What a Directional Control Valve Is Doing in Each Position
Why this matters
Never put any part of your body under a load held up by hydraulics. Lower it, or set a mechanical stand or blocking rated for the load, before anyone reaches in. A raised load is stored energy and 29 CFR 1910.147(d)(5)(i) requires it be relieved or restrained after the isolation devices are locked out, not merely left where the valve is holding it. The reason this article opens there rather than closing there is that its central finding is that no directional valve center condition is a load-holding device, and the field belief that a "closed center holds it" is what puts people under raised platforms.
A three-position four-way directional valve has three positions, and only one of them is a design decision. The two shifted positions do the obvious thing. The center is where the engineering lives, because the center is what the machine does when nobody is commanding it, which is most of its life.
The circuit the center sits in
reservoir --- pump --+--- relief --- reservoir
|
P
directional valve
A B
| |
cap end rod end
+-- cylinder -+
the valve's T port also returns to reservoir
The relief valve is teed off the pump line ahead of the directional valve, so it sees pump pressure no matter what position the valve is in. That is the whole reason the center condition matters: with the valve centered and no path to tank through it, the pump still has to put 100 percent of its flow somewhere, and the only door left is the relief.
The two shifted positions
Shift one way and P connects to A while B connects to T. Shift the other way and P connects to B while A connects to T. The cylinder extends or retracts, the pump sees whatever pressure the load requires, and the return oil goes back to the reservoir. There is very little to choose here between one valve and another. The sibling card on what pressure and flow each decide covers what the gauge reads during these two positions and why it is usually well under the relief setting.
The four common centers
Open center. All four ports interconnected. Pump flow goes straight to tank at low pressure, and both cylinder ports are open to tank, so the actuator is free to be pushed by its load. Pump unloaded, load not held. Common on mobile equipment with fixed pumps and a single function.
Closed center. All four ports blocked. Pump flow has nowhere to go through the valve, so pressure rises to the relief setting and the full pump flow crosses the relief as heat unless something else unloads it. Both cylinder ports are blocked, so the actuator is held by trapped oil, subject to the leakage this card exists to warn about. This is the center you use when several functions share one pressure supply, and it is the center that requires either a pressure-compensated pump or a separate unloading arrangement, because a fixed pump deadheaded on relief through a whole shift is the heat case worked in the card on what a hydraulic system is trading away.
Tandem center. P connected to T, A and B blocked. The pump unloads to tank at low pressure while the actuator stays held by trapped oil. This is the center that gets a fixed-pump circuit out of the heat problem without giving up holding, and it is why the same machine can be a temperature disaster or entirely uncooled depending on one spool choice.
Float center. P blocked, A and B connected to T. The actuator floats freely while the pump deadheads. Sometimes called a motor spool, because on a hydraulic motor it lets the motor coast to a stop instead of stopping against trapped oil.
Spool naming and port patterns vary between manufacturers and there are more centers than these four; the valve's own catalog symbol, not a name, is what tells you which one you are holding.
What no center does: hold a load
A directional valve spool is a ground cylinder sliding in a ground bore with a small clearance between them. That clearance is what lets it move. Oil crosses it, always, and the amount is a published specification in cubic inches or cubic centimetres per minute at a stated pressure, a stated fluid and a stated temperature. It is a number, not a zero.
Two things move that number, both in the direction people do not expect:
- Leakage rises roughly in proportion to the pressure difference across the spool, because flow through a small annular clearance at these dimensions is laminar. A valve quoted at rated pressure leaks proportionally less at a lower holding pressure.
- Leakage rises as viscosity falls, so a hot machine drifts faster than the same machine cold, and the datasheet figure applies at the fluid and temperature the datasheet names.
Holding a load is a job for a device that seats rather than slides: a pilot-operated check valve or a counterbalance valve, mounted at the cylinder port so a hose failure between the valve and the cylinder cannot drop the load. Those seal on a poppet against a seat, which is why they can be specified as essentially leak-free while a spool cannot. Neither one is a substitute for mechanical blocking when a person goes underneath.
Worked example: the platform that drifted two inches in ten minutes
A maintenance platform on a single cylinder, 3 inch bore, so 7.07 square inches of piston area. Load 6,000 lbf. The valve is closed center. The operator reports it settles about 2 inches over a coffee break and wants the valve replaced.
Before measuring anything, the platform is lowered to its stops and a mechanical stand rated above 6,000 lbf is set under it for any work that puts hands in the travel path. The drift measurement itself is taken with a rule and a standoff, from beside the platform, with nobody underneath and nothing below it that matters.
