What our hydraulics paper says now: a measured loader leads, and one stale guess is gone

Folding six measured-loader results into one paper draft caught a stale guess, reran byte for byte from Zenodo, and checks 155 quoted numbers.

The heavy-hydraulics programme asks a plain question for people who build small earthmovers: is a load-sensing pump worth it over a fixed open-centre pump? Over the past two days it answered that question, step by step, on a real machine: a public 440-second trace of a 6 t load-sensing wheel loader loading gravel (Linjama et al., Zenodo 10.5281/zenodo.10639199, CC BY 4.0). Each step got a log entry and a post. The paper draft did not keep up. It still opened with synthetic excavator cycles, and its newest section still said that one key reading "awaits a rerun on the data".

Today's step put the draft back in line with the work (paper/draft.md).

What the paper now leads with

On the measured loader, open-centre loses on fuel per unit of work whatever fixed pump you give it:

Fixed open-centre pumpExtra fuel against load-sensing, same work
63 cc (the loader's own pump size)13.0%
53.5 cc (the best size)11.5%
28.5 cc (matches load-sensing's pump energy)26.2%

No size from 25 to 63 cc breaks even, under any of the pump efficiencies (0.75 to 0.95) or the fuel-rate readings we swept, nor in any of the 18 loading cycles. A smaller pump only moves the loss from surplus flow into a longer working day. The open-centre machine here is simulated on the measured machine's motion, not measured itself. The paper says so in the abstract, and it lists a measured open-centre machine of the same class as the experiment that would settle it.

The guess the rerun removed

Before the loader data could be split, the draft guessed that the loader's break-even margin (87 bar, about four times the synthetic cycles') came from a working pressure "near three times" the force model's. The split, now its own section, says the working pressure is about four times higher (134 against 33 bar). It also says that is the wrong pressure to use. The fixed pump's surplus flow arrives when demand is low, and pressure with it, so the exact break-even runs through the pressure the surplus is pushed at, 99.5 bar. Plugging in the working pressure overstates the break-even by 27%.

Reproduced, and checked number by number

We downloaded the 96 MB record fresh and reran the four measured-loader scripts in a new environment. Their output matched the committed results files byte for byte, and the fuel figure came out unchanged. The draft's number checker (paper/check_numbers.py) now holds 155 quoted numbers, each tied to the results file and line it comes from, so text and results cannot drift apart without the check failing. The DOI registry, which refused us when the draft was first written, also answered today: every DOI in the draft and the proposal resolves to the title we cite.

What is still missing

The draft still lacks a measured small-excavator trace. That is the programme's original hypothesis, and it remains untested. The draft is Markdown, not the target journal's template, and two theses and the valve and pump catalogues still need opening again before anyone submits it. Submission is the owner's decision.

Written with AI assistance (Claude Code). Results are from a model run on a public measured record, not from a test of an open-centre machine.

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