BLOG

Fuel Readiness in Data Centre Commissioning

FMPS

Fuel Readiness in Data Centre Commissioning

Why does fuel belong in the commissioning scope?Because the load bank test proves the generators, the switchgear and the controls, and it proves almost nothing about the fuel. The first fill is new, clean and only weeks old on the day of the test. The fuel that has to work at the first real outage may be two years into storage in a tank nobody has opened since handover. Commissioning is the only point at which the difference can be designed out.

Fuel Readiness in Data Centre Commissioning

The gap between the witness test and the real event

Data centre commissioning is thorough in a way that few other building projects are. Levels of testing build from factory acceptance through to integrated systems testing, generators are run against load banks, transfer sequences are proved, and a client witness signs the result.

At the end of that process the facility has demonstrated that its emergency power system works. What it has demonstrated specifically is that the system worked on that day, with that fuel.

The fuel is the one input to the test that will not be the same input at the real event. Everything else in it, the engine, the alternator, the switchgear and the control logic, is a physical asset that will be maintained on a schedule and will still be recognisably itself in three years. The fuel will have spent those three years quietly changing.

This is not an argument that commissioning is done badly. It is an argument that fuel condition is a different kind of thing from the assets around it, and it needs its own line in the scope.

What to establish at commissioning, and record

Five items. None of them are expensive at this stage and all of them are awkward to reconstruct later.

A baseline fuel analysis from the first fill. Draw a sample from the tank after filling, not from the delivery hose, and have it tested. Water and sediment, particulate, oxidation stability and a microbial check. That result is the reference point against which every later test is read. Without it, a test in year two tells you the condition of the fuel without telling you whether it has changed.

The actual biodiesel content of the delivered fuel. Both major specifications permit a biodiesel component: up to 7 per cent FAME under EN 590, and up to 5 per cent biodiesel meeting ASTM D6751 within ASTM D975, with B6 to B20 covered separately by ASTM D7467. FAME content changes how the fuel behaves in storage, so it is worth having on the record rather than in someone’s memory of a conversation with the supplier.

A sampling arrangement that works. Prove that a bottom sample can physically be drawn from every tank, by a technician, without a shutdown. This is a five-minute check at commissioning and a very expensive discovery afterwards. ASTM D4057 covers sampling practice; the commissioning question is simply whether the tank allows it.

A drain that reaches the water. Confirm that the low point of the tank and the drain point are the same place. If the tank is flat-bottomed or the drain is not at the low end, water will accumulate somewhere the drain cannot reach, and no amount of procedure will fix it later.

The fuel maintenance system, proved as a system. If a fuel polishing and maintenance system is installed, commission it properly: run it, verify flow through every tank it serves, confirm that the valve line-up is correct, and record the filter differential pressure with new elements so there is a clean baseline. A polishing skid that was energised but never proved against each tank is a common finding, and it is always found later than it should be.

Autonomy, and what a large fuel store implies

Resilience frameworks and client requirements drive on-site fuel autonomy upward. The Uptime Institute’s Tier Standard sets a minimum on-site fuel store for the concurrent maintainability and fault tolerance topologies, and in practice many operators specify considerably more than the minimum, particularly where resupply is uncertain or the site is remote.

That is a sound decision for availability. It has a consequence that is rarely stated alongside it: a larger fuel store means a longer residence time. Fuel bought for seventy-two hours of running that is only ever used for a monthly test regime is fuel that will sit in a tank for years.

So autonomy and fuel condition pull in opposite directions, and the resolution is not to hold less fuel. It is to accept that a large store is a stored asset that needs maintaining, and to fund that as part of the resilience decision rather than as an afterthought to it.

The monthly test regime is worth thinking about in the same frame. Running generators regularly is good practice and it does consume some fuel, but a short monthly run against a fraction of the store does not turn the tank over. It draws from the top and leaves the bottom untouched, which is where the problems are.

Commissioning is when the operating regime gets written

The most valuable thing that can come out of commissioning is not a test certificate. It is a fuel regime that the operations team inherits and can actually follow.

That regime has four parts:

  1. A sampling and testing interval. NFPA 110 sets an annual fuel quality test as the minimum for emergency and standby power systems. For a data centre in a hot or humid climate, quarterly is the interval that catches degradation while it is still correctable, because the mechanisms run faster there and the sampling has to follow.
  2. Named tests, with the baseline attached. Not “test the fuel” but the specific set, with the commissioning figures alongside so the result can be read as a trend rather than as an isolated number.
  3. A water drain schedule, with a record of what came out. A drain that produces nothing is information; a drain that produces a litre is information; a drain that is never recorded is nothing at all.
  4. A polishing system operating schedule, and the filter change trigger expressed as a differential pressure rather than as a month.

All four belong in the operations and maintenance documentation at handover. The recurring pattern in this industry is that they leave the site in the commissioning engineer’s head instead.

Make the fuel system visible to the operations team

The final piece is that none of this survives contact with a busy operations team unless it is visible in the systems they already watch.

Connecting the fuel system to the BMS or SCADA platform puts tank levels, transfer pump state, water detection, filter differential pressure and polishing system status into the same alarm and trending environment as everything else in the facility. A fuel system that reports into the platform gets looked at. One that reports to a local panel in a fuel yard is looked at when someone happens to walk past.

Trending is the part that pays back over years. A slow rise in filter differential pressure across six months is the earliest available warning that fuel condition is deteriorating, and it is invisible to anyone reading a gauge once a quarter.

The short version for a commissioning scope

Add these lines and the fuel stops being the assumption in an otherwise well-proved system:

Commissioning proves the generators. It does not prove the fuel, because the fuel it was run on is not the fuel that will be there at the first real outage. Baseline it, make it samplable, make it visible, and hand over a regime rather than a certificate.

Standards and references: NFPA 110 for emergency and standby power systems; the Uptime Institute Tier Standard for on-site fuel requirements; EN 590 and ASTM D975 for diesel requirements; ASTM D6751 and D7467 for biodiesel and blends; ASTM D4057 for manual sampling. Read more on Diesel Fuel Maintenance Systems and Turnkey Fuel Engineering.

Have a project, or a tank that needs looking at?

Tell us the facility, the tank and the generator, and an engineer will come back to you.

Speak to an engineer