For a standard fixed-displacement hydraulic power unit (HPU) without servo valves, the oil cooler should be sized to reject about 0.25-0.35 kW of heat for every 1.0 kW of installed electric motor input power, with one-third (33%) the most cited rule of thumb across industry references [S4][S5]. A 100 kW input running at 80% overall efficiency, for example, generates 20 kW of heat that must be removed, but real circuits with throttling valves and return-line losses routinely push that figure to 25-35 kW [S2].
The ratio is not a universal constant: piston-pump/motor loops can run with 25% loss while gear-pump/motor loops can need 28-35% rejection, and proportional or servo circuits can demand 50-90% of input power as cooling capacity [S2]. Cooler selection therefore starts with the circuit's actual heat load, not a catalog face-area.
The 1/3 Rule of Thumb, and Where It Breaks Down
The "one-third of input horsepower becomes heat" rule applies to general-purpose mobile and industrial hydraulics using fixed-displacement pumps with modest throttling losses [S4]. A 75 HP (≈56 kW) prime mover on an excavator or skid-steer therefore needs roughly 25 HP (≈18.6 kW) of cooler capacity, with the reservoir making up the residual margin in a stationary package [S4][S5]. The same ratio is restated as 33% of installed power in independent fluid-power guidance, which treats installed motor nameplate power (not pump output) as the cooling basis [S5].
The ratio diverges sharply once the system contains accumulators, proportional valves, or servo valves: published examples show installed cooling capacity of 50-90% of input power for those circuits, because the throttling losses occur across continuously-modulating spools rather than across on/off directional valves [S2]. A 100 kW servo press loop, for instance, can need 50-90 kW of heat rejection, which is the range where engineers stop trusting a 1/3 rule and start calculating per-component loss [S2]. The same logic is reflected in broader hydraulic power unit sizing references, where thermal capacity is treated as a function of pump type, valve type, and duty cycle rather than a single fixed coefficient.
From Pump/Motor Efficiency to a kW Number
The cleanest first-pass calculation converts input power to waste heat via round-trip efficiency. A piston-pump driving a piston-motor combination is 0.92 × 0.92 = 85% efficient at best case, leaving 15% as heat, and a gear-pump driving a gear-motor combination is 0.85 × 0.85 = 72% efficient, leaving 28% as heat, before any valve or line losses are added [S2]. With a 100 kW input, the piston system needs a cooler that can reject about 25 kW and the gear system about 35 kW, once conductors, valves, and a safety margin are layered in [S1][S2].
Those numbers line up with the practical formula: heat load (kW) = input power (kW) × (1 - overall efficiency), with the efficiency drawn from the pump/motor pair, then de-rated further for throttling. Independent cooler-sizing guidance confirms that 25 kW is a realistic reject figure for a piston-pump/motor HPU and 35 kW for a gear-pump/motor HPU of comparable input [S1]. The same input can produce wildly different cooler duty if a proportional valve is added, which is why the heat-load figure must be measured or estimated before any model is selected.
Air-Blast vs Liquid-Cooled: What the kW Number Drives
Once the heat load in kW is fixed, the choice between air-cooled and liquid-cooled heat exchangers is a packaging decision, not a thermodynamic one. Air-cooled cores (with electric or hydraulic-motor-driven fans) are the default on mobile machinery, where a hydraulic motor driving a fan is often the simplest install if spare flow is available [S4]. Electric-fan cores are common on smaller mobile platforms and on indoor skids where 24 VDC or 480 VAC is available, with the control electronics cheap and the fan curve easy to modulate [S4].
Water- or glycol-cooled exchangers (shell-and-tube, brazed-plate, or plate heat exchanger blocks) make sense when ambient air is too hot for an air blast to do the job, when the package must be sealed against dust or paint-shop contamination, or when a plant chilled-water loop is already available. Cross-references in cooler selection literature note that water-cooled options are usually more efficient per unit face area, but introduce water-side pumping, water-treatment, and (if the discharge is regulated) waste-water compliance cost [S4]. For stationary indoor HPU skids, the question often comes down to whether the heat can be dumped into a heat-treatment-furnace cooling loop or a process chilled-water circuit that already exists, versus installing a new radiator and fan stack.
