Question 6 of 7: Communication Between Design and Manufacturing

Question 6: Communication Between Design and Manufacturing (15 marks)

Question text not reproduced: the examination questions are © Engineers and Geoscientists BC. Open the official past paper (linked at the top of this page) to read the question, then follow the worked solution below.

Part A — What the designer must communicate (6 marks)

The information divides into four layers, and teams that transmit only the first layer are the ones that have quality problems.

Layer 1, the product definition. The controlled 3D master model or fully dimensioned drawings; the datum reference frame and the geometric tolerances built on it; material specification with condition and heat treatment; surface finish and coating; the bill of materials with revision and effectivity; and the joining specification — weld sizes and symbols, fastener grade, torque and sequence, adhesive type and cure. This layer answers "what is it".

Layer 2, the process and quality requirements. Which characteristics are critical to quality and therefore need capability demonstration; the control plan and the gauging or inspection method for each; the acceptance sampling scheme; traceability and lot-control requirements; packaging, handling and protection so the part arrives as made; and any regulatory or certification obligation that flows onto the plant. This layer answers "how will we know it is right".

Layer 3, the design intent. Why the tolerance is what it is, which dimensions are functional and which are for convenience, what the part must survive in service, and which requirements are inviolable against which are negotiable. Without this layer, manufacturing cannot propose a cheaper alternative without risking function, so it proposes nothing, and the plant's process knowledge — which is usually deeper than the designer's — never enters the design. A tolerance stack analysis handed over with the drawing is the clearest single example of transmitting intent.

Given. An assembly gap has a total specification width of 0.60 mm ($\pm 0.30$ mm) and is closed by five independent contributors, each currently drawn at $\pm 0.12$ mm interpreted as $\pm 3\sigma$. Manufacturing is to be held to $C_{pk}\ge 1.33$ on the stack.

Find. Whether the current tolerances support the requirement, and what must change if they do not.

1. Evaluate the current stack. The worst-case sum is $5(0.12)=\pm 0.60$ mm, which is twice the specification, so the design cannot be defended on a worst-case basis at all. The statistical stack is $$T_{\text{RSS}}=0.12\sqrt{5}=\pm 0.268\ \text{mm}$$ which fits inside $\pm 0.30$ mm, but only just.
2. Work back from the capability requirement rather than forward from the drawing. A capability of 1.33 requires the half-specification to be four standard deviations of the stack, so $$\begin{aligned}\sigma_{\text{stack}}&=\frac{0.30}{4}=0.075\ \text{mm} \\ \sigma_{i}&=\frac{\sigma_{\text{stack}}}{\sqrt{5}}=0.0335\ \text{mm}\end{aligned}$$ and the per-contributor tolerance that supports it is $\pm 3\sigma_i = \boxed{\pm 0.101\ \text{mm}}$. The drawings must therefore tighten from $\pm 0.12$ to $\pm 0.10$ mm, or the design must remove one contributor from the stack — which is almost always the cheaper answer, and is a design change rather than a manufacturing one. That conversation cannot happen unless Layer 3 was transmitted.

Layer 4, the change process. Engineering change orders with effectivity, break points and disposition of stock; the deviation and concession route for parts already made; and the escalation path when the plant finds something the design did not anticipate. Design releases are events in the plant's life, not just in the design team's, and communicating them badly is expensive in a way that mis-tolerancing a single feature is not.

Part B — Challenges when manufacturing is far away (3 marks)

The first challenge is structural and is worth computing rather than asserting, because it defeats the reflexive answer of "have more meetings".

Given. Design in Halifax (UTC minus 3), engineering support in Stuttgart (UTC plus 2), manufacturing in Shenzhen (UTC plus 8), all working 09:00 to 17:00 local.

Find. The three-way and pairwise overlap of the working days.

