Behind-the-meter batteries for data centers solve a specific problem. AI load moves faster than rotating machinery can follow, and a battery energy storage system (BESS) responds in a fraction of a second. That makes the BESS the natural partner for gas turbines on an islanded or grid-light facility. It also has hard limits. A battery is an energy-limited device, and the value it adds depends on whether its controls agree with the turbines’ controls. SISO Engineering integrates BESS into hybrid power systems. Battery supply stays with the manufacturers, and this article explains where the seams sit.
What jobs does a behind-the-meter BESS do?
A BESS on a data center facility typically carries four roles. One battery may hold several of them, but each role sizes the system differently.
- Load smoothing. The battery absorbs fast load steps so that the turbine governors see a lower-bandwidth, flatter power demand. This is the role most specific to AI facilities, where training clusters can swing by large fractions of block load in well under a second.
- Bridging. The battery carries load during the gap between a disturbance and the moment another source is ready. That gap may be a generator start, a unit trip or a step the turbines cannot yet follow.
- Black start. The battery supplies the initial energy to restart generation with no external power. It moves the first turbine start from an outside dependency to an on-site capability.
- Peak support. The battery supplements generation during short demand peaks, which lets the turbine fleet be sized closer to average load than to the worst spike.
The roles compete for the same stored energy. A battery that is fully discharged bridging a trip has nothing left to smooth the next step. Someone has to decide, in writing and before commissioning, which role takes priority at which state of charge.
How do you size a battery: power or duration?
Two numbers define a BESS, and they answer different questions. Power capacity, in megawatts, sets how large a step the battery can absorb or deliver. Energy capacity, in megawatt-hours, sets how long it can hold that output. The U.S. Energy Information Administration notes that utility-scale batteries are designed to maximize either power or energy capacity depending on the application, and that many installations serve more than one use (EIA, Utility-scale battery storage uses).
For a data center, the sizing questions run in a predictable order.
- What is the largest credible load step, and how fast does it arrive? This sets power rating and inverter response requirements.
- How long must the battery hold while the turbines move? Turbine ramp rates set this duration, so it is a controls question as much as a battery one.
- What is the ride-through requirement for a unit trip? Bridging a trip until another unit picks up load asks for more energy than smoothing does.
- How many cycles per day will the battery see? Repetitive small steps wear a battery differently than an occasional large event, and the supplier’s warranty terms may reflect that.
A battery sized for smoothing is not automatically sized for black start, and the reverse is also true. The role that gets sized for is the role that works.
Why can’t batteries alone carry an AI facility?
Energy. A facility running at hundreds of megawatts around the clock consumes far more energy per day than any reasonable battery holds. A BESS moves energy in time; it does not create it. Something has to charge the battery, and on a behind-the-meter site that something is the generation fleet.
The reverse limit applies to turbines. Gas turbines carry a steady load well, but they follow fast repetitive steps poorly. Each step asks the fuel control, compressor and combustor to respond faster than their design envelope allows, and the thermal and mechanical consequences accumulate. Neither technology alone matches AI load. The pairing does, if it is controlled as one plant.
That condition is the hard part. For a conventional power plant, lost generation is a commercial problem. For a data center, loss of generation can become a mission-critical infrastructure event. Consider the risk stack when the condition is not met.
- Two control systems with overlapping authority. The BESS controller and the turbine controller may both respond to the same frequency or power error, and their responses can interact destructively.
- Mismatched response assumptions. The battery supplier assumes a turbine ramp rate. The turbine team assumes a battery response. Neither assumption is written where the other party can see it.
- Protection that trips each other. A protection setting valid for one device in isolation can open a breaker during a legitimate transient on the other.
- State-of-charge drift. Without a coordinating strategy, the battery may sit full or empty exactly when the plant needs it the other way.
The risk is rarely that one supplier is incompetent. It is that their assumptions do not align, and no one owns the seams.
What should a BESS specification say about controls?
Most battery specifications describe the cells, the inverters and the enclosure in detail. They say little about behavior at the plant boundary. A specification that supports integration should state at least these items.
- Response time and ramp limits, with the test conditions that produced them.
- Setpoint authority. Which controller commands active and reactive power, and which one may override it.
- State-of-charge windows for each role, with the reserve held back for black start.
- Protection settings and ride-through behavior, shared with the turbine and electrical teams before studies are frozen.
- Data points exposed to the plant control system and to monitoring, including alarm and event history.
Each item is a place where two suppliers’ assumptions can quietly diverge. Writing them down early costs a meeting. Discovering them at commissioning costs schedule.
Where is the integration seam, and what is SISO Engineering’s role?
SISO Engineering’s background is gas turbine controls and automation: GE Vernova Mark VIe, ControlST and multi-OEM platforms. The company is headquartered in Houston and works across North America, South America, and Europe, wherever the equipment runs.
The offer starts with SnapShot, SISO Engineering’s fleet intelligence layer. On a hybrid facility, the turbines, the BESS, and the balance of plant often come from different original equipment manufacturers (OEMs) on different control platforms. SnapShot provides asset-level intelligence, including trends, alarms, and deviations for the battery and each turbine. It adds fleet-level visibility and a cross-OEM view of how the pair performs together. That visibility helps teams find developing deviations and recurring alarm conditions earlier, which supports mission-critical reliability.
Plant control and asset dispatch stay with your distributed control system (DCS) and energy management system (EMS). Full plant optimization remains with your control and operations teams, supported by SnapShot’s insights. Controls run the plant. SnapShot helps you understand it.
Integration services are how SISO Engineering gets into a project. That work includes integrating the interface between the BESS and turbine controllers, control narratives for the transitions, and commissioning scenarios that test the pair together. The OEMs commission their own equipment, and SISO Engineering works alongside the engineering, procurement, and construction (EPC) contractor, the owner’s engineer, the utility, and the BESS supplier. SISO Engineering helps integrate the controls and data across the medium-voltage UPS, STATCOM, and BESS equipment, and analyzes the resulting system behavior. The hybrid power systems page describes the service, and the behind-the-meter data center power page covers how it fits the wider facility architecture.
Has this kind of integration worked in the field?
A relevant proof point comes from outside the data center world, and it is labeled that way. At the Sentinel Energy Center in Southern California, SISO Engineering worked on a battery-supported black start system for a gas turbine facility. The project received the ENR West 2024 Best Energy/Industrial Project award. SISO Engineering’s controls work included control narratives for the battery, turbine and balance-of-plant interaction, operator displays, and coordination logic with safeguards for critical transitions. The system has been activated twice after utility disturbances, and the facility restored generation independently both times. SISO Engineering’s six-year engagement with the site’s control systems informed the design (read the case study).
This is a black start application on a utility-connected plant, not a data center facility. What it shows is narrower and still useful. Battery and turbine controls can be integrated so the pair behaves as one system under real disturbances, not just test scenarios. SISO Engineering’s scope was the controls integration between the systems. The battery and the turbines came from their own suppliers.
Ready to test your battery and turbine assumptions together?
Before the critical decisions are locked in, bring your single-line diagram and your BESS specification to a conversation with someone who has integrated these systems. For a hybrid plant, the questions are the ones above: which role the battery holds at each state of charge, who owns each setpoint, and whether the protection schemes agree.
SISO Engineering’s goal is to strengthen the work of every partner on your project, making the entire power system more reliable, understandable, and operable.




