Data centre heat exchanger cleaning

Yes, and it has to be. Work is planned onto whatever redundancy exists, one chiller of N+1, one tower cell, one side of a dual circuit, so the hall never loses cooling and the load never moves.

What it does consume is redundancy. For the duration the site runs with less margin than usual. The work is planned into a period of lower ambient and lower risk. There is an agreed point at which it stops and everything returns to service.

Compare condensing approach temperature and kW per tonne against the machine's commissioning figures. Both drift predictably as the condenser fouls, and most BMS installations already log enough to show it.

The number that matters is not the energy. It is available capacity at design ambient: a fouled machine that makes nameplate at 18 degrees may not at 32, and that shortfall only becomes visible on the day it is least welcome. Modelling it from the approach data is a great deal cheaper than discovering it.

It is the first thing in the method statement rather than a consideration within it. Cleaning produces water, aerosol and dislodged debris, and none of those may reach live equipment or the air path serving it.

In practice that means containment, isolation of the unit from the airstream, control of the effluent, and often doing coil work behind temporary screening. On sites with fine-passage liquid cooling the bar is higher again, because particulate released into a loop is a contamination event in its own right.

If the site has evaporative cooling towers, yes, in full. They are a notifiable, controlled Legionella risk requiring a risk assessment, a written control scheme, and cleaning and disinfection at defined intervals with records kept.

Practically, that duty and condenser cleaning belong on the same visit. Access, isolation and permits overlap almost entirely, and doing them separately means arranging the same reduced-redundancy window twice.

The cleaning itself is usually a day or less per machine, whether chemical or mechanical.

The elapsed time is set by the approval and isolation process around it. Change approval, the isolation sequence, witness and verification on return to service, and the requirement to have everything back before the next change window all take longer than the work. Plan them as the larger part.

It matters more than most sites expect. A closed circuit filled once at commissioning and left alone corrodes for years, and the iron oxide it produces ends up in the interchanger plates, the chiller and every strainer in the loop.

Cleaning it without then dosing and filtering is temporary. Correcting the water treatment on a closed circuit is one of the cheaper interventions available and it prevents the fouling instead of removing it repeatedly.

The cleaning is priced like any other commercial plant of the same size. What is different is the overhead: method statements, change approval, escorted access, out-of-hours working and phased isolation all add to it, sometimes substantially.

State the topology and the change process in the enquiry. A quote written without knowing whether the work is single-path or phased across N+1 will not survive contact with the site.

Condition-driven, from data the site already has. Condenser approach temperature and kW per tonne against baseline are the triggers, and open circuits foul considerably faster than closed ones.

Cooling towers are the exception and are cleaned at the frequency the Legionella control scheme sets, independent of condition. Air-side coil cleaning tends to follow filter regime and site conditions, and needs bringing forward during and after any construction work on a live site.