Across Europe and North America, bioenergy plant operators still summarise a year of work in one figure for total energy produced, and in 2026, with legacy support schemes winding down and methane rules tightening around them, that figure looks increasingly like the least informative thing they measure.
Seven other numbers would show them where the losses sit. Most plants already generate the raw data and never assemble it into anything a manager can act on.
In This Article:
What Gross Output Keeps Out of Sight
Gross output tends to flatter whoever reports it. It climbs when you push more material through the gate, whatever that material cost to buy and haul, and it holds steady while an engine sheds a point of efficiency every quarter.
The sector can no longer count on growth to cover the difference. European biogas and biomethane production reached 22 billion cubic metres in 2024 against 21.7 the year before, a margin the industry association itself calls modest and attributes in part to regulatory uncertainty. When new capacity stalls, returns have to come from the plants already standing.
Net Exported Energy, Not Gross Generation
Start here, because a digester eats its own output before anyone else gets a share. Farm-scale guidance from Scotland puts the parasitic electrical load of an anaerobic digester at around six kilowatt hours per tonne of feedstock, covering mixing paddles, pumps and the rest of the moving parts, with roughly a fifth of the recovered heat going back into holding the tanks at temperature.
Plants running on a high proportion of slurry need more heat than that, and colder sites need more again in winter. None of it appears in the headline number, which may be why the headline number keeps rising in years when the export meter does not.
The correction is arithmetic rather than capital. Meter the import alongside the export, subtract one from the other and publish the result every month. A site that has never seen its own consumption drawn as a line often finds at least one pump that has run continuously since commissioning for a reason nobody present can still explain.
Borrowing a Discipline From the Factory Floor
There is nothing novel about this problem, and process manufacturers worked out the measurement half of it decades ago. The methods used to increase productivity in manufacturing break a single output figure into availability, running rate and quality, so that a bad month points at a specific machine on a specific shift rather than at bad luck. A digester has all three of those dimensions and rarely reports them apart.
The translation is not exact. A biological reactor answers more slowly than a filling line, and its quality dimension is gas composition rather than reject rate. The discipline still carries across, because in both settings the value of a measurement lies in whether it names a cause.
Unplanned Hours, and What Caused Each One
Second, count unplanned downtime in hours and attach a cause to every one of them. The EPA’s operator guidebook for on-farm digesters treats uptime as a profitability lever rather than a reliability statistic, and it is unusually blunt about how few sites are equipped to measure much of anything.
Most facilities estimate their flows instead of metering them. Meters on the influent, the effluent and the separator line cost little set against the price of a digester, and without them the operating data a plant collects cannot be converted into mass per unit of time, which is the form in which it starts to be useful.
The warning signs also tend to arrive in a consistent order, and many sites watch the wrong one. A falling methane concentration in the biogas usually shows up first. Alkalinity in the digestate moves next, and it moves ahead of pH, which makes pH the reading that confirms a problem you should already have caught.
Foaming is the case that is hardest to argue with. A sudden jump in organic loading floats the solids, traps gas, blocks meters and can push a tank into overflow, and the production loss then runs for weeks. Anyone reading the annual total afterwards sees a soft year and blames feedstock quality.
The sector does keep a record of how this ends. The EPA’s database of livestock digester projects maps the ones still operating, while the plants that shut down appear only if you open the spreadsheet. That is a defensible design choice for a promotional tool and a poor one for anybody trying to learn from failures.
Methane Loss as a Share of Methane Made
Third, track methane loss against methane produced. Every kilogram that leaves through a seal is product you paid to make twice over, and the flanged joints where much of it escapes are among the cheapest components on the site to put right.
The scale appears to be worse than the sector’s own working assumptions. A 2026 survey of thirty-one plants in Germany, Poland and the United Kingdom measured average losses of 5.4 per cent of the methane produced, with individual sites running past twenty per cent, and estimated that fifty-nine per cent of those emissions could be eliminated at no net cost.
That estimate goes a long way towards settling the argument. Where a repair pays for itself in recovered gas, an operator who declines to measure losses is arguably not saving money, only deferring the moment of finding out. Plant age, size and feedstock all failed to predict emissions across that sample, which points more at how a site is run than at how it was built.
The distribution matters as much as the average. Seven sources across the entire survey, two per cent of everything measured, accounted for a fifth of total emissions, and annual leak detection proved cost-effective at roughly three quarters of the facilities where fugitive losses turned up at all. Chasing every seal with equal energy looks like the expensive way to do this.
The Regime That Does Not Cover You Yet
The regulatory line currently sits in a revealing place. Regulation (EU) 2024/1787 obliges fossil gas operators to run leak detection surveys, repair leaks inside fixed deadlines and justify venting events, with first routine inspections due by May 2026. Its scope covers oil, fossil gas and coal. Biogas production sits outside it.
