PG&E applies ICE 2 in 2027 GRC filing to disaggregate reliability values, urges system‑level reporting

Pacific Gas and Electric Company told the California Public Utilities Commission Safety Policy Division technical working group that it used ICE 2 to calculate separate residential and nonresidential dollar‑per‑customer‑minute‑interrupted (dollars‑per‑CMI) values and then applied those values within its tranche‑level risk models for the 2027 test‑year general rate case.

PG&E said it ran ICE 2 to derive an $0.08 per residential CMI figure and $23.11 per nonresidential CMI figure, then multiplied those per‑CMI figures by the share of residential and nonresidential customers at circuit and tranche levels to produce tranche‑level dollars‑per‑CMI inputs for its GRC reliability models. PG&E filed those parallel calculations on June 20 as ordered by an administrative law judge, the company said.

Elijah Devon, enterprise risk strategy, Pacific Gas and Electric Company, summarized the process and timing: "We did attempt to transition from ICE 1 to 2, and did achieve some results using the ICE 2 calculator," he said, while noting the team had limited time between ICE 2's release and GRC filing deadlines.

PG&E engineers and risk staff told the TWG they did not run ICE 2 independently for every small circuit segment because small sample sizes and missing granular input data — industry mix, precise income distributions, and event‑level parameters — can produce large swings in modeled values. The company said it aggregated customer counts by circuit segment into tranche‑level figures and used the system‑level ICE 2 outputs to weight tranche estimates.

Justin Sadler, who identified himself as working on PG&E's undergrounding program, said the change did not prompt immediate alterations to the utility's wildfire system‑hardening strategy: "This has not really changed our system hardening strategy," he said, explaining that project selection was driven by their Wildland Distribution Risk Model and a 1‑to‑N risk ranking used to nominate projects for the GRC.

PG&E and other stakeholders discussed specific policy consequences for SB 884 reporting and the draft SB 884 project‑list data template (Table 6). PG&E recommended that the reliability inputs a utility provides should correspond to the version of the ICE calculator the utility used, and that certain Table 6 items be reported at the system level rather than at highly granular levels. The company argued system‑level reporting reduces instability caused by thin samples and prevents local income distributions from producing counterintuitive valuation results that could devalue disadvantaged communities.

Stakeholders pressed PG&E on the equity tradeoffs. Vincent Lowe (PG&E) and other attendees asked whether income buckets should remain in the model; PG&E said it would supply system‑level percentage breakdowns of residential customers by income if required, but cautioned that using income buckets at very small geographic units can yield highly variable results. LBNL presenters acknowledged that the ICE 2 outputs smooth many localized effects because the published damage functions represent average conditions and interpolate between surveyed outage durations.

Why this matters: the dollars‑per‑CMI used in reliability valuation affect how utilities prioritize investments — particularly in wildfire‑prone High Fire Threat Districts (HFTD) covered by SB 884. PG&E's choice to apply system‑level ICE 2 outputs to tranche models, rather than to compute fully independent circuit‑level ICE 2 runs, affects both how project benefits are counted and how Table 6 in the SB 884 data template should be specified.

Next steps: the CPUC Safety Policy Division expects to finalize the SB 884 data template by 2025‑07‑24; stakeholders were invited to submit written comments. PG&E said it will continue experimenting with ICE 2 and refine inputs as phase 2 survey data (including California IOUs) are incorporated into the tool.