On-Farm Dairy Digester in Lancaster County, Pennsylvania: A USDA REAP Model Study

Model study: an illustrative feasibility analysis built on public data for Lancaster County, Pennsylvania; not a client engagement.

Project summary

The credit question

A dairy digester is three businesses at once: a biology plant, an electricity seller and a small waste-handling business. The loan has to survive all three. A lender reviewing a file like this must be convinced of five things.

First, that the feedstock is real and controlled. In Lancaster County, six of the seven operating digesters in EPA's AgSTAR database co-digest food wastes [3]. The modelled project follows that pattern. A quarter of its revenue and about a third of its energy depend on residuals the dairy does not produce and must contract for.

Second, that the electricity has a buyer at a known price. Under PPL's net metering rider, a customer-generator's output is credited at the full retail rate up to the electricity delivered to it, excess carries forward, and kilowatt-hours left at the end of the PJM planning year on May 31 are paid at the rate schedule's price to compare [4]. That is valuable but regulated, and PPL's generation charge resets twice a year [5]. The modelled plant generates about 227% of the farm's load, so more than half its kilowatt-hours earn the price to compare.

Third, that the plant will run. AgSTAR lists 98 U.S. livestock digesters as shut down. Among the 87 with opening and closing years, the median life was five years, and "too costly to operate" is among the recorded reasons [3]. A stopped digester pays no debt and has almost no resale market.

Fourth, that the federal program fits the cash needs. USDA's October 1, 2026 rule makes REAP grants post-completion: the project must be built and operating before the owner applies, with 12 months of actual production data [2]. The grant can no longer fund construction. USDA's preamble points applicants to the REAP guaranteed loan under 7 CFR part 5001 for up-front costs [2].

Fifth, that the debt is sized to what the plant earns. REAP allows a guaranteed loan up to 75% of eligible project costs [1]. At that level this plant would cover debt service 0.87 times. The gap between the program ceiling and the bankable loan is the central finding.

Location and county economy

The county economy matters indirectly here. A digester does not sell to households. But the economy explains why the county has so many livestock farms, why food residuals are available to co-digest, and why land near the City of Lancaster may not count as rural for the grant.

Population. The county grew about four times faster than the state after 2020 [6]. In 2025, natural increase (1,428) and international migration (975) outweighed a domestic outflow of 1,319 [6]. Growth supports farmland values and local food processing, but it also brings homes closer to farms. The state's digestion general permit sets a 300-foot setback between waste storage or processing areas and occupied dwellings unless the owner waives it [10].

Income and labor. Median household income is about 11% above the state's [7]. Food manufacturing alone employed 8,697 people at county establishments in 2023 [8]. Staffing is a cost risk for a plant that needs a trained operator every day, so the model budgets $35,000 a year for dedicated digester labor.

Feedstock base. The state's list of the county's 50 largest employers includes Pepperidge Farm, Tyson Poultry, Turkey Hill Dairy and Four Seasons Produce [11]. That shows only that food processing is a large local industry. This model study assumes no relationship with any named firm, and a lender would need executed supply agreements.

Rural status. REAP's rural definition excludes cities over 50,000 people and the urbanized area contiguous to them [2]. The City of Lancaster's April 2020 estimates base was 57,983 and its July 2025 estimate 58,366 [9]. A farm near its edge could fail the grant's rural test, although part 5001 allows an agricultural producer to be located in either a rural or a non-rural area [1].

Market analysis

For a digester, market analysis means two questions: is there enough feedstock, and is there a buyer for the energy?

Feedstock supply: the farm's own herd

The feedstock is the farm's own herd. The 2022 Census of Agriculture county tables, the standard measure of a county's dairy herd and its trend, were not used in this model study, so it makes no claim about the size or direction of Lancaster County's herd. The model assumes a 1,500-cow lactating herd. That is larger than any dairy listed with an operating digester in the county in EPA's AgSTAR file, where dairy herds range from 250 to 1,225 cows [3].

Three credit points follow. The herd's scale is what makes on-farm generation plausible. Imported manure from other farms is not a dependable backup, because it depends on other owners' decisions. And the herd itself is the first-order risk: AgSTAR records closures where "the dairy operations ceased" [3]. The loan has to be analyzed as a dairy credit, not as a stand-alone plant.

Existing supply: digesters already operating

The per-cow figures are this model study's calculations from EPA's project file: annual electricity divided by dairy herd and 365, for operational projects listing only dairy animals [3]. EPA notes the database is not exhaustive [12].

Pennsylvania's farm digesters are almost all power projects (30 of 31 active entries), while the national construction pipeline has moved to renewable natural gas [3]. For the six dairy-only Lancaster projects, this model study calculates 2.29 to 8.36 kWh per cow per day, with a median near 4.84. The modelled design of 6.5 kWh per cow per day sits between the national co-digestion median and upper quartile [3]. The break-even analysis shows the loan breaks even at 4.78, above the manure-only median of 4.18. A manure-only version of this plant would not carry this loan.

