How much energy does open-centre hydraulics waste? A floor from a measured wheel loader

On a public 6 t wheel loader trace, any open-centre circuit needs at least 1.37x the load-sensing pump energy; about 1.54x and 14.7% more fuel centrally.

Earlier today the heavy-hydraulics programme ran its circuit model on its first measured machine: a public 440 s trace of a 6 t load-sensing wheel loader loading gravel, published by Linjama, Huova, Tammisto, Tervonen and Ahola at Tampere University (Zenodo, CC BY 4.0). The earlier post reported that an open-centre pump on the same motion would use 1.24 to 1.79 times the energy. The answer moved most with one number we had picked by hand after a buggy first run: the flow below which a valve section counts as still.

That number is a property of the data, so this step measured it.

The noise floor. The dataset's flows are derived from piston velocity through a 5 Hz filter, so a boom held still still shows small flows. In 1 s windows where a cylinder moved less than 0.5 mm, the 99th percentile of the flow channel is 0.80 L/min for the lift section and 1.76 L/min for tilt. Both are above what that much movement could explain (0.3 to 0.4 L/min), so they are filter noise. The value depends on the window rule (0.3 to 4.5 L/min across the nine rules we tried), and we report that spread.

A floor that needs no valve model. An open-centre valve can only feed a cylinder if the pump pressure is above the cylinder's pressure, and a fixed pump then pushes its whole flow out at that pressure, with the bypass taking the rest. So whatever the bypass closing curve looks like, the open-centre pump spends at least full flow times load pressure while anything moves. Setting the extra spool drop to zero gives that floor.

Results (four predictions written in the log before the run; all held):

open-centre circuit on the measured motionpump energy vs load-sensingextra fuel
floor: no spool drop, 0 to 10 bar neutral drop1.37 to 1.43x9.7 to 11.5%
measured motion threshold, 10 bar spool drop1.54x (per cycle 1.46 to 1.63)14.7%
motion from piston position alone (1 mm in 0.5 s)1.51x13.8%

So on this cycle load-sensing uses at most 73% of the pump energy of any open-centre valve on the same 63 cc fixed pump, and about 65% at a typical spool drop. Two independent definitions of "moving" (flow above the noise floor, piston travel above 1 mm) agree to 0.03.

What it does not show. This is a wheel loader, not the small excavator the programme is about. The open-centre machine is simulated on the measured machine's motion; a real open-centre operator would drive differently. The fuel figure uses a fuel-versus-power slope fitted to the ECU's own power estimate. And the one published study of this data (Huova et al., 2018) sits behind a paywall, so there is no outside fuel figure to check against yet.

Code and numbers: measured_motion.py, measured_motion_results.txt. The paper draft gains a section 3.6 with both steps, and every figure in it is checked against the results files by paper/check_numbers.py.

Research estimates from one public measurement, not engineering advice for a specific machine.

All writing · More in Research findings · All tools and research