Key takeaways
- Most assembly cost is decided by the part count and the handling time, before any tooling is cut.
- Design the build for one direction of approach and one hand.
- Symmetry and keying cut orientation errors faster than any work instruction.
- A design rule that adds a part to save a process is usually a net loss.
Where assembly cost comes from
Assembly cost is time multiplied by the number of times something has to be picked up, turned, aligned and fastened. The time is not in the drawing, but it is in the build: a part that arrives the wrong way up costs a second and a half on every unit, and a screw that needs a second hand costs more.
Four activities dominate. Handling is picking a part up and getting it into position. Orientation is turning it until it fits. Insertion is pushing it home, often against a seal or a clip. Fastening is turning a screw, closing a latch or applying a clip. Every design rule below attacks one of those four.
The rules are not free. Each one has a boundary where it stops paying, and a design review should say where that boundary sits for the programme.
The ten rules
The list is ordered by the size of the effect, though the order shifts with the product. The first three usually move the number more than the last seven combined.
- 01
Remove the part
Ask what the part does and whether the part next to it could do the same job. A single moulding that replaces a bracket and two screws removes handling, fastening and inventory.
- 02
Build in one direction
Stack the assembly from above so every part approaches from the same side. A build that needs access from two sides needs a fixture or a flip, and both cost time.
- 03
Make orientation obvious
Where symmetry is not possible, exaggerate the asymmetry. A chamfer, an offset rib or a keyed boss removes the trial and error that a symmetrical part invites.
- 04
Self locate before fastening
Chamfers, pilots and lead ins guide a part into place so the operator does not have to hold it. The fastener then holds a part that is already in position.
- 05
Design for one hand
If the part needs a second hand to stay in place while a screw is driven, the design has added a person to the station. A clip, a rib or a snap feature can hold it instead.
- 06
Standardise fasteners
One screw size and one length across the product removes a decision at the station and a bin at stores. Mixed fasteners are a reliable source of a wrong build.
- 07
Give tool access
A driver needs clearance for its body, not just its bit. Check the swing and the approach angle on the model, especially at a wall or under an overhang.
- 08
Control flexible parts
Cables, braids and gaskets tangle and need two hands. Captive clips, stiffened ends and routed channels turn a flexible part into a predictable one.
- 09
Build in poka-yoke
A feature that only fits one way prevents the error the instruction would otherwise have to describe. Colour coding helps, but a physical key is harder to ignore.
- 10
Test where the work happens
A test station downstream of the build finds the fault after the unit has been closed. A test at the station that created the feature finds it in time to fix the cause.
Part count: the rule that pays most
Removing a part removes its drawing, its supplier, its incoming inspection, its stock location, its handling time and its failure mode. That is why part count reduction is the first rule and why it is the hardest to apply late.
The test is not whether two parts could be one moulding, but whether the combined part can still be moulded, assembled and serviced. A housing that integrates a cable guide may be cheaper to build and impossible to service. The answer depends on the product life and the repair strategy.
Apply the test at concept stage, when a change costs a drawing revision. At tooling stage the same change costs a new mould, and the business case usually disappears.
A second test is the handling count. If a part has to be picked up, turned and placed, it costs time even when it is cheap. A part that can be delivered in a strip, a tray or a reel removes handling that the drawing never shows.
Fastening and tool access
Screws are flexible and familiar, and they are often the most expensive way to hold two parts together. A snap fit, a press fit or a heat stake can replace a screw and remove the driver, the torque check and the loose screw risk. Each alternative has its own limit: a snap fit needs draft and a tool that can form the undercut, and a heat stake needs access for the horn.
Count the fasteners and the tool changes in the build. Two screw sizes mean two drivers or two bits, and a station that swaps bits loses time on every unit. A single fastener type across the product is worth more than a small saving on the fastener price.
Check access on the assembled model, not on the component. A boss that looks clear on the part can sit under a rib once the cover is on, and the driver then approaches at an angle that strips the head.
Orientation and poka-yoke
Orientation errors are the most common defect in a manual build, and they are the easiest to design out. A part that can go in four ways will eventually go in the wrong one, usually on a Friday afternoon.
Make the wrong way impossible or obvious. A keyed boss, an asymmetric mounting pattern or a connector that only mates one way removes the decision. Colour coding is a second line of defence, because it relies on the operator noticing.
Test the poka-yoke on someone who has never built the product. If they can assemble it wrong without force, the feature is not doing its job.
Where a part is genuinely symmetrical, make it symmetrical in every axis that matters, including the mounting holes. A part that is symmetric except for one hole invites the error the symmetry was meant to prevent.
Test during assembly, not after
A test at the end of the line tells you that the unit failed. A test at the station tells you which step failed, while the part is still accessible. For a sealed housing, the difference is a scrapped unit against a two minute fix.
Put the check where the feature is made: continuity after the terminal is seated, a push check after the clip is closed, a torque trace at the screw station. That turns the test into a process control rather than an inspection.
Keep the checks short. A station that adds thirty seconds to a sixty second cycle doubles the labour content, however good the intent.
What the rules do not fix
Design for assembly improves the build. It does not fix a component that cannot hold its tolerance, a supplier that ships the wrong revision or a test that measures the wrong thing. Those are quality system problems, and no clip design solves them.
The rules also trade against other requirements. A part count reduction can hurt serviceability. A snap fit can reduce the strength of a joint that must survive vibration. A single direction build can make a housing taller. The design review should record the trade, not just the saving.
Record the trade in the design file with the reason. A future engineer who sees a snap fit and no note will assume it was chosen for cost, and may replace it with a screw that adds the cost back.
| Cost element | What drives it | Design response |
|---|---|---|
| Handling | Number of parts and how they arrive | Fewer parts, parts that stay oriented in the tray |
| Orientation | Symmetry and visual similarity | Keys, chamfers and exaggerated asymmetry |
| Insertion | Alignment, force and seal drag | Lead ins, pilots and controlled interference |
| Fastening | Screw count, tool changes and torque checks | Snap fits, one fastener size, clear access |
| Test | Where the check sits in the route | Check at the station that makes the feature |
| Rework | Access after closure | Serviceable covers and replaceable modules |
The largest element in most manual builds is handling, which is also the one a drawing change can remove.
Frequently asked questions
What is the most effective design for assembly rule?
Removing a part. It eliminates handling, fastening, inventory and a failure mode at the same time. Most other rules reduce the time a part takes to fit; part count reduction removes the part.
Do snap fits replace screws safely?
For many covers and clips, yes, provided the undercut can be moulded, the deflection stays within the material strain limit and the joint does not see repeated separation. For a joint under vibration or a high clamp load, a screw or an insert may still be the right choice.
When should design for assembly start?
At concept stage, when the part count and the build direction are still open. A review at tooling stage can improve access and fastening, but it cannot remove a part without a new mould.
How do I measure whether the rules worked?
Count the parts, the fasteners, the tool changes and the assembly minutes per unit before and after. A design review that cannot show those four numbers has not measured anything.
Sources and further reading
- ISO 9001:2015, Quality management systems, requirements
- VDI 2860, Assembly and handling technology, recommendations
- Qumek design for assembly reviews and build time studies
Qumek Engineering Desk
Engineering team
The engineering team behind Qumek drawings, DFM notes, tolerances and inspection standards. Every guide is reviewed against the drawings and processes used on live programmes.
Want an assembly reviewed against these rules?
Send the assembly drawing and the build sequence. The engineering desk will reply with the parts to remove, the access problems and the tests to move upstream.
Send an assembly drawing