The ASHRAE technical committee Datacom book series makes mention of ice storage for consideration. While ice storage is useful if you have the up-and-downs that a normal office building has, it has much less application for a data center where load runs much more evenly throughout 24 hours. The amount of ice stored would amount to the solar & building envelope heat gains, which, for a typical data center, is a small fraction of the overall load.
The use of ice storage may end up being a detriment for modern data centers, as the ice storage would require lower temperatures to create the ice. Unless the ice storage was a separate system this may put critical machines at increased risk of failing. The ice storage would need to be able to be cut out of the cricital systems operation, which means there is cost of controls and
Ice storage pushes the chillers harder, which results in a higher kW/ton. While this happens at night during the utility’s off-peak hours, the increase in energy use would wipe out any savings during night time use. The savings would then need to be derived solely from the day time operation when the ice is melted to reduce loading. Unless the ice storage system is sufficiently large, the number of chillers operating during the day and night remains the same. The only difference is the decreased load to the chillers as the ice is melted, and this would need to be examined against the reduction in the kW/ton and the cost during that time. The extra pump energy should also be considered, and if the ice storage system should have a dedicated pump. A simple payback/return on investment for a stand-alone ice storage system would likely be long.
Lastly, how would the ice storage system release it’s stored capacity? It takes a much longer time to create the ice versus how quickly it would be to melt. While the ice storage system has all night to create a batch of ice, the data center heat load could wipe it out in minutes, if the ice storage could melt fast enough. Trying to use a trickle flow during the day’s hottest hours would be one strategy, but doing so would be difficult to do and actually see the chiller load lowered. If the ice storage system were immense as compared to the overall load it may work. However, generating that amount of ice inside of 12 hours of night operation would require more chillers than the original load.