TRY THIS TODAY
The Physics of the Failure
Identify a problem’s physical aspects to find its root cause
by Matthew Barsalou and Robert Perkin
When developing a cause and effect diagram, it is important to start with the physics of the problem by identifying exactly how such a situation could have happened while also digging down as deep as possible into the fundamental physical mechanisms that could have caused the problem.
If two parts that shouldn’t have contact are rubbing against each other during operation, for example, the clearance between the parts must have gone to zero, meaning part A moved, part B moved or both moved.
Which of these three movements is physically possible or likely under the circumstances of the rubbing? More broadly, what hypotheses are compatible with your evidence? To determine this, you must look at the parts—or as Dorian Shainin said, "Talk to the parts."1
Typically, a cause and effect diagram is created by:
- Defining the problem.
- Identifying major categories of causes of the problem (typically method, material, employee, machine, measurement and environment).
- Drawing the diagram.
- Brainstorming potential causes of the problem under the major categories.
- Identifying potential subcauses for each cause.2
However, such a cause and effect analysis is opinion driven and "A root cause analysis (RCA) should be empirical."3 Understanding the physics of the failure, creating a cause and effect diagram and investigating the failure based on this understanding helps achieve the necessary empiricism to correctly identify a root cause.
Suppose a cross-functional team is investigating the cause of repeated broken welds in a welded frame. A typical cause and effect diagram would depict method, material, employee, machine, measurement and environment. Often, all six aspects must be considered when evidence is scant and the root cause is wide open.
When brainstorming to create a cause and effect diagram, the first question should be, "What are the physics of the failure?" In this example, the RCA team moderator or leader should ask the subject matter experts what exactly could explain the evidence.
One possible cause of a bad weld could be the welder’s poor welding technique, but how do you test for poor technique? Training records can be checked and the welder can be observed while welding additional parts, but what if it was just a momentary lapse in proper welding technique? Days of watching the welder’s future performance won’t reveal what exactly happened when the failure occurred.
To explain the evidence, the RCA team must start by explaining the physics of the failure. A poor welding technique may result in poor contact between the material that is being welded, so this should be listed in the cause and effect diagram, as shown in Online Figure 1.
There’s no need to dig deeper into causes of an imperfect contact at this point in the investigation. Possible causes should be listed as subbranches of the diagram, as shown in Online Figure 1, if the team already has identified them, but this is just to ensure the ideas are not forgotten over time. The team should consider method, material, employee, machine, measurement and environment as potential causes, but branches shouldn’t be added for the sake of adding them.
This approach will eliminate many root cause hypotheses that seem reasonable on their own, but in context cannot or are unlikely to explain the problem you have.
- Keki R. Bhote, World Class Quality: Using Design of Experiments to Make It Happen, Amacon, 1991.
- Michael S. Perry, "A Fish(bone) Tale," Quality Progress, November 2006, p. 88.
- Matthew Barsalou, "More Than Just Opinion," Quality Progress, March 2016, pp. 38-43.
Matthew Barsalou is a statistical problem resolution Master Black Belt at BorgWarner Turbo Systems Engineering GmbH in Kirchheimbolanden, Germany. He has a master’s degree in business administration and engineering from Wilhelm Büchner Hochschule in Darmstadt, Germany, and a master’s degree in liberal studies from Fort Hays State University in Hays, KS. Barsalou is an ASQ senior member and holds several certifications.
Robert Perkin is the chief engineer for problem resolution and statistical methods at BorgWarner Turbo Systems. Perkin has master’s degrees in management of technology and engineering management from Washington University in St. Louis. He is a Smarter Solutions-certified lean Six Sigma Master Black Belt.