Hydraulic fixture power unit sizing: the number everyone gets wrong is flow, not pressure

The first fork: high pressure with small cylinders, or low pressure with big ones

Both families are legitimate. They are not interchangeable, and the choice is made by two facts about your part, not by preference.

The decision rule that holds up on real projects: if the fixture is crowded, go high pressure; if the fixture is open and the clamps are few, low pressure is easier to live with. A tombstone with twelve swing clamps has nowhere to put twelve 100 mm cylinders, so the pressure has to go up and the bores have to come down. A single large hydraulic vise holding a casting has room, and running it at 60 bar keeps the seals and the hose simple.

What you must not do is mix. A 60 bar cylinder on a 300 bar circuit is a burst hazard, not a bargain. Every element in the circuit must be rated at or above the relief valve setting — and the relief valve setting, not the pump’s nominal pressure, is the number that defines the system.

Size the pump from the cycle, not from the cylinder

Here is where the arithmetic goes wrong, and it goes wrong in a very specific way: people compute the force for one clamp, pick a cylinder, and then pick a pump that matches that cylinder. The pump then has to move eight of them at once.

A worked example, using round numbers that are easy to check. Eight swing clamps, each with a 50 mm piston and a 20 mm total stroke:

Two things follow. First, 12.6 L/min is at the very top of what a machine-tool clamping power unit is built to deliver — the catalogue envelope for this class of unit tops out around 12 L/min, and the common configurations sit at 1–3 L/min. Second, reaching 12.6 L/min with a pump alone means a motor several times larger than the duty cycle actually requires, because that flow is needed for 1.5 seconds every two minutes.

That is the whole argument for an accumulator, and it is a flow argument, not a pressure argument. The pump is sized for average demand; the accumulator covers peak demand. Confusing the two produces either a pump that is absurdly oversized or a clamp cycle that visibly stalls.

Two details in the table are worth flagging because they are commonly missed. The stroke must be the full stroke: on a swing clamp, part of the travel is the swing, which moves no clamping force but still consumes oil. And the cylinders that matter are the ones that move simultaneously — not all eight if the circuit is sequenced in two banks, but certainly not one either.

Motor power then follows from pressure and flow, not from intuition: kW ≈ (bar × L/min) ÷ 600 ÷ pump efficiency. At 300 bar and 1.2 L/min with 85% efficiency that is about 0.7 kW, which is why the 0.75 kW frame is the smallest standard unit in this class and why a 3 kW unit on a fixture circuit is usually a sign that somebody sized for peak flow with the pump.

Pressure at the pump is not clamping force at the part

Between the pressure gauge and the workpiece there is a chain of losses, and each link is a few percent that nobody budgets for:

• Line loss. A long run of small-bore tube with several quick couplings drops pressure even at low flow. On a fixture with a quick-disconnect at the pallet, expect the drop to be concentrated at the couplings rather than spread along the tube.
• Check valve and sequence valve cracking pressure. A pilot-operated check valve to hold the clamp closed has its own opening behaviour; a sequence valve holding clamp pressure until the work supports are loaded costs a few bar by design.
• Return-line restriction. Double-acting cylinders need somewhere for the oil to go. A return line that is too small slows the stroke and, in the worst case, lets the rod side go sub-atmospheric and draw air in.
• Mechanism loss inside the clamp. This is the largest single item and the least visible. A swing clamp converts piston force into arm force through a swing mechanism and a lever, and the rated clamping force is roughly 40–45% of the theoretical piston force. A 50 mm piston at 250 bar is about 49 kN of theoretical force; the clamp delivers roughly 20 kN at the arm.

So the sizing pressure is not the cylinder’s rated pressure and not the pump’s maximum. It is:

p_set = (required clamp force ÷ total effective piston area ÷ mechanism efficiency) + estimated line loss

and the relief valve is set slightly above that, with every component rated above the relief setting. If you find yourself setting the relief valve at the maximum the elements allow in order to reach the required force, the cylinder is undersized — raise the bore, do not raise the pressure.

The accumulator: two different sums, and only one of them is a peak-flow sum

An accumulator on a fixture circuit does two jobs that are calculated in completely different ways, and this is where most sizing goes wrong. The two jobs use different values of the polytropic exponent n in the gas law, and picking the wrong one undersizes the vessel by roughly a quarter to a third.

The working form of the relation is:

V₀ = ΔV ÷ [ (p₀/p₁)^1/n − (p₀/p₂)^1/n ]

where V₀ is the nominal gas volume, p₀ the nitrogen precharge, p₁ the higher working pressure and p₂ the lower working pressure, and ΔV the oil volume that has to be delivered or absorbed.

Job one: covering the clamp stroke — use n = 1.4

A clamp stroke that finishes in about a second is an adiabatic discharge; the gas has no time to exchange heat. Taking the same eight-clamp example: ΔV = 0.314 L, precharge 225 bar (90% of a 250 bar minimum), working band 250–300 bar, n = 1.4.

Run the same numbers with n = 1.0 — the isothermal form that gets used because it is the one people remember — and the answer comes out at 2.1 L. A 2 L bladder would then be ordered, and it would be about a third short of what the stroke needs. The symptom is a pressure sag on every clamp stroke, a relief valve that keeps cracking, and a circuit that hammers.

Job two: holding against leakage — use n = 1.0

Holding is the opposite case. The clamp sits closed for twenty minutes of cutting, oil leaks past seals at some small rate, and the accumulator has to give up that volume slowly enough that heat can move and pressure stays inside a band. Over that timescale the process is isothermal. With 0.02 L/min of leakage over a ten-minute interval, ΔV = 0.2 L, and the same 225 bar precharge gives roughly 2 L of vessel.

