Mulbury Six Sigma

Mulbury Six Sigma

Six Sigma · Lean · TRIZ

Six Sigma Calculator V3.0 Download a FREE e-book containing this calculator and basic DPMO tables

Version 3 (August 2002) - includes shift and short-long term data collection

Measure a sample of events (e.g. customer applications) over a period of time. For each event measure the Critical To Quality factor (e.g. time taken to process application), and find out how many fail the customer requirements (e.g. take too long).

You will need

1. total possible number of events (i.e. the whole population)
2. time over which these events were measured
3. number of events sampled out of the whole population
4. number of defects - count two defects on one event twice
5. number of possible defects per event (usually one)

Note that the above calculation is only approximate, and is not accurate above 6 sigma. Negative values for sigma are meaningless, and are shown as zero. This calculation includes by default the standard 1.5 sigma shift for short-term sigma.

Form designed and coded by G Tennant, Mulbury Consulting, February 2000. The calculation uses an approximation for the inverse normal cumulative distribution, for which I take no personal credit. If you have found this calculator useful, please let us know!

You may use this calculator FOR PERSONAL AND NOT COMMERCIAL BENEFIT. For use off-line please download the Sigma Calculator e-book rather than saving this page!

Sample a subset of the entire population, preferably randomly selected and fully representative of the whole set of all possible events.

Population the entire set from which the sample has been taken. Theoretically the population is all entities that have, are, or will be produced, however we often take the population to be all entities from which the current sample has been drawn.

The period of the sample is used here to calculate the annualized defect rate and cost, both of which are useful in evaluating the customer and business experience of any defective process.

Defects are any critical characteristic of the sampled entity that fails to meet customer expectations.

Opportunities are critical characteristics of the sampled entity that could either meet or fail customer expectation. In many manufacturing situations there may be multiple opportunities per entity, whereas transactional situations often either meet the customer's requirements or they don't. When evaluating potential opportunities (and defects) it is important to work at the customer-significant functional level. Defect opportunities must be critical  to the customer, be independent of each other, and only increase numerically with increased complexity. For example, in a book, defects could be misspelled words, grammatically incorrect sentences, missing text or pages and so on, however readers may regard the 'sentence' as the functional entity of interest, and a defect is a spelling or grammatical error in each sentence.

Cost per defect is an easy way to work with Cost Of Poor Quality. Evaluate the sum total of all costs from the current defect rate, and divide out to obtain a unit cost per defect. This allows for easy calculation of the improvement, and enforces the Six Sigma concept of customer defect reduction as the end goal!

Shift is a complex subject in Six Sigma, and traditionally includes a value of 1.5 sigma short to long term shift in the calculations. The shift is the degradation experienced between the short-term best process capability and the long-term process performance when all possible process time-related causes of variation have been added in. From work undertaken at Motorola an empirical figure of 1.5 sigma is now taken as 'standard' for the worst case shift, although some today now advocate that this figure could be lower.

In theory:  Process Sigma (short term) = 1.5 + Process Sigma (long term)

Short term - is the process capability without any time related variation, for example one shift, one operator, one material batch, one machine setting and so on. This is often taken as the best capability figure, and by convention the Process Sigma value of every process is stated as the best short-term value.

Long term - is the process capability with every possible time related variation included, for example every shift, every operator, every material batch and supplier, every machine setting and so on. This is often taken as the worst capability figure, and by convention the short-term Process Sigma is related to the DPMO at the long-term state, typically shifted by 1.5 sigma from the short term state.

If your data is short term, then the DPMO figure can be related back to (short term) process sigma without any shift, however the customer will experience the DPMO defect rate for the long term process state.

If your data is long term, then the DPMO figure must have the 1.5 sigma shift added in to return to the short term Process Sigma value. Transactional process data is typically long term.

How this calculator works:

The sample and population sizes are used to extrapolate the total defects in the population. The population and defect opportunity are used to calculate the total number of opportunities. DPMO is then calculated using

DPMO = (total defects * 1,000,000) / total opportunities

The DPMO is then converted to the number of standard deviations for the equivalent right-hand tail fraction of the normal distribution. This number of standard deviations is the base process sigma value.

The base process sigma is then adjusted to obtain the long term process sigma value - if the data is long term then no adjustment is needed, however for short term data the assumed short-long term shift is added in. This is 1.5 sigma by default, but can be adjusted to any value of choice. Intermediate states between short and long term apply this shift in proportion.

The (official) short term process sigma value is then calculated by adding back a standard 1.5 sigma shift.

For example - our data is very short term, we anticipate a maximum 1 sigma shift between short and long term, and the DPMO works out to 6210. This is equivalent to the right tail of a normal distribution at 2.5 sigma (short term). We subtract 1 sigma to get our anticipated 1.5 sigma (long term) [we expect that over the longest term, the customer would experience 66800 DPMO], and then add back the standard 1.5 sigma to arrive at the (official) Process Sigma value of 3.0

Where the process shift is not known the default 1.5 shift should be used.

Since 6 sigma = 3.4 DPMO as standard, the 1.5 correction factor will always return any long term DPMO back to an 'official' short term Process Sigma Value

Mulbury Six Sigma

Publisher and reseller of Six Sigma books

© 2001-2008 G Tennant Mulbury Six Sigma