Base build

Base build testingLB500LB600LB700ACLB 200Standard or solid-state switching

End-of-line

End-of-line testingLCLB 160LCLB 240ACLB 60

Applications

Cooling

Chillers and cooling plantStatic and dynamic load testing

Power

UPS systemsGeneratorsSwitchgear and power qualityProtection settings

Explained

Liquid coolingGPU load testingWater qualityLoop monitoringCommissioning

Equipment

Commissioning equipment

Manifold trolleyFlushing & refill cart

Tools

The full rangeSizing calculatorDatasheets

Services

Commissioning supportLoad bank hireHigh load simulationBespoke load banksSupply and installationMaintenance and repairDeployment and managementBuy or hireProjects

Company

AboutPartnersProjectsAcademyNewsEventsFAQ Speak to the team
FAQ

Every question, answered

233 answers on liquid-cooled load banks, GPU emulation, commissioning levels, sizing, hire and what it takes to prove an AI data hall. Search it.

233Answers
9Topics
Factory to ISTCommissioning covered
UKManufacturer, not a reseller

233 answers

Load bank basics

What the equipment is and what testing proves

11 answers

A load bank applies a controlled electrical load so the power and cooling chain can be proven before live IT equipment is installed. For AI data centres, PowerLoad end-of-line units go further and replicate true rack-level heat and electrical load, so switchgear, UPS, generators, CDUs and cooling are tested against how the GPUs will behave before they are installed.

More on this: Liquid-Cooled Load Banks for AI Data Centres

Usually. Not every component in a server is liquid cooled, and some of the heat still needs air cooling. The ACLB 60 replicates that heat at the rack.

Read the full answer: Chiller and Cooling Plant Testing for AI Data Centres

Applying a steady load to prove a data hall can carry its rated capacity. It is the traditional load test and a required one.

Read the full answer: Static and Dynamic Load Testing for AI Data Centres

Changing the load quickly, up and down, to show how the cooling and power respond. It uses solid-state switching, which steps the load far faster than contactor-based banks.

Read the full answer: Static and Dynamic Load Testing for AI Data Centres

A load bank is a device that applies a controlled, measurable electrical load to a power system so the system can be proven without connecting the real equipment. It turns the energy it draws into heat and rejects it, either into the air or into a coolant loop. In a data centre it stands in for the IT load, so generators, UPS, switchgear and cooling can be tested before a single server arrives.

More on this: Testing Liquid-Cooled AI Data Centres

Load bank testing means applying a known load to a power or cooling system and measuring how it responds. It proves rated capacity, exercises generators and UPS under real heat, checks that protection and controls behave, and produces the logged evidence a commissioning witness pack needs. In an AI hall the test is only worth the time if the load profile matches the way the racks will actually draw.

More on this: Applications: What You Test With PowerLoad

A liquid-cooled load bank rejects its heat into a fluid circuit rather than blowing hot air into the room. The circuit usually carries a water and glycol mix or a dielectric fluid, which is why the data centre industry says liquid cooled.

More on this: Read the full page

An air-cooled load bank puts its heat into the room, so the room has to be able to take it. A liquid-cooled load bank connects into the coolant circuit and rejects heat the way the racks will, through the CDU, the manifold and the secondary loop. An air-cooled unit alone leaves the cooling path of a liquid-cooled hall unproven, so the two are used for different parts of the job.

More on this: Read the full page

A static load bank applies a fixed load and holds it there. A dynamic load bank moves the load on a controlled profile: steps, ramps and repeated cycles. Static testing proves capacity. Dynamic testing proves behaviour. When the eventual load is AI compute that swings hard and fast, behaviour is the thing that has to be proven.

More on this: Static and Dynamic Load Testing for AI Data Centres

Fan heaters put heat into a room, which is enough for a basic thermal soak and nothing more. They do not present a controlled, measurable, repeatable electrical load, they cannot be connected into a coolant loop, and they produce no data you can hand to a commissioning authority. For anything that has to be witnessed and signed off they are not a substitute.

A load reactor adds inductive load, so the test draws current at a lagging power factor instead of a purely resistive one. It matters when a generator or UPS has to be proven at its rated kVA and power factor rather than just its kW. Whether you need it comes down to what the witness scripts call for, which is one of the things a survey settles.

AI and GPU testing

Transients, emulation and testing for AI density

32 answers

The ACLB 60 is a rack-format air heat load emulator in the end-of-line range. It sits at the rack and puts 60kW of heat into the air path so room cooling and containment can be proven. The ACLB 200 is a portable 200kW resistive load bank in the base build range, for proving 400V three-phase power systems with fine control from a 1kW step.

More on this: Air-Cooled Load Banks: ACLB 60 and ACLB 200

A liquid-cooled load bank that replicates how a GPU rack will behave, for heat and for electrical load, at the rack. The load moves quickly in both directions, so commissioning teams can see whether the CDU, pumps, valves, controls, cooling plant, UPS and generators respond as they should before the GPUs are installed.

Read the full answer: AI Liquid GPU Emulators for End-of-Line Testing

The LCLB 160 and LCLB 240 both support GB200 and GB300, and the LCLB 240 is Rubin ready. Give us your design and IT load and we will size the emulation to match.

More on this: AI Liquid GPU Emulators for End-of-Line Testing

A base build load bank gives a dependable, controllable heat load that proves capacity. A GPU emulator adds solid-state switching so the load moves the way real compute moves, up and down, and the whole chain has to respond.

More on this: AI Liquid GPU Emulators for End-of-Line Testing

Yes. The emulator sits at the rack position instead of at the building entry, so power and cooling are tested where the GPUs will run.

Read the full answer: AI Liquid GPU Emulators for End-of-Line Testing

Yes. The AI liquid GPU emulators are available to hire. Speak to the team about availability for your project.

More on this: AI Liquid GPU Emulators for End-of-Line Testing

AI has changed the behaviour of the load, not just its size. GPU racks swing from roughly 25 to 100 percent, up and down, and the whole power and cooling system has to respond each time. A flat load shows capacity but not that response, so PowerLoad tests how the system responds in both directions.

More on this: Liquid-Cooled Load Banks for AI Data Centres

Base build testing uses the LB500, LB600, LB700 and ACLB 200 to prove the cooling plant, CDU and risers at capacity. End-of-line testing uses the AI liquid GPU emulators to replicate the heat and electrical load of the racks, including transient electrical spikes, so you see how the system responds before the GPUs are installed. The commissioning equipment flushes, fills and distributes coolant first.

Read the full answer: Base Build and End-of-Line Load Banks

It needs to step the load fast with solid-state switching, drive dynamic pressure shifts through the cooling circuit and produce flow reactions that mirror a live environment, rather than applying a steady load and generating heat.

Read the full answer: AI Data Centre Commissioning: What Changed

Each LCLB 160 stands in for an individual rack load, positioned on the chilled water loop where that rack will sit, with CDUs at each end of the system.

Read the full answer: How GPU Emulator Units Sit on the Cooling Loop

No. A static load holds the infrastructure at a steady state and never triggers the failure modes that matter. AI workloads swing very quickly, and it is the rate of change that stresses UPS, generators and cooling.

Read the full answer: GPU Load Testing for AI Data Halls, Explained

Solid-state switching uses semiconductor devices rather than mechanical contactors to apply and remove load. With no moving parts it steps the load far faster than contactor-based banks, which allows the load bank to reproduce the ramps and drops of a real GPU rack.

