Six-Week BESS Commissioning: How Houston IV's 160 MW / 320 MWh Came Online as ERCOT Hit 91 GW

SMT Energy and FlexGen have put a 160 MW / 320 MWh battery energy storage system (BESS) into operation in Houston, Texas — and they did it in six weeks. According to the companies, the Houston IV project was commissioned in one-sixth of a year, "cutting the time of deployment by 3x from the original timeline," as reported by Energy-Storage.News. The announcement lands days after ERCOT set a new hourly peak load record of 91.1 GW on 22 July 2026.

Those two facts — a 320 MWh asset energizing in a month and a half, and the Texas grid touching 91.1 GW — are not coincidental. They describe the same structural reality: a rapidly growing, weather-stressed grid that now pays for storage capacity on the timescale of a single construction season.

The Houston IV project at a glance

Houston IV is a 160 MW / 320 MWh lithium-iron-phosphate (LFP) grid-scale BESS located in the Houston area, interconnected to the ERCOT grid and jointly developed by SMT Energy (developer/owner) and FlexGen (system integrator and software provider). Engineering, procurement and construction was handled by Irby Construction Company. The project joins more than a dozen SMT–FlexGen assets built to serve ERCOT — the pair's twelfth-plus such deployment.

The economics behind Houston IV were set in motion well before commissioning. SMT secured US$135 million of funding for the project in February 2025, and broke ground with utility CenterPoint Energy in May. That chronology matters: the financing and interconnection groundwork was done up to 18 months ahead, which is precisely what allowed the physical build and commissioning phase to be compressed to six weeks. Fast commissioning is not improvised — it is planned.

What "six-week commissioning" actually compresses

The phrase "three times faster than planned" is worth unpacking, because it says something concrete about where BESS schedule risk now lives. A 320 MWh system at 160 MW — 2-hour duration — is modular by design. In round terms, that is on the order of 80–100 containerized DC blocks (4–5 MWh each depending on the cell/module generation), tied to a fleet of power conversion systems and medium-voltage transformers.

On a conventional schedule, commissioning occupies a long tail of discrete tests, each with its own hold points:

• Mechanical completion and continuity — torque checks, DC bus integrity, insulation resistance on every string/rack.
• Subsystem functional tests — HVAC/climate control, fire detection and suppression, battery management system (BMS) communications.
• PCS energization and power-path verification — first energization, open-loop checks, auxiliary loads.
• Performance / capacity test — a full charge–discharge round trip to verify usable energy (typically against the AC-side MWh guarantee), round-trip efficiency (RTE), and auxiliary consumption.
• Grid-compliance and control validation — protection settings, ride-through behavior, plant controller / EMS functional verification with the utility.
• Site acceptance test (SAT) — often spanning several days, including 24-hour monitoring windows.

Compressing all of that into six weeks means the project ran tests in parallel rather than in series, used pre-commissioned factory block-level testing to shrink field work, and sequenced the grid-interconnection steps against a fixed utility window. FlexGen's role as integrator plus its software stack (the plant controller and energy management layer) is a key enabler — when the same vendor supplies and commissions the control layer, the commissioning test matrix has fewer hand-off interfaces to reconcile.

This is the practical frontier of BESS delivery in 2026. Balance-of-plant and interconnection queues — not battery supply — are the dominant sources of schedule delay, which is why a project that already secured its $135 million funding and its interconnection path can convert engineering-complete kit into an operating asset with unusual speed.

Why ERCOT's 91.1 GW peak changes storage economics

ERCOT is an energy-only market — there is no separate capacity payment to soak up fixed costs. Revenue comes from energy arbitrage, ancillary services (regulation up/down, responsive reserve, ECRS), and increasingly from fast frequency response and grid-support services. That structure makes duration, dispatch flexibility, and availability timing decisive.

The 22 July 2026 record of 91.1 GW, reported by the US Energy Information Administration, was set during a heat wave and was 6% above the prior record of 85.5 GW from August 2023. EIA noted the record hour was met with 48% natural gas and 32% solar. Two implications follow for storage:

• Solar carries more of the peak. With roughly a third of record-hour supply from solar, the evening ramp — when PV output collapses just as air-conditioning load persists past sunset — is where scarcity pricing concentrates. That is exactly the window a 2-hour, 320 MWh asset is designed to capture.
• The peak is growing faster than the fleet. A 6% step in the all-time peak over three years, in a market whose installed BESS capacity has grown several-fold in the same period, keeps the arbitrage spread between midday charging and evening discharging wide enough to clear project hurdles.

