Using Simulation at Andell Associates

Jonathon Andell is President of Andell Associates, an independent consulting firm specialising in the technical, organisational, and interpersonal aspects of modern Quality Management.

One of the earliest certified Six Sigma Black Belts, Andell has written extensively on statistical methodologies, quality management, and the business ramifications of Six Sigma. For Andell Associates, Crystal Ball is a significant weapon in their substantial arsenal against process and product variation.

A Crystal Ball user since 1993, Andell related an early application of the software. 'I was working for an electronics assembly house that used a robot to mount integrated circuit (IC) devices onto printed wiring boards (PWBs). Each device lead had to penetrate a mating hole in the PWB. Andell and his colleagues were struggling to establish specification limits for component geometries and for placement equipment. Up to that time, the complex twodimensional geometries made it difficult to predict how variation would propagate. For example, placement machines had target Xaxis and Y-axis locations, along with rotational orientations (called the ? axis).' Andell and his colleagues could mo del how a single combination of X, Y, and ? errors would impact whether a pin could penetrate its mating hole, but they lacked the ability to know how distributions of X, Y, and q would stack up. 

Andell and his colleagues addressed this uncertainty by developing an Excel spreadsheet model that computed the clearance(the output, or dependent variable) based on part dimensions, pin diameters, andhole diameters, as well as placement errors in two horizontal axes (X and Y) and one rotational one (?). Collectively these inputs were called the independent variables. The analysis consisted of using statistical Design of Experiments (DoE) to dictate settings for the mean and standard deviation of each independent variable. 

Once the distributions of the independent variables had been defined, the team used Crystal Ball to simulate 1,000 trials for each experimental run, which in turn yielded the probability of all leads penetrating their mating holes on the first attempt. The higher the probability, the more closely they could approach the Six Sigma objective. 

As a result of the simulation, they were able to establish tolerances for component geometries and for placement capabilities. 'For the first time,' Andell explained, 'these tolerances were based on objective data, rather than the educated guesses that had preceded this approach. The upshot was that we could relax some tolerances, resulting in cost savings. Other tolerances had to be tightened, but we had hard evidence to drive home the need with those vendors.' 

'Crystal Ball didn't actually solve any problems,' Andell noted. 'However, it gave us the knowledge to sharpen our focus in crucial ways. Our resources were applied where we knew there would be rewards. The bottom line: higher quality and lower costs. This problem might have been solved without Crystal Ball, but not nearly as fast!'  

For more information, please contact: 

Kevin Weiner
1.800.289.2550
303.534.1515
info@crystalball.com www.crystalball.com

Jonathon Andell, President
4610 East Desert View Drive
Phoenix, AZ 85044-1148
480.893.9004
jandell@hotmail.com