Read the full answer: GPU Load Testing for AI Data Halls, Explained

A conventional load bank applies power and produces heat. A GPU emulator reproduces how that power and heat arrive and disappear, using fast switching to mirror the surge, dip and cycle pattern of a real accelerator under workload.

Read the full answer: GPU Load Testing for AI Data Halls, Explained

Yes, and it is the point at which testing is most useful. Equipment that emulates the electrical and thermal behaviour of GPU racks lets the cooling, flow and power infrastructure be stress tested months before any compute arrives.

Read the full answer: GPU Load Testing for AI Data Halls, Explained

It tests everything upstream of the rack under rapid load change: the UPS, the generators, protection and the cooling system. Those components respond to transitions rather than to steady draw.

Read the full answer: GPU Load Testing for AI Data Halls, Explained

By matching behaviour rather than just output. Most test equipment can generate heat, but a GPU does not sit at one power level or one temperature. It surges, dips and cycles constantly with the workload. Switching modules produce rapid, repeatable power transitions that mirror that electrical draw, and the design pushes heat into the cooling loop in the same dynamic pattern a real GPU cluster produces.

More on this: GPU Load Testing for AI Data Halls, Explained

By simulating the electrical and thermal behaviour of real GPU racks in advance, so every part of the infrastructure is stress tested before any hardware is delivered. You see how the cooling loops respond, how stable the flows are and whether the system copes with sudden power transitions. The result is walking into deployment day already knowing the environment works.

More on this: GPU Load Testing for AI Data Halls, Explained

There are two ways to switch load in a load bank, and only one of them can follow a GPU. Traditional resistive switching uses mechanical contactors that step load in fixed increments, which is fine for steady, predictable infrastructure. GPU transitions happen very quickly. Solid-state switching has no mechanical components and steps the load far faster than contactor-based banks, so it reproduces real GPU behaviour at full load and, critically, at low load where resistive systems lose resolution entirely.

More on this: GPU Load Testing for AI Data Halls, Explained

Commissioning for the AI era means proving infrastructure against load that moves rather than load that sits still. AI changed the behaviour of the load, not just its size, so the whole cooling ecosystem and the power chain have to respond when the load changes quickly, in both directions. That takes solid-state load steps, dynamic pressure shifts and flow reactions that mirror a live environment.

More on this: GPU Load Testing for AI Data Halls, Explained

Because a static load test passes a building that would fail in service. AI load patterns switch very quickly, and most traditional load banks cannot keep up. Testing properly means simulating fast transitions, sudden surges and precise load steps, and it does not just test the rack. It shows how the UPS, the generators and the cooling all respond under rapid load change, which is where the real failure modes sit.

More on this: GPU Load Testing for AI Data Halls, Explained

Multiple 160kW load bank units are distributed across the chilled water loop, with CDUs positioned at each end of the system. Each 160kW unit represents an individual rack load. Flexible hose connections allow rapid deployment and reconfiguration, and isolation valves at every connection point mean a single unit can be isolated or removed without affecting the rest of the system.

More on this: GPU Load Testing for AI Data Halls, Explained

Air cooling reaches its limit well below AI rack densities. Traditional halls were designed around 10 to 15kW racks, where air performed well, but high-density AI racks such as Blackwell draw around 120kW, which is far beyond what air can carry.

Read the full answer: Why AI Data Centres Moved to Liquid Cooling

Yes. The cooling system responds to heat rather than current, so the emulator rejects heat into the loop in the same changing pattern as a real GPU cluster, not as a steady thermal load.

Read the full answer: How a GPU Emulator Matches Real Rack Behaviour

Contactors step the load in set increments, much more slowly than solid-state switching. They suit steady load tests, but cannot reproduce the quick swings of a GPU workload.

More on this: Solid-State Switching in Load Banks Explained

How the cooling loops respond under real heat, how stable flows are in service, and whether the infrastructure copes with surges and sudden drops rather than only sustained load.

Read the full answer: How to Test a Data Hall Before the GPUs Arrive

Because flow and temperature are not uniform across a hall. A hall average can pass while one rack runs outside its limits.

Read the full answer: Chiller and Cooling Plant Testing for AI Data Centres

By applying the transient electrical load of the racks, repeatedly, and checking that protection stays closed on a surge and operates on a fault.

Read the full answer: Protection Settings Testing for AI Data Centres

Voltage, frequency and harmonic distortion on the distribution while the load changes, from the first base build load through to rack-level surges.

Read the full answer: Switchgear and Power Quality Testing for AI Data Centres

Yes. The end-of-line GPU emulators replicate the electrical load of the racks at each rack position, so the supply is tested months before any accelerator arrives.

Read the full answer: Switchgear and Power Quality Testing for AI Data Centres

Rated load and battery runtime on the base build, then repeated fast load steps at rack level, logged throughout, with the generator in circuit.

Read the full answer: UPS Load Testing for AI Data Centres

A steady load proves a hall can carry a number. It does not show whether the hall can handle the rate of change. AI racks move between near idle and full draw far faster than traditional IT load, and many failures come from that rate of change rather than the peak.

Read the full answer: High Load Simulation

With solid-state switching, which steps the load far faster than contactor-based banks. It is optional on the LB500 and LB600, built into the LB700 and the LCLB GPU emulators, and lets the load bank follow a profile rather than a staircase. Profiles can be set before you arrive and repeated identically for comparison.

Read the full answer: High Load Simulation

Liquid cooling and the loop

CDUs, manifolds, flow, fluid and heat rejection

52 answers

Yes. Not every component inside a liquid-cooled server is cooled by liquid, so the ACLB 60 runs alongside the LCLB 160 and LCLB 240 AI liquid GPU emulators and loads the air path and the coolant loop at the same time.

More on this: Air-Cooled Load Banks: ACLB 60 and ACLB 200

It splits one coolant feed into regulated outlets, so each circuit can be balanced independently. A digital flow meter and pressure sensor on every outlet measure flow and pressure drop, feeding data back over I/O Link to a BACnet interface that integrates with the load bank system.

Read the full answer: Commissioning Equipment for Liquid Cooling Loops

New pipework carries swarf, debris and trapped air that will foul cold plates and CDUs once the system is live. The refill cart circulates coolant through a 25 micron filter to remove that contamination and de-aerates the loop, so the load bank proves a clean, representative system rather than a dirty one.

Read the full answer: Commissioning Equipment for Liquid Cooling Loops

The cart is built around a 250L 316-grade stainless tank with vacuum relief, an auto air vent and a sight glass. A high-volume circulation pump and a dedicated fill pump with wand handle both the flushing and the filling stages.

More on this: Commissioning Equipment for Liquid Cooling Loops

A flushing and refill cart fills, cleans and drains a liquid cooling loop at controlled pressure. The PowerLoad cart is built around a 250L 316-grade stainless tank with dual pumps and a 25 micron filter. It circulates coolant to strip out swarf, weld debris and mill scale, vents trapped air, then charges the loop so the readings taken during load testing reflect the system rather than the dirt in it.

Read the full answer: Flushing and Refill Cart for Coolant Loops

In a direct-to-chip system the coolant passes within millimetres of the die through channels as fine as a human hair, so anything carried in the fluid has somewhere to lodge. Less than half a millimetre of scale on a heat transfer surface can cut heat transfer by around a third. Particulate in circulation also erodes pump internals and damages valve seats, so a loop that was never cleaned properly degrades in ways that only appear months into service as reduced flow and rising temperatures.

