Land near a line is only useful if power can reach it in time. So next to the facts about a Texas site, Voltane shows an estimate of how long a new grid connection is likely to take. This page says what is estimated, how, and how well the estimate held up against real projects.
What is estimated
- Months from an interconnection request to a signed interconnection agreement.
- Months from an interconnection request to energization, the day the connection is switched on.
- The chance that a request filed now is energized by the end of a given year.
- The completion rate: the share of past projects in the group that reached operation. The rest withdrew.
Each estimate is a range, not a date: a low, a middle and a high value, taken from the lower quartile, the median and the upper quartile of past projects.
Estimates are made by group
Voltane does not model a single site's grid connection. It places the site in a group by its ERCOT zone, its voltage class and how crowded the interconnection queue is nearby, then reports what happened to past projects in that group. Every estimate names its group and the number of past projects behind it. A group with few past projects gives a weak estimate, and the count is shown so you can see that.
Facts and estimates are kept apart
Estimates are shown separately from sourced facts. A fact has a source and a date. An estimate is labelled "Estimate" and shows its range, its group and its method. The two never share a column.
The backtest
An estimate like this should be tested against what happened. We fitted the groups on older projects only, then compared what they predicted with newer projects the method had not seen. The result is printed here whether it flatters the method or not.
| Measure | Result |
|---|---|
| Projects in the test set | 135 |
| Median error | 11.0 months |
| Actual results inside the predicted range | 57% |
| Fitted on projects through | 2019-12-31 |
| Tested on projects from | 2020-01-01 |
| Data as of | 2026-08-31 |
| Source | Berkeley Lab "Queued Up" interconnection queue data file, 2026 edition (ERCOT requests through 2025-12-31, CC BY 4.0), compiled from ERCOT's monthly Generator Interconnection Status reports; ERCOT GIS report August 2026 (MIS report type 15933) |
Median error is the typical gap, in months, between the predicted middle value and what happened. The predicted range runs from the low to the high estimate. Data (JSON)
- Cohort hold-out. Groups and their quantiles are fitted on ERCOT generation requests filed 2015-01-01 to 2019-12-31 (outcomes as known at the as-of date), then compared with requests filed 2020-01-01 to 2021-12-31. For each test project that was energized, the error is its actual months from request to energization minus its group's predicted p50; coverage is the share of those projects whose actual months fall inside the group's p25 to p75 band (50% would be ideal). A group needs 30 energized training projects, else the zone or ERCOT-wide numbers are used.
| Group | Test projects | Predicted low | Predicted middle | Predicted high | Actual middle | Actual inside the range |
|---|---|---|---|---|---|---|
| North zone | 49 | 42.3 | 54.5 | 65.7 | 45.6 | 61% |
| South zone | 29 | 45.5 | 60.1 | 71.1 | 47.2 | 55% |
| ERCOT-wide | 17 | 39.0 | 52.2 | 66.4 | 49.0 | 71% |
| Coastal zone | 10 | 46.9 | 53.6 | 67.4 | 45.4 | 40% |
| West zone | 18 | 33.6 | 46.7 | 59.2 | 35.1 | 50% |
| West zone, 345 kV | 12 | 34.0 | 48.9 | 60.5 | 40.4 | 50% |
Months. Low, middle and high are the lower quartile, the median and the upper quartile. "Actual inside the range" is the share of the group's test projects that landed between the predicted low and high. Unknown means the file has no value. Data (JSON)
What this does not tell you
- Fitted on ERCOT generation interconnection history. Large-load requests follow a different process (and Texas SB 6 rules); no public site-level large-load history exists yet.
- Censoring: 282 of the 564 test requests were still in the queue at the as-of date. They are neither failures nor successes and are left out of the error and coverage numbers. The test projects that are counted are the ones that finished within 48 to 72 months, so the actual medians here run short and will rise as slower projects finish.
- Check without censoring: the model gave the 564 test requests a 16.7% average chance of energization within 48 months; 14.4% of them were. The estimate ran optimistic for those request years.
- The training projects' outcomes are as known today, not as known at the cut-off, so this is a hold-out by request year, not a forecast that was made in the past.
- Energization means ERCOT's energization or synchronization approval where the August 2026 GIS report lists it, and the operational date in the Berkeley Lab file for everything else; the operational date usually comes a few months later.
- The public history ends 2025-12-31 (Berkeley Lab file). Projects that left ERCOT's list between January and July 2026 are counted as still open at 2025-12-31 because the monthly reports for those months were not available to this run.
- Withdrawals before August 2018 carry no date in the public file; they count in completion rates and are left out of the crowding measure.
- Withdrawn projects are missing a county or voltage more often than energized ones in the public file; group completion rates are corrected for that by spreading the unlabelled projects in proportion to the labelled ones.
- Small samples: a group's numbers rest on a few dozen projects. Solar, wind, storage and gas are pooled.
Use it in practice: Texas solar site selection · battery storage screening · reading ERCOT grid data