Experts say physics of failure is overtaking MTBF as the go-to model.
A large part of the contract manufacturing session that I co-chaired with Circuits Assembly Editor-in-Chief Mike Buetow at this year’s SMTA International focused on systems and processes related more to reliability than quality. Our panel discussion delved into that topic more deeply. At the center of that discussion was the question, “Can an industry with margins as thin as those found in electronics manufacturing services afford the liability that comes with recommendations related to reliability?” A decade ago, that answer would probably be no. Today, that answer is far less clear.
Two trends are responsible for changing the playing field. New software tools take a “physics of failure” approach to model and predict reliability far more accurately than mean time between failures (MTBF) mathematical models. And the EMS model continues to migrate toward greater value-add through engineering services and identification of cost-reduction opportunities. Leaving reliability evaluation and recommendations totally in the realm of customers can open the door to greater risk, if those customers aren’t worrying about that part of the equation until field returns show up.
From the modeling standpoint, several available tools identify potential reliability risks based on analyzing the product’s environment against the performance ratings of its various components. The University of Maryland’s Center for Advanced Life Cycle Engineering (CALCE) developed a package known as Simulated Assisted Reliability Assessment (SARA) that supports its Design for Reliability (DfR) assessment process. DfR Solutions has also developed an automated design analysis tool, known as Sherlock Automated Design Analysis, which takes data from existing design activities and simulates the environment where the product will be used, including temperature, vibration and shock, and then predicts the life of the product based on several key failure mechanisms.
Why the change from the MTBF model?
“The underlining problem with MTBF is the assumption of a constant failure rate. That doesn’t give you an answer that is valid because failure rates are rarely constant. Today’s methods give you a better idea of how many units will fail at different points in time and the effect of certain types of stress failure rates,” said Fred Schenkelberg, owner of FMS Reliability, a reliability engineering and management consultancy.
Citing his prior experience at a consumer electronics firm, he pointed out that the number one reason quarterly numbers were missed was because warranty cost estimates were off.
“When you use a physics of failure model, such as the models developed at CALCE, you can see patterns of failure over time. This type of analysis works in tandem with failure modes and effect analysis (FMEA) processes. You typically use an FMEA approach to start developing a model for a new product because the physics of failure models require that you know the failure mechanisms,” he added.
“I think the potential is there for broad adoption of physics of failure analysis tools,” said Cheryl Tulkoff, senior member of the technical staff at DfR Solutions. “The overhead and expertise needed to do it is low. A working model can be set up in as little as two hours. In the early days it took much more time and experience. Numerical modeling was less focused on the prevention side. The new models use empirical data to generate a more accurate result.”
According to Tulkoff, the majority of users of DfR’s software have been OEMs. But, EMS companies are now starting to look at it as well. However, there are two distinct classes of users in EMS. Some are using it as part of their NPI process, but others use it more as a tool to evaluate the potential litigation risks of products with potential reliability issues.
“We are seeing more litigation between OEM and EMS companies on reliability-related issues. Figuring out whether a defect is likely the result of workmanship or an inherent reliability issue can be very critical,” added Tulkoff.
Another potential area discussed in the SMTAI session was the benefit of this tool in evaluating the right materials for the product application.
Tulkoff pointed out that while this type of reliability analysis is often used to help engineers avoid selecting material not robust enough to withstand the intended operating environment, it could also help them avoid specifying higher cost materials and components when a lower cost option would suffice.
“This type of analysis helps engineers pick the most appropriate, cost-effective solution. It also helps minimize test resources. You can build reliability in and do the minimum testing required, because you’ve modeled some of the issues for which you’d normally test,” she said.
However, no discussion on this topic would be complete without considering the legal implications.
“In the right context with the right customer and the right written agreement this could work,” said Jeff Roth, current shareholder of F&B Law Firm, PC and an attorney with over three decades of experience in EMS-related contract law. “However, a company needs to assess risk vs. rewards. There needs to be an agreement with very clearly written terms on what the customer is paying for. The contractor needs to bracket limitations and exceptions. There need to be warranty disclaimers to ensure there are no implied warranties given in that situation. The agreement also needs to limit consequential damages and any damages that can flow from that.”
Roth says he isn’t seeing a trend toward this in the EMS industry because he believes the return doesn’t support the potential liability.
He pointed out that while risk-mitigating agreements could be put in place, one big risk contractors would be taking would be whether their employees would be doing that.
“In small companies, the owners tend to keep their eyes on the ball and get this done. In larger companies, it can slip through the cracks,” he added.
Customer solvency is also an issue, since plaintiffs often seek to chase all involved parties in product liability suits, and a bankrupt OEM customer may not have resources to follow through on an indemnification agreement.
“These types of recommendations should never be the centerpiece of a customer’s picture of reliability. Opportunities should be carefully evaluated. The type of product makes a difference. And, ultimately, the OEM needs to validate and approve recommendations in writing. My impression is that the risk doesn’t justify the reward,” he said.
Like religion and politics, reliability discussions tend to spark a variety of opinions. Whether physics of failure software tools will become widely used in EMS is still up for debate. However, for contractors constantly debating with customers whether field failures are the result of workmanship or poor design, there may be some light at the end of the tunnel.