A prototyping process is a planned sequence of builds in which each one answers a question the last one left open. The reason to plan it is timing: the cost and delay of a change grow the later it arrives, and a prototype is the cheapest way to find a problem early. That leaves three things worth settling before the first build: how much structure the sequence needs, what skipping it costs, and what each stage should produce.
Why Does a Prototyping Process Need Structure?
Without a prototyping process, prototyping drifts. One model gets built, someone marks it up, and the next version heads toward production without anyone having tested what the first one was supposed to prove. Stanford Biodesign’s prototyping guide describes the better pattern as iterative and progressive: early prototypes stay simple, quick, cheap, and incomplete, and more time and expense go in only as the project advances. Focusing each build on one essential question, the guide adds, is what retires a project’s biggest technical risks one at a time.
A prototyping process also keeps you between two failure modes. Over-building spends difficult, expensive materials or extra features on a question that did not need them, while under-building yields a prototype too flimsy to teach anything. The same guide notes that without a record of each round it is difficult to tell which features improve performance and which degrade it.
We would write down the question a build is meant to answer before making it, since a build with no question attached can only be judged on whether it looks good. Stanford’s guide also asks for a written test plan that maps each test to the requirement behind it, and the question you wrote down becomes the first line of that plan.
What Does Skipping Prototyping Cost?
A change made in CAD (computer-aided design) costs the time it takes to edit a model. A change that lands after tooling is cut can mean a new tool, and the price gap is wide: In Formlabs’ guide to casting, a silicone mold runs $200 to $1,000 while a steel injection mold runs $5,000 to $100,000. A flaw that a foam model would have shown can end up in the second bracket if nobody looks until the steel is cut.
The pattern holds well beyond molds. Research on engineering change finds that changes become more expensive and more time-consuming the later in a product’s life they occur. Manufacturing-software maker aPriori’s design guide puts it concretely for tooling: once tooling has begun, the options for re-engineering become increasingly limited.
That is why skipping a stage rarely removes its problems. It moves them to the next stage, where they cost more to fix.
What Marks a Stage as Finished?
Each stage should end with something you can check, not a feeling that the model is good enough. Write the finishing test before the build, because a criterion like holding a stated load can be checked and a feeling of strength cannot. We would call a stage finished only when it has produced the answer it was built for:
- Concept model: foam, cardboard, or a rough print. Done when the team agrees the shape matches the design intent well enough to justify detailed CAD.
- Functional prototype: tests a specific performance question. Stanford’s guide expects a works-like prototype to be rough, because most barely work and each only needs to settle one key question. Done when it passes the tests it was built for, not when it merely moves.
- Appearance model: finished to look like the product so it can be shown to the people who sign off. Done when it can stand in for the product without a caveat.
- Pre-production prototype: parts made the way production will make them, done when one meets the drawing without hand-fitting.
The last two stages are the expensive ones, and Stanford advises holding off on a fully integrated, final-looking build until well after you think the time has come. Let each stage earn the next. When a part nears the end of the sequence and you need to know whether it is truly ready, the checks for that call are in the post on whether a prototype is ready for production.
Who Should Review Each Stage?
We would put a different person in front of each stage, because each is the person most likely to find that stage’s problem. A concept model goes to a prospective user, and Stanford notes that certain looks-like builds are ready to put in front of users for sizing and design feedback. A functional prototype goes to whoever owns the performance requirement, and an appearance model goes to the people who will sign off on it. On a small team the reviewers overlap, and that is fine, as long as someone other than the builder sees each stage before the next one starts.
Pre-production parts go to the manufacturer, who will read the drawing and the part against each other. When users do see a build, watch what they do with it. That is Nielsen Norman Group’s guidance for user research too, since what people say about their own behavior, and what they predict they will do, is unreliable.
Where Do Printing and Machining Fit?
3D printing suits the early stages since no mold, special tooling, or fixtures are involved, which keeps setup cost minimal. For one-off and small-batch prototypes it is typically cheaper than machining, which carries programming and fixturing setup that gets spread over larger runs, although a large, solid, simple part can be the exception.
Which method fits which stage, with realistic tolerances and what to expect from each, is covered in how rapid prototyping works. Printing, routing, casting, and scanning all happen at our Brooklyn location, so a round does not have to wait on a vendor.
Not sure where you are in the prototyping process? Getting started on prototyping covers the first question to ask, and if you are not an engineer, prototyping for non-engineers covers what to have ready before a first conversation.
Tell us which stage you are at and what the next build has to prove, and we will say what we would build next. Reach us through the contact page or at 718-557-9578.
