Metal and Plastic in One Assembly: Somebody Has to Absorb the Error

An industrial actuator assembly arrives as two drawings from two processes. An aluminum alloy flange and support ring. A plastic actuator housing.

Each drawing is achievable. Each part can be made comfortably within its own tolerance. And the assembly can still fail to go together — because the question that decides whether it fits was never asked at the quoting stage:

In which direction does the error flow?

Every assembly spanning two materials has to answer that. One part is the reference. The other absorbs. If nobody chooses, the assembly chooses for you, usually at the bench, usually late.

Two Processes, Two Different Sets of Tolerances

Aluminum and plastic are not two versions of the same material with different labels. They behave differently at every step that matters to an assembly.

Achievable tolerance is different. A machined or formed aluminum part holds a tight band as a matter of course — and holds it consistently across a run, because the process is subtractive or deterministic. An injection molded part holds a realistic band that is wider, and one that moves with shrinkage, wall thickness and gate position.

Stability is different. Aluminum does not creep. Molded plastic does, under sustained load, especially at the boss where a fastener clamps it. A joint that is tight on day one and loose in month six is normally a plastic joint, not a metal one.

Expansion is different. Aluminum moves roughly 23 µm per metre per degree. A glass-filled plastic can run half again as much or more. On a 200 mm part through a 40 °C swing, that difference is on the order of a tenth of a millimetre — an illustration, not a measurement of this project, but the right order of magnitude for why a precision fit across two materials behaves differently warm than cold.

Cost of change is different. Moving a metal dimension means touching tooling, or re-cutting, or scrapping. Moving a plastic dimension often means moving a boss in the mold — real work, but cheaper, and it does not disturb the metal.

Put those four together and the design rule writes itself.

Pick the Datum Before You Pick the Supplier

The datum should be the part that cannot move. Everything else is measured from it and absorbs the difference.

In a metal-and-plastic assembly, that is almost always the metal. It is the stiffer element, the more dimensionally stable one, and the more expensive one to change. Making the metal the datum and letting the plastic absorb has a second, practical advantage: the corrections happen on the side where a fix is affordable.

Reverse it — make the plastic the reference and require the aluminum to come to it — and you are asking the material that moves most to define correctness. Every thermal cycle and every batch of resin becomes an argument about the metal.

This is not a design preference. It is the difference between an assembly where variation has somewhere to go and one where it has nowhere to go but into the fit.

Variation does not disappear. It is either absorbed by a part that was designed to absorb it, or it becomes a rework ticket.

What CTE Does to a Bolted Joint

The expansion mismatch does not only show up as a dimensional problem at temperature. It shows up as a clamp load problem.

A fastener passing through plastic and threading into aluminum is a joint between two materials moving at different rates. When the assembly heats, the plastic grows more than the metal beneath it, and the joint geometry changes. When it cools, the plastic does not fully recover — it creeps. The bolt does not loosen because it was under-torqued. It loses preload because the material under the head relaxed.

Two design answers, both ordinary and both decided at the drawing stage:

  • Spread the load. A boss large enough in diameter and filleted at the root carries clamp load without cracking. The thin, sharp-rooted boss is the one that splits.
  • Move the load into metal. Where the joint is structural or repeatedly serviced, a metal insert or a metal-to-metal clamp path takes the preload out of the plastic entirely.

Neither is expensive before tooling. Both are expensive after.

The Multi-Vendor Failure Mode

Here is what this looks like when the assembly is split across suppliers, which is the default structure in this category.

The flange comes from a metal shop. The housing comes from a molder. Both parts are inspected against their own drawings and both pass. They arrive at the client’s bench and do not fit — and at that moment there is no arbitration.

The metal shop is correct: its part is in tolerance. The molder is correct: its part is in tolerance. The assembly is wrong. Nobody’s paperwork shows it, and the fix requires a conversation between two companies that have no contract with each other, mediated by a customer who is now paying for the delay twice.

This was the client’s actual situation on this project. Their own description of it: parts from different suppliers created coordination difficulty and inconsistent quality, and dimensional mismatches between components from separate vendors caused assembly problems and delays. They were not buying bad parts. They were buying correct parts that did not fit together.

That failure mode is structural, not accidental. It is what a split supply chain produces when nobody owns the stack.

Surface Treatment Is a Spec, Not a Default

The same principle applies to finishes, and it is where prototyping projects most often lose a week.