Cooler Type, kW Density, and Pressure Drop
Three cooler architectures dominate HPU duty, and they trade kW-per-unit-volume against allowable pressure drop. Bar-and-plate air-cooled cores are the cheapest and lightest, suitable for the 5-30 kW reject range typical of small-to-mid mobile HPUs. Brazed-plate liquid-cooled blocks reject 30-100+ kW in a much smaller envelope, but add a 0.5-2.0 bar pressure drop on the oil side that must be budgeted into the return line. Shell-and-tube oil-to-water exchangers are the workhorse for large stationary HPUs and can be cleaned mechanically, which matters in dirty steel-mill or marine duty. [S3]
Pressure drop is the most underrated constraint in this comparison. Return-line back pressure must stay below the suction port's allowable limit on most piston pumps (typically 0.3-0.7 bar absolute for an open-circuit pump at sea level), and an over-sized cooler that pushes the return line at 2 bar can de-rate the pump or shorten hose life. A sand cooler or similar process-side reference exchanger follows the same pressure-drop rule of thumb: the oil-side ΔP is a function of flow, core geometry, and oil viscosity at operating temperature, not of the kW nameplate.
Sizing Procedure, Field Numbers, and Acceptance Margins
A defensible cooler selection walks five steps. First, measure or calculate the heat load in kW from input power and round-trip efficiency, with explicit adjustment for proportional/servo valves. Second, confirm the oil flow through the cooler: most air-cooled cores need 1-3 L/min per kW rejected, depending on the ΔT target between oil inlet and outlet. Third, check the available ΔT: a 10°C oil-side temperature rise is typical for mobile equipment, while a stationary HPU running close to a 60°C oil target often sizes for 5-8°C rise to stay under the viscosity limit of the ISO VG 32 or 46 oil in the tank. Fourth, verify the return-line pressure budget. Fifth, add 20-30% margin on kW, because fouling, altitude, and summer ambients will erode nameplate performance within a year. [S3]
A worked example: 75 kW input through a piston-pump/motor loop with proportional valves. Round-trip efficiency before valves might be 85%, so 11.25 kW of pure conversion loss, but the proportional valve throttling can add another 10-15 kW, putting total heat load at 25 kW. Applying the 1/3 rule would undersize this cooler at 25 kW exactly, leaving zero margin; a safer specification is 30-35 kW of air-cooled capacity, which is consistent with published sizing examples for similar loops [S1][S2][S3]. Independent cooler-sizing tutorials echo the same margin posture, recommending an oversize rather than an undersize for the same reliability reason.
What Goes Wrong When the Number Is Wrong
Undersized coolers show up as oil temperature creep after 20-30 minutes of continuous duty, falling cylinder speed, and rising pump noise as viscosity drops and internal leakage climbs. Seals start to leak within weeks once bulk oil temperature exceeds 80°C, and the hotter oil accelerates oxidation, which in turn fouls the cooler core and starts a positive feedback loop. The system is sometimes wrongly diagnosed as a "weak pump" when the real cause is 5-10 kW of missing heat rejection [S3].
Oversized coolers have a different failure mode: an air-cooled core that is too large for the loop's flow can fail to dump enough heat per pass, because the oil-to-air ΔT collapses and the unit operates in a low-effectiveness region. The cure is not a bigger face area but a more accurate flow and ΔT match, or a thermostat-bypassed cold bypass line that protects the core at startup. For low-duty intermittent systems, a simple rule of thumb is to verify that the reservoir alone can absorb the short-term heat load, and only then spec a cooler for the continuous-load kW figure, which is the same logic used to size buffer capacity on a degassing unit or any other thermal-mass-buffered process skid.
Trackable Signals for the Next Sizing Decision
Two signals tell you the next cooler decision is approaching. First, measured bulk-oil temperature at the tank return above 70°C under continuous duty, which is the point where seal life and oxidation rate start to drop sharply on mineral hydraulic oil. Second, the kW-per-kW reject ratio of your installed cooler: anything under 0.20 on a non-servo fixed-displacement HPU is unusually good and worth checking for a misread, while anything above 0.50 on the same circuit type is a flag for a throttling or return-line problem that a bigger cooler will not fix. [S3]
See also our earlier report, Dual-Slope vs Single-Slope Grade Lasers: Choosing by Axis Count, Fall Direction, and Job.