Peak Load Shaving for Data Centers with Peak+

What is Peak Load Shaving? 

Electric utilities bill for demand and consumption.  Demand is driven by the peak power draw at any given moment within a 15-minute time interval and is measured in kilowatts (kWs).  Consumption is the total amount of energy used within a given time and is measured in kilowatt-hours (kWh).  In a data center, most energy conservation measures (ECMs) such as: containment, VFDs, lighting, EC fan upgrades, raising temperature set points, etc…reduce electric consumption, but many utilities prefer to reduce peak demand. The illustration below shows an example of power draw throughout the day also known as a data center’s load profile.  A data centers (and most other facilities) peak power use will likely come during the hottest hours of the afternoon when their cooling load is at its highest output. Performing the ECMs mentioned above will reduce the load profile across the board, which will certainly help the utilities, but what the utilities really care about in this context is flattening the load profile (or, put another way, reducing variance and making grid demand more predictable).  Flattening the profile may even lead to a lower consumption rate due to the pressure it helps take off the grid.

Reducing Peak demand is very valuable for utilities (less power generation needed) and they incentivize their customers (e.g. Data Centers) to reduce overall demand on the grid.  They do this through two mechanisms: (1) the price of a kW of electricity and (2) the amount of money they pay data center customers to make improvements that reduce electric demand.  (1) A data center electric bill will always include a demand ($ / kW) charge and that charge can fluctuate during different seasons with summer demand charges sometimes three times the winter demand charge.  This pricing signal tells data centers that the utility wants you to reduce summertime demand.  (2) Most utilities also pay out rebates for demand reduction technologies and these rebates are often priced on a $ / kW basis.  Austin Energy in Texas, for instance, pays a $350 / kW rebate for Peak+ technology. 

How does Peak+ lower the Peak Load

To Illustrate this point, the power needs for a Data Center break down into three buckets:

  1. IT Load:  servers, etc.

  2. Non-IT Load: Other:  things like lighting, refrigerator, coffee maker, whose draw is fairly predictable.

  3. Non-IT Load: Cooling: chillers, pumps, fans, DX systems, heat rejection energy use will increase as outdoor ambient dry bulb increases.

The Peak+ system attaches to virtually any air-cooled heat rejection equipment (chillers, condensers, etc.) and pre-cools the air through evaporative cooling (you can observe this temperature lowering effect in our published site data). On hot days, Peak+ will lower the air inlet temperature of your HVAC, increasing its efficiency, and lower the amount of power needed to cool your facility. The effect is the greatest the hotter and the dryer that outdoor temperature which aligns with peak demand hours.

What this means for a Data Center:

Let us assume that the graph above Illustrates a 3MW leasable data center with a peak PUE of 1.83.  With fully loaded IT in the worst-case weather the facility was designed for the data center would pull 5.5MW of power from the grid. Climate change related heat waves are causing major data center markets to hit new historic peak demand highs the facilities were not designed for.

The table below illustrates this scenario at different peak load conditions:

The Baseline scenario in the above table shows the breakout of Peak Demand across the different buckets without Peak+ installed. The PUE reduction created by the installation has different implications depending on the scenario:

In a Retrofit Scenario, where Peak+ is installed on a datacenter that has already been designed & constructed, there is a 9%[1] improvement in PUE (1.83 > 1.67). Here the datacenter operator will benefit from substantial electric utility bill savings by reducing their peak demand charges from the utility and the additional safety buffer of increased peak demand capacity on very high temperature days.

In a New Construction Scenario, where the datacenter design has not yet been finalized, the datacenter owner can reallocate (up to) 0.5MW of peak capacity freed up by installing Peak+ into IT load for the datacenter.  Increasing their top-line revenue potential and improving their peak PUE by 14% over the baseline scenario (1.83>1.57).  A retrofit owner may also be able to take advantage of the load redistribution in this way as well through an infrastructure reallocation project.

In summary, among the benefits of Peak+ is that it will reliably shave off 8%+ of peak electricity demand. The datacenter owner can benefit from this by keeping the status quo and realizing substantial savings on their energy bill, or reallocating load to their IT equipment and increasing top line revenue potential.

Peak+ has 20+ active Data Center installations across 11 states with several key partners who recognize these benefits. If you have questions or are interested in our product please submit a contact form and we will get back to you shortly.




[1] Assuming a 25% reduction in HVAC load at peak conditions, a figure derived from a real installation on a data center in Irvine, California.

Previous
Previous

Why You Shouldn’t Spray Water on you HVAC

Next
Next

Regional Site Analysis: Las Vegas