More on this: Flushing and Refill Cart for Coolant Loops

A new system is the dirtiest it will ever be on the day it is finished. Miles of freshly cut and welded pipework leave slag, filings and mill scale behind, and the moment the loop is filled that debris travels towards the pumps and the cold plates. Commissioning is the only window in which the loop can be flushed, drained, dried and charged properly, because once the racks are live you cannot drain it again.

Read the full answer: Flushing and Refill Cart for Coolant Loops

The cart is built around a 250L 316-grade stainless tank with vacuum relief, an auto air vent and a sight glass. A high-volume circulation pump handles the flushing stage and a dedicated fill pump with a wand handles filling and topping up.

Read the full answer: Flushing and Refill Cart for Coolant Loops

Yes. The same connection is used at the end of the job to flush and condition the system again, so the load banks are not left standing with used fluid in them and are clean and protected before the next commissioning programme.

More on this: Flushing and Refill Cart for Coolant Loops

Yes. The cart is available to hire alongside the load banks and the manifold trolley. Speak to the team about availability for your programme.

More on this: Flushing and Refill Cart for Coolant Loops

Yes. PowerLoad supplies liquid-cooled load banks and GPU emulators for purchase and for hire across the UK and Europe, alongside commissioning equipment: manifolds, flushing and refill carts, and remote monitoring.

More on this: Liquid-Cooled Load Banks for AI Data Centres

A liquid-cooled load bank rejects its heat into a coolant loop rather than into the air, so it connects directly to the data hall cooling infrastructure. In base build testing it proves the capacity of the chillers, CDU and risers while the hall is being built. The base build range runs from the 200kW ACLB 200 to the 700kW LB700.

Read the full answer: Liquid-Cooled Load Banks for Base Build Testing

Yes. The liquid-cooled load banks are available to hire. Speak to the team about availability for your project.

More on this: Liquid-Cooled Load Banks for Base Build Testing

In a sequence. Clean and prove the coolant loop first, then fill it and confirm flow and pressure at every rack, then apply electrical and heat load together, then change the load quickly, up and down. The base build range proves capacity. The end-of-line range then replicates the racks. The tests are run by the client's commissioning agent or accredited partners.

Read the full answer: Testing Liquid-Cooled AI Data Centres

Yes. A liquid-cooled load bank rejects its heat into the coolant loop, so it exercises the CDU, the manifold and the loop as the racks will, before any accelerator is on site.

Read the full answer: Testing Liquid-Cooled AI Data Centres

A coolant distribution unit is the interface between the facility water and the technology loop serving the racks. It has to follow the heat as it rises and as it falls, so it is tested with a load that changes.

Read the full answer: Testing Liquid-Cooled AI Data Centres

A new loop can carry manufacturing debris, and it ends up in the narrowest passages, which are the cold plates. Flushing and proving coolant quality before load is applied protects the IT equipment.

Read the full answer: Testing Liquid-Cooled AI Data Centres

A liquid-cooled load bank rejects heat into the coolant loop, which proves the CDU, the manifold and the loop. An air-cooled unit rejects heat into the room. Even with direct liquid cooling, some components in a server still need air cooling, so the end-of-line range includes the ACLB 60 to recreate that heat.

Read the full answer: Testing Liquid-Cooled AI Data Centres

Enough to reach the design condition. The liquid-cooled units range from 160kW to 700kW each, and up to 250 units run on one controller for hall-scale testing.

Read the full answer: Testing Liquid-Cooled AI Data Centres

Yes. The 700kW LB700 suits chiller and larger cooling plant testing as well as CDU and riser proving.

Read the full answer: Testing Liquid-Cooled AI Data Centres

No. Most of our work is liquid cooled, but air-cooled space still has to be tested and the ACLB 200 and ACLB 60 cover it.

Read the full answer: Services for Data Centre Commissioning Teams

Yes. In both the base build and end-of-line ranges, liquid-cooled or air-cooled, the units are plug and play.

Read the full answer: Networked Load Banks Across a Whole Data Hall

One cart services up to five LB500 units from a single connection point.

Read the full answer: Connecting a Flushing Cart to Load Banks

No. It runs the clean flushing cycle before operation and is used again after testing to flush and condition the system, which is what keeps the units protected between jobs.

Read the full answer: Connecting a Flushing Cart to Load Banks

It is how closely the equipment's heat output and thermal response match a real GPU environment, including how quickly the cooling circuit reacts when electrical load rises or falls. It is set by internal flow path design, heat exchange surfaces and sensor placement.

Read the full answer: Designing Load Banks for Thermal Accuracy

A steady electrical load produces a steady thermal state. Faults such as slow loop response, imbalance between circuits and inability to hold temperature only appear when conditions change, so a static test never triggers them.

Read the full answer: Designing Load Banks for Thermal Accuracy

Each circuit has its own isolation valve, accurate flow metering and quick-connect fittings, so it can be controlled, measured and reconfigured independently of the others.

Read the full answer: Fluid Distribution Points: Per-Circuit Control

It fills, flushes, filters and drains a liquid cooling circuit at controlled pressure and flow, removing trapped air, debris and contaminants so the loop is stable before load testing begins.

Read the full answer: How a Coolant Flushing and Refill Cart Works

Consistent pressure and flow are what actually carry debris out and clear trapped air. Flushing at uncontrolled conditions can leave pockets of air and deposits in place while appearing to have run correctly.

Read the full answer: How a Coolant Flushing and Refill Cart Works

Water carries far more heat per unit volume than air. It can also be delivered directly to the chip and controlled at rack level, which air cannot do at any speed.

Read the full answer: Why AI Data Centres Moved to Liquid Cooling

As soon as the main pipework, risers and cooling infrastructure can be run under genuine liquid-cooled conditions, which is typically months before any compute hardware is delivered. The value of the test comes from having time to act on what it finds.

Read the full answer: Why AI Data Centres Moved to Liquid Cooling

Because a cooling system that is not ready announces itself at the worst possible moment, when the real load arrives and the programme has no slack left. Testing early puts genuine liquid-cooled conditions through the main pipework, risers and cooling infrastructure months ahead of any hardware, so anything unbalanced, restricted or undersized surfaces while it is still straightforward and cheap to correct.

More on this: Why AI Data Centres Moved to Liquid Cooling

By designing the internal flow paths, heat exchange surfaces and sensor placement so the equipment's thermal response matches what a real GPU environment produces. When electrical load spikes or collapses, the cooling circuit has to react cleanly and consistently rather than lag behind. That fidelity is what exposes slow loops, imbalance between circuits and places where the system cannot hold temperature under dynamic conditions.

More on this: Why AI Data Centres Moved to Liquid Cooling

Because air cooling has a hard physical limit and AI racks are well past it. Air handled comfortably the 10 to 15kW racks it was designed around, reaches its limit well below AI rack densities, and a Blackwell rack draws around 120kW. Water carries far more heat per unit volume than air and can be delivered directly to the chip, which is why liquid is the baseline for AI density rather than an upgrade to consider.

More on this: Why AI Data Centres Moved to Liquid Cooling

Flow, temperature and pressure across every hose, in real time, so each circuit can be assessed individually rather than as part of a system average.

Read the full answer: Commissioning Manifolds and Loop Monitoring

Each outlet is a 1 to 2 inch hose pair with its own metering and control, all running from a single connection to the secondary fluid network.

More on this: Commissioning Manifolds and Loop Monitoring

Advanced fluid distribution points are where each circuit is individually controlled and measured. Every circuit gets its own isolation valve, accurate flow metering and quick-connect fittings, so the system can be configured or rebalanced without shutting everything down. They connect directly to the LB500 and give one point from which several load banks are managed.

More on this: Commissioning Manifolds and Loop Monitoring

Because at scale the hard part is not generating load, it is controlling and understanding what happens across every circuit. Most setups rely on guesswork: readings checked in different places, flows balanced by hand, and hope that nothing drifts as the test scales up. A bespoke manifold trolley puts flow, pressure and temperature for multiple circuits in one place, so lines can be isolated, balanced properly and problems spotted instantly.

More on this: Commissioning Manifolds and Loop Monitoring

The manifold runs 1 to 2 inch hose pairs from a single secondary fluid network. Each outlet carries an electronic flow meter reporting real-time flow rate and temperature back to the controller, and a target pressure drop can be set per valve. It is how each rack is fed and measured individually from one connection to the secondary loop.

More on this: Commissioning Manifolds and Loop Monitoring

Each outlet represents a provisioned server rack and has its own isolation valve, so every rack can be controlled on its own, with flow returning through the CDU.

More on this: Two 500kW Load Banks on One Manifold

Not properly. Once compute is installed and the system is live you cannot drain and flush the loop without taking the hall out of service, so the water quality you go live with is what you keep. The flush belongs before the racks arrive.

Read the full answer: Water Quality and Flushing in Liquid Cooling

Cold plates use micro-channels to move heat away from the die. Debris and scale narrow or block those channels, which reduces the heat the plate can carry away and causes the chip to throttle or trip out.

Read the full answer: Water Quality and Flushing in Liquid Cooling

Air displaces coolant, restricts flow and moves around the circuit under pressure. It produces unstable temperatures and inconsistent readings that are easily mistaken for equipment or control faults, so it wastes diagnostic time as well as degrading cooling.

Read the full answer: Water Quality and Flushing in Liquid Cooling

Before load testing begins. The point of flushing is to make the readings you take during commissioning trustworthy, so it has to happen ahead of the tests that depend on those readings.

Read the full answer: Water Quality and Flushing in Liquid Cooling

A flushing and refill cart fills, cleans and drains a liquid cooling loop at controlled pressure, using a precision flow meter and a valve layout an engineer can work quickly inside a live hall. It removes trapped air, debris and contaminants before load testing begins, so the flow and temperature readings taken during commissioning reflect the system rather than the dirt in it.

More on this: Water Quality and Flushing in Liquid Cooling

Because a dirty or air-filled circuit does not announce itself. It shows up later as unstable temperatures, restricted flow and readings that drift for no visible reason, and the commissioning team ends up chasing faults that are not really there. Starting from a flushed, balanced, air-free loop removes an entire category of phantom problems before the first load test runs.

More on this: Water Quality and Flushing in Liquid Cooling

In a direct-to-chip system the coolant runs through micro-channels inside the cold plates and through the CDU, so contamination always has somewhere to lodge. Blocked manifolds reduce flow and degrade thermal performance, and left unchecked debris erodes pumps and damages valve seats. Water quality is not a housekeeping detail in a liquid-cooled hall. It decides whether the cooling system performs.

More on this: Water Quality and Flushing in Liquid Cooling

It finds the same faults, but while correcting them is still possible. Imbalance, flow restriction and undersized infrastructure all need design changes or long-lead components, which will not fit the time left once hardware is installed.

Read the full answer: When Should Load Testing Start on a Data Hall?

It is offered as an upgrade, but most teams working at scale or against tight handover dates rely on it, because per-circuit data is what allows performance to be validated rather than assumed.

Read the full answer: Why Instrumented Commissioning Manifolds Matter

Yes. The liquid-cooled load banks reject heat into the coolant loop. The 700kW LB700 suits chiller and larger cooling plant testing where a heavier heat load is needed.

Read the full answer: Chiller and Cooling Plant Testing for AI Data Centres

A liquid-cooled load bank rejects its heat into the loop as the racks will, so the CDU, manifold and risers are exercised under real heat months before any accelerator arrives.

Read the full answer: Chiller and Cooling Plant Testing for AI Data Centres

Yes. Both ranges are available for hire: the LB500, LB600, LB700 and ACLB 200 for base build, and the LCLB 160, LCLB 240 and ACLB 60 for end-of-line, with the manifold trolleys and flushing and refill carts that go with them.

Read the full answer: Load Bank Hire for Data Centres, Base Build and End-of-Line

Sizing and specification

Working out what a hall actually needs

9 answers

Because AI has changed the behaviour of the load, not just its size. The cooling and power systems have to respond when the load changes quickly, and in both directions, when it rises and when it suddenly reduces.

Read the full answer: Applications: What You Test With PowerLoad

The LB700 is rated at 700kW and is the highest-capacity single unit. It runs on 5 inch flow and return, and optional solid-state switching gives 1kW switching for fine load resolution.

Read the full answer: Liquid-Cooled Load Banks for Base Build Testing

The same units cover a single rack, a full row or an entire data hall. Up to 250 units run on one controller, with live flow, temperature and pressure and exportable test data across the fleet. The GPU emulators are rack format, so they sit where the servers will go.

More on this: What Commissioning an AI Data Hall Involves

It helps most at scale, where manual balancing across many circuits becomes unreliable, but the same single point of control helps on smaller runs too.

Read the full answer: The Bespoke Commissioning Manifold Trolley

Start from the design load of the hall, not the number of racks. Work out the total kW the power and cooling chain has to prove, decide whether you are testing the whole hall at once or block by block, then divide by the rating of the units and add margin so you can hold full load with a unit out. Testing block by block needs fewer units but takes longer, which is a programme decision rather than a technical one.

More on this: Read the full page

The LCLB 160 and LCLB 240 support GB200 and GB300 racks, and the LCLB 240 is Rubin ready. Each unit is a liquid-cooled GPU emulator in rack format, with 160kW or 240kW of load, so a high-density AI rack can be replicated at rack level with the transient behaviour to match.

More on this: Read the full page

Both, at different stages. Room-level capacity proves the plant: generators, UPS, chillers and the primary loop carrying design load. Rack-level testing proves the last few metres, where the CDU, the manifold and the secondary circuit either deliver at the rack or do not. A hall that passes at room level can still fail at a rack, which is the failure that costs the most to find late.

More on this: Chiller and Cooling Plant Testing for AI Data Centres

Partial load proves the systems run. Full design load proves they run at the point they were sold on, which is the only number anyone will hold you to later. Partial testing also hides the failures that only appear at the top of the range, where cooling margin disappears and protection sits closest to its trip settings.

More on this: Applications: What You Test With PowerLoad

As a rough shape, a 1MW block needs 1MW of load bank capacity to prove at once, so two 500kW liquid-cooled units, plus whatever air-cooled capacity the base build side needs. Most programmes carry a little more than that so a unit can drop out without stopping the test. The exact count depends on the split between liquid and air, and on whether you are proving the hall in one go or in blocks.

More on this: Read the full page

The commissioning process

Test stages, IST, test scripts and evidence

30 answers

The base build range proves capacity while the hall is built, which is where the specification only requires static load testing. The end-of-line range replicates true rack-level heat and electrical load, so the response of the whole system is proven before the GPUs are installed. Most halls use both.

