Sensors Support New Data-Driven Design Workflows

According to a PwC report, The Internet of Things: What it Means for the U.S. and Manufacturing, 38% of manufacturers are already embedding sensors in products to collect usage and health data. Their goal is to leverage it in a variety of ways, including predictive and preventive maintenance as well as to drive future product innovations. In fact, 34% of respondents in that group said it is “extremely critical” that U.S. manufacturers infuse an IoT strategy into their digital operations as a means of maintaining a competitive edge.

For engineering, the promise of IoT is generating the data-driven feedback that can help guide future product directions, resulting in designs that are optimized for materials costs, quality improvements and more efficient manufacturing. That’s a big shift from today’s workflows, where engineers are often removed from any direct insights into product performance in the field other than spotty anecdotal discussions.

“For a long time, most products have been fire and forget,” says Stan Przybylinski, vice president of Research for CIMdata, a market research firm focused on the engineering sector. “Engineers were lucky if they got some warranty data, which is typically old and very after-the-fact, or if a product needed to be maintained, they might get some better information. One of the promises of IoT is a better opportunity to keep track of what the physical state of a product is.”

As engineers map out the next product iteration, most typically get answers to questions like, “can the product use less power,” or “can this component handle more vibration,” via informal discussions with people on the ground, perhaps support engineers, sales people or in the optimal case, customers. However, this feedback is based on personal experiences, not data, which means it is skewed by individual preferences and emotions, notes Brian Thompson, senior vice president of the CAD segment for PTC. “Invariably, an engineer would feel a lot more comfortable knowing and answering those questions if they had insight into how their existing product was behaving in the environment they’re being asked about,” Thompson says. “If you could learn from the product directly, there is no filter and you get real information.”

Still Early Days of IoT

Just because IoT devices have the capacity to collect a lot of potentially useful design data doesn’t mean it’s going to be easy to tap into that data, especially establishing workflows that loop it back into engineering and specifically, into existing 3D models created in CAD and simulation tools, experts say. For one thing, products in use in the field need to have been properly instrumented with sensors to collect the kind of data that can actually help advance subsequent designs—a bit of a “chicken and egg situation” for engineering groups just getting their feet wet with IoT, Thompson says.

Beyond sensors and connectivity, data plays a huge role in delivering design insights, and there is currently a huge gap in the skills and tools necessary to meld this element into traditional engineering solutions and workflows, experts say. Engineering, with support from the IT organization and design tool vendors, will need to build out solutions that aggregate and normalize data coming in from multiple products in the field. They will also need to construct the analytics and presentation layer necessary for transforming the data deluge into valuable insights while delivering them in a way that’s compelling and readily accessible to the average design engineer.

While CAD and simulation vendors are doing their part to evolve their product lines to address these issues, most of the early IoT-led design workflow is falling to internal engineering and IT groups, which are being tasked with a lot of heavy lifting in the areas of integration and customization. “This is not mainstream—it’s for bleeding edge early adopters who are willing to do a lot of integration work themselves,” says Mark Hindsbo, vice president of marketing for ANSYS. “It’s not out of the box, and it’s a project every single time.”

An example of how GE Predix and ANSYS can work together in assessing the environmental, health and safety impact of a plume dispersion from a potential gas pipeline rupture (left); detecting anomalies (top right) and realizing the digital twin of a blowout preventer (bottom right). It’s virtual diagnostics/root cause analysis and prognostics/lifting. Images courtesy of ANSYS.

Understanding the strategy for the smart connected product as well as exactly what you want in terms of collected data is the first step towards mapping out these new design workflows, notes PTC’s Thompson. Knowing, for example, that the thermal performance of a product in the field might be critical for future design enhancements is central to creating a sensoring strategy and leads directly into the next step, which is figuring out what to do with the data once it’s collected.

Consider a design team that takes its lead from marketing requirements that specify that a front loader product portfolio be able to universally support a load of 1 ton, says Thompson, providing an example to illustrate his point. Data collected from front loaders out in the field, however, reveals that units used in specific parts of the world rarely carry a load greater than 500 pounds, which means the product is being over-designed. “It starts with getting your feet wet, understanding the value of the data and then using it to inform a strategy for designing the next iteration,” Thompson explains. “Now you can design a more optimized front end load assembly to handle that load, which reduces cost and still meets customer requirements.”

IoT Platform Vendors At Work

To help facilitate such workflows, PTC is actively building “plumbing” into Creo that will make a connection to the ThingWorx IoT platform and API (application programming interface), Thompson says. Slated for delivery in Creo 4.0, these capabilities will formalize what PTC has been demonstrating on the road: A sensor-equipped bike feeding real-time usage data such as wheel speed and suspension pressure as it’s ridden directly into a Creo digital model of the bike by way of the ThingWorx platform. While the demonstration shows the power of a digital twin collecting data from one specific bike, PTC envisions data being collected from hundreds, maybe thousands, of bikes in the field, brought into ThingWorx and condensed down to a couple of critical design insights, which are then fed back to the engineering team via CAD.