Demand method: feedstock-to-energy build

The 139.2 kWh per ton increment is an assumption. A REAP technical report for a digester must include tabular values or laboratory biodegradability analysis to produce daily, monthly and seasonal gas estimates [1]. A lender should require that lab work on the actual residuals.

The energy buyer: net metering, AECs and the policy stack

Two 2026 changes shape this stack. EPA's RFS rule for 2026 and 2027 removed renewable electricity as a qualifying fuel [14], which ends the prospect of RINs for a power digester. And the REAP rewrite turns the grant into a reward for performance rather than a construction source [2]. Both changes push value toward gas projects or toward debt sized without subsidy.

Net metering is the anchor. For a small general service farm account, PPL's variable charges are mostly generation and transmission, $0.12746 per kWh in July 2026 [5]. Distribution is billed largely on demand ($5.156 per kW for GS3) [5], which a generator does not reliably reduce. The model values offset at $0.125 in 2028 with 1.0% escalation and excess at a flat $0.100, a 22% discount for the semiannual price reset [5] and for policy risk on 1.6 million kWh of yearly excess. AECs are the most volatile line, ranging from $6.41 to $31.01 per MWh over seven years [13], so the model uses $15.

The alternative pathway: renewable natural gas

RNG earns more per unit of gas because RINs are worth far more than AECs. But it adds an upgrading plant, a pipeline connection whose cost depends on distance to a main, and revenue set by annual EPA rules. The 2025 waiver credit, computed as an inflation-adjusted $3.00 less average wholesale gasoline [15], can act as a ceiling on D3 prices in waiver years. This model study keeps the power case because it matches what Lancaster dairies run [3] and rests on a tariff rather than a commodity price. An RNG option would need its own pro forma and RIN stress cases.

Forecast

Year 1 (2028) runs at 75% of potential, with residual contracts at 70% of tons. Year 2 reaches 88% and 90%. From Year 3 the plant holds 90% of potential and full tons. Revenue grows only through escalation: 1.0% on offset power, 2.0% on tipping fees and 2.5% on heat and bedding. Excess power and AECs stay flat. No herd growth is assumed.

Site and technical

Site. The digester has to sit near the freestall barns to keep manure transfer short, with room for a 7-day residual storage tank, a separator building next to the bedding area, and the engine near three-phase service. The general permit keeps waste storage and processing out of the 100-year floodplain, at least 100 feet from wetlands, 300 feet from occupied dwellings unless the owner waives it in writing, and generally 50 feet from property lines [10]. In a growing county, the dwelling setback often decides the site.

Permits. Co-digestion requires DEP general permit WMGM042, issued August 18, 2022 and expiring August 18, 2032. It authorizes on-farm digestion of animal manure mixed with grease trap waste and pre-consumer and post-consumer food waste [10]. Operators must register under the permit, incoming municipal and residual waste may be stored up to seven days, and processed material must be sampled every 80 to 100 days with records kept five years [10]. If fats, oils and grease push digestate above 15,000 mg/l, it cannot be used as a soil additive [10], which matters for a farm that spreads digestate on its own fields. The budget also covers an engine air permit or exemption, an updated nutrient management plan, township zoning and the PPL interconnection application. Those were not verified for a specific site.

Technology. A heated complete-mix vessel is the dominant design in Pennsylvania (21 of 31 active entries) and in all seven Lancaster projects [3]. It handles the variable solids of co-digestion better than plug flow. Gas passes through a biological hydrogen sulfide scrubber, carbon polishing and a chiller to a 499 kWe lean-burn engine with heat recovery. Recovered heat warms the digester first, so only the surplus displaces farm fuel, worth $20,000 a year. A screw press produces bedding, and a flare burns gas during outages.

Construction. A fixed-price engineering, procurement and construction contract with a performance guaranty fits what the loan's technical report must cover for projects of $200,000 or more: design and engineering, project development, procurement and installation, and operations and maintenance [1]. For the grant, the application must document the installer's state license, how many similar systems it has installed and how the key service provider was selected [2]. Because the grant now comes after completion, an owner who ignores those requirements during construction can weaken the grant application before it is filed.

The Project Period may not exceed 24 months [2]. This schedule keeps it to about 15 months.

Development budget and sources and uses

Capital costs are assumptions; no public source prices a Lancaster co-digestion plant. The total is $2,910 per cow and $8,749 per kW. A lender should require two fixed-price bids and an independent engineer's review.