Read the full answer: Applications: What You Test With PowerLoad

The client's commissioning agent or accredited partners. PowerLoad supplies the equipment, supports the team remotely on every project and trains them.

Read the full answer: Applications: What You Test With PowerLoad

Applying the load where the racks will sit, rather than at the riser. It proves the busway, the tap-off, the local protection, the manifold branch and the flow at the rack itself.

Read the full answer: Testing Liquid-Cooled AI Data Centres

PowerLoad's engineers are contactable and support the team remotely, and accredited CxA partners can attend site. Fault finding during live commissioning is part of the support.

Read the full answer: Beyond the Load Bank: What Clients Actually Buy

Density and dynamics. Rack power densities now require liquid cooling, and the load itself moves quickly rather than holding steady, so commissioning has to validate how a fluid circuit responds as well as what it can carry.

Read the full answer: AI Data Centre Commissioning: What Changed

PowerLoad equipment is used across the programme, from early-stage integration on the rising main through to end-of-line GPU profile simulation, with support from accredited CxA partners trained on the systems.

More on this: What Commissioning an AI Data Hall Involves

Long before the test window. They are typically set by late planning: equipment booked at the last minute, unclear site layout and unresolved questions about connections, all of which surface as lost time once engineers are on site.

Read the full answer: What Commissioning an AI Data Hall Involves

Anyone can deliver a load bank. What clients also need is a team that understands the data centre and what they are trying to prove, supporting every stage from early planning and setup to on-site testing and fault finding. When deadlines are tight, what matters is knowing someone will pick up the phone.

More on this: What Commissioning an AI Data Hall Involves

Earlier than most people assume. Commissioning stalls because planning was left too late: kit booked at the last minute, an unclear site layout, and nobody certain about the connections. Engineers then stand around fixing things that should have been solved weeks before. Thinking about the full deployment from day one, including access routes, hose runs and power requirements, is what removes that.

More on this: What Commissioning an AI Data Hall Involves

No. PowerLoad designs, builds, sells and hires the load banks used from early-stage integration on the rising main through to end-of-line testing. It supports commissioning teams remotely on every project, gives full on-site support through accredited CxA partners, and trains clients' teams through its accreditation programme. The commissioning agent or contractor does the commissioning.

Read the full answer: Commissioning Support: What PowerLoad Provides

Early-stage integration connecting to a rising main can detect cooling and power issues weeks ahead of traditional load testing, which is considerably earlier than most commissioning plans assume.

Read the full answer: Commissioning Support: What PowerLoad Provides

Because monitoring what happens inside the cooling loop is the part of commissioning most often overlooked. A custom-built, fully instrumented manifold measures flow, temperature and pressure across every hose in real time, so you are not just applying loads, you are seeing exactly how the system performs under them. It feeds straight into remote monitoring, which lets engineers spot bottlenecks before they become failures.

More on this: Commissioning Manifolds and Loop Monitoring

Because hyperscale sites grow and change faster than a fixed test environment can follow. Modular units can be expanded, repositioned or combined as phases overlap, so the setup adapts instead of being rebuilt.

Read the full answer: Modular Load Bank Systems for Hyperscale

Because the local protection sits at the rack, and a hall proven only at the riser has not exercised it.

Read the full answer: Protection Settings Testing for AI Data Centres

A controlled load is applied to the set in steps while frequency and voltage are measured, to show how it picks up load.

Read the full answer: Generator Load Acceptance Testing for AI Data Centres

AI load can reduce suddenly as well as rise. The set has to respond in both directions, and the fall is the half most often left untested.

Read the full answer: Generator Load Acceptance Testing for AI Data Centres

Because it can mask a slow set. With the UPS in circuit the test shows what each part contributes.

Read the full answer: Generator Load Acceptance Testing for AI Data Centres

Because the system has to respond in both directions. A sudden fall in load is its own event for voltage, frequency and the generator, and it is rarely tested.

Read the full answer: Switchgear and Power Quality Testing for AI Data Centres

Where the specification only requires static testing, the base build range covers it. For AI halls, dynamic testing at the rack adds the response that capacity alone does not show.

Read the full answer: Static and Dynamic Load Testing for AI Data Centres

Because AI racks apply steps continuously. A UPS that handles one step cleanly is shown to work once, so the test repeats the step and logs the response.

Read the full answer: UPS Load Testing for AI Data Centres

Yes. It can cover a generator that is slow to pick up load, which is why the generator stays in circuit.

Read the full answer: UPS Load Testing for AI Data Centres

Continuous logged data exported in the format your witness pack or client reporting needs, produced by the equipment rather than transcribed by hand.

Read the full answer: Load Bank Fleet Deployment and Management

Yes. That is end-of-line testing with the LCLB 160 and LCLB 240: true rack-level heat and electrical replication at each rack position, so the busway, tap-off, local protection and cooling are tested at the point the GPUs will sit.

Read the full answer: High Load Simulation

Base build testing with the LB500, LB600 and LB700, and end-of-line testing with the LCLB 160 and LCLB 240 GPU emulators. Both are tests the commissioning team runs; PowerLoad supplies the kit and supports the team.

Read the full answer: Commissioning Support, Remote and On Site

Yes. The controller logs the run continuously and exports it, so results drop into the format your client or authority expects rather than being transcribed off a clipboard.

Read the full answer: Commissioning Support, Remote and On Site

Factory witness testing happens at the equipment manufacturer, before anything ships. Site acceptance confirms the kit arrived correct, undamaged and correctly installed. Base build testing proves the cooling plant, CDU and risers have the capacity while the hall is built. End-of-line testing replicates the live rack's heat and electrical load at each rack position, before the GPUs are installed. Integrated systems testing is where everything runs together.

More on this: Testing Liquid-Cooled AI Data Centres

IST is the last commissioning stage, where power, cooling, controls and life safety are run together under load while faults are deliberately induced: utility loss, generator start, CDU failure, pump changeover. It is the only stage that proves the building behaves as one system rather than as a set of systems that each passed on their own. It needs a load that can be held steady and then moved on command, which is why fleet control matters more at integrated systems testing than anywhere else.

More on this: Commissioning Support, Remote and On Site

As soon as the power and cooling chain is energised, and before the IT fit-out closes off the access routes. Testing early means faults surface while the contractors are still on site and the hall is still open. Leaving it to the end compresses the fix window into the handover period, which is exactly where commissioning programmes slip.

More on this: Read the full page

Logged data tied to the test scripts, in a form the commissioning authority can put straight into the witness pack: applied load against time, temperatures, flow and delta-T, and how the system responded at each step and each induced fault. The heat is a by-product. The evidence is the deliverable.

More on this: Commissioning Support, Remote and On Site

Yes. Because the units are designed and built in house, a factory witness test can be arranged as part of factory testing rather than treated as a favour. It is the cheapest place in the whole programme to find a problem.

More on this: Read the full page

Equipment and products

The units, the carts and the control

39 answers

The ACLB 200 loads in stages of 1, 2, 2, 5, 10, 20, 40, 40 and 80kW, so the minimum step is 1kW. It runs from local manual control or from PC software over RS485, with automatic loading sequences and exportable records.

More on this: Air-Cooled Load Banks: ACLB 60 and ACLB 200

The controller networks up to 250 load banks, so a whole hall is driven, logged and reported from one point.

