Statewi de # Enroll ed 154,7 31 3,016 14,06 5 171,8 11 Load % Ref. Impac Load Load t Impac (MW) (MW) t Event Temp 1,439 28.8 2.0% 93.1 630 14.5 2.3% 89.4 786 8.8 1.1% 88.4 2,855 52.0 1.8% 90.3 Overall state level reduction of 52 MW PG&E contributes 55% of impacts Per participant percentage impacts are low across all three utilities 1-2% Impacts by Size # Enroll ed 5,174 47,527 119,11 0 171,81 1 Load Impac t (MW) 1,377 45.0 1,233 7.1 Ref. Load (MW) % Eve Load nt Impa Tem ct p 3.3% 90.5 0.6% 91.3 243.9 (0.1) 0.0% 93.0 2,855 52.0 1.8% 91.6 Applied Energy Group · www.appliedenergygroup.com Large customers contribute more than 86% of the impacts but make up only 3% of the participants Small customers essentially contribute zero 26

KEY FINDINGS Ex Post Analysis (Cont.) Notification is critical to improving participant response They can’t respond to an event if they don’t know about it Additional support and communication around events improves response further PG&E’s enhanced communication customers that receive post event feedback performed better than other customers across the board Aggregate Type No Communication Communication # Enrolled (MW) % Load Impact Avg. Event Temp. Ref. Load Load Impact 26,614 762 4.4 0.6% 90.3 145,197 2,093 47.7 2.3% 91.0 Applied Energy Group · www.appliedenergygroup.com 27

KEY FINDINGS Ex Ante Analysis Despite increased enrollment from additional defaults forecasted impacts dropped dramatically from 121 MW to 40 MW New RA window only includes 2 program operating hours (PG&E and SDG&E) while the other three hours are post event hours Updated assumptions about impacts for SCE’s small and medium default customers resulted in much smaller impacts Assumptions were based on PG&E’s experience which showed that the defaulted small and medium participants had low impacts Applied Energy Group · www.appliedenergygroup.com 28

PROJECT CONTRIBUTORS IOU Contributors Gil Wong, PG&E Overall Project Manager [email protected] Lizzette Garcia-Rodriguez, SDG&E SDG&E Project Manager [email protected] m AEG Contributors Kelly Marrin Project Director [email protected] Katie Chiccarelli Project Manager [email protected] Abigail Nguyen Analysis Lead Edward Lovelace, SCE SCE Project Manager [email protected] [email protected] Anthony Duer Senior Analyst [email protected] Applied Energy Group · www.appliedenergygroup.com 29