1. Reduce every window to one axis and intersect. On a UTC axis the windows are Halifax 12:00–20:00, Stuttgart 07:00–15:00 and Shenzhen 01:00–09:00. The intersection of the three is $$[\max(12,7,1),\ \min(20,15,9)]=[12,9]=\boxed{\varnothing}$$ There is no hour of any day on which all three sites are at work. Pairwise, Halifax and Stuttgart share 3 h, Stuttgart and Shenzhen share 2 h, and Halifax and Shenzhen share nothing at all. Extending everyone to a ten-hour day, 08:00 to 18:00 local, still gives a three-way intersection of zero.

That result forces the conclusion: on a three-site team of this geometry, the working method must be asynchronous by default, with Stuttgart used deliberately as a relay between the two ends, and any process step that requires all three parties in one room must be planned as an exception with someone's evening spent on it. It also shows why adding people makes coordination harder rather than easier — a twelve-person team has $n(n-1)/2 = 66$ possible pairwise channels, and without an explicit hub-and-spoke structure most of them carry stale information.

The remaining challenges follow from distance rather than time. Tacit knowledge stops flowing: the designer no longer walks the line, does not see the operator's workaround, and learns about a problem from a report rather than from a part in the hand. Feedback latency lengthens, so a defect that would have been caught on the first article is discovered after a container has shipped. Standards diverge — unit systems, and more subtly the difference between ASME Y14.5 and ISO GPS interpretations of the same geometric callout, which can change what a tolerance means without changing what it looks like. Measurement systems differ between the design house's laboratory and the plant's gauges, so the two sides disagree about the same part. There are commercial and legal frictions: intellectual property exposure, export control, and customs. And cultural difference affects escalation directly — in some working cultures reporting bad news upward is discouraged, so problems surface late, and the design engineer must build a route that makes reporting safe rather than assume one exists.

Part C — Three tools that facilitate the communication (6 marks)

Tool 1: the controlled digital master, under PLM with model-based definition. A single product-lifecycle-management system holds the 3D model, the annotated model-based definition, the bill of materials, the revision history and the release state, and every site works from it rather than from copies. What it solves is the ambiguity of "which version is current", which is the single most expensive failure in a distributed team; the release state is a fact in the system rather than a claim in an email. It carries the engineering change order workflow as well, so effectivity and break points are visible to the plant the moment they are approved. Its limitations are that access must be provisioned across companies and borders, that it disciplines only what is inside it — a decision agreed on a call and not entered is still lost — and that it does not carry design intent unless the team deliberately attaches the stack analyses and design records to the release.

Tool 2: the structured design-for-manufacture review and the first-article approval that follows it. A DFM and DFA review before design freeze puts the plant's process engineers in front of the model while change is still cheap, with a checklist that forces specific questions — draft, wall thickness, access for tooling, fixturing, handling, gauging — rather than an open invitation to comment. Its formal counterpart after freeze is the first-article inspection and production part approval process, in which the supplier demonstrates on real parts from real tooling that every characteristic is met and that the critical ones are capable. Together they convert communication from opinion into evidence: the review is where manufacturing changes the design, and the approval is where the design is proved makeable. Their limitation is that both are events, so they catch what is visible on the day and nothing that arises between them.

Tool 3: shared asynchronous visual collaboration, anchored on the model. A lightweight 3D viewer that both sides can mark up, annotated screen captures, short recorded walkthroughs of the assembly sequence, and a shared issue tracker in which every question is a numbered item with an owner and a state. This is the tool that actually answers the time-zone geometry above, because it lets a question raised at the end of the Shenzhen day be answered during the Halifax day and read at the start of the next Shenzhen day — a full turn in 24 hours rather than a week of missed calls. A recorded three-minute walkthrough of an awkward assembly step transfers more than an hour-long call at an antisocial hour, and it can be re-watched by the operator who was not on the call. Its limitation is that it is only as good as the discipline of writing things down, and it does not replace the small number of genuinely synchronous events — the design freeze review, the run-at-rate — which should be scheduled as exceptions and, where the volume justifies it, done in person.

A fourth tool worth naming in passing, because it is the one that spans all three, is a jointly owned process FMEA and control plan: design and manufacturing must sign the same document, which forces the two teams to agree on what can go wrong and who is guarding against it.