More on this: Air-Cooled Load Banks: ACLB 60 and ACLB 200

Cooling: chillers and cooling plant, and static and dynamic load testing. Power: UPS systems, generators, switchgear and power quality, and protection settings. The same units load the cooling and the power at once.

Read the full answer: Applications: What You Test With PowerLoad

Commissioning equipment is the supporting kit that prepares a liquid-cooled loop before a load bank ever applies load. It flushes and fills the pipework, then distributes and regulates the coolant across each rack circuit. PowerLoad supplies a flushing and refill cart that cleans and charges the loop through a 25 micron filter, and a commissioning manifold trolley that splits one feed into regulated, individually metered outlets.

Read the full answer: Commissioning Equipment for Liquid Cooling Loops

Yes. Both pieces share the FD83, Tri-clamp and UQD connection standards used across the PowerLoad load banks, so the equipment connects straight to the load banks.

Read the full answer: Commissioning Equipment for Liquid Cooling Loops

Yes. The manifold trolley and the flushing and refill cart are available to hire alongside the load banks. Speak to the team about availability for your project.

More on this: Commissioning Equipment for Liquid Cooling Loops

No. Published datasheets download straight from this page with no form in the way. If you want a revision we have not published yet, ask us and we will send it over.

Read the full answer: Load Bank Datasheets and Technical Documents

Everything published here is the current revision at the time of publishing. Equipment changes, so if you are specifying against a datasheet for a live programme, ask us to confirm the revision before it goes into a document that someone signs.

Read the full answer: Load Bank Datasheets and Technical Documents

Yes. Every product has a datasheet; not all of them are published on the site yet. Use the request link on the card and we will send the current version.

Read the full answer: Load Bank Datasheets and Technical Documents

Yes. Most of what we build is adjusted for the hall it goes into, so the published datasheet is the baseline rather than the limit. Send us the constraint and we will tell you what is possible.

Read the full answer: Load Bank Datasheets and Technical Documents

One cart connects directly to the load bank units and services up to five 500kW units from a single position, so a full bank of test equipment is prepared from one connection point rather than unit by unit. That ratio decides how much of the commissioning window is spent on preparation and how much is left for testing.

More on this: Flushing and Refill Cart for Coolant Loops

The cart carries a manifold with five inlet and five outlet ports and shares the FD83, Tri-clamp and UQD standards used across the PowerLoad range, so it connects to the load banks and the commissioning manifold trolley without adaptors.

Read the full answer: Flushing and Refill Cart for Coolant Loops

They are the same design. The LB500 is rated at 500kW and the LB600 at 600kW, both on 4 inch flow and return with stainless steel internals and PG25 compatibility. Solid-state switching is an optional extra on either, not a separate model.

Read the full answer: Liquid-Cooled Load Banks for Base Build Testing

The ACLB 200 is a portable 200kW air-cooled load bank with load steps from 1kW. It is part of the base build range. The full specification is on the air-cooled load banks page.

Read the full answer: Liquid-Cooled Load Banks for Base Build Testing

The PowerLoad controller networks up to 250 load banks, monitored and driven from a single point with live flow, temperature and pressure and exportable test data.

Read the full answer: Liquid-Cooled Load Banks for Base Build Testing

PowerLoad equipment is built for a range of site conditions, including outdoor testing. The exact ingress-protection specification is being confirmed, so speak to the team about your site conditions before assuming a specific rating.

Read the full answer: Testing Liquid-Cooled AI Data Centres

PowerLoad supports the commissioning team across early planning, setup, on-site testing and fault finding. Remote support is included on every project, and full on-site support is available through accredited CxA partners, rather than support that ends when the equipment is delivered.

Read the full answer: Beyond the Load Bank: What Clients Actually Buy

Up to 250 units can run on one controller, with live flow, temperature and pressure and exportable test data across the whole fleet.

Read the full answer: Networked Load Banks Across a Whole Data Hall

By making the test environment modular enough to keep pace with the site. Hyperscale projects grow fast, and equipment that is fixed, awkward to scale or slow to reconfigure holds the programme back. Each unit can operate alone or as part of a larger system, and the connections, fittings and layouts are designed so engineers can extend or reposition the setup quickly as the site moves through commissioning stages.

More on this: What Commissioning an AI Data Hall Involves

Yes. Per-circuit isolation and quick-connect fittings mean the system can be configured or rebalanced without shutting the whole setup down.

Read the full answer: Fluid Distribution Points: Per-Circuit Control

Yes. Isolation valves at every connection point allow an individual unit to be isolated or removed without affecting the rest of the system.

Read the full answer: How GPU Emulator Units Sit on the Cooling Loop

Individual connections spread control and measurement around the hall, so balancing is manual and readings are taken at different times. A manifold trolley brings flow, pressure and temperature for every circuit into one place, where they can be compared and adjusted together.

Read the full answer: Commissioning Manifolds and Loop Monitoring

In this configuration two LB500 load banks connect into a manifold trolley, which distributes chilled water to individual outlets. Each outlet represents a provisioned server rack within the data hall, and isolation valves at every outlet allow each one to be controlled independently. Flow from each outlet returns through the CDU, so multiple racks are simulated and tested at once with full control over flow and return.

More on this: Commissioning Manifolds and Loop Monitoring

One connection to the secondary fluid network, fanning out into 1 to 2 inch hose pairs that are each independently metered and controlled.

More on this: The Manifold Trolley Outlets: Each One Metered

Each outlet has an electronic flow meter reporting live flow rate and temperature to the controller, and a target pressure drop can be set on its valve.

More on this: The Manifold Trolley Outlets: Each One Metered

Yes. Each unit operates alone or as part of a larger coordinated system, with connections and layouts designed so the setup scales up from there.

Read the full answer: Modular Load Bank Systems for Hyperscale

One flushing cart connects directly to the load bank units and can service up to five 500kW units from a single position. It runs a full clean flushing cycle before operation, and the same connection is used again after testing to flush and condition the system, so the load banks are left clean and protected between commissioning jobs.

More on this: Water Quality and Flushing in Liquid Cooling

Timelines, access routes, hose runs, power requirements and how each unit will apply load in the specific space, so the setup is understood before anything is delivered.

Read the full answer: Where Commissioning Delays Actually Start

Rarely. The device is usually doing what it was set to do. The settings need adjusting against the measured load.

Read the full answer: Protection Settings Testing for AI Data Centres

Yes. We design and manufacture in the UK, so connections, footprint, control interface, ratings and finish can be built around your site rather than worked around on it. Bespoke work runs alongside the standard range, not as a separate business.

Read the full answer: Bespoke Load Banks

Yes. It is built in the same factory by the same engineers and covered by the same maintenance and repair service.

Read the full answer: Bespoke Load Banks

Often, yes. Tell us what you have and what it needs to do. Where a modification is not sensible we will say so rather than sell you one.

Read the full answer: Bespoke Load Banks

Yes. Up to 250 units network to a single controller, so a bank of units across a hall is driven, monitored and logged from one place, with live flow, temperature and pressure. That makes testing at hall scale practical.

Read the full answer: Load Bank Fleet Deployment and Management

Yes. Remote monitoring gives live visibility of the units and the test in progress, with continuous logging and export, so a project manager does not have to be standing next to a unit to know what it is doing.

Read the full answer: Load Bank Fleet Deployment and Management

It depends on the fleet and the window. Tell us the dates early. Demand clusters around handover seasons, and the halls that book earliest get the kit they need.

