01
Data and model
Report §2–3: HIFLD transmission lines, EIA generator and demand data, the utilities' Ten-Year Site Plans and FEMA's National Risk Index. The additional analyses add EIA-923 and EIA-930 hourly data and the FRCC 2026 plan.
UF / EDF report · March 2026 · interactive edition with additional analyses
Congestion is, and will continue to be, most prevalent in northern Florida, the western Peninsula and south Florida, and south Florida is the biggest regional driver of system operating cost. Transmission projects and grid-enhancing technologies can provide significant economic benefit to grid operators and ratepayers.
The findings, recommendations and the figures above are the submitted report's own, quoted with the section
each comes from. The maps and the additional analyses come from a full-year, hour-by-hour model of the same
county network, run on representations of the network and its
dispatch that all reproduce 2023 (gas generation within
of EIA-923). Every number in them is generated by one command from public data and stamped with commit
.
The submitted report · Executive Summary and Results
Condensed from the report, with its figures and the section each comes from. Each one links to its step on the map.
Beyond the report
Scenarios and deeper studies run since the report, several of them from its Future Work: a full-year, hour-by-hour model checked against 2023, several representations of the network, gas supply, storm replays, large-load hosting and grid-enhancing technologies.
01
Report §2–3: HIFLD transmission lines, EIA generator and demand data, the utilities' Ten-Year Site Plans and FEMA's National Risk Index. The additional analyses add EIA-923 and EIA-930 hourly data and the FRCC 2026 plan.
Report §2.1 · additional analysis: the county network
Hover a line to see its substations, owner and the two counties it connects.
These are the HIFLD transmission lines whose two ends lie in different counties, from 69 kV to 500 kV. Line width is each line's thermal rating (a voltage-class estimate). Turn on "Lines inside counties" to see the lines with both ends in one county: they are real, but the county model does not represent them. The study concerns transfers between counties.
Report §3 Model · additional analysis: full-year validation
The model combines the lines between each pair of neighboring counties into one corridor with their combined rating. It keeps bulk-system circuits ( had no voltage in HIFLD and were identified from the voltages at their ends). Generators come from EIA-860, with plant heat rates and delivered fuel prices from EIA-923; hourly demand is EIA-930 adjusted demand. Over all 8,760 hours of 2023:
02
Report §4.1: today's system under normal conditions.
Report §4.1 Current state: base case
Additional analysis
The map shows the corridors the report names. The full-year model measures a different quantity, the hours a corridor is at its full rating (voltage-class ratings, 8,760 hours of 2023), and does not re-test the report's snapshot results; that layer is in the explorer below ("Hours at limit").
Additional analysis · transfers between utilities
Hover an interface to compare the model with EIA-930.
Each county is assigned to the utility serving most of its load, so a territory is a set of whole counties. The interfaces between territories are the corridors that cross their boundaries; line width is the interface capacity (the sum of its corridors' ratings) and color its average loading. Arrows show the model's average 2023 flow. EIA-930 reports the interchange between each pair of connected utilities, the only public data on flows: with statewide pooled dispatch the model's flow direction matches the reported one on of interfaces, and with utility self-dispatch on . Flow magnitudes differ more, and hourly correlations are low: the county model has no utility boundaries inside a county and does not model bilateral schedules. reported ties, including Gainesville's and Homestead's, lie inside counties served mostly by another utility and are not represented.
03
Report §4.4–4.5. Additional analysis: the FRCC 2026 plan, and summer peak, solved for every hour.
Report §4.4–4.5 2035 outlook
Additional analysis
Blue counties curtail solar because the corridors out of them are at their limit; red counties call interruptible load or have unserved energy. With unlimited transfers the planned fleet serves every hour ( unserved). Across the credible representations interruptible load is – a year. The map shows the base network.
Additional analysis · 2035 summer peak day
Faint line: corridor capacity. Colored line: the flow, colored by loading. Use the hour slider.
2023 weather, base network. At 1 p.m. the corridors out of Hendry, DeSoto and Baker are at their limit and their solar is curtailed (blue), while the southwest coast and Southeast Florida import as much as the paths allow. Across network representations Broward's shortfall over the year is –, and none in of (); facilities inside a county, such as FPL's Andytown–Oasis project, are not represented. Batteries charge at midday and discharge in the evening.
