Skip to main content

Removing hypotheses for fault-finding in Six Sigma to revolutionise quality management

Six Sigma

Written by Dan Somers (pictured, right) CEO of Warwick Analytics

Dan Somers.jpg

“DMAIC” is the five-step approach that makes up most quality processes such as the Six Sigma tool kit, to help drive costly variation from manufacturing and business processes.  The five steps in DMAIC are Define, Measure, Analyse, Improve, and Control.  “8D” is similar.

Each of the steps is linear meaning that a bottleneck in one step directly impacts the others. If issues are ‘easy’ to resolve (in terms of finding the root cause) then the process can be relatively quick. However with the increasing complexity of new products and processes, combined with the lack of clean and reliable data, the process can be a lengthy or even insolvable one, particularly if the problem is a new one.

Hypothesis testing in this context can mean anything from an educated guess to using one of the well-trusted root cause analysis frameworks such as Fishbone Diagrams and ‘5-Whys’.  The engineer needs to act as a data scientist, setting up hypotheses and using sophisticated statistical tools such as multivariate analysis or logistic regression to try and solve such issues.

But what if there was a way of obviating hypotheses from the Define, Measure and Analyse phases?

This hasn’t been possible until now, but with current advancements in analytics, in particular non statistical techniques and root cause analysis software, the options have now opened up.

Software such as SigmaGuardian from Warwick Analytics is not dependent on hypotheses and analyses any and all of this data from across the entire enterprise and supply chain - although it doesn’t need to be clean or complete. The software identifies the ‘fault region’ corresponding to the root cause of costly faults and provides recommendations based on the most economical fix. Because it is based on information theory and is ‘non-statistical’, it won’t give a wrong answer - as with the case with an incorrect hypothesis or with dirty data - it always reduces the search space of the problem.

Rather than taking months to identify faults, this new breed of analytics can work in quasi-real time, in just a few seconds.

Manufacturers already using advanced analytics software such as SigmaGuardian, in conjunction with their Six Sigma techniques, are showing up to a 75% reduction in the cost of fixing yield issues and a 50% reduction in warranty resolution lead time. If this were rolled out across a manufacturing enterprise it could save between 1.5% and 8% of overall sales.

Notably, the technology from Warwick Analytics was applied at Motorola (i.e. the home of Six Sigma) to support their quality processes. It was used to eliminate two of their most prominent audio and battery No Fault Found quality issues for a particular mobile phone model. 

Previously, even after the Analyse phase and extensive hypothesis tests, the issues were costing Motorola a significant amount in terms of returns, replacements and reputation –the typical costs associated with COPQ.

The results of the SigmaGuardian non hypothesis analysis picked out the key parameters – all within tolerance – that were contriving to cause the root cause of the issue. Furthermore, the ‘fault region’ was also quantified, meaning that it was possible for the manufacturing engineers to easily identify and predict when the failure would occur again. The product was also redesigned in the next generation to improve the yield inside the factory. As a result of using the advanced analytics, the issues were no longer in Motorola’s top 50 warranty failures.

The software can also be used in the final ‘Control’ phase of DMAIC as it can also act as an Early Warning and Prevention system (“EWAP”). This helps to prevent defective products from being produced and predicts maintenance requirements. By identifying fault regions, engineers can work out corrective actions such as remanufacture, redesigning the product or to specify precisely the products in the field which require corrective action, and the lowest fix, without having to recall the entire fleet or batch, or implement an expensive workaround.

Many Six Sigma specialists are already realising the opportunities and advancements they can make by incorporating non hypothesis software into the Analyse phase of their existing techniques and quality processes. Used on a stand-alone basis the software will effectively identify fault regions in manufacturing processes, but used in the right hands as part of a sophisticated Six Sigma toolkit it will positively transform the manufacturing industry forever.

Facebook Conversations



Supply Chain Digital Weekly