The Overlooked On-Site Fuel:
Biogas Assets at Industrial
and Water Facilities

At thousands of industrial and water facilities in the United States, a flare burns around the clock on fuel the site itself produced and has already paid to handle. This paper treats that flame as what it is: an energy decision renewed every day by default, and one that deserves evidence instead.

Section 01The fuel that is already there

Anaerobic digestion is not an energy technology. It is a waste technology that happens to make fuel. Municipal treatment plants run digesters because digestion stabilizes sewage sludge, cuts its volume, and controls odor and pathogens. Food and beverage plants run anaerobic reactors and covered lagoons because high-strength process wastewater is expensive to discharge, and anaerobic treatment strips the organic load at a fraction of the aeration energy that aerobic treatment burns. Dairies and other livestock operations build digesters to manage manure, nutrients, and relations with neighbors. In every case the gas arrives as the byproduct of a process that was justified on other grounds: compliance, odor, solids handling. That is precisely why the fuel gets overlooked. At most facilities, nobody's job title includes it.

The fuel is real. Digester biogas typically runs 50 to 75 percent methane, with the balance mostly carbon dioxide, along with hydrogen sulfide, water vapor, and trace constituents, per the U.S. Environmental Protection Agency's description of the process.¹ Because heating value tracks the methane fraction, raw biogas carries very roughly half to three quarters of the energy of pipeline natural gas per cubic foot. It leaves the digester saturated with water vapor and brings its impurities with it. It is, in other words, a genuine fuel with genuine handling requirements, and both halves of that sentence belong in the analysis.

The scale of the fleet is easy to underestimate, and so is the scale of the neglect. In its 2011 market analysis, the EPA's combined heat and power program counted 1,351 wastewater treatment facilities with flows above one million gallons per day operating anaerobic digesters, and 104 of them using the biogas for combined heat and power.² Most of the rest used some gas to heat their own digesters and sent the balance to a flare. On the agricultural side, the EPA's AgSTAR program counted 400 operational livestock digester systems nationwide as of June 2024.³ The 2011 census is dated and the installed base has grown since, but a 2019 U.S. Department of Energy characterization of the same sector was still, as its title says, estimating opportunities rather than describing a solved problem.⁴ The structural fact has been durable for fifteen years: digestion is common, and beneficial use of the gas is the minority case.

For screening, the 2011 federal analysis used a rule of thumb of roughly 26 kilowatts of electric capacity, plus about 2.4 million Btu per day of recoverable heat, per million gallons per day of influent flow.² That heuristic is useful for exactly one purpose: deciding whether a stream is worth measuring. A machine is sized from a meter, never from a rule of thumb.

Section 02Flaring is the baseline

Begin by defending the flare, because the flare deserves it. Methane is both a safety hazard and a potent greenhouse gas, and venting it raw is not an acceptable practice at a permitted facility. A flare destroys it reliably, at low capital cost, with almost no operating attention, through every digester upset and every maintenance outage. Facilities flare for sound reasons, and every beneficial-use system ever built keeps a flare standing behind it for the hours the machine is down. The flare is not the villain of this paper.

It is the baseline. Keep-flaring is the no-project case: near-zero incremental capital, zero captured fuel value, full exposure to purchased-energy prices. Every conversion path is an investment whose return is measured against that row. Treating the flare as a priceable option, rather than as either an embarrassment or an inevitability, is what turns a vague sense of waste into a decision.

Two facts sharpen the baseline. First, flares are themselves permitted combustion sources, with emission limits, source tests, and inspection obligations of their own; the do-nothing case is not an unregulated case. Second, in California's strictest air basins the direction of regulatory travel is explicit. The South Coast district's Rule 1118.1, adopted in 2019, addresses emissions from non-refinery flares, including digester-gas flares at water and organics facilities, and is written to push facilities toward alternatives to routine flaring.⁵ Nothing in that rule compels any particular technology. It does mean that the assumption behind many flares, that a large stream can be burned off indefinitely with no questions asked, is weakening in at least some jurisdictions.

The flare is not the failure. The unmeasured flare is.

Which points at the cheapest energy instrument most facilities never install: a totalizing gas meter on the flare line. A surprising share of sites that flare cannot say, within tens of percent, how much they flare or when. Until that line is metered, every conversation about engines, fuel cells, or pipeline injection is a conversation about a rumor.

