Don't Fail-to-Scale (Part 1 of 2)
The Five Critical Design Questions to Answer Prior to Scaling Your Medical Device
By: Peter Holst, Partner, Aproio LLC
Picture this… You’re in a phase gate design review with your Executive Team. You’re about to make a major commitment to tooling and the question is asked: “How do you know this is a good concept? How do you know this design will be successful through scale-up?”
A silence falls on the room. All eyes turn to you.
How do you confidently answer this question?
Waiting for 100% certainty is not an option but spending time and money on a concept that is fundamentally flawed is career-limiting.
All engineers must make decisions without knowing all the facts. If you wait until everything is known you will likely have missed a market window, or watched your competitors get to the market first. But rushing to tooling without getting basic insights into your design is bad, and so is delaying tooling until you have all the facts. Finding the right balance between speed and risk is critical.
You’ve been diligent. You’ve gotten inputs from your peers. You’ve thought about this a lot, and you think you have a nicely integrated design. You many even have a prototype in hand that seems to work. What else can you do? It’s time to build parts at scale, put them together and do some testing. But are your design and prototype robust enough to perform to specification when subjected to all of the variability in the components and subsystems that are inherent in manufactured components and that your subsystems will naturally encounter as you scale?
Wouldn’t you like the assembled device to confirm that your design works? Or are you wishing that the assembled device shows that your design works? This device is important to the company’s future. The company doesn’t have time or money to waste. The VP of Engineering is under pressure from the CEO to get to market with a reliable device.
Evaluating Your Design
In our experience, the best way to answer this reliability question is by answering these five fundamental questions:
1. Have I clearly defined the product’s performance requirements and are these current?
In the most specific terms possible, what are the inputs and outputs expected from the product? What are the tolerances on these factors?
Sources of this information vary – from Market Requirements Documents, Quality documents, an executive’s latest idea, etc. And requirements change as the project proceeds. Regardless of the source, the engineering team needs to translate these into specific, measurable criteria. Only in this way can clear “success criteria” be defined & understood by all constituents.
This may seem basic but many engineering teams, while caught-up in the daily grind of project needs, do not engage in the rigor required to translate all requirements into a complete set of testable specifications. This applies, even more so, when it comes to incorporating any changes that may have arisen since the project kicked-off.
Key Takeaway: Prior to committing to tooling – take the time to confirm that all critical requirements have been captured and turned into hard pass/fail criteria. And do a “once-over” check with all key constituents (executive team, marketing, quality, compliance, etc.) to assure that these requirements and specifications meet current project needs.
2. Will the design work - will it perform the essential functions – if everything is perfect?
If all the parts are perfect – dimensions on nominal, elastomers at exactly the nominal durometer, motors, actuators and sensors perfectly to specification – does the design work? Because if it doesn’t there is no reason to go any farther. This can often be answered with quite simple, low-cost, “scoping” modeling and experiments. Furthermore, answering this question sets us up for answering the next question. We can use the same analytical tools – and they really are tools that will serve us throughout the development process – that give us insight into on-nominal performance to help us understand what is going to happen when there is variation.
You might say “I know it works, I have a prototype in my hand”, which then raises the interesting question “does it really work, and if you build another one will that one work?”. The model helps us to understand basic things like higher order dependency on certain terms, which terms directly affect performance, and which terms inversely affect performance. And the model may reveal some completely unexpected phenomena that are critical to performance.
3. How sensitive is the design to variation in one or more inputs, and how does the product behave at the limits of the inputs?
For example, how sensitive is the design to external factors – like temperature and humidity? How does the product behave over the range of conditions it may encounter? And, how does it behave at the limits?
Product performance is typically characterized according to a Gaussian distribution. What is the shape of this curve – does it have a plateau at the top where performance is stable across a wide range of inputs, or peaked where small changes in inputs or operating conditions result in rapid degradation of performance?
Key Takeaway: Prior to committing to tooling ask yourself: Have I done the necessary amount of modeling at the component/sub-system level to determine where on the performance surface my current design sits? Am I in the “sweet spot” on the performance surface or am I near a performance cliff that almost certainly will be encountered due to the variability that my components/sub-systems will experience as we scale-up?
Very often, simple modeling can assist you in determining the answer to these questions and save you and your team significant time, money, frustration (and maybe your job!).
4. Which parts – individual components or subsystems - have the greatest impact on this performance & sensitivity? Which are most likely to fail?
Every design has a weakest link. What is it? What conditions will trigger it? How many other vulnerabilities does it have? How significant are they? And to complicate matters, failure is often multi-dimensional – a combination of factors and conditions may contribute to a component or subsystem failure.
5. What are the implications of a failure if it were to occur?
Does the product crash catastrophically, or in some recoverable way? Does it “failsafe”, or does it fail in a way that can result in user or patient harm? What design considerations and efforts are necessary to ensure a predictable, expected use and/or failure path.
Answers here typically derive from the product’s intended use. Is the product an inexpensive, throw-away part or a critical care medical device where failure has catastrophic consequences?
Key Takeaway: Prior to committing to tooling - ask yourself: Have I done a thorough risk analysis and does this analysis include an adequate risk mitigation plan?
How Aproio’s Optimal Design Process™ can help your design team:
We are a medical device design company focused on delivering Client Value by providing a device design approach that assures that our Clients can bring their products to market in the most efficient manner possible.
Over our 120 + years of combined industry experience we have continuously refined this approach. We call it our Optimal Design Process™. This is a design philosophy that focuses on Knowledge-Driven and Model-Based design and employs time-tested project management and client communication tools to streamline the overall design experience. Optimal Design Process™ allows our Clients to provide the relevant answers to the Five Questions listed above and to avoid Wishful Engineering by applying the necessary up-front rigor to specification development, virtual design iterations and sub-system testing.
We strongly believe that the appropriate level of modeling at the component and sub-system level enables development of robust sub-systems, and that confirmation testing at the sub-system level leads to a robust total device design solution that Will Work versus one that Can Work.
This approach speeds-up the overall development process by avoiding costly and time-consuming trial-and-error engineering. The output of the Optimal Design Process™ is a robust device design that is completed on-time and on-budget.
We wish you and your team the best of luck in realizing your design concept! If you would like to find out more about our Optimal Design Process™ or if we can be of service in any way to help make your device concepts become reality – click here.
Also – keep an eye out in the coming weeks for our next Blog Post entitled “A Reliable Design Approach” that will describe the FOUR KEY ELEMENTS that we believe all design programs must have to get a final product that WILL work versus one the MAY work. This Post will serve as a guideline to achieving a robust design that scales well into manufacturing.