Three lines carry program meaning. The guarantee fee is loan times guarantee percentage times fee rate [1]; for fiscal year 2026 REAP loans the rate is 1.0% with an 80% guarantee, plus 0.50% if the guarantee issues before construction is complete [19]. The model assumes issuance at completion, so the lender carries 2027 construction risk itself. The feasibility study is a loan-eligible cost [1] but an unallowable grant cost [2]. And the $80,000 of utility-owned upgrades counts in total project cost, but the grant treats equipment the recipient does not wholly own as unallowable [2].

The guaranteed portion is $1,828,000 and the lender's unguaranteed exposure is $457,000. The REAP grant is left out of sources: it cannot be sought until 12 to 24 months after completion [2], it is competitive and subject to funds [2], and the funding notice for fiscal years 2025 to 2027 was rescinded in April 2026 [20]. Equity of 47.7% is well above the 25% matching funds REAP requires [1].

Revenue build and operating expenses

Assumptions

Ramp and expense structure

Year 1 revenue is $479,065, Year 2 $564,028 and Year 3 $592,245. The offset line is capped by farm load, so ramp gains land in the lower-valued excess line. A one-cent change in the excess value moves Year 3 revenue by about $16,466; a one-cent change in offset value moves it by $13,000.

Operating expenses take 36.6% of revenue and reserves 7.8%, leaving a 55.6% cash margin. Per gross kilowatt-hour, the plant earns $0.185 and spends $0.068 on operations and $0.014 on reserves. The retention fee is 0.25% a year on the guaranteed outstanding balance [19]; lenders may pass it through [1], so it is treated as an operating cost.

Two choices matter more than their size. Reserves come out before coverage because part 5001 defines DSCR as EBITDA less reasonably expected replacement capital expenditures, divided by debt service [1]. An engine running about 6,400 equivalent full-load hours a year will need overhauls within the loan term. And scrubber media ($15,759) protects the engine: sulfur that gets through corrodes heads and degrades oil, which is why the stress tests treat a scrubber failure as a severe case. The pro forma is pre-tax on the assumption that the dairy owns the plant through a pass-through entity. Tax credits are excluded (see Program compliance).

10-year pro forma

Ten-year revenue totals $5,967,492 and cash flow after debt service $642,757. Every row ties: revenue less expenses is NOI, NOI less reserves is CFADS, CFADS less debt service is cash flow. Coverage stays flat because expenses escalate at 2.5% while the main revenue lines grow 0% to 2%. The project never grows out of its debt, so any cushion has to come from structure.

Debt structure

Why $2.285 million and not $3.3 million. At the program ceiling, a $3,275,000 loan (74.1% of the $4,421,856 cost after fees and interest scale up) would carry $376,796 of debt service against $329,104 of CFADS: 0.87x. At 65% of cost ($2,840,000), coverage is 1.00x. The 1.25x sizing target is this model study's lender-policy assumption; part 5001 sets no REAP-specific coverage minimum and relies on the lender's credit evaluation [1].

Equity and the grant. If the grant is won, combined REAP funding would be $2,785,000, or 63.8% of cost, inside the 75% combined limit [2]. Equity net of the grant would be $1,580,674 (36.2%). A loan at the 75% ceiling would have made any grant application ineligible, because combined REAP funding would exceed 75% [2]. Sizing to cash flow keeps the grant option open.

Rate basis. Prime was 7.00%, the effective federal funds rate 3.88%, and the 5-year and 10-year Treasury yields 4.67% and 4.88% on November 9, 2026 [21]. The stress tests run from 6.75% to 9.75%.

DSCR and break-even

The break-evens say the credit depends on co-digestion. Output break-even exceeds the manure-only median, and the residuals must pay at least $8.03 a ton. The 68.1% performance break-even gives the biology some room, but the stress tests show a single bad event can use it up. A $131,447 liquidity covenant, six months of amortizing debt service, would cover the Year 2 gap to 1.25x and the three-month outage case. It is modelled as a covenant on the dairy's cash, not a funded project reserve.

Returns

Unlevered flows exclude financing fees and the retention fee; Year 10 includes the exit value. Levered flows are cash after debt service, with Year 10 adding $971,123 of net exit proceeds.

As a credit, the plant covers its debt. As an investment, it barely returns its capital: 3.5% to 4.4% unlevered, and negative over 10 years levered without the grant. The 7.75% loan costs more than the project earns, so debt lowers the owner's return. Even the $500,000 grant, 11.5% of cost, only lifts the 20-year levered return to 3.5%.

That matters to the lender. An owner earning a sub-market return on a plant that needs daily attention has weak economic reasons to keep it running through a bad year, and the AgSTAR shutdown record reflects that incentive [3]. Dairies also value odor control, nutrient handling and manure storage capacity, which this pro forma does not count. Those motives help explain why an owner keeps operating, but a lender should not lend against them. Hence the dairy guaranty in the conditions below.

