A Qualification Framework for Non-OEM Drivetrain Components

2. The qualification gap

2.1 Why it exists

A machine manufacturer developing a travel drive works from an internal specification that encodes accumulated experience: an alloy and heat-treatment recipe validated against its own field returns, a case depth chosen against its own duty spectrum, an accuracy grade set by its own noise and durability history, a bolt preload derived from its own joint testing. That specification is a competitive asset, and it is not published. Neither, in general, is the drawing.

A supplier outside that relationship — an aftermarket manufacturer, a remanufacturer, a distributor specifying to a contract manufacturer — cannot obtain those criteria, and cannot demonstrate conformity to them. The obvious response, and the common one, is to demonstrate conformity to the part instead: to reproduce the dimensions of the item being replaced. That establishes something real, and section 3 sets out precisely what. It does not establish the thing the buyer actually wants to know.

2.2 Why the gap matters commercially

The consequence is an information asymmetry that penalizes the careful supplier. Two components can be dimensionally identical, fit identically and function identically on installation, and differ by a factor of several in service life, for reasons — case depth, core hardness, inclusion content, grinding burn, accuracy grade, bolt grade — that are invisible on the finished part and are the subject of DYCO-TP-101, DYCO-TP-102, DYCO-TP-109 and DYCO-TP-110. Where the buyer cannot see the difference, the buyer cannot pay for it, and a market in which quality is undetectable selects on price.

A framework that makes the difference statable is therefore not merely a technical convenience. It is the mechanism by which a supplier who has done the engineering can demonstrate it, and by which a buyer can require evidence rather than assurance.

3. What "equivalent" can and cannot mean

The word carries at least four separable claims. Compressing them is the root of most of the confusion in this market, and separating them is the first thing the framework does.

3.1 Dimensional conformity

The component occupies the same envelope, presents the same interfaces and articulates the same way as the item it replaces. This is the claim that "it fits", and it is verified by the geometric methods of DYCO-TP-110: interface geometry referenced to the datum frame of the original, splines verified as fits rather than dimensions, bolt circles and mounting faces controlled to that frame.

It is necessary. It establishes nothing whatever about how long the component lasts.

3.2 Material and process conformity

The component is made of what it is claimed to be made of, and was processed as claimed: the stated alloy, the stated case depth on the stated surfaces, the stated core hardness, an acceptable microstructure, no grinding burn. This is verified destructively on a sample and by process control on the population.

This claim is where most of the difference in service life actually resides, and it is the claim most often absent.

3.3 Calculated capacity equivalence

The component's rated capacity, calculated by a published method from its own geometry and material condition, is not less than that of the item it replaces, calculated the same way, under the same duty assumptions. This is a calculation, not a test, and it is only as good as its inputs — but it is a calculation any competent engineer can reproduce and challenge, which is exactly what an assertion is not.

3.4 Demonstrated durability

The component has been run, on a rig or in the field, under a defined duty, for a defined duration, with a defined acceptance criterion, and passed. This is the only claim that is evidence rather than inference. It is also the only one that costs a test program, and it is therefore the one most rarely available.

The levels are cumulative in evidential weight but not in kind, and a higher level does not subsume a lower one: a component that has passed a durability rig but was never dimensionally verified against the correct datum frame will still install crooked. They are also not a grading of suppliers. They are a description of what has been shown.

4. Evidence by component class

The framework is only useful if it resolves to specific methods. Table 2 maps each component class to the published standards that supply them. The list is not exhaustive and is not intended to be; it is the minimum set that makes each claim checkable.

4.1 The seal row is the honest one

Table 2 has one cell that says a capacity cannot be calculated, and it is worth dwelling on rather than glossing. There is no published method that predicts the service life of a floating face seal from its geometry and materials. Seal qualification is empirical: it rests on demonstrated durability, or on a supplier's accumulated field history with a specific design in a specific duty, and DYCO-TP-103 sets out why the governing variables — face cleanliness, load-ring working height, squareness and load-ring dryness — are largely process variables rather than design ones.

A framework that pretended otherwise would be worse than useless, because the seal is a common initiating failure and the one whose consequences propagate furthest. Where a claim cannot be calculated, the framework's job is to say so and to require the empirical evidence instead.

5. Why L2 is worth the effort

Calculated capacity equivalence is the level most often skipped, on the grounds that a calculation proves nothing. That objection misunderstands what it is for. Its value is not that it is certain; it is that it is reproducible and falsifiable. A rating calculation states its inputs — geometry, material grade, accuracy grade, load spectrum, service factor — and any engineer with the same standard can repeat it and disagree with a specific number rather than with a claim.

It also has a property that makes it unusually informative in this market: the load-life relations are steep. In the finite-life region, the pitting life curve ISO 6336-2¹ gives for case-carburized steel places calculated life proportional to contact stress raised to about the power −13.2 — the fit DYCO-TP-102 derives from that curve — and because contact stress goes as the square root of load, calculated life goes as load to the power −6.6. Every factor that multiplies load — the application factor, the mesh load factor of DYCO-TP-107, the face load distribution factor — therefore enters life at that exponent.