Read the full answer: Load Bank Hire for Data Centres, Base Build and End-of-Line

Tell us immediately. A unit down during a live window is the priority, and because we built the equipment we can usually diagnose it on a call before anyone travels.

Read the full answer: Load Bank Maintenance, Repair and Calibration

Talk to us. We will look at it and tell you honestly whether it is worth maintaining.

Read the full answer: Load Bank Maintenance, Repair and Calibration

Yes. PowerLoad delivers the equipment, supports positioning and connection, and proves it under load, rather than leaving it at the gate. Load banks are heavy, carry coolant and connect to live distribution.

Read the full answer: Load Bank Supply and Installation

Yes. Manifold trolleys, hoses, valves and flushing and refill carts are part of the range, so the connection side comes from the same supplier as the load and nothing is left to be improvised on site.

Read the full answer: Load Bank Supply and Installation

Buy or hire

Routes, cost drivers and what is included

17 answers

Hire if you have a defined test window and no ongoing programme. Buy if you test repeatedly across a portfolio and want the kit available whenever a hall is ready.

Read the full answer: Buy or Hire Load Banks

Often that is the sensible order. Hire for the first hall, find out what the programme actually needs, then buy against real requirements rather than a guess. Talk to us about how a hire relationship can lead into a purchase.

Read the full answer: Buy or Hire Load Banks

With hire, PowerLoad owns the equipment and your team runs the test. With purchase, you own the equipment, backed by a service agreement.

Read the full answer: Buy or Hire Load Banks

Yes. The load banks, GPU emulators and commissioning equipment are all available to hire. Speak to the team about availability for your project.

More on this: Base Build and End-of-Line Load Banks

Yes. Buying makes sense where you test repeatedly across a portfolio. Hiring makes sense for a defined window. There is a page comparing the two routes.

Read the full answer: Services for Data Centre Commissioning Teams

Because circuits that perform correctly on their own can still interact when they all draw simultaneously. Testing circuits together is what reveals starvation and imbalance between them.

More on this: Commissioning Manifolds and Loop Monitoring

Because circuits that perform correctly on their own can still interact when they all draw simultaneously. Testing outlets together is what reveals starvation and imbalance between them.

More on this: Two 500kW Load Banks on One Manifold

Most commissioning hires run from a fortnight to a few months, matched to the test window rather than a fixed term. Phased projects often hire across several windows as each hall comes online.

Read the full answer: Load Bank Hire for Data Centres, Base Build and End-of-Line

Power at the rating we agree, a coolant connection point, access for the units and somewhere to run hoses. We survey all of that before anything is booked, so nothing turns up that cannot be connected.

Read the full answer: Load Bank Hire for Data Centres, Base Build and End-of-Line

Talk to us. Where a client is standardising on PowerLoad across a portfolio, we would rather find a route that works than lose the standard.

Read the full answer: Load Bank Hire for Data Centres, Base Build and End-of-Line

Yes. PowerLoad trains clients' teams through its accreditation programme, and the PowerLoad Academy has operator lessons for every unit.

Read the full answer: Commissioning Support, Remote and On Site

Cost is driven by rating, cooling type, the connection arrangement, the control and monitoring you need, and whether the unit is a standard build or built around a specific hall. Rather than a list price that would be wrong for most projects, the number comes off the survey so it reflects the connections and constraints you actually have. Hire is quoted against the commissioning window.

More on this: Buy or Hire Load Banks

It depends on whether the unit is a standard build or bespoke, and on the connection and control arrangement. Send the hall, the IT load and your dates in the first message and you will get a real date rather than a range. We come back with a proposal within five working days of the survey.

More on this: Bespoke Load Banks

Hire is set against the commissioning window rather than sold in fixed blocks, because the useful unit for a data hall is the programme, not the day. Tell us the dates and the number of halls and it is priced against that.

More on this: Load Bank Hire for Data Centres, Base Build and End-of-Line

Both. Buying makes sense where you test repeatedly across a portfolio and want the kit and the knowledge in-house. Hiring makes sense for a defined window on a single programme. There is a page that compares the two routes properly rather than pushing you at one.

More on this: Buy or Hire Load Banks

Planned servicing and emergency response are available for the life of the unit, from the team that built it. The exact terms come with the quote, because they depend on where the unit will live and how hard it will be worked.

More on this: Load Bank Maintenance, Repair and Calibration

Yes. UK hire is where a lot of our work starts, with delivery, connection and remote support as required, and we deploy the same fleet across European programmes.

More on this: Load Bank Hire for Data Centres, Base Build and End-of-Line

On site, delivery and support

What we need from you and what you get from us

34 answers

The same on both. PowerLoad supports your team remotely on every project, on-site support is available through accredited partners, and training is offered through the accreditation programme.

Read the full answer: Buy or Hire Load Banks

Yes. Purchased equipment is covered by preventative planned maintenance, calibration and emergency response, from the people who manufactured it.

Read the full answer: Buy or Hire Load Banks

We survey first and aim to send a proposal within five working days of the survey. Sending the hall, the IT load and the dates in the first message is the fastest way to a useful answer.

Read the full answer: Services for Data Centre Commissioning Teams

Head office is in Witham and manufacturing in Bridgwater in the UK. Projects run across the UK and Europe, with partners covering India and the UAE.

Read the full answer: Services for Data Centre Commissioning Teams

A supplier's obligation ends at delivery. A partner stays through the programme, supporting the commissioning team with engineers who know the equipment, so the equipment and the expertise to use it arrive together.

More on this: What Commissioning an AI Data Hall Involves

No. Support is built into how PowerLoad works rather than sold as an add-on, including working with your engineers and test plans from day one, remote support on every project and on-site support through accredited CxA partners.

Read the full answer: What Commissioning an AI Data Hall Involves

Support is built in rather than added on. From day one PowerLoad works with your engineers, timelines and test plans, with remote support on every project and on-site support through accredited CxA partners, because things do not always go to plan. Adjusting in real time and solving problems early keeps the schedule on track and gives the client more confidence at handover.

More on this: What Commissioning an AI Data Hall Involves

Remote support is included on every project. Full on-site support is provided through accredited CxA partners, or by your own team once PowerLoad has trained them. Talk to us about what your project needs.

More on this: Load Bank Support That Continues After Delivery

Teams save time because issues are adjusted on the spot rather than escalated, investigated and rescheduled.

More on this: Load Bank Support That Continues After Delivery

Most often the connection side: coolant fittings, hose runs, manifold arrangements, electrical terminations and the physical footprint so units fit the access route and the space available. Control and reporting interfaces are the next most common.

Read the full answer: Bespoke Load Banks

It depends entirely on how far from standard it sits. A change of connection or finish is short. A new rating or a new form factor is a programme. Tell us the date you need it on site and we will tell you honestly whether it is achievable.

Read the full answer: Bespoke Load Banks

As a fleet rather than a series of hires. Kit is allocated across the programme, profiles are shared so results are comparable between halls, and servicing is scheduled into the gaps rather than interrupting a window.

Read the full answer: Load Bank Fleet Deployment and Management

Yes, and it should be. Stored profiles replayed on each hall are what turn a set of individual tests into a portfolio standard that means something.

Read the full answer: Load Bank Fleet Deployment and Management

Yes. PowerLoad engineers and manufactures in the UK and has supported projects in London, Frankfurt, Helsinki, Paris, Munich and Sweden, with partners in India and the UAE.

