Linear Generators and New Entrants:
Evaluating Young Operating Bases

Every mature generating technology was once a young machine with a short record and a confident brochure. The owner's question is never whether newer classes such as linear generators deserve consideration; it is what a short operating history should cost in the comparison, and that price has to be set from the owner's side of the table.

Section 01The entrant's dilemma, stated plainly

California's time-to-power problem has done what shortages always do: it has invited new machinery into the market. The most visible newer class is the linear generator, a machine that converts the reciprocating motion of free pistons directly into electricity through electromagnetic translators, using a low-temperature reaction of fuel and air in place of conventional flame combustion. As a class, these machines are typically presented with a wide manufacturer-stated fuel range, modular unit sizes, fast start, and a favorable air-emissions posture. They are not alone. Newer fuel-cell chemistries, long-duration storage classes, and compact variants of familiar combustion equipment arrive with versions of the same profile: a genuine engineering idea, early commercial deployments, and a fraction of the operating base their mature competitors carry.

Owners tend to respond in one of two mistaken ways. The first is reflexive dismissal. No one is ever criticized for buying the incumbent, so the entrant is screened out before it is priced. This forfeits real value, because the newer classes exist precisely where mature equipment leaves problems unsolved: combustion machinery faces demanding air-district treatment in California's strictest basins; some mature classes cycle poorly or lose efficiency at part load; fuel flexibility is a genuine hedge when the future price and carbon treatment of any single fuel is uncertain; and modularity changes both redundancy math and delivery logistics. An owner who refuses to examine entrants has quietly decided that none of those problems matter at the site. Sometimes they are the site's binding constraint.

The second mistake is buying the brochure: accepting projected availability, projected degradation, and projected service cost from a party that has not yet operated a large fleet through a full life cycle, then modeling those projections as if they were history. That is not optimism. It is a transfer of unpriced risk onto the owner's balance sheet.

This paper sets out the middle discipline: five pillars of diligence that convert technology-maturity risk from a mood into a line item. The framework applies to any young class, in any direction. Linear generators are simply the current archetype, and nothing here is an argument for or against them. It is an argument for grading them the way an owner should grade everything else.

Section 02What an operating base proves, and what it cannot

Reliability is not a property a machine has on paper. It is a property a fleet demonstrates over hours, and the demonstration has a known shape. Reliability engineering describes equipment failure rates with the bathtub curve: an early period of elevated but declining failures as latent defects surface, a long flat region of low and roughly constant failure rates, and a final rising region as materials wear out.¹ A young fleet has mapped only the first region, and only partially. The flat region is an extrapolation. The wear-out region is a guess, because no unit has reached it yet.

Two consequences follow. First, early-life data cannot be dismissed as teething trouble; it is the only empirical record that exists, and the way a manufacturer handles early failures, in root-cause transparency, retrofit speed, and fleet-wide fixes, is itself diligence evidence. Second, serial-defect risk is specific to young fleets: a design or supplier fault replicated across every unit shipped, discovered only once the fleet accumulates hours. Mature classes have largely burned that risk down over decades. Entrants have not had the hours to do so. Neither statement is an accusation. Both are arithmetic.

The information environment differs as well. For large central-station equipment, North American practice includes a pooled, mandatory statistical base: conventional generating units of twenty megawatts and larger report event and performance data into the Generating Availability Data System administered by the North American Electric Reliability Corporation, under reporting instructions that have made participation compulsory for units of that size since 2013.² An owner comparing mature utility-scale equipment can consult decades of pooled availability statistics. No comparable public pool exists for distributed-scale machines of any class, mature or young. The practical difference is the surrounding ecosystem: mature distributed classes are graded continuously by independent service firms, insurers with claims files, secondary markets, and thousands of owners who have already made every mistake. For a young class, nearly all of the performance record sits with the seller. Diligence is how the owner rebalances that asymmetry.

A short operating history is not a verdict against the machine. It is a fact with a price, and the study's job is to name the price.

Section 03Pillar one: fleet hours, counted honestly

Cumulative fleet hours is the headline number every young manufacturer offers, and it is the beginning of the analysis rather than the end. The number means little until it is decomposed:

• Hours on the current configuration. Products evolve quickly in their early generations. Hours accumulated on superseded models, earlier control systems, or prior component suppliers transfer only partially. Ask what fraction of the claimed fleet hours belongs to the hardware actually being quoted.
• The distribution across units. As illustrative arithmetic: a million fleet hours accumulated as two thousand hours on each of five hundred units demonstrates manufacturing repeatability and early-life behavior; the same million hours including a lead unit deep into its service life demonstrates something different, namely how far into the wear-out unknown the fleet has scouted. Both facts matter, and they are different facts. Ask for the lead unit's hours and current condition.
• Duty-cycle match. Hours at steady baseload do not certify cycling duty, and hours in a mild coastal climate do not certify an inland summer. Ask for hours at a duty profile and environment resembling the site's.
• Fuel match. For fuel-flexible classes, the stated fuel range is a capability claim; hours are accumulated one fuel at a time. Operation on pipeline natural gas says little about sustained operation on a site's actual biogas quality or on blends. Ask for hours on the fuel the site will run.
• Degradation, measured rather than modeled. Ask for measured efficiency and output degradation against hours from operating units, with the dataset's fleet coverage stated. A degradation curve produced by simulation is a hypothesis, not a record.