The consequence is worth stating numerically, because it explains why small specification shortfalls are not small. A component whose real load-side factors are ten per cent worse than assumed has about half of the calculated life; twenty per cent worse, about thirty per cent. A specification that is "close enough" on four independent load-side factors is not close at all.

This is also the argument against qualifying by dimensional copying alone. Copying reproduces geometry, which fixes some of the load-side factors and none of the material-side ones. It cannot reproduce a case depth it cannot see, and sections 7.1 and 7.2 of DYCO-TP-110 set out why that particular property is invisible on a finished part.

6. Documentation and traceability

6.1 Evidence that cannot be checked is not evidence

Each of the levels above is a claim about a physical population, and the claim is only meaningful if it can be traced to that population. Case depth is determined on a sectioned specimen, so the item tested is by construction not the item shipped; the inference from sample to shipment rests entirely on batch identity and process control. Without a heat-treatment batch record, a destructive test result attaches to nothing.

6.2 Grades of document

The vocabulary of EN 10204²⁸ is the one generally used in this trade, and the distinction it draws between a test performed on the delivered items, or on a test unit of which they are part, and a test performed on material merely representative of them is the distinction that matters. A type 2.2 report and a type 3.1 certificate are both legitimate; they make different claims. A destructive test does not by itself rule out the 3.1 form: a case-depth section taken from the test unit the delivered parts belong to — usually their own heat-treatment batch — can support one. Where the result comes instead from process control that cannot be tied to the shipped batch, a 2.2 report is the most it can support, and for a destructively determined property such as case depth that situation is frequent. What the framework asks is that the grade be stated rather than left ambiguous. Where an external laboratory is used, the scope of its accreditation under ISO/IEC 17025²⁹ names the specific methods it is competent to perform, and an accredited result outside that scope is not an accredited result.

6.3 Sampling

Destructive evidence is sample evidence, and a sample plan under ISO 2859-1²⁷ states the relationship between sample size, acceptance quality limit and the confidence a result supports. A supplier who tests "periodically" has not stated a plan; a supplier who tests to a stated AQL has.

7. The declaration

The framework's practical output is a statement a supplier can make and a buyer can check. It has five parts, and its usefulness comes from the third and fifth as much as the first two.

1. Identification. What the component is, what it is offered as a replacement for, and on what basis the correspondence was established.
2. Levels claimed. Which of L0 to L3 are claimed, for which characteristics, by which method from Table 2.
3. Levels not claimed. Stated explicitly. A declaration that claims L0 and L1 and is silent on L2 and L3 is honest; one that lists only what it has is read as claiming everything.
4. Evidence grade. The document grade, sampling plan and traceability behind each claim, per section 6.
5. Duty assumptions. The load spectrum and service conditions under which any L2 calculation was performed, since a capacity claim outside its assumed duty is not a claim at all.

Point three is the one that distinguishes a declaration from marketing. A supplier who states what has not been demonstrated is a supplier whose statements about what has been demonstrated can be believed, and the trade in these components has for a long time been short of exactly that.

8. Limits of the framework

Four limits should be stated plainly, because a framework that oversells itself reproduces the problem it was written to address.

It is not a standard and confers nothing. There is no issuing body, no ballot, no conformity assessment and no mark. A component qualified against this framework is not certified, and the framework should not be cited as though it were an authority.

It does not substitute for OEM approval. Where a machine manufacturer's warranty, service policy or contractual terms require an approved part, this framework has no bearing on that requirement.

L2 is only as good as its duty assumptions. The load spectrum a component actually sees is set by the machine, the operator and the application, and a calculation performed against a nominal duty can be reproducible, honest and still wrong about a particular fleet.

It says nothing about safety-critical certification. Braking, structural and operator-protection requirements are addressed by their own standards and regulatory regimes, and nothing here displaces them.

What the framework does is narrower and, we would argue, still worth having: it gives a claim of equivalence a shape that can be examined, and it makes the difference between a component that has been engineered and one that has been copied into something a buyer can ask about.

9. Summary

An original equipment manufacturer qualifies a component against internal criteria that are not available to anyone else, which leaves every supplier outside that relationship without a citable definition of qualification and every buyer without a basis for comparison. The word "equivalent" compresses four separable claims — that it fits, that it is made of what is claimed, that its calculated capacity is not less, and that it has been run and survived — and the value of separating them lies as much in what each leaves open as in what each establishes. Each claim resolves to published methods that already exist, and the framework proposed here adds no method of its own; it arranges the ones that exist and requires that the level claimed, the level not claimed, the evidence grade and the duty assumptions all be stated. Calculated capacity equivalence is the level most often skipped and the one worth most, because the load-life relations are steep enough that a ten per cent shortfall in the load-side factors costs about half of calculated life, and because a calculation, unlike an assurance, can be reproduced and disputed. The framework is a proposal, not a standard, and it confers nothing; its purpose is to give a supplier who has done the engineering a way to show it, and a buyer a way to ask.