Case Study - Process Capability - Oxygen in Package

This is an example of how statistical and problem solving techniques has been used to address process issues.

The content of oxygen in a packaged liquid can be important as if present it can oxidise organic material give rise to off flavours, much in the same way that bread if left out goes stale. This can have a significant impact both on sales and shelf life.

In this example a packaging company identified that it had problems with packaged oxygen on a new bottled product that was sensitive to oxygen. The desired specification measured as total in package oxygen (or TIPO) is <0.8pp desired and <1.5ppm operational).

Testing of eighty results taken at random over 11 days showed a skewed distribution to the right with a mean of 1.2 ppm and a 95% confidence level between 1.07ppm and 1.32ppm.
However there were results as high as 6 ppm - above the operational limit.
The filling machine has 60 filling heads and testing by head showed no particular correlation with TIPO (i.e. the effect appeared random by head).

The filling process was observed. The process works by a rinsed new bottle entering the machine and first is evacuated by vacuum and then counter pressured with CO2. Once the presure is equalised with the filling machine, product is allowed to run down the outside of the filling tube in the neck of the bottle and is deflected by a rubber sleeve so that it runs down the inner wall of the bottle. CO2 and any residual air goes back up the inside of the filling tube until flow is cut off when the product reaches the bottom of the filling tube and blocks it. After that a jet of high pressure hot water is injected into the neck to make the beer foam and expel any oxygen there. Finally a closure - a crown cork or a twist cap is applied to seal the bottle

The critical process outputs were:

Clean bottle
No foreign objects
No liquid in bottle prior to filling
Minimum (zero) oxygen in finished bottle.
Filled to correct level.
Beer of correct quality (e.g. Colour, gas, alcohol, micobiologally)
Crown crimp tight enough to seal but not so tight as to damage neck and cause splinters.

The process inputs to achieve low packaged oxygen were laid out in the cause and effect diagram to the right

Next the filling process was observed closely (from behind a transparent protective screen) and the following point were noted:-

Some of the bottles entering the machine still contained small quantities (2-3ml) of water from the bottle rinser. When this was sucked into the vacuum system it was causing sporadic freezing and lack of vacuum. The fix was to adjust the rinser so the bottles had time to drain out properly.
The evacuation system was a single evacuation system. Good practice is a double evacuation system where the bottle is evacuated, flooded with CO2, evacuated again, and then pressurised with CO2. A double evacuation system was fitted to the machine.
The filling tubes were incorrect with internal bore that was too large - allowing the bottles to fill quickly and with turbulance. The spreader rubber on the outside of the tube was also in the wrong place - high up in the bottle neck - and acted as a choke, giving rise to further turbulence.
The tubes were changed for tubes of the correct design for the new bottle.
The sealing rubbers that seal the bottle to the filling machine were worn, allowing bottles to brim fill on some heads when the machine stopped.
The sealing rubbers were replace with new.

After these changes another 80 samples were taken - this time over 5 days.
This showed a mean of 0.70ppm with little skew and with a 95% confidence interval of 0.65ppm to 0.76ppm. The maximum result was 1.45ppm.

This brought performance in line with specification measured as total in package oxygen (or TIPO) is <0.8pp desired and <1.5ppm operational).