10.2 Design for X (DFX)
Key Takeaways
- Design for X is a family of design constraints, each optimizing the design for a downstream lifecycle stage.
- The four constraints ASQ names are design for cost, manufacturability or producibility, test, and maintainability.
- Around 70% of lifecycle cost is committed during design, while only a small fraction has been spent at that point.
- Design for manufacturability reduces part count, standardizes components, and designs for the actual process capability rather than an ideal tolerance.
- DFX constraints frequently conflict, so they must be weighted and traded explicitly rather than optimized one at a time.
The cost commitment curve
The single most important argument for DFX is that decisions made in design commit costs that later phases can only manage, not avoid. The conventional figures: roughly 70% of a product's lifecycle cost is committed during design, while only about 5% to 10% has actually been spent by that point.
| Stage | Cost committed (cumulative) | Cost spent (cumulative) | Cost of a change |
|---|---|---|---|
| Concept | 40-60% | 2-3% | 1x |
| Detailed design | 70-80% | 5-10% | 10x |
| Process design | 85-90% | 20-30% | 100x |
| Production | 95%+ | Rising | 1,000x |
| Field | 100% | Full | 10,000x |
Once tooling is cut, the design's manufacturability, testability, and maintainability are essentially fixed for the product's life. This is also why the design FMEA must be complete before production drawings are released for tooling.
The four constraints named in the Body of Knowledge
Design for cost (DFC)
Designing to a target cost derived from the market price and the required margin, rather than costing the design after it is finished.
The target is then allocated down to subsystems and components, and each design decision is evaluated against its allocation. Supporting techniques:
- Value engineering: for each function, ask what it costs and whether a cheaper means would deliver the same function. The ratio of function to cost is the measure.
- Should-cost modelling: build up the theoretical cost from material, process time, and overhead rather than accepting a quoted price.
- Total cost of ownership: include the cost of scrap, warranty, and service, not just the piece price. A component that is $0.30 cheaper and doubles the field failure rate is not cheaper.
Design for manufacturability and producibility (DFM)
Designing so the product can be made reliably with the processes actually available.
| Principle | Effect |
|---|---|
| Minimize part count | Every part is a purchase, an inventory item, an assembly step, and a failure mode |
| Standardize components | Fewer part numbers, better volume pricing, less inventory |
| Design for the process's real capability | Specify tolerances the process can hold at the required Cpk, rather than tolerances an ideal machine could hold |
| Design symmetric or clearly asymmetric parts | Symmetric parts cannot be assembled backwards; strongly asymmetric parts cannot be assembled at all in the wrong orientation |
| Design for automatic handling | Avoid parts that tangle, nest, or need reorientation |
| Minimize fasteners and reorientations | Assembly time is dominated by handling and fastening, not by the value-adding operations |
| Use existing processes and tooling | Novel processes carry capability risk that only appears at volume |
The tolerance point connects DFX directly to the rest of the Body of Knowledge. A designer who specifies $\pm 0.05$ mm on a process whose short-term capability is $\pm 0.09$ mm has designed a defect. Process capability data belongs in the design review, and a producibility review that does not include Cpk data for the specified characteristics is not a producibility review.
Design for assembly (DFA) is the closely related discipline that scores a design on assembly efficiency, asking of each part whether it must move relative to others, must be of a different material, or must be separable for service. Parts that fail all three tests are candidates for elimination or combination.
Design for test (DFT)
Designing so the product can be verified economically and completely.
- Test access: provide test points, connectors, and probe access; in electronics, ensure boundary scan or in-circuit test coverage.
- Testability by design: built-in self-test, diagnostic modes, error codes that identify the failing subsystem rather than merely reporting failure.
- Fault coverage: what proportion of credible failure modes the test suite actually detects. Untestable failure modes escape to the customer regardless of how good the test is at what it does cover.
- Test time and cost: a test that takes four minutes on a 30-second takt time becomes the constraint.
- Measurement capability: the test must pass a gage R&R against the tolerance it is verifying. A test designed without an MSA is a source of escapes and false rejects.
Design for maintainability (DFM-a)
Designing so the product can be serviced quickly and correctly in the field.
| Principle | Measure it affects |
|---|---|
| Accessibility of high-wear items without disassembly | Mean time to repair |
| Modularity and line-replaceable units | Mean time to repair, spares cost |
| Standard tools rather than special tools | Field feasibility |
| Diagnostics that identify the failed module | Time to diagnose, no-fault-found rate |
| Keyed and colour-coded connectors | Reassembly error rate |
| Consumables reachable without removing other parts | Preventive maintenance compliance |
The key measures are MTTR (mean time to repair) and availability:
Availability can be improved either by making failures rarer (reliability) or by making repairs faster (maintainability), and maintainability is frequently the cheaper lever.
Other DFX constraints
The family is open-ended. Others commonly applied: design for reliability, for safety, for the environment and end-of-life recycling, for supply chain and sourcing, for serviceability, for compliance, and for six sigma (DFSS) itself.
Managing the conflicts
DFX constraints conflict routinely, and pretending otherwise is how designs get re-worked late.
| Conflict | Trade-off |
|---|---|
| Cost versus maintainability | A sealed, bonded assembly is cheaper to build and impossible to service |
| Manufacturability versus function | Fewer parts may compromise a performance requirement |
| Test versus cost | Test points, connectors, and self-test hardware add material cost |
| Manufacturability versus reliability | A cheaper material is easier to mould and fails sooner |
The methods for resolving them are already in the Body of Knowledge:
- Quality function deployment weights customer requirements and carries them into technical requirements; the roof matrix is specifically where negative correlations between technical requirements are recorded and resolved.
- Prioritization matrices score design alternatives against weighted DFX criteria.
- Pugh concept selection compares alternatives against a datum concept on each criterion.
- Robust design finds settings that are insensitive to the noise the trade-off introduces.
The governing discipline is that trade-offs are made explicitly, early, and against weighted criteria rather than implicitly and late by whichever function shouts loudest.
Approximately what proportion of a product's lifecycle cost is committed during the design phase, and why does this matter for DFX?
A designer specifies a tolerance of plus or minus 0.05 mm on a characteristic produced by a process whose short-term capability is plus or minus 0.09 mm. Which DFX principle has been violated?
A product has an MTBF of 900 hours and an MTTR of 6 hours. Which DFX approach offers a second, often cheaper lever for improving availability besides increasing MTBF?