Report §4.5 Heavily loaded corridors
Additional analysis
We ran representations of the network and its dispatch, varying voltage-layer coupling, ratings, the 69 kV layer, normally-open ties and how utilities dispatch. Two of them () shed load in 2023, which did not happen, so 2023 rules them out. A corridor counts as a robust need only if it binds at least 500 hours a year in every surviving representation.
Both are solar-export constraints on kV lines: at least
of their binding hours fall in daylight
solar hours, with power flowing toward . Their ratings are
voltage-class estimates; the owners' actual ratings are the first thing to check (every binding circuit is
listed in binding_elements.csv). Add 3 GW of metro data centers and the list grows to include
. Double the planned solar and storage and it adds northern paths,
including the ties to Southern Company: .
FPL's Andytown–Oasis 500 kV project, approved by the Florida PSC in May 2026 (Docket 20260020-EI), is in service in every 2035 case. Most of its route lies inside Miami-Dade, below county resolution, so it is not evaluated here.
04
Report §4.2 and §4.5: line, generator and county outages, and 100 MW load spikes. Additional analysis: every single-circuit outage at three stress hours, and the new load each county can take.
Report §4.2, §4.5 Contingency analysis
Additional analysis
Each bulk circuit was removed in turn at the summer peak, a design-cold winter morning and the solar-heavy midday, and the system re-dispatched within emergency ratings, in each of the network representations (the screen re-dispatches the whole fleet, so utility self-dispatch is not a separate case). In 2035, outages on the 500 kV paths into Southeast Florida (highlighted) cause load loss in Broward (red), each in at most of the representations, so their size depends on how the network is represented; the utilities' detailed planning models can size them.
Report §4.5 Load spike
Additional analysis: a stricter test, firm load with any circuit out and no interruptible load
Each county is shaded by its hosting capability: the largest new firm load it can take at the 2035 summer peak with any single circuit out, without shedding existing customers or calling interruptible load, in the worst network representation (hover a county for the best). It answers a different question from the report's 100 MW cost test. In Southeast and Southwest Florida it ranges from 0 to across representations (3,000 means 3,000 or more). Placing the same 3 GW of data centers at the six strongest nodes instead of six metro counties avoids – of unserved energy and – of interruptible load a year, the multi-load case the report flags. For new generation at a sunny midday, can each take at most more.
05
Report §4.2 and §4.5 Risk Score; §5 Future Work. Additional analysis: a design-cold week, a gas-supply disruption and six historical storms.
Additional analysis · the report's future work (§5): gas supply
Design-cold week: the , peaking at , is served in every credible representation (unserved energy at most , interruptible load at most ). A Southeast-wide cold snap strains Southern Company too: with imports held to the firm contract block, the week has unserved with unlimited transfers and at most unserved and of interruptible load on the networks.
Gas-supply curtailment: half of the FGT-only, gas-only capacity is lost (shaded counties), and dual-fuel units burn oil for the lost gas, with no limit on oil inventory. With unlimited transfers the summer peak week has unserved; with no oil backup either, (cold week: ). On the modeled networks the summer week has – unserved, all in Broward, but – of it occurs with full gas supply: it reflects south Florida's import limits, where the report finds contingencies cost the most, and its size depends on how the network is represented. Without oil backup every network representation falls short in the summer week: –, mostly in Broward, whose large gas plants are listed FGT-only and switch to oil when gas is short.
The map draws EIA's pipeline routes and every gas plant with the pipelines EIA-860 lists for it (hover a plant). EIA-860's coded pipeline fields never name Sabal Trail or Florida Southeast Connection, and they list FGT alone at FPL's Martin and Manatee plants, which Gulfstream also serves. Keeping the plants with a documented second pipeline on gas ( that cannot switch to oil), the summer week has – unserved, practically unchanged: Martin and Manatee are outside the Broward pocket.
