Why size on the peak rather than the average?

Because the load is present whenever the lamp is. A chiller sized on a half-duty average meets half the demand while the lamp is struck, so loop temperature climbs until the system trips on high pressure or the lamp trips on coolant temperature. Duty cycle is a real and useful quantity, for energy, for the plant room, for the bill. It is not a capacity.

Is there ever a case for sizing below the peak?

Yes, with a properly sized buffer tank. A buffer stores the excess heat during the on period and rejects it during the off period, which is exactly what lets a smaller chiller serve an intermittent load. That is an engineered solution with a calculable tank volume, not the same thing as multiplying a peak load by a duty fraction and hoping.

What ΔT should I design around?

It is bounded from both ends. A wide ΔT means less flow, smaller pipes and a cheaper pump, but higher return temperature and more variation at the lamp. A narrow one gives better temperature stability at the cost of flow, and the lamp itself usually specifies a minimum flow and a maximum pressure drop, which is the constraint that decides it in practice. Start from the lamp manufacturer's figures rather than a rule of thumb.

Does the 500 constant apply to my glycol loop?

No. 500 is 8.34 lb/gal × 60 min/hr × 1 BTU/lb·°F, which is water. A glycol mix has a lower specific heat and a different density, so it carries less heat per gallon. Commonly 5–15% less depending on concentration, and needs proportionally more flow for the same duty. Use the manufacturer's fluid data rather than this page's constant when the loop is not plain water.

What else should I add to the chiller's duty?

Pump heat, which goes straight into the fluid; ambient gain along an uninsulated loop; and any derating the chiller suffers at your maximum ambient temperature, which for a rooftop or unconditioned plant room can be substantial. A modest margin over this page's figure is normal practice, and the honest place to put it is on top of the peak rather than hidden inside a fraction.

What actually kills lamp cooling jackets?

Water quality, not capacity. Scale, corrosion products and biological growth restrict the narrow passages inside a lamp's cooling jacket, and the first symptom is usually a lamp failing early for reasons blamed on the lamp. A treated, filtered closed loop with the manufacturer's recommended fluid outlasts an oversized chiller on untreated water every time.

UV Curing calculator

UV Cooling Calculator: Size on the Peak, Not the Average

Size the cooling a water-cooled UV system needs, and the flow to deliver it. Enter the system's electrical input, the share of it the cooling loop must remove, the runtime fraction, and the coolant temperature rise the loop is designed around. The calculator returns the peak capacity, what the chiller must reject while the lamp is on, the coolant flow that implies, and separately the average heat rejected over the hour. A runtime fraction below one reduces the second and not the first, which is the distinction that decides whether a chiller holds or trips.

What this calculator does

Size the cooling a UV system needs on the load it rejects while running, and convert that into the coolant flow the loop has to carry.
Use it for specifying a chiller for a water-cooled UV system, checking whether an existing loop can carry a larger lamp, sizing pump and pipework from a heat load, separating chiller capacity from plant-room heat balance, testing whether a tighter ΔT is affordable in flow terms.
Size the cooling a UV system needs on the load it rejects while running, and convert that into the coolant flow the loop has to carry.

Formula used

Heat to remove while running (kW) = system input × heat-to-cooling fraction
Cooling capacity required (BTU/hr) = heat to remove × 3412.142
Coolant flow (GPM) = cooling capacity ÷ (500 × ΔT°F)
Average heat rejected (BTU/hr) = cooling capacity × runtime fraction
Applied fraction × duty = heat-to-cooling fraction × runtime fraction

Inputs explained

UV system electrical input: Total electrical input to the system while running, including the power supply.
Heat-to-cooling fraction: Share of input power the cooling loop has to remove. The rest leaves as UV, as exhaust air, or as radiation to the room and the part.
System runtime fraction per hour: Fraction of the hour the system runs. Sets the ENERGY rejected over the hour; it does not reduce the capacity the chiller needs while the lamp is on.
Coolant temperature rise (ΔT): Difference between the loop's return and supply temperatures at design flow. Turns a heat load into the flow the pump and pipework must actually deliver.

How to use the result

Best suited to specifying a chiller for a water-cooled UV system, checking whether an existing loop can carry a larger lamp, sizing pump and pipework from a heat load, separating chiller capacity from plant-room heat balance, testing whether a tighter ΔT is affordable in flow terms.
The 500 constant is water-specific. A glycol mix carries less heat per gallon. Typically 5–15% less depending on concentration, so the flow figure is optimistic for any loop that is not plain water. Ignores loop losses, pump heat and ambient gain, all of which add to the chiller's real duty. Assumes no buffer tank. A well-sized buffer genuinely does let a smaller chiller serve an intermittent load, which is the one legitimate case for sizing below the peak. Gives no allowance for fouling, scaling or the derating a chiller suffers at high ambient temperature. Says nothing about coolant quality, which is what actually ends most lamp cooling jackets.

Common questions

Why size on the peak rather than the average? Because the load is present whenever the lamp is. A chiller sized on a half-duty average meets half the demand while the lamp is struck, so loop temperature climbs until the system trips on high pressure or the lamp trips on coolant temperature. Duty cycle is a real and useful quantity, for energy, for the plant room, for the bill. It is not a capacity.
Is there ever a case for sizing below the peak? Yes, with a properly sized buffer tank. A buffer stores the excess heat during the on period and rejects it during the off period, which is exactly what lets a smaller chiller serve an intermittent load. That is an engineered solution with a calculable tank volume, not the same thing as multiplying a peak load by a duty fraction and hoping.
What ΔT should I design around? It is bounded from both ends. A wide ΔT means less flow, smaller pipes and a cheaper pump, but higher return temperature and more variation at the lamp. A narrow one gives better temperature stability at the cost of flow, and the lamp itself usually specifies a minimum flow and a maximum pressure drop, which is the constraint that decides it in practice. Start from the lamp manufacturer's figures rather than a rule of thumb.
Does the 500 constant apply to my glycol loop? No. 500 is 8.34 lb/gal × 60 min/hr × 1 BTU/lb·°F, which is water. A glycol mix has a lower specific heat and a different density, so it carries less heat per gallon. Commonly 5–15% less depending on concentration, and needs proportionally more flow for the same duty. Use the manufacturer's fluid data rather than this page's constant when the loop is not plain water.
What else should I add to the chiller's duty? Pump heat, which goes straight into the fluid; ambient gain along an uninsulated loop; and any derating the chiller suffers at your maximum ambient temperature, which for a rooftop or unconditioned plant room can be substantial. A modest margin over this page's figure is normal practice, and the honest place to put it is on top of the peak rather than hidden inside a fraction.

Last reviewed 2026-08-25.