Read the full answer: Load Bank Fleet Deployment and Management

High load simulation tests a data hall with a load that behaves like the equipment that will occupy it, rather than a steady resistive draw. For an AI hall that means reproducing the swings a rack of GPUs creates, so the switchgear, generators, UPS and cooling are proven against real behaviour.

Read the full answer: High Load Simulation

Yes. Profiles are stored and replayed, so hall two is tested against the same shape as hall one and the results are genuinely comparable rather than being a matter of who ran the test.

Read the full answer: High Load Simulation

Yes. Base build testing with the LB500, LB600, LB700 and ACLB 200 proves the cooling plant has the capacity while the hall is built. Solid-state switching is an optional extra where the specification asks for transient testing. End-of-line testing with the LCLB 160, LCLB 240 and ACLB 60 then replicates the live rack.

Read the full answer: High Load Simulation

Yes. The system has to respond in both directions. A hall spends most of its life below peak, and switchgear, generators and cooling can behave just as unpredictably stepping down as stepping up. PowerLoad units step down as well as up, so partial-load conditions are tested rather than assumed safe.

Read the full answer: High Load Simulation

Remote support is included on every project. On-site support is available through accredited CxA partners, and PowerLoad trains your team through its accreditation programme.

Read the full answer: Load Bank Hire for Data Centres, Base Build and End-of-Line

Yes. Preventative planned maintenance is available for PowerLoad equipment, scheduled between your test windows rather than during one.

Read the full answer: Load Bank Maintenance, Repair and Calibration

Electrical and control checks, coolant side inspection including seals, fittings and internals, calibration of the instrumentation the results depend on, firmware and controller checks, and a report on anything that will need attention before it becomes a failure.

Read the full answer: Load Bank Maintenance, Repair and Calibration

Because the point of a test is the number it produces. An uncalibrated instrument can put a wrong result into a witness pack.

Read the full answer: Load Bank Maintenance, Repair and Calibration

Yes. Where equipment is deployed across several halls or several countries, servicing is planned as a fleet so units come back into the calendar in a sensible order rather than all at once.

Read the full answer: Load Bank Maintenance, Repair and Calibration

Remote support on every project, and on-site support through accredited CxA partners. That covers guidance on positioning and connecting the units, support on flushing and filling the loop, support on the load steps the test scripts call for, and logged data for the witness pack.

Read the full answer: Commissioning Support, Remote and On Site

Yes. PowerLoad engineers and manufactures in the UK and has supported projects in London, Frankfurt, Helsinki, Paris, Munich and Sweden, with partners in India and the UAE.

Read the full answer: Commissioning Support, Remote and On Site

Offload and positioning, coolant connection into the loop, electrical termination, start-up, a functional proof under load, and handover with documentation. Where the site needs it, flushing and filling the loop comes first.

Read the full answer: Load Bank Supply and Installation

Yes. Retrofit into an operating facility is a different job from a new build, with access, isolation and containment agreed before anything arrives.

Read the full answer: Load Bank Supply and Installation

Datasheets, test certification, connection and operating documentation, and the logged output from the functional proof.

Read the full answer: Load Bank Supply and Installation

Somewhere to connect, a route for the cables, physical space and access to get the unit into position, and on a liquid-cooled unit a connection into the coolant loop at the flow and temperature the unit is rated for. Establishing exactly this is what the survey is for, before anything is quoted.

More on this: Load Bank Supply and Installation

Flow rate and supply temperature inside the unit's rated window, connections that match the manifold arrangement, and a loop that has been flushed and filled to the right fluid quality. A loop carrying debris or the wrong mix will foul the test before it proves anything, which is why flushing and refill carts sit alongside the load banks rather than as an afterthought.

More on this: Read the full page

Agreed at survey. PowerLoad can deliver and offload the units and support their connection and proof under load before handover, or deliver to a site team that handles its own connection. The commissioning team then runs the test.

More on this: Load Bank Supply and Installation

Your commissioning team or the commissioning agent runs the units through the test scripts. PowerLoad does not commission the hall: it supports the team remotely on every project, provides on-site support through accredited CxA partners, and trains your own team to run the equipment through its accreditation programme.

More on this: Commissioning Support, Remote and On Site

Handover training on the equipment comes with any supply or hire. PowerLoad also trains client teams through its accreditation programme, so commissioning teams can run their own fleet properly rather than learn it one project at a time.

More on this: Read the full page

Yes. The units are designed and manufactured in the UK, so spares come from the factory that built them. That also means a unit can be modified to suit a hall rather than the hall being worked around the unit.

More on this: Load Bank Maintenance, Repair and Calibration

About PowerLoad

Who we are, where we work, how we deliver

9 answers

Nine units in three groups. Base build testing: the 500kW LB500, 600kW LB600 and 700kW LB700 liquid-cooled load banks, and the 200kW ACLB 200 air-cooled load bank. End-of-line testing: the 160kW LCLB 160 and 240kW LCLB 240 AI liquid GPU emulators, and the 60kW ACLB 60 rack-format air heat emulator. Commissioning equipment: the manifold trolley and the 250 litre flushing and refill cart. Up to 250 units network on a single controller.

Read the full answer: Base Build and End-of-Line Load Banks

PowerLoad manufactures load banks and provides the services around them: hire, remote support on every project, on-site support through accredited partners, training, high load simulation, bespoke design and manufacture, supply and installation, maintenance and repair, and deployment across a portfolio of halls. On-site commissioning is carried out by the client's commissioning agent or accredited partners.

Read the full answer: Services for Data Centre Commissioning Teams

A supplier delivers hardware and moves on. A support partner stays involved from start to finish, which matters because you need to know the infrastructure performs under real-world conditions rather than on paper. The equipment is modular, quick to deploy and engineered in house, but the difference clients remember is how hands-on the support is, from remote help to accredited partners on site.

More on this: What Commissioning an AI Data Hall Involves

A supplier's obligation ends at delivery. A partner stays involved through the programme. PowerLoad supports commissioning teams remotely on every project, gives on-site support through accredited CxA partners and trains client teams, so the equipment and the know-how arrive together. The client's CxA or contractor runs the commissioning.

More on this: Load Bank Supplier or Partner

Yes. It is engineered in house, from the bespoke manifolds to the solid-state switching, and hand built and tested in the UK.

Read the full answer: Load Bank Supplier or Partner

PowerLoad is a UK manufacturer of load banks for data centre commissioning, with head office in Witham, Essex and manufacturing in Bridgwater, Somerset. We design and build our own liquid-cooled load banks, GPU emulators and commissioning equipment, and we supply and support them across the UK and Europe, with partners in India and the UAE.

More on this: Read the full page

Yes. PowerLoad is the same company under a new name. The engineering team, the manufacturing and the project history all carry across. If you worked with DC-PPM, you are dealing with the same people.

More on this: Read the full page

In the UK, at our own facility in Bridgwater, Somerset, with design and engineering in Witham, Essex. Both are ours, which is why a unit can be changed to fit a hall and why the people supporting your project know the equipment.

More on this: Read the full page

The physics is the same wherever a controlled load is needed, and load banks are used across generation, marine, industrial and standby power. Our work is data centre commissioning, and specifically the liquid-cooled, high density end of it, which is where the problems are hardest and where being the manufacturer counts for most.

More on this: Read the full page

Still not answered?

Send the hall, the IT load and the dates. You get a real answer from an engineer, and if it is a question other people ask it ends up on this page.