How to optimize product design for IoT
Back in the 80’s when computer-aided design, manufacturing and engineering (CAD/CAM/CAE) was in its early stages, a favorite topic of mine was around the idea of push-button product design optimization. Not the more esoteric mode of delivering the ‘best’ design, but actual product shape driven by the physics of the product in use. The idea was that the projected operating environment would provide the design variables that drove the optimization algorithms that would automatically modify the CAD shape parameters to optimize the product.
Déjà vu all over again?
Well, thirty years later, this idea of shape optimization has been largely realized, but the topic of design optimization is still hot. However, in today’s discussion the design variables are not related to shape, but related to business processes and outcomes.
Huh? Ok, it’s not push-button, but the potential of using business objectives to drive product design in an algorithmic sense is interesting to my ‘engineering’ side.
Using algorithms to drive product design
To apply any sort of algorithm to drive product design decisions, there need to be models that can be executed. The difference between today and the early 80’s lies in what is being modeled. Today, we want to model not only the product in operation, but also the processes to manufacture and maintain the product. Ultimately we want to model and understand the optimal manner in which to drive revenue, satisfy customers, increase profitability—and the list goes on.
The models that we use today are increasingly sophisticated digital representations of real-life designs or processes—called digital twins. We didn’t use this term in the early years, but the intent is roughly the same—to use data to drive design decisions.
Except in today’s scenario, we also use operational data, customer and market sentiment data, and business results. We capture operational data from instrumented devices, analyze it, and use the results of the analysis to optimize the design. We can analyze social network data to determine customer sentiment. We can look at failure data, warranty data, purchasing data—all sorts of data—and use the insight from that data to update and refine design requirements.
Digitally instrumented feedback
This digitally instrumented feedback loop—from design through operations to analysis and back to design—is called a digital thread. The components required to instantiate the digital thread include instrumentation within products, an IoT platform to collect the data, analytic modeling tools to derive insight from which requirements can be created and, of course, fundamental product design tools. These design tools include capabilities to capture and manage requirements, model architectures and designs, build, test and deploy the product.
For this feedback loop to be effective, stakeholders must organize their activities to be in support of the overall objective, i.e. business results. Providing guidance and incentive to these stakeholders requires an overarching framework that, in the early stages of IoT, isn’t yet well understood or defined.