The heavy-hydraulics programme asks which hydraulic circuit a small shop could build to get a compact excavator closest to stock fuel use. A plain fixed pump with an open-centre valve wastes oil: it pumps its full flow whenever the machine moves, and the work uses only part of it. Earlier today we put a number on that for a real excavator. On ten truck-loading recordings of a 7.5 t class XCMG machine (the public fuxi-robot/excavator-motion dataset, CC BY-NC-SA 4.0), a fixed pump sized for peak demand delivers about 2.4 times the oil the cylinders use.
The obvious fix is a smaller pump. Then the cylinders slow down whenever the operator asks for more than the pump gives, so you pay in cycle time instead. On a measured wheel loader, matching load-sensing's pump energy that way cost 39% more time. This run asked what the same trade costs on the excavator.
What we did
We wrote down six predictions before reading any data, then ran the script as registered. Demand comes from the machine's joint speeds and the published cylinder sizes of the XCMG XE80E (method in the earlier post). Where demand is above the pump's flow, that moment stretches until the pump has delivered the oil, so the work is the same and only time is added. Load-sensing is charged its own 219 L/min pump limit. The "equal-energy" pump is the size at which open-centre's waste equals what a load-sensing pump loses to its 20 bar margin.
What came out
| Fixed pump | Break-even margin | Time vs load-sensing |
|---|---|---|
| 219 L/min (stock) | 118.8 bar | 1.000 |
| 175.2 L/min | 77.7 bar | 1.016 |
| 131.4 L/min | 38.4 bar | 1.056 |
| 109.5 L/min (half) | 21.1 bar | 1.108 |
| 87.6 L/min | 7.8 bar | 1.232 |
- The equal-energy pump is 108 L/min, 45% of peak demand and about half the stock pump, and the machine then takes 1.114 times load-sensing's time. Across margins of 14 to 25 bar and pressures of 100 to 174 bar the range is 1.07 to 1.19.
- Across 384 lever, bucket-sign and swing assumptions it is 1.05 to 1.69 (median 1.20). The loader's 1.39 is inside that range, but well above the central case.
- Each step down buys less. The first 10% off the pump buys 33.8 bar of break-even per percent of extra time; the step from 131 to 110 L/min buys 3.3.
Three of the six predictions failed: we expected the excavator to lose more than 25% in time, more than the loader, and more than 10% under every assumption. It lost less. The reason is the shape of its demand: a median of 90 L/min against peaks of 240. A half-size pump covers most of the record, and only the peaks slow down.
What it does not show
Fuel. On the loader, the record's own fuel signal showed that the extra time burns more than the pump energy saves, so no fixed pump won. These excavator files have no fuel or engine signal, so the same price can't be read here. The pressure we used comes from a different, larger excavator in the same dataset. The 10 Hz samples miss short peaks, and those would make a small pump slower. The break-even rule is a straight-line fit from the loader, and here it is used further from where it was checked.
So the honest summary is narrower than a headline. On pump energy, a small shop's open-centre excavator can get close to load-sensing for a modest time cost. On fuel, it's still open.
Script: measured_excavator_pump_size.py, output: measured_excavator_pump_size_results.txt, paper section 3.7c of the draft.