Holdfast Training Services Trial Blog

With three data centres on one campus, and residential properties approximately 50m away, the acoustic limits on site are very tight. With over 70x 2.4MW generators and 1.6MW chillers (amongst far more plant), the noise output is significant. We’re pretty much one shout away from our planning permission limits.

Load banks (LB) are required within data centres to test the generators on-load twice annually. Typically, the LB will be fixed in place and have busbar run to each generator, testing via the flick of a switch. Value engineering, in all it’s glory, led to the removal of a fixed LB for a far cheaper mobile unit, sat on a trailer and driven to each of the generators. The CAPEX savings are in the millions (though the OPEX is questionable). The issue with mobile LBs on this site is that the units can’t be routinely attenuated for noise, due to the centre of weight on the trailer and ventilation requirements.

The pertinent info for the mobile LB (not yet procured) is:

• 88dbA noise emissions, likely upwards.
• At least 2m space required to the sides to allow air intake.
• No cover above to allow exhaust of air (fire risk).
• Broadly 3.5x2x2.5m dimensions (HxWxL, on trailer).

From multiple vendors and data sheets, we will significantly breach noise limits. Some work with the acoustics team has shown that we need to reduce the dBA to 82@1m and 56@10m to the side. Modelling shows for the closest test location to the nearest noise receptor (residential), an acoustic screen would need to be over 7m tall, 2m away to the side, to screen the upwards trajectory of the noise emissions. This would then have to be wheeled around with the generator, stored, windproofed, etc… Highly unpractical.

I have also looked at:

• Procuring a compliant load bank. Tests show the load bank would need to emit 72dBA – a huge acoustical gap on a logarithmic scale.
• Buying the nearest receptor. Unsavoury and questions remain about breaching noise limits at other receptors.
• Screening the large fence line in front of the receptor – deemed ineffective, due to soundwave trajectory.

My preference and recommendation is to retrofit the fixed load bank and attenuate in place. A business case is being made elsewhere, but there may be no way back. So, keen to hear if anyone has any mobile acoustic solutions/suggestions to this kind of problem?