Report §4.2, §4.5 Risk Score
Additional analysis: six historical storms replayed
Irma, Michael, Ian, Idalia, Helene and Milton each failed lines in proportion to the wind they brought. The map draws each storm's track, colors the corridors it put at risk by how often they failed in 300 draws, and shades counties by the load cut off the bulk system. Ian does the most damage. Averaged over those eight seasons, the transmission-caused shortfall is about a year, but between and if line fragility is 15% better or worse. Pick a storm:
06
Report §4.3 and §4.6: line expansion, dynamic line rating and new connections. Additional analysis: candidates, each valued in full-year combinations of scenario and network representation.
Report §4.6 2035 solution analysis
Additional analysis: grid-enhancing technologies, full-year evaluation
Every candidate targets a corridor that binds in the full-year runs: advanced reconductoring, dynamic line ratings and power-flow control (technologies the report lists as future work), new 230 kV circuits, more Florida–Southern import capability, and the utility-filed projects. Each is valued over full 2035 years in every credible network representation of the utility plans and in the other 2035 scenarios on the base network: Order No. 1920 benefits 2, 3, 4 and 7 in dollars over 20 years, benefits 1, 5 and 6 as indicators. Options are ranked by their worst case across those evaluations. This valuation (full years, discounted at 7%, worst of 10 evaluations) differs from the report's and is not comparable with its 20-year net benefits. keep B/C above 1 in all of them; on the map they are the highlighted corridors.
The submitted report · Insights, Future Work and Conclusion
Condensed from the report; each links to the step that supports it.
Pick two cases and a layer, then drag the divider to compare. Hover a line or county to trace it; click to inspect.
Click a county or a line on the map.
Additional analysis · FRCC 2026 plan, published after the report
The report's 2035 outlook used the utilities' Ten-Year Site Plans: Florida Power & Light, Duke Energy Florida and Tampa Electric plan 23.2 GW of new solar and more than 9 GW of storage by the end of 2034 (§4.4). This chart tracks utility-scale capacity in operation (EIA-860) and in the newer FRCC 2026 plan, in which of the solar planned by 2035 has no public site yet; where it lands decides which corridors fill.
| County | MW |
|---|
| Quantity | Model | Reported | Error | Type |
|---|
Calibrated quantities were fitted to 2023 data; validated ones were not.
The only public flow data are each balancing authority's hourly net interchange (EIA-930). Least-cost pooled dispatch moves more power between Florida utilities than they actually do: Tampa Electric exports on average in the model against reported. Letting each utility serve its own load first cuts that to and the total error by , so that "utility self-dispatch" representation is one of the credible networks every robust finding must survive. Load is split among utilities within for FPL and at most for the five largest.
| Balancing authority | Reported | Model | Hourly correlation, all credible representations |
|---|
Network build: circuits modeled,
with voltage inferred from their end substations,
below 100 kV left out of the base network (kept in the
69 kV representation) and
excluded because no voltage could be determined. Every line's
fate is listed in network_log.csv.
The report solves average-load, summer-peak and winter-peak snapshots of a county-level production-cost model (§3). The additional analyses extend it to a DC optimal power flow over 67 county buses and a Southern Company bus, solved hour by hour for a full year with batteries co-optimized day by day (HiGHS). Unserved energy is priced at VOLL. Interruptible load and firm plus economic imports are explicit resources.
The report's 2035 outlook follows the utilities' Ten-Year Site Plans (§4.4). The additional analyses use the newer FRCC 2026 Load & Resource Plan (, ) and its listed additions and retirements. High Load adds 3 GW of data centers in metro counties. Accelerated Clean doubles the planned solar and storage and retires the remaining coal.
Constraints inside a county are invisible. Ratings are voltage-class estimates. Hurricane fragility is a screening assumption (±15% shown). Results are screening-level and say where detailed studies should look; they do not replace them.
Generated · commit
· code .
Sources: HIFLD, EIA-860/923/930, EIA natural gas pipeline routes (U.S. Energy Atlas), FRCC 2026 Load & Resource
Plan, FPL 2026 Ten Year Site Plan, NREL SLOPE, NOAA HURDAT2, MISO MTEP24 cost guide, DOE DLR report.