ERCOT reinforces this with market design. In May, ERCOT initiated a programme to encourage battery storage and other inverter-based resources (IBRs) to adopt grid-forming technology — a one-time advanced grid support incentive (NPRR-based) aimed at resources that would not otherwise be required to provide advanced grid support, with implementation expected within 18 months of award notification. Paying for grid-forming capability signals that the value of storage is shifting from pure energy shifting toward system-strength services that a plain grid-following inverter cannot provide.

Competitive context: a record Texas deployment wave

Houston IV is not an isolated milestone — it is one entry in a dense sequence of Texas grid-scale BESS announcements during Q3 2026, and any honest read of the market has to place it in that frame:

The pattern is clear: 2026 is the year Texas storage transitions from single-asset milestones to industrial-scale deployment tempo. The distinguishing variable in this cohort is no longer cell supply or chemistry — it is execution speed and market access. Houston IV's six-week commissioning is a claim about schedule, and schedule is where competitive advantage now concentrates.

Modeling the schedule: where simulation earns its keep

For engineers specifying assets like Houston IV, the six-week story has a direct modeling analogue. The questions a compressed schedule forces are all quantitative:

1. Guaranteed vs nameplate energy

A 320 MWh nameplate system does not deliver 320 MWh at the AC meter on day one, let alone in year ten. Usable energy at point of interconnection is derated by depth of discharge (DoD) limits, RTE, and auxiliary loads (thermal management typically dominates). A 2-hour asset with a 10% AC-side loss and 90% usable DoD will deliver meaningfully less than nameplate — and the performance test that closes commissioning is where that number gets pinned to the contract.

2. Degradation and augmentation over the contract life

LFP cells lose capacity with cycling and calendar time. The energy a system can actually arbitrage in year 8 is not the year-1 value, and the difference flows straight into revenue projections. This is why degradation must be modeled as a function of both throughput (cycles/day, C-rate) and age — the three-dimensional SOH/RTE surface Energy Optima uses (year × C-rate × cycles/day, with trilinear interpolation across 16,000+ data points), rather than a single flat annual percentage.

3. Dispatch optimization against the actual spread

Whether Houston IV earns its return depends on when it charges and discharges. A 2-hour asset must charge during the midday solar trough and discharge into the evening net-load ramp — and the width of that spread is a function of PV penetration and load, not a fixed price. Simulating dispatch against an 8,760-hour price/load series (rule-based, economic dispatch, or MILP hybrid) is the only way to size duration correctly. Choose the wrong duration for the spread, and neither the fastest commissioning nor the cheapest cells will save the IRR.

Engineering takeaways

Four points worth carrying into the next specification cycle:

• Schedule risk has moved upstream. Six-week commissioning is downstream of interdiction — funding, interconnection, and procurement are what make it possible. Compressed delivery is a reward for earlier rigor, not a substitute for it.
• Two-hour duration is the ERCOT arbitrage sweet spot — for now. As solar penetration grows and the evening ramp lengthens, the marginal value of duration rises. Expect the 2-hour standard to migrate toward 3–4 hours as the spread structure evolves.
• Grid-forming capability is becoming a paid product. ERCOT's advanced grid support incentive explicitly values inverter capability that a basic grid-following design doesn't provide. Specify it early; retrofitting controls after COD is far costlier.
• Model the asset, not the brochure. Nameplate MWh, nameplate MW, and nameplate RTE all get derated in the field. Build the simulation around AC-side usable energy, degradation, and actual dispatch — then hold the guarantee to that basis.

Houston IV is a useful marker precisely because it is unremarkable in technology — an LFP BESS, integrated and controlled by a vendor with a dozen similar ERCOT projects behind it — and remarkable only in tempo. In a market that set a 91.1 GW peak in July 2026, the ability to bring 320 MWh from ground-break to grid support inside six weeks is a capability, and the industry will be measured against it.

Sources