Queue-cohort survival

Kaplan–Meier estimate of the share of each cluster's projects not yet withdrawn, by months since queue date. Event = withdrawal. Active projects are censored at the CAISO report run date (2026-10-12); completed projects are censored at their on-line date — completion is success, not an event. S(t) is reported while at least 5 projects remain at risk. Withdrawal is not failure and the files carry no cause.

Summary

s12…s60 = S(t) at 12…60 months; _mw = each project weighted by net MW. n/a = not observed that long.

cohort	n	events	censored	mw_total	follow_up_months	median_survival_months	s12	s24	s36	s48	s60	s12_mw	s24_mw	s36_mw	s48_mw	s60_mw
C10	89	66	23	15,984	113	12	0.45	0.45	0.33	0.29	0.28	0.60	0.60	0.49	0.44	0.43
C11	94	67	27	22,403	102	12	0.38	0.37	0.35	0.34	0.33	0.32	0.31	0.28	0.27	0.27
C12	136	99	37	42,108	90	12	0.38	0.38	0.35	0.34	0.32	0.38	0.38	0.36	0.35	0.34
C13	147	108	39	42,129	78	13	0.51	0.44	0.43	0.37	0.33	0.51	0.40	0.40	0.33	0.29
C14	359	255	104	105,452	66	44	0.94	0.57	0.56	0.48	0.35	0.95	0.62	0.62	0.54	0.40
C15	170	84	86	59,013	17	not reached	0.64	n/a	n/a	n/a	n/a	0.63	n/a	n/a	n/a	n/a
C13-C15 standalone storage	352	260	92	92,296	78	35	0.78	0.51	0.50	0.42	0.29	0.78	0.50	0.49	0.43	0.30
C13-C15 solar+storage	252	149	103	79,959	78	35	0.77	0.51	0.49	0.46	0.41	0.77	0.54	0.53	0.50	0.46
C13-C15 other	72	38	34	34,340	78	36	0.74	0.52	0.50	0.37	0.26	0.74	0.53	0.53	0.35	0.19

Findings

Among cluster cohorts with at least 24 months of follow-up, C11 shows the steepest 24-month attrition: 63% of its 94 projects had withdrawn by month 24, against 43% for C14, the shallowest. Cluster 15 has 17 months of follow-up so far; at month 12 its unweighted S is 0.64 and its MW-weighted S is 0.63, versus S(12) of 0.94 for C14 and 0.51 for C13 at the same age. Median time to withdrawal is reached for C10 at month 12, C11 at month 12, C12 at month 12, C13 at month 13, C14 at month 44; C15 has not yet crossed S = 0.5 within its follow-up. For C14 the MW-weighted S(24) of 0.62 is at or above the unweighted 0.57, so the projects that withdrew in the first two years were on average smaller than those that stayed. Within C13–C15 by technology, standalone storage has the lowest S(24) at 0.51 and other the highest at 0.52 (n = 352 and 72). Across the six cluster cohorts 995 projects enter the analysis, 679 withdrawals are events, 316 rows are censored (active at the report run date or completed at their on-line date — completion is success, not an event), and 0 rows were excluded for missing or inconsistent dates. Attrition compares regimes, not only nodes. C10–C13 entered under the standard cluster process; C14 entered in April 2021 at 2.4x the prior year's volume under special procedures that made Phase I cost estimates advisory and refunded deposits in full on early withdrawal when Phase II costs came in 25 % or more above Phase I; C15 is the first cohort scored and capped by zone under the 2023 IPE, and its queue date (2025-02-12) is the post-scoring date, so the 541 → 170 intake cut happened before its month 0. A flatter C14 or C15 curve is partly the rule change, not only the projects or the nodes.

Process regime by cohort

Read the curves against this table. A cohort that had to score, pay more, or prove site control to be in the queue at all will withdraw less for that reason alone.

cohort	queue window	process the cohort entered under
C10	2017-05	standard cluster study process
C11	2018-04	standard cluster study process
C12	2019-04	standard cluster study process
C13	2020-04	standard cluster study process
C14	2021-04	373 requests / ~150 GW — 2.4x the prior year. Special FERC-approved procedures (Sept 2021): Phase I cost estimates advisory only; full deposit refund on early withdrawal if Phase II costs exceeded Phase I by 25 % or more; Phase I results delayed to Sept 2022.
C15	2025-02	First cluster scored under the 2023 IPE (commercial interest 30 %, viability 35 %, system need 35 %) and capped at 150 % of available capacity per constraint. 541 requests / 347 GW filed in the April 2023 window; 255 resubmitted Oct–Dec 2024; 145 / 68 GW proceeded to study (CAISO, June 2025). The queue_date in the file is 2025-02-12, so month 0 here is AFTER that cut: this cohort is already a filtered survivor set and its early months are not comparable to C10–C14's. (The file holds 170 requests; CAISO's June 2025 summary counted 145 proceeding to study — the difference is requests that withdrew during validation and are in the file as withdrawn.)

Definitions: DATA.md (survival_by_cluster.csv, survival_summary.csv).

No figure on this site states a cause. CAISO files carry no withdrawal reason beyond "IC Request".

Screening tool, not engineering advice. Built from public filings by an engineer who is not a licensed PE. Verify every figure against the source filing named in its row before you rely on it for any decision. No warranty of any kind. Not affiliated with CAISO, EIA, or any utility. No engineering services are offered.