Section 03Measure the resource before pricing a machine

Biogas projects fail more often on resource honesty than on equipment. The resource statement has three dimensions. All three are measurable, and all three are routinely assumed instead.

Quantity

Twelve consecutive months of metered production is the defensible minimum, because digesters breathe with the calendar. Where the piping allows it, meter three streams separately: gas produced, gas consumed on site in digester heating and boilers, and gas flared. Then cross-check the meters against the process itself. Volatile solids destroyed at a sludge digester, or chemical oxygen demand removed across an industrial anaerobic reactor, imply a gas yield within a known range; a meter that disagrees sharply with its own process is reporting on the meter, or on a leak, and either answer is worth having. If there is no meter, the honest sequence is: install one now, bracket the interim with process-based estimates, and label every bracketed number as an estimate.

Quality

The composition panel decides more of the project than the headline volume does. Methane content sets heating value. Hydrogen sulfide sets corrosion rates and the size of the cleanup train, and it varies with feed. Siloxanes, the volatile silicon compounds that ride into digester gas from municipal sludge and from any waste stream touched by consumer products, burn into a glassy silica that coats pistons, valves, turbine hot sections, and heat-transfer surfaces. Moisture is a given; the gas arrives saturated. Oxygen and nitrogen are the tell for air intrusion into the collection system: they dilute the fuel, and they are expensive to remove if pipeline injection is ever the goal. Sample across seasons and operating modes. One grab sample tells you what the gas was doing on the morning somebody visited.

Seasonality

The annual average is the least useful number in the file. A cannery's anaerobic system produces on the pack calendar. A winery's load arrives at crush. An ambient-temperature covered lagoon swings with the weather in a way a heated tank digester does not. Municipal plants breathe with wet weather, temperature, and the slow drift of what arrives at the headworks. On top of the seasonal pattern sit the interruptions that belong in any honest model: digester cleanouts every several years, process upsets, feed changes. The defensible presentation is a duration curve: how many hours per year the stream can support a machine of a given fuel appetite. Machines sized to the flat base of that curve run loaded. Machines sized to the peak, or to the average, spend their lives hungry.

One forecast deserves its own caution: hauled-in organics. Co-digesting fats, oils, grease, and food waste can multiply gas production, which is why it appears in so many pro formas. It also brings receiving infrastructure, supply contracts, quality swings, odor exposure, and permitting questions, because purchased organics are a business, not a byproduct. A resource statement built on hoped-for hauled waste is a forecast, and it must be labeled and stress-tested as one.

Before any conversion path is priced, six facts belong on one page:

1. Metered production history.Twelve months of production, on-site use, and flare volumes from meters, or a meter installed today plus labeled interim estimates.
2. The composition panel, across seasons.Methane, hydrogen sulfide, siloxanes, moisture, oxygen, and nitrogen, sampled in more than one operating mode, not once.
3. The duration curve.Hours per year the stream supports a given fuel appetite, drawn from the meter, not from the annual average.
4. The thermal balance.What the digester and the process need back, month by month. Heat the process requires is not surplus, and winter is the test.
5. The priced counterfactual.Current flare volumes beside the site's actual purchased-energy prices, taken from bills and the tariff, not from recollection.
6. The commitment horizon.The production plans, sludge strategy, and co-digestion intentions the facility will stand behind for the life of the equipment.

Section 04Six ways to use a fuel you already own

With the resource stated, the paths can be compared. Six recur at digester sites. None of them wins everywhere, and each earns its row honestly.

Two footnotes to the table. Combustion turbines and steam cycles appear at the largest streams and are priced the same way. And low-pressure gas storage deserves a sentence of its own: a gas holder smooths hours, not seasons. It buys the prime mover a steadier diet and buys the operator sleep. It does not manufacture annual volume, and it should never be allowed to disguise a seasonal resource as a flat one.

The fleet evidence says the comparison is genuinely open. The 2011 federal census of the 104 plants then running combined heat and power on digester gas counted 54 on reciprocating engines, 29 on microturbines, and 13 on fuel cells, with the remainder on turbine and steam cycles.² That distribution is not a verdict. It reflects capital costs, district air rules, and gas quality as they stood, site by site, over decades. It is evidence that more than one machine class can carry this duty, which is exactly why the comparison has to be run per site rather than settled by fashion.

