The convergence of simulation, data analytics, and AI is transforming the process of tracing warranty claims back to flaws in product design.

Accessing all data is critical, as is using machine learning algorithms to make data-driven decisions on design, manufacturing, and more.
Accessing all data is critical, as is using machine learning algorithms to make data-driven decisions on design, manufacturing, and more.

By Sudhir Padaki, Director, APAC Data Analytics, Altair

Manufacturers can typically expect to spend two percent of their product sales revenue resolving warranty claims. What’s more, in addition to the immediate costs of rectifying or replacing faulty items, enterprises need to factor in the less tangible damage inflicted on corporate and brand reputations. In an era of ultra-competitive markets and thin profit margins, alleviating these wide-reaching consequences represents an obvious priority. Reflecting this, manufacturers are increasingly responding to the challenge of warranty claims by applying an emerging technology mega-trend: the convergence of simulation, data analytics, and artificial intelligence (AI). Crucially, the integration of these capabilities can enable a virtuous, automated circle that applies the lessons learned from warranty claim analysis right back to initial product designs and the assumptions around which they were built. As a result, continual improvement is embedded in the entire journey, securing relentless reduction in warranty costs, sustained improvement in product quality, and stronger customer loyalty.      

This new approach leverages the rich array of field-based product data now available to manufacturers. Traditionally, channel partners such as dealers and resellers have been the primary source of warranty claim information. This can involve CRM and ERP system outputs, as well as plain text and audio formats. In addition, manufacturers can now draw on the flow of data downloaded by engineers during maintenance operations. Moreover, the connected car is just one example of how the Internet of Things (IoT) is facilitating a continuous stream of real-time information from products in the field. Spreading the net even wider, social media offers a different kind of insight. Here the raw information tends to revolve around consumer perception, but often provides advance notice of future warranty claims.

The sheer diversity of this data raises problems. AI-driven data analytics needs clean, consistent inputs. Deploying tools that prepare all the available data is therefore a cornerstone of any solution looking to exploit the new technology convergence. With effective data preparation in place, AI-enabled analytics can quickly draw invaluable insight from this wealth of information. The most obvious strategy is to link it back to smart manufacturing systems. By comparing data related to product faults with the granular feedback generated by these systems, it becomes possible to automate and expedite root cause analysis. Modern manufacturing is a complex process, and faults can often be attributed to a particular combination of factors. Remedial action can therefore be targeted with a high degree of precision and cost-efficiency.

However, the new intersection of simulation, data analytics, and AI enables manufacturers to go a step further. Warranty claims can be traced right back to the original design. Are the assumptions on which it was based being borne out by use in the field? How do the loads being applied compare to designers’ initial assumptions? What impact is this having on performance? How can the design best be modified to resolve problems?

By running new simulations based on the insight afforded by warranty claim analysis, designers can test and verify new iterations rapidly. Issues exposed by warranty claims can be designed out of the product, dramatically reducing future costs incurred due to faults.     

Essentially, simulation, data analytics, and AI form the building blocks of an expert system, one that draws on a continually evolving data reservoir to guide decision makers quickly towards the right course of action. And speed is critical. Today, even complex products are made in a matter of seconds. That means warranty costs caused by an unidentified fault will multiply fast, as will the savings realized by swift rectification.

Prioritization is another important part of the equation. Applying algorithms in line with the principles of recency, frequency, monetary value will ensure efforts are focused on the most pressing and costly issues.

Experience suggests that initial deployment is best targeted at specific elements of a product design. Solutions can then be scaled up. It should also be stressed that sophisticated warranty analytics is not the preserve of larger OEMs. The roots of smart manufacturing lie in the German ‘Mittelstand’ of SMEs. With relatively narrow design and manufacturing focus, such enterprises are well suited to this approach. And with limited human resources, the high degree of automation is particularly valuable.

Warranty claims are never going to be welcomed as good news by manufacturers. But now more than ever, they need to be recognized and embraced as an opportunity to affect profound change. By harnessing the convergence of simulation, data analytics, and AI, enterprises can address the causes of warranty claims, rather than the symptoms. And the sooner they use these technologies to hardwire continuous improvement into their design and production processes, the sooner they will start reaping the benefits.

sudhir padaki altair
Sudhir Padaki

About the Author:
Sudhir Padaki manages the business development and go-to market strategy of Altair’s data analytics portfolio for the manufacturing sector. He has several years of experience in building and delivering customized solutions for manufacturing and has enabled various organizations break technology barriers to adopt new product development strategies across EMEA and APAC.

Padaki earned a master of science degree in computational engineering from the RUHR University, Germany and a bachelor of science degree from the University of Madras.