A data center water request usually arrives as an annual volume or an average daily figure. Distribution systems do not fail on annual volume. They are constrained by the hour of greatest draw, the pressure that holds at the far end of a pressure zone, and the storage and pumping that carry the system through that hour. A utility that answers a large request with an average-day number has answered a question the pipe network does not ask.
The American Water Works Association has released a report, Cooling the Cloud: Considerations for Water Utilities Evaluating Data Center Development, intended to help utilities work through these requests. Water Finance and Management summarized the AWWA report, which the outlet describes as a tool for navigating these conversations. The report sets the frame for the sector. The capacity question still has to be answered inside your own system, against your own hydraulic model and your own records.
Why the average day misleads
A cooling load that runs continuously adds draw around the clock. That raises the average, but the more important effect is on the peak. A flat, round-the-clock block sits on top of whatever the surrounding residential and commercial customers are doing in their own peak hours. Whether the combined peak grows by the full continuous rate or by a smaller amount depends on the cooling design and the site's operating schedule, and neither is fixed until the applicant commits to a design.
Evaporative cooling needs attention. Its water draw tracks ambient heat and humidity, so the hottest afternoons may be the highest draw days for the site. Those are also the days when the municipal system is already at its highest demand because of lawn irrigation, air conditioning in homes, and commercial cooling. The two peaks can coincide. That coincidence is the mechanism to test, and an average-day figure will not reveal it.
Map the request onto the zone
Begin with the connection point. Identify the pressure zone it sits in, the diameter and age of the main that will serve it, the nearest booster station, and the storage that feeds that zone. Then compare the requested peak against the margin the zone has at its current peak hour. The test is zone margin, not system total. A system can have ample capacity on paper and still have a zone at the edge of its pressure envelope.
Fire flow reserve belongs in the same run. The fire flow requirement for the zone must remain satisfied with the data center load added at its peak. Model that condition separately from the normal peak-hour run. A hydraulic model is only as reliable as its demand allocation and pipe data. Older records may carry unknown diameters, undocumented closed valves, or demand assignments that no longer match the service area. Where the record is weak, say so in the file rather than presenting the output as settled.
What the applicant must supply
The utility cannot model a load it has not been given in usable form. The request should include an hourly or sub-daily profile, a peak-day figure, a peak-hour figure, the cooling type, and the seasonal variation the applicant expects. Ask for the phase schedule as well. Data center loads often arrive in stages, and the capacity commitment should follow the phases rather than the final build-out. A single final figure encourages a capacity decision for a load that will not exist for years, and it can hide the stages when the zone would first be stressed.
The tradeoffs
When the model shows a constraint, the utility has several options, and each carries cost and reliability consequences. The utility can require the applicant to fund a main or pump upgrade through a contribution arrangement. It can phase the connection so that capacity is added with the load. It can write curtailment terms that limit draw during the system's peak hours. Or it can decline the peak at that point and offer a different connection or a later date.
Curtailment reduces the capital burden, but it depends on the customer honoring the limit and on the utility holding a supply plan for the case where curtailment fails during a heat event. The uncertainty is real. If the upgrade timeline for pumping or mains does not match the load timeline, the utility may carry the risk for years. Write the timeline gap into the record, with the assumptions behind it, so that later staff can see which conditions the decision depended on.
What to keep before the first number is public
Keep the hydraulic model run, its inputs, the run date, and the assumptions with the request file. Separate the figures the applicant supplied from the figures the utility verified against its own records. If a public number is later quoted, staff should be able to show which run it came from and what conditions it assumed. The file should also record the zone margin at the peak hour, the fire flow result, and any phase at which the margin falls below the level the utility has set for itself.
The question for the board is not whether the system can deliver the water over a year. It is whether the system can deliver it at the hour that matters, through the zone that receives it, while keeping the fire flow reserve the code requires. Answer that question with records you can defend, and state the uncertainty where the records run out.