A Grid Inno product

gpf

gpf is a steady-state power system tool: interconnection screens, contingency analysis, optimal power flow, and transfer capability on one solver.

Read the docs Install

Single-case solve time

Fastest on every case, from 2,000 to 82,000 buses.¹

Power-flow solve time by case and solver, with each solver's time as a multiple of gpf's.
Solver	Solve time	Relative to gpf
ACTIVSg 2000 2,000 buses
gpf	3.2 ms
Other commercial tool	19.6 ms	6.1×
MATPOWER	24.3 ms	7.6×
PEGASE 9241 9,241 buses
gpf	15.4 ms
Other commercial tool	40.3 ms	2.6×
MATPOWER	177 ms	11.5×
ACTIVSg 10k 10,000 buses
gpf	13.3 ms
Other commercial tool	32.8 ms	2.5×
MATPOWER	138 ms	10.4×
ACTIVSg 25k 25,000 buses
gpf	43.8 ms
Other commercial tool	76.7 ms	1.8×
MATPOWER	382 ms	8.7×
ACTIVSg 70k 70,000 buses
gpf	265 ms
Other commercial tool	414 ms	1.6×
MATPOWER	2.94 s	11.1×
Synthetic USA 82,000 buses
gpf	294 ms
Other commercial tool	418 ms	1.4×
MATPOWER	3.14 s	10.7×

AC contingency screening (N-1)

Faster on every case, from small systems to interconnection-sized ones.²

Power-flow solve time by case and solver, with each solver's time as a multiple of gpf's.
Solver	Solve time	Relative to gpf
ACTIVSg 2000 2,000 buses, 3,206 outages
gpf	955 ms
Other commercial tool	1.34 s	1.4×
PEGASE 9241 9,241 buses, 16,049 outages
gpf	29 s
Other commercial tool	47.2 s	1.6×
ACTIVSg 10k 10,000 buses, 12,706 outages
gpf	21.9 s
Other commercial tool	31.9 s	1.5×
ACTIVSg 25k 25,000 buses, 32,229 outages
gpf	2.62 min
Other commercial tool	6.87 min	2.6×
ACTIVSg 70k 70,000 buses, 88,207 outages
gpf	32.4 min
Other commercial tool	63.4 min	2.0×

DC contingency screening

Every branch outage screened in seconds, even at interconnection scale.³

Power-flow solve time by case and solver, with each solver's time as a multiple of gpf's.
Solver	Solve time	Relative to gpf
ACTIVSg 2000 2,000 buses, 3,206 outages
gpf	62.3 ms
Other commercial tool	361 ms	5.8×
PEGASE 9241 9,241 buses, 16,049 outages
gpf	737 ms
Other commercial tool	20.3 s	27.5×
ACTIVSg 10k 10,000 buses, 12,706 outages
gpf	501 ms
Other commercial tool	7.68 s	15.3×
ACTIVSg 25k 25,000 buses, 32,229 outages
gpf	2.82 s
Other commercial tool	out of memory
ACTIVSg 70k 70,000 buses, 88,207 outages
gpf	31.3 s
Other commercial tool	out of memory

An agent runs the whole study

gpf ships an MCP server, so an AI agent can load a case, screen an interconnection, sweep N-1, and hand back an auditable answer. The agent orchestrates; the solver computes, so the same study returns the same numbers every time.

In wecc_10039.raw, screen a 200 MW solar + 80 MW battery hybrid at the Maple Creek 230 kV POI. Run N-1 and report the worst line loadings.

agent tool glob("cases/*.raw")

cases/wecc_10039.raw

mcp tool call load_case("cases/wecc_10039.raw") via gpf.load_case("cases/wecc_10039.raw")

10,039 buses · base case solved · 0.18 s

agent tool write("maple_creek_hybrid.toml")

base_case = "cases/wecc_10039.raw"

[[project]]
label = "maple_solar"
kind = "solar"
poi_bus = 45021
mw = 200.0
power_factor = 0.95

[[project]]
label = "maple_bess"
kind = "storage"
poi_bus = 45021
mw = 80.0
power_factor = 0.95

mcp tool call run_study("maple_creek_hybrid.toml", delta=true) via gpf.Study.from_toml("maple_creek_hybrid.toml").run()

1,912 N-1 contingencies swept · 2 violations added · 2.6 s

mcp tool call get_report(label="thermal") via gpf.scenario.report.project(result)

N-1 line violations attributed to the hybrid

branch	contingency	loading
Maple Creek–Deer Ford 230 kV	Deer Ford–Salt Flat 230 kV ckt 1 out	108.4%
Maple Creek–Bergen Tap 230 kV	Maple Creek–Deer Ford 230 kV out	103.1%

Base case is clean at the full 280 MW; both violations bind on the emergency rating.

mcp tool call export_bundle("maple_creek_hybrid.raw", "maple_creek_hybrid.zip") via gpf.export_bundle(case_path, out_zip)

✓ auditable bundle saved: tamper-evident ledger, every version reconstructs

What gpf does

Deep device and control models

Two- and three-winding transformers, LCC and VSC HVDC, FACTS, switched shunts, tap and phase-shifter controls, area interchange, and impedance-correction tables are represented and enforced during the solve.

Reliability studies

Batched N-1 contingency analysis with AC-verified corrective-action proposals, interconnection headroom at a point of interconnection, and pre-queue siting screens.

Markets and economics

DC optimal power flow with LMPs, preventive SCOPF verified against AC, multi-period security-constrained dispatch, transfer capability, FTR auction clearing and portfolio valuation, and PTDF/LODF/OTDF sensitivities.

Built for automation

A typed Python API, a scriptable CLI, and an MCP server, so the same studies run from a notebook, a pipeline, or an AI agent.

Pricing

gpf is available across four tiers that differ by supported network size, from a free tier to enterprise.

See pricing →