The cleanup train is the project inside the project

Every row after the flare buys some amount of gas conditioning, and the tolerance ladder is steep. Direct thermal use is the most forgiving: drop the condensate, manage sulfur enough to protect the equipment and the stack, and burn. Engines tolerate moderate sulfur and pay for it in oil analysis, shortened oil life, and wear metals. Microturbines are notably sensitive to siloxanes in the hot section. Fuel cells need near-complete sulfur removal, because sulfur is a poison to them rather than a nuisance, and they hold the strictest fuel specification of the power-making rows. Pipeline injection is strictest of all, because the specification governs the bulk gas itself: carbon dioxide out, oxygen and nitrogen tightly capped, heating value inside the utility's band, with the gas utility as the judge.

The decision consequence is simple. Cleanup capital and cleanup operating cost belong inside each path's price, never in a shared footnote. Ranked on brochure efficiency, the rows arrange themselves one way. Ranked on delivered cost after each row's own conditioning train, they can arrange themselves differently, and the second ranking is the only honest one.

Section 05The economics, stated plainly

A captured cubic foot can do one of three things: displace purchased electricity, displace purchased fuel, or leave the site as a product. The channels are not equal. Displacing retail electricity is usually the highest-value use per unit of energy, and it demands the most capital and the deepest cleanup. Displacing purchased gas in a boiler is the cheapest conversion and earns commodity value. Selling upgraded gas earns wholesale value plus whatever environmental credits attach, at the price of the most complex project on the menu. The per-unit value ranking and the capital ranking run in the same order, which is why no path dominates on arithmetic alone.

Model net, not gross. Compression, cleanup skids, chillers, and controls all consume part of the output before the site sees it. Maintenance is a first-class line, not a contingency; digester gas is harder on machines than pipeline gas, and the operating history of each machine class on this specific fuel is knowable and should be asked for. Availability assumptions should be defended with fleet history on comparable gas, not asserted.

Tariffs decide more than brochures do. On-site generation changes the utility relationship: standby provisions, demand charges, and metering arrangements vary by utility and by rate, and a project's value is computed under the tariff as written, not under a generic savings model. For pipeline injection, the governing documents are the gas utility's interconnection agreement and quality specification, and both carry site-specific costs that enter the model at quoted values, not assumed ones.

On incentives: current federal law as of August 2026 provides a 30 percent investment tax credit for qualifying energy property under 26 U.S.C. §48 and §48E.⁶ Whether a specific biogas configuration qualifies, which vintage of the statute governs it, and whether any statutory adder applies are determinations for qualified tax counsel on the specific facts. Adders exist in statute and must be individually qualified, never assumed. Separately, gas that reaches vehicle-fuel service can access federal and state credit markets; those are policy markets, their prices move, and a project underwritten at peak credit prices is a bet, not a plan. Label it as one.

Many water facilities, and some agricultural ones, are public agencies. Procurement and financing routes for public agencies are agency-specific and counsel-driven, and nothing in this paper is procurement or finance guidance for any public agency.

Finally, the labeling discipline that runs through this series applies with particular force here, because biogas pro formas mix measured facts with hopes more freely than most documents in this field. Every number is either sourced and dated, from a meter, a bill, a tariff, or a quote, or it is labeled an estimate. Illustrative means illustrative.

Section 06A decision, not a default

A defensible biogas decision has a consistent anatomy. A resource statement built from meters and a seasonal composition panel. Every credible path priced against continuing to flare, each row carrying its own cleanup train, its own tariff arithmetic, and its own permitting screen in the site's actual air district. Sensitivity on the assumptions that genuinely move biogas answers: gas volume, cleanup cost, and the prices of energy and credits. And a decision memo at the end: one recommended path, the conditions that would change the answer, and the next three actions with owners and dates.

The structural requirement is the one this series repeats deliberately: the party pricing the paths should have nothing to gain from which machine wins, because several rows in the table above pay a technology seller nothing at all, and those rows have a way of vanishing from sellers' spreadsheets. A facility that flares its gas is not failing; flaring is a lawful, safe, engineered default. A facility that flares without ever having metered the stream and priced the alternatives is doing something different: renewing a decision every day without ever having made it. The fix costs a meter, a sampling program, and one honest study. Start with the meter.