Data centers / Power generation control

On-site generation optimization.

A new turbine is roughly three years out. Until one arrives, the only megawatts you can add are the ones you recover from the plant you already have — and control optimization delivers them in weeks, with no capital upgrade to the generating assets.

  1. MW/minRamp rate gained so generation follows a load that steps rather than glides
  2. Heat rateLess fuel per megawatt delivered, without touching the turbines
  3. AvailabilityLatent control faults found before a real load profile finds them
  4. Permit headroomCompliance held through transients instead of bought with a derate
01 / What it costs you

Your generation sits inside the fault domain of your SLA.

A grid-connected plant leans on an infinite bus. An islanded plant leans on nothing — and the load it serves steps rather than ramps.

 What happens on siteWhat it actually costs
01
On site
Block load acceptance sags frequency and voltage
Cost
Underfrequency load shed on the IT floor — an outage caused by your own generation.
02
On site
Turbine hunting through fast load ramps
Cost
Degraded heat rate — fuel burned for megawatts you never delivered, on every ramp.
03
On site
Drum level swinging on a ramp or a feed pump transfer
Cost
The classic trip path — a unit down mid-run, at $200–300K per event in plants we have fixed.
04
On site
NOx or CO excursions during transients
Cost
Permit exceedance — and the derate you take to avoid it gives back megawatts you waited three years for.
05
On site
A commissioning punch-list carried into operations
Cost
The plant never reaches design performance, and nobody is given time to go back.
02 / What we optimize

The loops we take responsibility for.

Gas turbines, reciprocating engines, fuel cells and balance of plant — on whatever DCS or PLC the package arrives with. Select a control point.

 
Point 1 · Prime mover & load acceptance
Speed, droop & load acceptance

How the turbine answers a block load step — the behaviour nobody tests until the first real training run finds it.

Prime mover & load acceptance

The highest-risk group, and the one your IT load exercises hardest.

  • Speed, droop and isochronous control; load sharing and ramp-rate limiting across a multi-unit block
  • Block load acceptance — the step-response behaviour nobody tests until the first real training run finds it
  • Fuel valve characterization, IGV scheduling, and ambient-temperature compensation on available output
03 / The method

Ramp against what the plant can actually give you.

The instinct on a step load is to open everything and hope. We do the opposite.

On one plant, a fixed-rate heat-up routinely tripped the shared boilers feeding the rest of the site. We replaced it with multivariable control that fed live boiler capacity back into the ramp — as aggressive as the plant could support, backing off automatically when it could not.

Trips ended. Heat-up got faster. Coordinating demand against real-time capacity beat opening the valve and hoping — which is precisely the problem you are about to commission, with "turbine block" in place of "boiler."

The same discipline applies to emissions. A cascade strategy with layered limits and bumpless override lets the binding constraint take control cleanly, so the unit holds its permit through a transient instead of derating to feel safe.

Generation · block load step
 

Demand coordinated against live plant capacity: the block reaches the new load cleanly, as fast as it safely can.

04 / Proof

This is the plant we have been tuning for forty years.

Sixty-plus U.S. power plant installations. Client names withheld; full case studies on request.

30.3MW/min
from 22.8 · +90,000 MWh/yr · $670K/yr
575 MW 2×1 combined-cycle
California

Control changes alone made a legacy plant fast enough for a market it couldn't reach

Model predictive control on the duct burners and surrounding loops, with no capital upgrade to the generating assets. The station qualified for day-ahead, real-time dispatch and ancillary services it had been too slow to bid into. Delivered in eight weeks.

→ your siteBlock load acceptance. Ramp rate decides whether the plant holds when the load you serve steps rather than glides.
<1week on site
$200–300K per trip avoided
400+ MW unit
Major U.S. energy provider

The trip that only armed under conditions nobody had thought to test

Years of random startup trips, worked around by biasing the drum level setpoint rather than solved. Loop analysis across successful and failed startups isolated a misconfigured single-element controller that engaged only in a narrow band — and drove the drum straight into trip territory.

→ your siteEvery latent logic fault a behind-the-meter plant inherits at handover and does not discover until a real load profile finds it.
Permit held
through load swings, no derate
Gas turbine station
U.S. independent power producer

Holding a NOx permit through transients instead of derating to feel safe

A cascade emissions strategy on turbine water injection: feedforward on load, with NOx ppm, NOx lb/MWh and CO ppm as layered limits and bumpless override between them, so whichever constraint binds takes control without a bump at the handoff.

→ your siteThe permit conditions that decide whether a project gets built, and whether it keeps running at full output.
05 / How it works

Bring us in before commissioning, not after the first trip.

The cheapest place to fix a control strategy is on paper. The most expensive is at 3 a.m.

01on paper

Review

We read the control narrative, EPC controls scope, OEM package logic and your expected IT load profile, then hand back the loops that will bite you — ranked, before commissioning.

From $7,500 per site
02FAT → handover

Design & commission

Cascade emissions strategy, load-acceptance and ramp logic, multi-unit load sharing — corrected before it is built, witnessed at factory acceptance, then tuned across the whole operating envelope.

Through to handover
03ongoing

Monitor

Continuous loop scoring on the generation plant, so degradation shows up as a number on a dashboard rather than as a trip during a training run.

Per unit or block
06 / Request a consultation

Send us your control narrative.

Your EPC will hand you a plant that starts. We make sure it holds — through a 40 MW step, an ambient swing and a permit inspection. We are not the EPC and not the turbine OEM, which is exactly why the review is worth having.

Cooling closer to your critical path? Cooling control →
BTM Control Readiness Review
Duration
Two weeks
Inputs
Control narrative, EPC controls scope, OEM package logic, expected load profile
Stage
Design, FAT, commissioning, or an operating plant
Deliverable
Ranked list of the loops that will bite you, with the correction for each
Suggested price per site
Trend reviewRemote, using exported BMS trend data
$7,500
Core offerRemote, live data, 2 weeks, read-only
$15,000
On-site, generationOn site + live data, turbines, engines, fuel cells
$25,000–30,000+ travel and lodging