Published on 27th January 2015 by Ian Bitterlin
If you didn’t do Latin in school or are not a lawyer (Caveat Emptor is a common legal term) I should translate the title to ‘Buyer Beware’ but this blog builds on from my last on the subject of load diversity in data centre design.
Just in case you didn’t read that blog (!) it raised the question of whether or not the ‘actual’ load should be taken as the design load or something less – as partial load is endemic in data centres, particularly colocation. The principle is simple, for example; the facility design is for 8kW per cabinet but the average load is ‘probably’ going to turn out to be 4kW (or less) per cabinet and, therefore, the UPS system should be optimised for 50 percent of the ‘design’. I posed the question rather than suggesting an answer although the ‘risk’ of undercooking the solution probably will steer most folks away from such heresy.
Anyway a couple of days after posting the blog and reading the feedback it occurred to me that I had in fact seen a real example of diversity in the data centre – albeit in a somewhat underhand way. The subject matter was battery autonomy, i.e. how many minutes the battery can support the UPS at full load.
Unscrupulous vendors have long played one game when the client's specification is weak: If the client asks for 15 minutes of battery autonomy in an N+1 UPS system and is not careful enough to specify the 15 minutes for ‘each’ UPS module then they lay themselves open to getting 15 minutes with all the modules running – probably about 11 minutes for each module. Still ‘enough’ but that’s not the question, yet.
There is also the more subtle trick of rating the cell capacity at 25⁰C (77⁰F) when the life is rated for 20⁰C (68⁰F) – squeezing just a few tens of Watts per cell out of the selection and probably missing the 15 minute spec by 2-3 minutes.
But why ‘did’ (some still ‘do’) they do that? Well that’s the easy part – the typical 15 minute battery may represent 40-50 percent of the cost of the UPS system so shaving 20 percent off the battery cost makes them 10 percent more competitive without having to discount the UPS itself, which is the bit they manufacture instead of buy/sell at low margins, like the battery.
But all that historical rule-bending has been eclipsed by the example I will now share with you. This is ‘real’ and comes from a major player in the European UPS market. I only know for certain that the malpractice occurred in the UK but the data-sheet that fell into my hands didn’t restrict the geographic reach.
The ‘scam’ (which it mght not be if the client knows what they are being offered) is simple – a ‘typical data centre rating’ for the battery. Sounds application specific and specialised enough to look impressive on the quotation but in reality is simply ‘smaller’. Not a little ‘smaller’ but only rated for 60 percent of the kW load for the stated autonomy time.
So, to be clear: The client asks for a data centre UPS for X kW and Y minutes. He gets offered what he asked for under the ‘data centre rated’ tag but in reality is being offered 60 percent of kW for Y minutes.
Now we are talking about cutting the battery size in half, since autonomy is not linear with load etc. That provides a 25 percent cost reduction without discounting the UPS one penny/cent et al.
As far as I could tell from the battery selection table it would ‘always’ do the kW load for 5 minutes autonomy (which is the minimum practical design point for lead-acid cells) but there is no reason why this should be so.
So is this a scam? Is it ‘wrong’? It depends upon the client knowing what he is being offered. If he does not know then it is sneaky, underhand, unethical and anti-competitive – probably even illegal if not described in the small print. If he does know then it can be argued (and I would) that it is sensible CapEx minimisation and good for the environment as less will have to be recycled. The quote I saw didn’t spell it out and the client was being duped. What do you think?
I won’t name the vendor, so don’t ask, but they know who they are and I would ask them to be more transparent.
And Caveat Emptor? Specify the number of cells that suit the UPS, the length of the cables between the battery and the UPS, the input kW to the inverter, the minimum minutes of autonomy required and the minimum temperature that your system is designed for. That way you can’t get duped. Then look at the competing batteries by weight – there is no substitute for lead...