ARTICLES
# Time to replace Shewhart Control charts? Computer algorithms show the way, says Dr. Juergen Ude. (Part 2)

#### Stability Index

#### Commercial Acceptance

#### Resistance

It may seem hard to understand why Manhattan Control was forgotten, why statisticians and consultants have not promoted them. Little has been published on them. Indeed the first publication was one page only. I believe that the reasons are simple. When Manhattan Control was first introduced, personal computers were not available. They are of no interest to trainers and consultants as they require little training. The Shewhart control chart requires up to 3 days training, whereas the Manhattan Control only requires less than half an hour. The power of the tool stems from the computers number crunching ability, not sophisticated new statistics and hence statisticians have little interest in them.

Not having access to the original algorithm, I developed my own, which as it turned out later was different to the original. Having had unprecedented success in applications such as sales analysis, problem solving and process control I introduced Manhattan Control in Qtech International’s first software package, SPC2000. Since then companies such as Alcoa have reported success after success, whereas previously they achieved little with other control charts.

I have run seminars on the technology in many countries, including the USA. The new technology has been found newsworthy in India, a country noted for its statisticians and mathematicians. Everywhere, seminars were greeted by full houses. Practitioners present have agreed that it is time for new technology and that the old technology does not work.

It is worthwhile listing advantages of the Manhattan Chart over the other charts:

- They are robust to non-normality
- They can be applied to attributes data and variables data alike
- Training is a fraction of the time compared to other charts
- They show the onset and duration of a problem
- They show relative changes
- They can be used for fingerprint analysis to show relationships between other variables, where conventional regression analysis fails
- The technology is conducive to further technology advances

Figure 1: A typical Shewhart Control Chart

Figure 2: A typical Manhattan Control Chart

We have already introduced the D-Chart, as shown in figure 3, which using manhattan technology can show changes in the mean, dispersion, mean and conformance all on the same control chart.

Figure 3: A typical D Chart

We have introduced the stability index, using Manhattan technology, which I believe is the most powerful new statistic since the Cp, Cpk index. The Cp and Cpk index provide information on process capability in a concise way. Management does not have time to scan hundreds of capability histograms. Similarly management cannot look at hundreds of control charts. Management needs a convenient index.

Historically the number of out of control points was used to summarise the degree of control. However this statistic provides little useful information. Instead the stability index not only advises whether the process is out of control, but also the effect of the out-of-control situation. An index of 100 implies that the process is 100 percent in control, whereas an index of 8 means that 20 percent of variation is due to the process being out of control. Immediately management can tell how significant an out of control status is.

Management can now prioritise process improvement changes by relating the stability index to the Cp and Cpk index.

Manhattan control is being accepted by many of the worlds largest corporations. More and more organisations are embracing the technology. Naturally there is resistance. The most noteworthy resistance is from industrial statisticians. I understand that just like we have challenged Shewhart Charts for effectiveness, there will be those that will challenge the Manhattan Control chart.

I myself am continuously searching for problems and disadvantages, so that the technology can be improved. Currently the only significant disadvantage is that Manhattan technology requires a computer. Since we are in the computer information technology era, I even doubt whether this is a disadvantage.

With regards to resistance I will make a few comments. Contribution is good, resistance is bad. It is a positive attitude to point out deficiencies whilst at the same time suggesting a better alternative. It is negative to discredit new technology, on the basis of minor theoretical problems whilst on a practical level the technology is better and works. It is negative to make assertions about a technology without understanding it.

Several consultants and statisticians have commented that the ARL for an in-control process is inferior, without having access to the technology to establish this. Statisticians in Australia, China, Singapore and the USA have alleged that Manhattan Control charts are a disguise form of the Shewhart Chart. Nothing could be further from the truth. One world famous statistician has expressed scepticism on the basis that it uses Cusum theory. The original algorithm used the Cusum time series, but not Cusum theory per-se. Our algorithm which is more efficient does not even use the Cusum time series. On the other hand many statisticians from major universities throughout the world have supported the technology.

Some have been critical of the fact that our algorithm sits in a black box. Since we have developed our algorithm it is only fair that only we have access to it. However, we have not taken credit for the concept, we have referred competition to the same technology, so that they can copy what we have done. They too can develop their own improved algorithm. We provide access to simulation programs to test the algorithm, and we provide our customer with a choice.

Our customers can use old and new technology and use what they feel suits them. Ultimately it is not our word that counts. but that of our customers.

It is important to understand that the technology offers no statistical breakthrough, and that it is reliant on a computer algorithm. The time has come where new breakthroughs for quality technology will not be through advances in statistical theory but advances in computer algorithms. Computer algorithms can make things possible that statistical theory alone cannot. Indeed, much of today’s statistical advances are impractical in the real-world. Software technology is a means of converting theory into practical solutions.

I will continue to promote the technology in the spirit of continuous improvement and continue to improve the technology and welcome suggestions for improvements. I will continue to encourage that tools for the quality profession should be modernised instead of being in the situation where 100 year old technology is encouraged.

Improve product quality through smart data insights using new-age, machine-powered driven analytics, available to download through a suite of Apps!

- Apps for SPC, MSA, Process Performance, Inferences, Visualization, and much more
- No licencing or subscriptions! Pay ONLY per analysis billed monthly - don't use, don't pay!
- Always up-to-date with the latest features

TRY FIRST WITH A 30 DAYS FREE TRIAL!