Systems Engineering

Using the IoT for prescriptive engineering

Share this post:

In one of my earlier posts, I touched on the topic of engineering for business outcomes, and how engineers are being encouraged to think more broadly about the impact of their decisions. In a follow-on post, I outlined how a digital thread, in the form of an IoT feedback loop, can improve the overall results of an IoT implementation.

In this post, I’d like to combine both of these thoughts into something I’ll call prescriptive engineering. Given that we’re still in the early days of IoT, this concept is still somewhat aspirational, but I enjoy thinking about possibilities.

Prescriptive engineering

So what is prescriptive engineering? Take the concept of the IoT feedback loop. We see lots of value in being able to monitor assets and product in operation, to learn about their behavior—and their users’ behavior—to derive insight about how to make improvements in the next design iteration.

diagram of prescriptive engineeringThe IoT feedback loop

But if we really want to engineer for business outcomes, it would be great if we could be more proactive. Rather than simply react to how a product is operating, perhaps we could simulate operational behavior such that we can optimize it during the design stage.

Moving beyond design simulation

Ok, this is not new, you say, and yes, we’ve been simulating the behavior of products and systems for decades. But that simulation has been based on known laws of physics that, while valuable, don’t provide insight into business outcomes.

And yes, we’ve worked on Design for Manufacturing (DFM) for decades as well. But this idea is focused on lowering the cost of manufacturing, which is indeed a business result, but not a top-line result, and doesn’t predict operational costs.

Using IBM Watson to construct a digital thread

What if we could use the data we’re collecting from operational performance and construct some models on the impact of that performance on top-line business results? We could then construct a digital thread such that engineers could experiment with design alternatives with an understanding of their impact on revenue, market share, and not only manufacturing costs, but operating costs, maintenance and repair costs, and warranty costs.

As I mentioned in my previous post on the topic of designing for the right analytics, we’re using only a small fraction of the data that we’re actually collecting. Perhaps we can feed the reams of collected data to IBM Watson along with a history of business metrics, and let Watson help construct the models that we can use to predict business outcomes associated with various design alternatives?[JC1]  These models can be embedded into the digital thread such that engineers can focus not only on improving a design based on performance, but designing FOR performance in the first place.

So yes, I admit prescriptive engineering [JC2] is aspirational, but I believe it’s possible. We’re only just now tapping into the power of the data we can collect within an IoT implementation, and the power of IBM Watson to constructively use that data.

I’ll expand on some of these thoughts in my next post, but in the meantime, you can learn more about IBM Watson capabilities at http://www.ibm.com/watson/.

In one of my earlier posts, I touched on the topic of engineering for business outcomes, and how engineers are being encouraged to think more broadly about the impact of their decisions. In a follow-on post, I outlined how a digital thread, in the form of an IoT feedback loop, can improve the overall results of an IoT implementation.

In this post, I’d like to combine both of these thoughts into something I’ll call prescriptive engineering. Given that we’re still in the early days of IoT, this concept is still somewhat aspirational, but I enjoy thinking about possibilities.

Prescriptive engineering

So what is prescriptive engineering? Take the concept of the IoT feedback loop. We see lots of value in being able to monitor assets and product in operation, to learn about their behavior—and their users’ behavior—to derive insight about how to make improvements in the next design iteration.

But if we really want to engineer for business outcomes, it would be great if we could be more proactive. Rather than simply react to how a product is operating, perhaps we could simulate operational behavior such that we can optimize it during the design stage.

Moving beyond design simulation

Ok, this is not new, you say, and yes, we’ve been simulating the behavior of products and systems for decades. But that simulation has been based on known laws of physics that, while valuable, don’t provide insight into business outcomes.

And yes, we’ve worked on Design for Manufacturing (DFM) for decades as well. But this idea is focused on lowering the cost of manufacturing, which is indeed a business result, but not a top-line result, and doesn’t predict operational costs.

Using IBM Watson to construct a digital thread

What if we could use the data we’re collecting from operational performance and construct some models on the impact of that performance on top-line business results? We could then construct a digital thread such that engineers could experiment with design alternatives with an understanding of their impact on revenue, market share, and not only manufacturing costs, but operating costs, maintenance and repair costs, and warranty costs.

As I mentioned in my previous post on the topic of designing for the right analytics, we’re using only a small fraction of the data that we’re actually collecting. Perhaps we can feed the reams of collected data to IBM Watson along with a history of business metrics, and let Watson help construct the models that we can use to predict business outcomes associated with various design alternatives?  These models can be embedded into the digital thread such that engineers can focus not only on improving a design based on performance, but designing FOR performance in the first place.

So yes, I admit prescriptive engineering is aspirational, but I believe it’s possible. We’re only just now tapping into the power of the data we can collect within an IoT implementation, and the power of IBM Watson to constructively use that data.

I’ll expand on some of these thoughts in my next post, but in the meantime, you can learn more about IBM Watson capabilities at http://www.ibm.com/watson/.

More Digital Twin stories

AI is transforming the way engineering teams work

Written by Dibbe Edwards | August 8, 2019 | AI Assistants, Engineering, Requirements Management...

IBM helps its customers embrace the necessary and inevitable digital transformation with the deployment of RQA within their Requirements Management solution. RQA uses Watson AI to help engineers improve the quality of their requirements, in real time. ...read more


Your CIO and Engineering Manager want you to attend this MBSE webinar

Written by Mark Kovich | July 22, 2019 | Engineering, Events, Systems Engineering

On July 31, please join me for a special webinar that is part of the continuing IBM Engineering Academy Comes To You series. ...read more


IoT Exchange Sydney: Unlock new value from data with AI and IoT

Written by Joel Pinto | June 21, 2019 | Conferences, Engineering, Events...

IBM IoT Exchange is an educational event designed to teach and inspire attendees with the latest innovations and thinking. ...read more