The comparison is unflattering. The IEA’s global outlook for biogases puts today’s plants at somewhere between 2 and 5.5 per cent of output lost as methane, a level it describes as far above the average in the oil and gas industry, which is the sector now carrying the mandatory obligations.
An operator reading that might reasonably draw a practical conclusion rather than a comfortable one. The same regulation defines a super-emitting event as a source releasing more than a hundred kilograms of methane an hour, and it excludes from the definition of malfunction any failure caused in part by poor maintenance. Both of those ideas are likely to travel.
Yield Against What the Feedstock Could Have Given
Fourth, measure actual yield against the tested methane potential of what went in, not against last year’s figure for the same month. Laboratory assays tend to overstate what a full-scale digester will deliver, so the gap rarely closes to zero, but movement in that gap is among the clearest early signals of biological drift a plant produces.
Feedstock makes the comparison unavoidable. Cattle slurry gives up roughly twenty cubic metres of biogas per fresh tonne. Grass silage gives around nine times that. A plant that quietly changed its mix in March and reads only total output has no way to separate a process failure from a procurement decision, and the two call for opposite responses.
Two cheap tests carry most of the signal. Total alkalinity and volatile fatty acids can both be run on site for little more than the cost of glassware, and the acids measured against the alkalinity sit between roughly 0.1 and 0.25 while a digester is healthy. Watching that ratio climb tends to buy weeks of warning instead of days.
Set the sampling schedule before you need it. Solids, alkalinity and pH daily, the slower tests weekly once values settle, and a fresh potential assay whenever a new feedstock is offered rather than after it has already been through the tank.
The Half of Production Nobody Publishes
Fifth, report the heat, and report it beside the electricity. A combined heat and power set turns about forty per cent of the biogas energy into electrical output and makes roughly as much again as recoverable heat, so a plant publishing only its electricity figure is describing half of what it made.
Most of that heat goes nowhere useful once the digester has taken what it needs. The European survey found that lifting heat use by half, displacing natural gas at the point of use, would be worth the equivalent of a nineteen to thirty-six per cent cut in a site’s methane emissions. Very few of the plants studied used their heat for anything beyond keeping their own tanks warm.
Recovered well, a set can return something close to ninety per cent of the biogas energy as useful output. The distance between that benchmark and what a given plant can actually document is usually heat that nobody was made responsible for.
The practical move is to find a year-round heat customer before the engine is specified rather than afterwards. Grain floors, timber drying and digestate drying all work partly because they run outside the heating season, which is exactly when a space-heating contract stops paying.
Energy Per Tonne at the Weighbridge
Sixth, price feedstock at the gate per unit of energy it contains rather than per tonne delivered. Haulage is where that number moves, and because low-density biomass fills a trailer long before it reaches its weight limit, a load of baled straw can arrive carrying roughly a third of the tonnage of a load of pellets after the same journey, burning comparable fuel and paying the same driver.
Procurement systems record tonnage because tonnage is what gets invoiced. Volume decides how many trucks run, and every truck is a deduction from the energy the material eventually yields.
Recording both is the fix, and it costs a tape measure. Log deck volume alongside weight on every consignment and report utilisation against each, and the loads that looked cheap per tonne stop looking cheap.
When the Energy Arrived, Not Only How Much
Seventh, and least often tracked anywhere, record when the energy arrived. A unit used on site to displace an imported one is generally worth more than the same unit exported at tariff, and system operators increasingly pay for the ability to respond as readily as they pay for installed capacity.
A dispatchable plant that runs flat because flat is easier to operate is selling a premium product at the average price. Its export meter cannot see the difference. Neither can anyone reading the annual total.
Most digesters have more room to move than their operators use. Sixty to ninety days of retention in a mesophilic system is not a timetable that has to be run at a constant rate, gas storage in the roof buys hours more, and pairing generation with storage and smart dispatch can widen that window further.
Plot a month of output against a month of prices and the answer arrives quickly. If the two lines have nothing to do with each other, the plant has been leaving the timing premium on the table, and no part of that will show up in the annual figure.
Where to Start on Monday
None of these seven measurements needs hardware more exotic than flow meters costing less than a week of lost production. What they need is an operator willing to publish a number that goes down.
There is a reason so little of this gets measured, and cost does not appear to be it. Gross output is the only figure on the list that cannot embarrass anybody. Every other one names a component, a shift or a decision, and somebody then has to own it.
So pick two. Meter the parasitic load, log unplanned downtime with a cause attached to each hour, run both for a quarter and put the pair in front of whoever signs off the annual report. Two numbers with an owner will move more than seven without one.
The plant that reports its best year on record while venting a fifth of its methane through the joints has not lied to anybody. It answered the only question it was ever asked.
