Where the standard chilled-water flow-rate formula comes from, why 500 is the constant used, how design ΔT choices trade off pumping energy against pipe size, and its limits as a sizing tool.
The Formula and Where It Comes From
The standard rule-of-thumb relationship for chilled-water flow rate is GPM = BTU/hr ÷ (500 × ΔT°F), where ΔT is the temperature difference between chilled-water supply and return. The constant 500 isn't arbitrary — it's derived from water's physical properties: one gallon of water weighs approximately 8.33 lb, and water's specific heat is approximately 1 BTU per lb per °F, so 8.33 lb/gal × 60 min/hr ≈ 500 (lb·min)/(gal·hr) — the constant that converts a heat rate in BTU/hr and a temperature rise in °F directly into a flow rate in gallons per minute.
Working Through the Data Center Application
For a data center rejecting, say, 13,648,560 BTU/hr (equivalent to about 4 MW of IT load), with a typical chilled-water design ΔT of 10°F: GPM = 13,648,560 ÷ (500 × 10) ≈ 2,730 GPM. This is the flow rate the chilled-water pumping system needs to circulate through the cooling loop — cold plates, CDUs, or CRAH coils — to remove that much heat given a 10°F temperature rise across the loop.
Why the Design ΔT Choice Matters — The Flow/Pipe-Size Tradeoff
The formula makes the tradeoff explicit: for a fixed heat load, a larger design ΔT (bigger temperature difference between supply and return) reduces the required flow rate, and vice versa. A smaller flow rate for the same heat load means smaller pipes, smaller pumps, and less pumping energy — all real cost and efficiency benefits. This is why data center chilled-water systems are often designed with wider ΔTs (12-16°F or higher in some modern designs) than older commercial HVAC conventions (which commonly used a narrower 8-10°F ΔT) — the wider delta directly reduces both capital cost (smaller pipe and pump infrastructure) and ongoing pumping energy.
What Limits How Wide the ΔT Can Practically Go
Design ΔT can't be increased without limit — the practical ceiling depends on the cooling equipment's actual performance characteristics (a CDU or cold plate has a rated operating range for both flow and temperature rise) and the chiller plant's own design parameters. Pushing ΔT beyond what the connected equipment is actually rated for doesn't produce the flow-reduction benefit in practice — it just means the equipment underperforms its rated heat-rejection capacity. Real system design checks the formula's output against manufacturer equipment specifications, not just the arithmetic result in isolation.
Why This Is a First-Pass Estimate, Not a Final Design Number
This formula assumes a single, uniform ΔT across the entire loop and doesn't account for pressure drop, pipe sizing constraints, multiple parallel loops with different ΔTs (common when liquid and air cooling loops are served separately, as covered in the companion liquid/air split article), or the specific performance curves of the actual pumps and heat exchangers to be installed. It's the right tool for an early sizing check — confirming an order-of-magnitude flow rate before detailed mechanical design — not for final chilled-water plant specification, which requires full hydraulic modeling of the actual piping layout.