overages for power usage

What kind of typical overage costs have you seen when a customer/you use more than you’ve committed to?

I’m especially interested in datacenter power situations, where maybe you sign up for 5kw or 500kw and use more than that in a given month. Is it billed at the same rate? Is it billed at a higher rate? What’s the % increase of the higher rate versus the regular rate?


The fuses might match what you ordered. If you go over you might lose power due to a blown fuse.

When there is A and B power for redundancy, you need to make sure that one side can take the whole load without blowing any fuses. Otherwise you have no redundancy.



fre. 21. sep. 2018 04.12 skrev Alan Hannan <alan@routingloop.com>:

What kind of typical overage costs have you seen when a customer/you use more than you've committed to?

Telehouse London is 0.75 (GBP) per KWH of overage. Obviously it will depend on datacentre/country. Telehouse increase this annually at 2% above inflation measured against the RPI (last increase was 5.27%)

We would typically order a 20 or 30-amp 208v circuit per rack for a flat fee then install a metered PDU to make sure we didn’t overload it. The flat fee per-circuit seems pretty standard in the US. Using your own metered PDU would help predict the usage if you’re being billed by kWH.


If we saw a dc customer approaching their power limit we would typically have a conversation or message them to make sure they were aware. If they anticipated needing more sustained power we would update their agreement accordingly. Sometimes it would be a one-time occurrence so we would typically let it slide. If they needed additional power usually it was in 10 Amps per cabinet or 5-10kW increments if in a larger footprint (cage or suite). We typically did not punish our customers for overages because we wanted them to continue to grow with us and we had the additional power capacity available. Some colo providers will put a premium on power overages 5-50% (I have seen all over the board) in order to make additional revenue on power spikes, make the customer more accountable to staying within their power commit or maybe the power is limited at that dc or on that floor. Hope this helps. Feel free to call with additional questions - 816-213-7731.


We see lots of different approaches to this, depending on the datacenter operator:

This isn’t a data center specific answer but it may help explain upstream cost drivers that may factor into rates seen. For commercial and industrial customers, utilities will often measure and bill power demand in kW separate from energy in kWh. Energy may only be $0.04/kWh but demand charges can be tens of dollars per peak kW during a bill cycle (averaged over 15 minute intervals).

The demand fees are a result of grid capacity requirements…the utilities need to maintain enough readily available generation capacity (turbines, storage, etc.) at all times to handle the peak loads, even if those loads are only seen once a year. If demand ever exceeds capacity, then brownouts and blackouts are on the plate.

Assuming a $10/kW demand charge, a customer consuming 5000 kWh energy and drawing a 15 min peak average of 40 kW would pay $200 for energy + $400 for demand, for a monthly electric cost of $600. If the same customer could do load leveling to get those peaks down to 10 kW, they’d pay a $100 demand charge instead of $400, for a monthly electric cost of $300.

With perfectly constant loads, 5000 kWh consumed over a 30 day month (720 hours) translates to 6.94 kW. 20 amp 240v circuits can each supply up to 3.84 kW continuous using the 80% loading rules. At 208v, that drops to 3.328 kW continuous.

Said another way, with level loads, approximately 5000 kWh per month could be pulled through two 20 amp circuits running close to their capacity limits 24x7.

Here is an example rate sheet for Xcel Energy in the US state of Colorado: