Hi all,

My company have recently purchased a steel tube washing machine. This machine flushes a water/chemical mix through straight cut tube to remove cutting and deburring swarf and basically brings the tubes up to a certain cleanliness standard before further processing.

The machine holds a constant cleaning fluid level of 1500 liters, 2% of which are the above mentioned chemical, and 98% water (Ideally).
The chemical supplier has recommended to monitor the performance of the chemical in order to determine when it needs to be brought back up to strength, and when it has to be changed out completely.

Here's my problem...

How do I statistically determine when the chemical stops working???

I am collecting the following data:

1. Chemical strength , by titrating and toping up the tank to keep the ratio at 2%
2. Tube contamination after cleaning operation through gravimetric test. Since samples vary in length and diameter, I normalise samples by recording contamination weight per square meter.
3. Chemical contamination, by filtering and weighing a 50ml fluid sample.

I initially thought that when my contamination figures on both, tube samples, and chemical rise, the chemical has reached the end of its performance life.
However, I have found that varying tube diameters give different cleaning results (for example a short tube with large diameter comes cleaner than a long tube with small diameter eventhough both are the same square meter surface area.
Cleaning pressure and temperature stay the same and do not vary.

How do I take those additional variables into consideration and how can graphically display and monitor my findings?

hope someone can help.

cheers
Floyd

Floyd,

I would recommend analyzing your data by performing a regression analysis using chemical contamination, tube diameter and tube length as x-variables and tube contamination as your response or y-variable.

Y (tube contamination) = function (chemical contamination, length, diameter)

A long tube with a small diameter probably does not get as good a flow of liquid through it to rinse and leaves more deposits than the short tube with a large diameter. This may mean that when the chemical contamination reaches one level it can no longer be used for the long, small diameter tubes, but may continue to be used for the short, large tubes until a second higher level is reached.

If this is not feasible, the lower level would be the useful life.

Get a model from the regression analysis. Set Y to the max tube allowable contamination, the longest tube, the smallest diameter, then calculate the resulting chemical contamination for the useful life.

Floyd,

I would recommend analyzing your data by performing a regression analysis using chemical contamination, tube diameter and tube length as x-variables and tube contamination as your response or y-variable.

Y (tube contamination) = function (chemical contamination, length, diameter)

A long tube with a small diameter probably does not get as good a flow of liquid through it to rinse and leaves more deposits than the short tube with a large diameter. This may mean that when the chemical contamination reaches one level it can no longer be used for the long, small diameter tubes, but may continue to be used for the short, large tubes until a second higher level is reached.

If this is not feasible, the lower level would be the useful life.

Get a model from the regression analysis. Set Y to the max tube allowable contamination, the longest tube, the smallest diameter, then calculate the resulting chemical contamination for the useful life.

There are other variables to consider:

- temperature of the wash fluid
- flow of the wash fluid
- washing time
- drying time and air flow during drying
- contamination level before washing
- tube design (straight, bending, angles, fittings)

Floyd,

I would recommend analyzing your data by performing a regression analysis using chemical contamination, tube diameter and tube length as x-variables and tube contamination as your response or y-variable.

Y (tube contamination) = function (chemical contamination, length, diameter)

A long tube with a small diameter probably does not get as good a flow of liquid through it to rinse and leaves more deposits than the short tube with a large diameter. This may mean that when the chemical contamination reaches one level it can no longer be used for the long, small diameter tubes, but may continue to be used for the short, large tubes until a second higher level is reached.

If this is not feasible, the lower level would be the useful life.

Get a model from the regression analysis. Set Y to the max tube allowable contamination, the longest tube, the smallest diameter, then calculate the resulting chemical contamination for the useful life.

Hello Miner,

Thanks for your help. I've been looking into regression analysis and it seems to be the way to go. I don't fully understand the process yet, but I'm sure I'll get there:).

Can I use this method to graphicaly monitor the development, or is it purely for analyzing collected data?

What do you mean by "get a model from the regression analysis"?

Thanks
Floyd

Can I use this method to graphicaly monitor the development, or is it purely for analyzing collected data?

What do you mean by "get a model from the regression analysis"?

A regression model is an equation such as Y = C + B1X1 + B2X2, where Y is the response (tube contamination), C is a constant, B1 and B2 are coefficients and X1 and X2 are factors (such as chemical contamination or tube diameter).

If your regression model is simple such as Y = C + BX, you could create a plot with the regression line and confidence limits, then plot chemical contamination points as you test. If your model is more complex such as having two independent variables (X1, X2), you will not be able to easily graph it unless you do something like creating different plots for different levels of one variable (such as X2).