Analytical software could solve mass product recall problems
A small British company has developed statistical analysis software which it said could uncover minor defects in manufactured products such as cars before they become massive safety problems.
The company, Warwick Analytics, claimed that its software could potentially prevent public relations disasters such as Toyota’s recall of more than 1.5 million cars worldwide over brake and fuel pump defects. It uses computational simulation techniques to study and illustrate the statistics related to the most unlikely combinations that can occur in a manufacturing environment. Root cause analysis can be carried out using this model, without having possession of the problem part, and statistical analysis for the assembly can be accomplished at a rate of several thousand parts per day, as opposed to one item in days or weeks.
Warwick Analytics is a spin-out from the University of Warwick, where its technology was developed at the WMG department in work led by Professor Darek Ceglarek. It uses this technology to provide computational product analysis services to the automotive industry in the area of warranty control and early warning systems related to product recall management.
“Warranty and recall issues have been particularly painful in North America, where annual costs run into billions of dollars,” noted Professor Ceglarek. “The striking thing about recent recalls is the mundane nature of the items recalled - throttle pedal assemblies (Toyota), seat belt latches (GM) and ignition switches (Honda).
“These are all low-cost elementary assemblies manufactured on a ‘build to print’ basis, for commodity level prices, in large volumes, by specialist suppliers. They could not be rated as complex devices by any stretch of the imagination, and yet replacing them under recall conditions has cost the industry vast amounts of money. This is because they all have to deliver a safety critical function with total reliability for the life of the vehicle.
“Under garage service conditions, defects are typically dealt with by changing out the item, with the defective item going back to the company that made the vehicle, maybe for further investigation. On many occasions for a low-value item, it simply gets flung in a bin as scrap and forgotten. The new parts and labour for fitting are charged to the manufacturer, and the whole process is repeated many times.
“The critical dimension that is missing is a thorough root cause analysis for the defective part, carried out by a knowledgeable technical expert. Someone in a white lab coat, if you like. Progress is slow and costly for this person, so if every defective item was to be subject to a thorough root cause analysis, a cast of thousands would be needed to support the work load. A fully professional root cause analysis will generally be only invoked as a mandatory requirement, if there is a fatality or serious accident with a lot of liability at stake.”
He added that problems arise when an assembly, which is right on the edge of specification when manufactured, becomes marginal due to usage and wear, and then unsafe under certain service conditions. He said that the Warwick team has extensively studied random variance in manufacturing at the micro level, and used this to create its computational simulation technique.