So the same accumulator has two answers — 2.8 L for the stroke, 2.0 L for the hold — and you specify the larger. What you must not do is compute one figure, apply it to both jobs, and assume it covers them.

There is also a diagnostic value in separating the two. If a circuit needs recharging far more often than the leakage calculation predicts, the accumulator is not the problem; the leak is. Chasing it with a bigger vessel hides a failing seal and gives you a bigger vessel that still cycles.

Precharge: the 90% rule and the 4:1 rule do not agree

Two rules circulate, and they produce very different numbers on the same circuit.

• Percent of minimum working pressure. Precharge to about 90% of the lowest pressure the circuit must hold at (95% for piston accumulators). On a 250 bar minimum, that is 225 bar.
• Quarter of maximum pressure. Precharge to about 25% of the maximum working pressure, a 4:1 ratio sometimes quoted for clamping and fabrication circuits. On a 300 bar maximum, that is 75 bar.

They cannot both be right, and on a fixture circuit of this type the percentage rule is the one that works. The 4:1 rule belongs to circuits where the pressure swings across a very wide band or where the accumulator is being used as a crude energy reservoir; applied to a 250–300 bar clamping band it puts the precharge so far below the working pressure that the gas has already compressed most of the way before the useful part of the stroke begins, and the usable volume collapses.

The hard constraint is simpler than either rule: precharge must be below the minimum working pressure, with enough margin that the bladder does not bottom on its seat. Below that ceiling, higher precharge gives more usable volume in a narrow working band, which is exactly the situation a clamping circuit is in.

And precharge is not a set-and-forget number. Nitrogen pressure moves with temperature by roughly 1% per 3 °C. An accumulator charged in a cool workshop in spring and run through a summer shift will not be at the pressure it was charged to, and the usable volume moves with it. Recheck annually, and recheck after any relocation.

A pressure switch does not prove the clamp closed

This is the most consequential misunderstanding in the whole circuit, and it survives because the wiring looks correct.

A pressure switch confirms pressure at the location where it is mounted. What the process needs to know is whether the clamp reached its closed position and is holding force at the workpiece. Those two statements are not the same, and there is a whole family of failures that sit between them: a clamp arm that stopped against a chip, a jaw that never contacted the part, a cylinder that stroked fully while the arm was blocked, a check valve trapping pressure in a branch whose clamp never moved.

Practically, that means:

• Mount the pressure switch so that the pressure it senses represents the state you actually care about — in the clamp branch, downstream of anything that could trap pressure, not on the pump outlet.
• Where seating matters, confirm seating independently with a position sensor or a proximity switch. Pressure alone will not tell you the part is down on its locators.
• Sequence supports before clamps with a sequence valve so that clamp force is not applied over an unlocked support — and coordinate that sequence pressure with the actual requirements of both circuits rather than picking a percentage of system pressure.
• Never use a timer as the primary confirmation. A timer set to 0.5 s will pass whether or not the clamp moved, because it measures only that half a second elapsed. Timers are a fallback for the case where a sensor fails, not a substitute for the sensor.

On the machine side this becomes an M-code sequence: command clamp, wait for the confirmation input, and only then enable the spindle and the feed. If the input never arrives, the control alarms rather than proceeding. The order matters more than the brand of the clamp.

Oil temperature moves everything downstream of it

Viscosity falls as oil warms, leakage rises, response slows, and precharge drifts. A circuit sized cold and run hot behaves like a circuit that was never sized at all. Two habits keep this under control: keep the reservoir sized generously enough that the oil has dwell time to shed heat (the 11 L and 27 L reservoirs in this class exist for a reason), and treat 50 °C as the point where you stop assuming and start measuring at the test point.

If the fixture runs three shifts, add a temperature switch to the monitoring rather than relying on somebody noticing that the clamp feels slower in August.

The sizing sheet

Five mistakes that show up on site

1. Sizing the pump from one cylinder. The pump serves the circuit, and the circuit moves several cylinders at once.
2. Using the isothermal exponent for a fast stroke. Undersizes the accumulator by a quarter to a third, and the symptom is a pressure sag you will misdiagnose as a weak pump.
3. Setting precharge by the 4:1 rule on a narrow pressure band. The gas is already compressed before the useful part of the stroke starts, and the vessel behaves like a pipe fitting.
4. Putting the pressure switch on the pump outlet. It confirms the pump is running, not that the clamp is closed, and it will happily pass a cycle with an unclamped part.
5. Raising the pressure to make up for an undersized cylinder. It works until it does not, and then it is a seal failure at the far end of the circuit rather than a specification error at the drawing.

The acceptance test almost nobody runs

Before the fixture goes into production, with a part clamped and the machine in a safe condition, cut the power to the hydraulic pump and watch. The part should stay clamped, held by the accumulator and the check valves, for long enough that a controlled stop is possible.

The rest of the test list is short and worth doing in the same session: trip the pump overload and check that the clamps hold; press unclamp during a cycle and confirm the control refuses; simulate loss of pressure mid-cycle and confirm the program feed-holds rather than continuing; try to release a pallet while it is unclamped and confirm it is inhibited. These are five minutes of checks that find the difference between a circuit that is drawn correctly and a circuit that is wired correctly.

The reason to run them is that a clamping fault is silent by construction. The program runs, the tool cuts, and the only signal that something is wrong arrives as a scrapped part or a broken cutter — or not at all, until the shift when the clamp lets go under load.