Section 04Pillar two: service depth and the parts question

A machine's availability is mostly a statement about the organization behind it. Mature classes are backed by dense ecosystems: factory and independent technicians within driving distance of most industrial sites, distributors with parts on shelves, rebuild shops, rental fleets to bridge an outage. A young class is backed, in the typical case, by the manufacturer alone. That is not disqualifying. It is a structure whose failure modes the owner should price.

The service questions are concrete. Where do trained technicians physically sit, and how many exist in total, as a roster rather than a coverage claim? What response time will the manufacturer commit to in writing, and what remedy attaches if it is missed? What is the technician training pipeline, and what happens to regional coverage if one field engineer resigns? Remote diagnostics are genuinely valuable and typical of newer classes, and they are also a dependency: they work only while the manufacturer's monitoring operation exists and is staffed.

The parts questions are harder and matter more. Which components actually fail in service, and what are current lead times for each, quoted from experience rather than intention? Which parts are single-sourced, and from whom? A young manufacturer with a young supply chain can have its entire fleet idled by one supplier's stumble. What on-site spares inventory does the manufacturer recommend, and what does it cost? That cost belongs in the model, not in a footnote. Modularity helps here, and honestly so: where the product is a fleet of identical smaller units, whole-module swap can substitute for field repair, and holding spare capacity is a legitimate availability strategy. But the swap pool must exist somewhere real, with logistics and pricing in writing.

Then the late-life question. Service agreements are typically written for the early years, while the investment case runs much longer. Ask what an independent service path would look like if it were ever needed: whether documentation is sufficient for a qualified third party to maintain the machine, whether special tooling is available for purchase, and what commitment exists to support discontinued models. Mature classes answer these questions with a market. Young classes answer them with a promise, and the diligence is about the quality of the promise.

Section 05Pillar three: the provenance of every number

Performance claims arrive with different evidentiary weight, and a study should sort them explicitly before any of them enters the model. A workable hierarchy, from weakest to strongest:

1. Specification sheets and marketing collateral. Statements of intent, not evidence. Useful for screening, never for modeling.
2. Manufacturer-reported fleet statistics. Real information with a known bias. Ask for the methodology: what fraction of the fleet reports, whether early units or bad months are excluded, and how availability is defined, because definitions move numbers.
3. Certification and listing tests. Genuine third-party or witnessed results with deliberately narrow scope. Safety listings and interconnection certifications establish that a machine can connect and operate safely; they say nothing about its economics. On emissions, California operates a distributed generation certification program under which manufacturers of electrical generation technologies exempt from air-district permit requirements must certify to criteria-pollutant emission standards before selling in the state.³ For a low-emissions entrant, presence on the current certification list is a real, checkable fact with permitting consequences, and the owner should verify the exact model on the current list rather than accept a summary claim. What no certificate establishes is availability, degradation, or service life. A certificate proves what it tested.
4. Independent measured performance at third-party sites. The strongest pre-contract evidence: metered output, fuel input, and availability at operating installations, reviewed by the owner's engineer rather than curated by the seller. Reference calls belong here, and they are chosen from the full fleet list, not from the manufacturer's shortlist. Three owners, each with more than a year of operation, will teach more than any document.
5. The owner's own acceptance test. The only evidence generated under the owner's contract: a written test protocol at site conditions, with measured output, efficiency, emissions, and availability over a defined period, tied to contractual remedies. For a young class this is not a formality. It is the moment projections become obligations.

The provenance rule is simple to state: every number that moves the model carries its source and its conditions, and any number the seller cannot support at the appropriate tier is modeled conservatively or excluded. This is the same rule an honest study applies to utility timelines and tariff inputs. Entrants are not being singled out; they are being included.

Section 06Pillar four: the counterparty behind the warranty

A young technology class is usually sold by a young company, and the two risks compound: the same purchase exposes the owner to an immature machine and to a counterparty whose own durability is unproven. This is normal in the history of energy equipment. Every mature manufacturer once looked exactly like this. It still has to be priced, because a warranty is worth precisely as much as the entity standing behind it.

For publicly traded manufacturers, a remarkable diligence resource exists and costs nothing: the risk-factor disclosure the company itself files with the U.S. Securities and Exchange Commission, which securities regulation requires to be a discussion of the material factors that make the investment speculative or risky.⁴ Those pages routinely address limited operating history, supplier concentration, dependence on additional capital, and warranty reserves, in the company's own words and under liability. An owner evaluating the machine should read the manufacturer's filings at least as carefully as its brochure. For private manufacturers, ask for the equivalent facts directly: financing runway, audited statements, order backlog, customer concentration. A seller may decline to answer. A decline is also information.