The 15% exit rate prices remaining life rather than a trading market. At Year 10 the vessel has about a decade left; ten years of cash flow at 8% is 6.71 times one year, close to the 6.67 implied by 15%. For an owner who runs the plant to the end, the 20-year figures are the better guide.

Sensitivity and stress tests

Grid 1. Year 3 DSCR by electricity value and performance factor

Grid 2. Year 3 DSCR by interest rate and performance factor

Grid 3. Year 3 DSCR by tipping fee and co-digestion tonnage (energy scales with tons)

Electricity value and performance each move coverage about 0.11x to 0.13x per 10 points. A one-point rate change moves it about 0.07x, and fixing the rate at closing removes that risk. Co-digestion outweighs both: half the tonnage at the base fee gives 0.81x, and full tonnage with no fee gives 0.71x.

The H2S case shows the danger is the repair bill, not lost output: the outage alone leaves 1.14x, but adding a $120,000 rebuild in the same year drops it to 0.68x. Gas conditioning, continuous H2S monitoring and an engine service contract with fixed overhaul pricing are credit protections. The co-digestion case shows the plant has no stand-alone fallback. Without residuals it earns about $97,000 against $262,894 of debt service, and the gap has to come from milk income.

Program compliance

Estimated grant score. Merit scoring has 90 points plus up to 10 priority points [2]. The modelled project would earn 5 points if the dairy employs at least three people other than its owners (assumed), and the full 25 points for energy per award dollar: about 21,856 BTU per grant dollar (computed from gross generation), against a scale of one point per 100 BTU capped at 25 [2]. It would get zero on the 20-point percent-replaced criterion, because projects replacing 109.5% or more of prior use score nothing and this plant produces about 227% of load [2]. Further points depend on criteria this model study did not score. Selection takes the top two applications in each State that meet the minimum score and then funds the rest in rank order until funds run out [2], so the grant stays uncertain.

Tax credits. None are valued. Section 45Z covers clean transportation fuel sold from January 1, 2025 through December 31, 2029 [16], so it applies only to RNG. Whether a power-producing manure digester can claim an electricity credit under current Treasury rules is not addressed here.

What changed in 2025 and 2026. USDA rescinded the REAP funding notice for fiscal years 2025 to 2027 in April 2026 and required applicants without an executed agreement to submit new applications under the new rule [20]. The October 2026 rule moved grants post-completion, set the RES maximum at ,000 with 75% Cost Sharing, folded the former technical merit appendices into the application requirements, and replaced State-level competitions with national selection [2]. The 2026 RFS rule removed eRINs [14]. Each change pushes construction onto the guaranteed loan and equity, with the grant a possible later prepayment.

Risks and mitigants

Model determination

Determination: financeable as structured, with conditions, at about half of cost. Not financeable at the REAP loan ceiling, and not a stand-alone project credit.

The modelled digester supports a $2,285,000 REAP guaranteed loan at 1.25x from Year 3, because the loan is sized to cash flow (52.3% of $4,365,674), the owner invests $2,080,674, and residuals pay the farm to take them. At the 75% ceiling coverage would be 0.87x, without co-digestion 0.37x, and in a single bad engine year 0.68x. The real repayment source is therefore the dairy, with the digester's cash flow as the main but not only support. Part 5001 requires a global credit evaluation where affiliated entities are involved [1], and the file should say so plainly. The owner's economics are thin, so the structure must make continued operation the owner's best choice even in a bad year.

The conditions a lender would likely set:

1. Five-year residual supply agreements for at least 9,125 tons per year at a net fee of at least $15 per ton, from at least two suppliers, with lab-tested yields supporting 139 kWh per ton.
2. A fixed-price EPC contract with a performance guaranty, an independent engineer's review, a technical report under 7 CFR 5001 appendix E [1] and an independent feasibility study.
3. A PPL interconnection agreement and written confirmation of net metering treatment for output near 227% of farm load.
4. WMGM042 registration [10], an amended nutrient management plan, and air and zoning approvals before the first draw.
5. A long-term engine service contract with set overhaul pricing and a reserve funded at $0.010 per kWh.
6. Guaranties from the dairy entity and its owners, cross-default with dairy loans, and a blanket lien on dairy real estate.
7. A $131,447 liquidity covenant, a 1.20x global coverage test and a cash sweep below 1.10x digester coverage.
8. A fixed rate at closing.
9. Any REAP grant applied to principal, with REAP application and Project Period requirements followed during construction.
10. Annual performance reporting to the lender, plus monthly production reports.

With these conditions the loan is a reasonable guaranteed credit: 80% is guaranteed, and the $457,000 unguaranteed portion is backed by a working dairy and its land. Without them, and especially without the residual contracts and the dairy guaranty, this model study does not support the loan.