NABERS bestows four stars on first data centre audit

Green Global Consulting CEO Bob Sharon said the first official NABERS audit resulted in a lot of lessons for all parties - most of which are scheduled to be discussed at an official de-briefing later this week.

1. Records must be validated

Fujitsu's facilities staff had originally expected printed reports of monthly meter readings would suffice to achieve the rating.

On several occasions, Fujitsu staff had to pull data from various systems to back their claims.

"I would recommend making sure you’re collecting as much data as possible that can be validated easily by an third party," Stewart advised.

“It wasn’t easy. I would recommend that anybody doing an infrastructure rating ensure their metering is in the right spots."

There were delays, for example, when it was discovered Fujitsu had been taking readings from the outbound side of PDUs (power distribution units) but not from master PDUs - which could have made the job much easier.

2. Meter your power and IT load separately

All parties learned that the use of in-row cooling infrastructure - which can drive down cooling costs - can also carry an audit overhead if the power it consumes isn't measured separately to the IT load.  

Sharon found at Noble Park a series of rear door extraction fans that - while powered within the same PDU as the IT load - needed to be measured and subtracted for an accurate reading.

If they aren't measured separately, auditors are faced with installing dozens of loggers on individual equipment (at significant cost), or must choose the theoretical maximum 'nameplate' power draw, which could overstate the draw of this infrastructure by up to 400 percent and harm the chances of a good NABERS rating.

“The ‘nameplate’ draw is basically what the manual says - which at full load, might consume 1.2 kW per hour, but some of those extraction fans might only be drawing 300 watts," Sharon said.

"The moral of the story is to have separate metering for all devices that you could record - if you’re not metering them separately, your rating might pay dearly.”

3. Test your equipment

There was a danger during the final stages of the process that Fujitsu may have fallen just short of a four star rating, if not for the discovery of several anomalies.

Fujitsu had recently installed new chillers that were serving both Buildings One and Two at Noble Park. As the data centre rating was limited to Building One, the energy these chillers required to cool the small amount of IT load in Building Two could be deducted.

Further, Sharon experienced difficulties when attempting to read data from one of the CT (current transformers) in a switch port within the distribution board. This would be telling as it would usually be the CT that - wrapped around the phase - is used to measure the current. Sharon confirmed that the white phase from one of the PDUs had not been read during the period of the fault.

Sharon raised the point with NABERS and sought to extrapolate historical data to gauge the impact of this anomaly.

“I argued that as long as we were within a five percent margin of error, the rating would stand," he said. "But - to their credit - I was advised that NABERS doesn’t allow any form of extrapolation."

"There are strict rules around what constitutes acceptable data," explained Robison. "For example, only actual readings can be used - not utility estimates."

Fujitsu's engineers again came to the rescue - pulling data from Branch Circuit Monitoring - a system that measures power use within each individual circuit - that had otherwise been used to charge customers for power usage.

These measurements - far further down the chain than the PDU - provided measurements for the missing white phase across 32 circuits within the range of the affected PDU.

To validate that data, Sharon captured the total power usage for 12 months of the blue and red phase as recorded by both the Branch Circuit Monitoring system and the PDU. They varied within 0.06 percent over 12 months.

Green Global Consulting now had the numbers to give Fujitsu’s Noble Park a four star rating.

Sharon said the long process tested his relationships with both his client (Fujitsu) and the NABERS scheme.

“It’s perfectly natural that the client will get a bit shirty sometimes at how much we needed to test the veracity of their numbers," he said. "The NABERS team gets just as shirty about any effort to extrapolate numbers - they wouldn’t budge an inch. Its all part of the learning curve.”