“The ThingWorx layer could represent a summary of performance of 10,000 bikes averaged out over five weeks,” Thompson says, explaining the need for a step that would boil that fire hose of IoT data into something useful for the design engineer. “A data aggregation step has to occur if you want true product population data instead of single, one-to-one connection.” As part of that effort, Creo’s ThingWorx API could also cull data from other enterprise systems like PLM (product lifecycle management) or CRM (customer relationship management) in addition to physical IoT-enabled products, he adds.

Creo 4.0 will have built-in connections to the ThingWorx IoT platform to feed live, in-use data about products like this bike example, directly into the engineering workflow. Image courtesy of PTC.

Siemens PLM Software plans to leverage its acquisition of the LMS testing and simulation software to retool its CAD and Teamcenter PLM products to handle in-coming sensor data, according to says Peter De Clerck, director of Business Segment at Siemens PLM Software. LMS’ experience with incompatible data formats due to range of testing devices the software supports, coupled with its ability to process massive quantities of test data, are key assets in Siemens’ plan for dealing with the disparity and scope of big data emanating from sensored products, De Clerck says.

“With that part of the Siemens portfolio, we are already working with sensor data, not coming from IoT devices, but coming from instruments,” he explains. “We do a lot of time data manipulations, we do durability testing, and we calculate the fatigue life of components—that type of analysis isn’t much different than the type of analysis with IoT.”

Siemens PLM Software’s Predictive Engineering Analytics strategy aims to leverage data from multiple sources coupled with analytics to improve product designs. Image courtesy of Siemens PLM Software.

In addition to the LMS simulation software, Siemens’ Omneo Performance Analytics cloud-based platform, gleaned from its Camstar acquisition, along with recently acquired CD-adapco simulation software and its Teamcenter PLM backbone, will be central in melding IoT data into design workflows. “To close the loop, you need a backbone that has traceability across everything, including the IoT data,” says Ravi Shankar, director for Simulation Product Marketing at Siemens. Instead of storing a massive database in a backbone like Teamcenter, you also need capabilities for reducing the data in intelligent ways and then only storing what’s necessary to help the next iteration of the design, he explains.

At Dassault Systèmes, the CATIA Systems modeling tool, EXALEAD Big Data search and analytics tool, and the 3DEXPERIENCE platform are the core building blocks of its strategy for integrating IoT data directly into the engineering workflow, according to Olivier Ribet, vice president of High Tech Industry. In addition, Dassault is committed to keeping an open platform approach and forging key alliances with partners in the IoT space, Ribet says. “IoT is fantastic in the sense that it continuously helps engineers do the right engineering—not under-engineer or over-engineer,” he says.

For its part, ANSYS is doubling the surface of its open APIs across its entire product line and steering more solutions to the cloud to foster IoT data-driven design workflows. “Today, CAD, simulation and PLM are more point solutions than an integrated workflow so you might be able to exchange data or import data, but having fully open APIs and pre-built workflows is different from sending flat files back and forth,” explains Hindsbo. By opening up APIs across 80% to 85% of its product set and by building pre-built connections to IoT platforms like GE Digital’s Predix, ANSYS hopes to facilitate a data-driven design workflow that automates a lot of the process, including feedback insights back to engineers, as opposed to requiring custom integration, he says.

Putting cloud capabilities in place to support on-demand simulation bursts, instituting new elastic licensing terms, and building additional security and encryption capabilities into its offerings to protect simulation model IP (intellectual property) are other areas ANSYS is pursuing to address existing challenges to IoT-driven design workflows.

TRUMPF, a manufacturer of general purpose machines that cut sheet metal, is already leveraging live data collected from its equipment in use in the field to inform future design decisions as opposed to operating off assumptions, according to Stephan Fischer, head of the Software Development for the firm. Leveraging an IoT connectivity platform from C-Labs, the company is collecting in-field data, which is providing reliable evidence about how products are actually used and how they perform. “Knowing which material the customer prefers helps us to develop more specialized machines,” he says.

TRUMPF enforces the new data-driven design workflow by upholding an expectation that engineers don’t make any product decisions without comparing assumptions against in-field data. “This evidence-based decision making ensures that we focus on customer value based on reality, not assumptions, when iterating product designs,” Fischer says.

More Info

• ANSYS
• CIMdata
• C-Labs
• Dassault Systèmes
• PTC
• Siemens PLM Software