Holding pressure. Nothing is flowing while the valve is centered, so there is no back-pressure term to add: pressure on the cap side is load divided by area, 6,000 lbf divided by 7.07 square inches, which is 849 psi. That is a static hold, and it is a different arithmetic from a moving stroke, where the return path's back pressure has to be added first.
What drifted. Two inches of travel on 7.07 square inches is 14.1 cubic inches of oil that left the cap side, over 10 minutes, which is 1.41 cubic inches per minute. In flow units a tech would recognize, that is about 0.006 gpm, or roughly 23 cubic centimetres per minute. It is a trickle, and it is a trickle that moved a platform two inches.
The comparison, corrected onto the same basis. Say the valve's datasheet quotes 6 cubic inches per minute of spool leakage at 3,000 psi, at its stated reference fluid and temperature. That figure cannot be compared to the measurement as printed, because the measurement was taken at 849 psi.
- Datasheet figure: 6.0 cubic inches per minute at 3,000 psi.
- Pressure correction, laminar clearance flow proportional to pressure difference: 6.0 times 849 divided by 3,000, which is 1.70 cubic inches per minute expected at this holding pressure.
- Temperature correction, in the direction that makes the valve look worse, not better: the machine had been running and the oil was above the datasheet's reference temperature, so the thinner oil should leak somewhat more than 1.70, and the expected figure is a floor rather than a ceiling.
- Measured: 1.41 cubic inches per minute.
The finding. Measured leakage is below what the valve is specified to do at that pressure, and further below what it is entitled to do hot. The valve is healthy. A new valve of the same model would drift the same two inches, and the shop would have spent a valve and a shutdown to reproduce the complaint.
What actually fixes it. A pilot-operated check valve at the cap port. It seats, so it holds; it opens only when pilot pressure from the retract side commands lowering; and being at the cylinder it also holds the load if the hose to the valve fails. The center condition of the directional valve does not change at all.
The failure mode of getting this wrong. The common version is not the wasted valve, it is what the wasted valve teaches. After the replacement fails to fix it, the next move is usually to raise the relief setting or add a check valve of the wrong type somewhere convenient, and meanwhile the platform is still being worked under on the strength of "we fixed the drift." The measurement above takes ten minutes and a rule, and it converts an argument about a part into a number with a specification beside it.
What flips the choice of center
- More than one function on one pump. Open and tandem centers connect P to T, so the first valve in a series circuit steals the supply. Multiple independent functions on a shared supply generally means closed center plus a pump that can hold pressure without dumping flow.
- A pressure-compensated pump. A compensator regulates by sensing pressure, so it needs a circuit that lets pressure build when nothing is moving. An open or tandem center routes pump flow to tank at low pressure during center, which is exactly the condition a compensator cannot act on. The card on what a pressure-compensated pump changes carries that pairing; open center and compensated pump is a mismatch, not a tuning problem.
- A load that must not be trapped. Some machines are actively dangerous with the actuator held. Anything that must yield to an external force, or must be movable by hand after a shutdown, wants a float or open center and a separate, deliberate holding device where holding is required.
- A hydraulic motor rather than a cylinder. A rotating mass stopped against blocked ports generates a pressure spike as its own inertia keeps driving the motor. That is what the float center exists for, and it is why the center choice on a motor circuit is a mechanical question, not a hydraulic preference.
How to verify you read the center correctly
- Center the valve with the machine running and read the pump gauge. At or near the relief setting means the center blocks P: closed or float. Near zero means P is open to tank: open or tandem.
- Then check whether the actuator is held. Push or load it, or watch it under its own weight. Held means A and B are blocked: closed or tandem. Free means they are open to tank: open or float.
- Those two observations together name the center, and they take under a minute with no schematic. If the schematic in the cabinet disagrees with what the machine does, the valve has been changed at some point and the schematic is the thing that is wrong.
References
- 29 CFR 1910.147, the control of hazardous energy, including the requirement at (d)(5)(i) that stored and residual energy, which includes a raised load, be relieved or restrained after isolation
- Valve manufacturer catalog for the spool symbol, the published internal leakage at a stated pressure, fluid and temperature, and the pressure rating, all of which own the numbers this card only shows how to use
- See related: What Pressure and Flow Each Decide in a Hydraulic Circuit; What a Pressure Compensated Pump Changes About the Circuit; What a Hydraulic System Is Actually Trading Away