Nine CVT Housings, One Machine: What Flexible Cleaning Requires

Overview

Flexible cleaning is a phrase that appears in every proposal and means very different things depending on what is being promised. At one end it means a machine that runs several variants with a program change and no tooling change. At the other it means a machine that accepts a different part if the fixture is rebuilt, which is not flexibility but a rebuild.

The distinction matters on transmission programmes, where a single platform can carry nine or more housing variants and the volume per variant is too low to justify a dedicated line for each. The machine has to amortise across all of them, which only works if the changeover is genuinely a program change.

Big Bird Industrial (formerly Shimada Big Bird Industrial) has been building cleaning and deburring systems for automotive powertrain components since 2001, and the case described here is a CVT housing line configured for nine variants at a 30-second cycle with dual robots.

1. What Flexible Has To Mean To Be Worth Paying For

A flexible machine earns its cost only if the change between variants is short compared with the production run per variant.

The Changeover Arithmetic

If a run lasts four hours and changeover takes two, the machine spends a third of its life not cleaning. The same machine with a fifteen-minute changeover spends six percent. That ratio, not the variant count, is what makes flexibility real.

Three Things That Have To Change Together

Identification, fixturing and the cleaning program all have to switch. If any one of them requires manual intervention, the changeover time is set by that one step regardless of how automated the others are. This is where most flexible proposals fail in practice.

2. The Nine-Variant Case

A CVT housing line configured for nine variants runs a 30-second cycle with two robots, holding the part at or below 3 mg overall with a dryness criterion measured by the solvent method.

Why The Variants Share A Machine

The nine housings differ in size and in the arrangement of bearing bores and oil passages, but they share a family resemblance: similar footprint, similar mounting faces, similar access for nozzles. That shared geometry is what makes one machine possible. Where variants differ structurally, the same approach produces a machine that is good at none of them.

The Dryness Criterion

On a housing with internal passages, residual water is measured rather than judged. A criterion stated in millilitres, verified by rinsing with a water-insoluble solvent and measuring the water it picks up, turns dryness into something that can be specified and checked across all nine variants in the same way.

3. Identification At Load

A flexible machine has to know which variant it is holding before it starts. The identification is either read from a code on the part or pallet, or inferred from the geometry.

Reading A Code Versus Recognising A Shape

A code is unambiguous and requires the code to be present, legible and correctly associated with the part. Geometric recognition requires no marking but is slower and less certain on variants that differ only in a bore position.

Palletised Loading With Visual Recognition

On a DCT housing line, housings arrive on pallets and are identified visually as they are loaded, after which the fixture adapts automatically. That arrangement combines the certainty of a code with the speed of a palletised transfer, and it also allows a housing to be presented without precise clamping, which reduces the loading time.

4. Fixture Adaptation

Fixtures are where flexibility is usually lost. A fixture that grips one variant precisely cannot grip another that differs by fifty millimetres in length.

Adjustable Supports Versus Variant Pads

Where variants differ only in length, adjustable supports cover all of them with a setting change. Where the mounting pattern differs, variant pads are unavoidable, and the changeover becomes a manual step measured in minutes. That is the honest boundary of flexibility on a given platform.

Why Adaptation Beats Precision Clamping

A cleaning fixture does not need to locate the part to machining tolerance; it needs to hold it against the jets and let liquid drain. Designing the fixture around adaptation rather than around precision is what makes a short changeover possible, and it is the opposite of what a machining fixture would do.

5. Program And Nozzle Boxes

The cleaning program carries the variant: which nozzles are active, what the robot path is, how long each stage runs and how the part is oriented for drainage. On a dual-robot cell at thirty seconds, the program also decides how the two robots share the work without colliding.

Where the variants differ in where the critical bores sit, interchangeable nozzle boxes are the alternative to a robot path per variant. A box costs fabrication and a changeover step; a path costs programming and validation. Which is cheaper depends on how often the variant changes over the machine lifetime.

6. What Still Costs Changeover Time

Even a well-designed flexible cell has irreducible changeover: the fixture check, the first-part inspection and the confirmation that the new program is running against the correct recipe. On a validated process those steps cannot be skipped, because a wrong recipe produces parts that look clean and fail sampling.

The practical target is to make all of it a defined sequence with a checklist rather than a judgement, so that the changeover is repeatable and its duration is known. A changeover that varies between fifteen minutes and two hours depending on who performs it is not a flexible machine; it is a machine with an undocumented process.

7. Where Flexibility Breaks Down

Three failure modes account for most of the flexible cells that stop being flexible within a year of commissioning.

Variant Creep

A machine specified for nine housings gets a tenth that is slightly longer, then an eleventh with an additional boss. Each addition looks small and each one reduces the margin that made the original nine work. Within two model years the fixture needs pads for four variants and the enclosure no longer accepts the largest one. The remedy is to state the envelope explicitly at specification, in length, width, height and mass, and to treat any part outside it as a different project rather than an extra recipe.

Recipe Version Drift

The program that runs variant five is edited to improve one feature and the edit is not documented. The next run of variant five uses an untested process, and the difference is invisible because the machine reports the same variant name. Stamping the recipe version into the production record, and requiring a first-part check after any edit, is what keeps a flexible machine inside its validated state.

The Skipped First Part

The first-part check after a variant change is the step that most often gets skipped under production pressure, because the part looks clean and the line wants to run. On a flexible machine this is the single check that confirms the correct recipe, the correct fixture and the correct active nozzles. A cell that has learned to skip it has traded its flexibility for throughput it cannot audit.

All three are management problems rather than engineering problems, and all three are why the changeover procedure belongs in the specification alongside the machine. A cell that documents its envelope, its recipe versions and its first-part check keeps the flexibility it was bought for. Those documents cost nothing to produce and are usually the first thing to disappear when the schedule tightens.

Frequently Asked Questions

How Many Variants Can One Cleaning Machine Handle?

Systems are routinely programmed for fifty to eighty variants where the parts share a family envelope. The limit is the enclosure and the fixture, not the program.

Does Flexibility Reduce Cleanliness Performance?

It can, if the shared nozzle arrangement is a compromise that suits none of the variants. The mitigation is per-variant active positions rather than a single universal spray pattern.

Is Visual Identification Reliable Enough?

On variants with distinct geometry, yes. On variants that differ only in a bore position, a code is safer, and the station should reject rather than guess when the identification is uncertain.

What Is The Realistic Changeover Time?

Minutes for a program change with a base fixture, and longer where variant pads or nozzle boxes are involved. The figure should be stated per variant pair rather than as a single average, because it varies.

Conclusion

Flexibility is the product of three switches working together: identification, fixturing and program. Nine variants at thirty seconds is achievable where the housings share an envelope and the fixture is designed for adaptation rather than precision.

A proposal should state the changeover time per variant pair and the parts of the fixture that must be changed. Without those, flexibility is a claim about the software rather than about the machine.