On a separate bracket programme for a US industrial equipment manufacturer — two custom steel mounting brackets for assembly validation — the review question was not dimensional at all. It was whether the parts needed sandblasting before anything else happened.

That question is easy to skip. A bracket can be made, shipped, and look like a bracket. But blasting is not cosmetic: it cleans and roughens the surface, it is the preparation step that makes a subsequent coating adhere, and it changes the surface you are handing to the next process. Decide it after the parts are made and you either accept a finish you did not want, or you rework parts that were dimensionally perfect.

The client’s own experience with previous partners was exactly that — samples produced without anyone confirming the surface treatment requirement, then rework, then delay.

Whether to blast is not a finishing decision. It is a drawing decision, and it belongs before the first cut.

The same logic covers anodising, plating, passivation and powder coating. Every one of them is a process with a dimensional and a functional consequence, and every one of them is cheaper to specify than to retrofit.

What One Supplier Actually Changes

It is tempting to sell this as convenience. Convenience is the smaller half.

The real change is that the tolerance can be moved to whichever side is cheaper to move. When one supplier owns the flange and the housing, an interference found at assembly is resolved by deciding where it is cheapest to fix — a boss in the mold, a clearance in the metal — and then doing it, once, with one owner of the answer.

When two suppliers own the two halves, each can only fix its own side. The cheapest fix may sit on the side that has no reason to accept it, and the customer arbitrates. That is why multi-material programmes with split supply chains tend to converge on the more expensive fix, later.

The Sequence We Ran

  1. All component drawings reviewed as one assembly. Not part by part — the flange, the support ring and the housing reviewed together for dimensional tolerances and assembly fit, with the datum decided before anything was cut.
  2. Material and surface specification confirmed. Aluminum alloy grade for the flange and support ring, resin for the housing, and the surface treatment question settled in writing.
  3. Dedicated tooling for the aluminum components. Tooling developed for the flange and support ring with dimensional control treated as a system requirement, since these parts define the datum.
  4. Injection molds for the plastic components. The housing and related plastic parts molded to a specification that includes the clearance to the metal, not just their own drawing.
  5. First samples with full dimensional checks and assembly validation. The parts were measured, then assembled — because a dimensional report does not prove a fit.
  6. Client inspection and design confirmation. Samples reviewed and approved before production, with any adjustment confirmed in advance.
  7. Mass production and delivery, coordinated as one programme across two processes rather than two programmes that happen to share a customer.

What Came Out of It

  • Datum and tolerance direction settled at drawing review, before tooling on either side
  • Aluminum flange and support ring tooled and sampled to the assembly requirement
  • Plastic housing molded to a clearance specification defined against the metal, not in isolation
  • Assembly fit validated on physical samples, not inferred from inspection reports
  • Both material streams approved by the client before mass production
  • Finished components delivered on schedule, from one programme owner

What the Client Said

“Having both aluminum and plastic components produced by a single supplier eliminated our coordination headaches. Assembly fit was perfect.”

“The quality of both the aluminum flange and the plastic housing is consistent and reliable. We will continue cooperation on future projects.”

Four Questions for a Multi-Material Assembly

  1. Which part is the datum, and who decided? If the answer is “each part is made to its own drawing,” nobody has answered it.
  2. Where is the variation supposed to go? There should be a named part that absorbs it, and it should be the cheaper one to change.
  3. Is the surface treatment on the drawing, or in someone’s head? Blast, anodise, plate, coat — specify before cutting, not after inspecting.
  4. Who arbitrates when two correct parts do not fit? If the answer is “the customer,” the structure is already costing you money.

Send the whole assembly, not the parts list

A metal-and-plastic assembly is decided by the relationships between its parts. Sent one drawing at a time, those relationships are invisible until they fail at the bench.

What to send

  • All component drawings together — metal and plastic
  • The assembly: what bolts to what, and where the load path runs
  • The operating environment: temperature range, chemicals, vibration
  • Surface treatment and coating requirements, or the performance they must meet
  • Volume per part, and whether the parts ship as a kit or as a finished assembly

What comes back

  • A datum recommendation, and where the tolerance should be absorbed
  • A clearance specification for the plastic side, defined against the metal
  • The surface treatment decision, stated explicitly, with its dimensional consequence
  • Tooling and part pricing separated by material, so the scope is visible across both processes

Send us the assembly drawings →

Share:

Facebook
Twitter
Pinterest
LinkedIn

Related Posts

Scroll to Top
You may add material , quantity or other requirements here.