Then structure the contract for continuity rather than hoping for it. The instruments are standard: escrow of design documentation and maintenance procedures sufficient for qualified third-party service; parts-supply commitments with defined triggers; transferability of warranty and service obligations to a successor; owner access to controls and monitoring data rather than seller-exclusive custody. None of this is adversarial. A manufacturer that believes in its machine can accommodate an owner who plans for the case where the manufacturer is wrong.

Section 07Pricing the risk into the model

Diligence only matters if it changes numbers. The table summarizes where the evidence typically differs between mature classes and young entrants; the paragraphs after it translate those differences into model discipline.

Run availability as a sensitivity, not a point value. Model the project at the manufacturer's projected availability and again at a conservative band, and state plainly whether the investment case survives the conservative case. Any derating chosen this way is an illustrative judgment and should be labeled as one. The point is not that anyone can compute the true number in advance; it is that the decision gets made with the downside visible.

Carry an operations contingency. Young-fleet maintenance costs are estimates built on short histories. A model that carries the manufacturer's service quote without contingency is treating a projection as history, which is the exact error this framework exists to prevent.

Buy redundancy deliberately. Modular classes make spare capacity a purchasable good. Pricing one increment of redundancy, along with the space and electrical provisions to use it, is often the plainest protection available against fleet-wide surprises. Whether it is worth buying is site arithmetic, not doctrine.

Bound the role. The entrant does not have to carry the whole site to earn its place. Hybrid configurations, in which the young machine serves a defined share of load alongside grid service or mature equipment, let an owner capture the entrant's advantages while capping the consequence of disappointment. For a first deployment, the bounded role is frequently the honest answer.

Put commitments where projections were. Availability warranties, response-time remedies, degradation warranties, and acceptance testing convert the seller's confidence into the seller's obligation. What a manufacturer will sign is evidence. What it will not sign is louder evidence.

Treat incentive eligibility as a determination, not a feature. Under current federal law, the investment tax credit for qualifying clean-energy property is 30 percent; statutory bonus adders exist but must be individually qualified, never assumed.⁵ Whether a specific new machine class, in a specific configuration and fuel arrangement, constitutes qualifying property for a given credit category is a question for qualified tax counsel, answered against the statute and current guidance rather than against a spec sheet. Young classes carry the additional wrinkle that guidance may not yet address them cleanly; a study should flag that uncertainty instead of resolving it by optimism.

Value the exit honestly. Mature equipment trades in secondary markets. A young class's resale value is unknown because the market for used units has not formed. Carrying residual value at or near zero until evidence exists is not pessimism; it is declining to book a market that does not yet exist.

Section 08The seven-question screen

The pillars compress into a screen an owner can run in a single meeting with any manufacturer of a young class. The questions are not hostile, and a confident seller will have crisp answers. Vagueness in response to any of them is itself a data point.

1. How many fleet hours exist on the exact configuration being quoted?Total fleet hours, hours on the current product generation, the lead unit's hours, and hours on the site's intended fuel and duty cycle, stated separately.
2. How many units operate commercially at third-party sites today?And, from the full fleet list, three owner references with more than a year of operation each, selected by the buyer rather than the seller.
3. What availability has the operating fleet actually achieved, and defined how?The definition, the reporting coverage, and the exclusions, in writing. A number without a definition is not a number.
4. What service presence and response time will be committed in writing?Technician locations and headcount, committed response times with remedies attached, and every remote-monitoring dependency named.
5. Which components are single-sourced, and what are the real lead times?Plus the recommended spares package, priced into the project, and whole-unit swap logistics if the class is modular.
6. Which performance numbers are third-party measured, and which are projections?Certifications identified by their actual scope, measured degradation data on the table, and an acceptance test at site conditions accepted in principle.
7. What happens to the owner if the manufacturer stops operating?Disclosed financial condition, escrowed documentation, warranty transferability, and a parts continuity plan, inside the contract rather than inside a conversation.

Section 09Symmetry, and the honest conclusion

One caution completes the framework: the scrutiny must run both ways. Mature classes get no free pass for being familiar. The incumbent's questions are different, not absent: an aging design's efficiency against current alternatives, tightening emissions rules and the control-technology spending they imply, part-load behavior against the site's actual profile, and the wear-out economics of equipment approaching the far end of its own bathtub curve. A study that interrogates the entrant and waves the incumbent through is not neutral. It is nostalgic.

Run honestly, this framework will sometimes select the entrant. A young class whose measured early record is clean, whose manufacturer signs real commitments, and whose fuel flexibility or permitting posture resolves the site's binding constraint can win the comparison on the merits, in a bounded role, at a priced risk. The framework will sometimes decline the entrant, when the available tier of evidence cannot support the projections the economics require. Both outcomes are correct when they come from the same arithmetic. What is never correct is deciding by temperament, in either direction, and calling it analysis.

That is the posture of an independent firm on the owner's side of the table: no machine to sell, no class to defend, and no stake in whether the winning answer is old or new. The operating base is young. The discipline for grading it does not have to be.