Illustration only, not data and not a forecast. Defaults are set so that the two pass rates multiply to about 19%, the share of 2000-2019 requests (by count) that Berkeley Lab's Queued Up 2025 edition shows operating at the end of 2024. The split between the two stages, the month split and the extra-withdrawal input are our assumptions. Replace them with your own.
Method: requests x first pass rate = signed agreements; agreements x second pass rate = projects in service; in-service month = request month + both stage durations + delay. The delay input moves dates for every project that survives. Only the extra-withdrawals input changes how many survive. Cohort-level averages: one real project either gets through or does not.