13 replies to this topic

[Bee5]

I have some data on wastewater that is to become the feed into a wastewater treatment plant. How does one specify the range that the plant is required to treat in a realistic way if the preliminary data is variable. Is it safe to simply use a factor and multiply the mean by this factor, or is it better to use three standard deviations from the mean?

[thorium90]

What do you mean by "the range that the plant is required to treat"?

Perhaps you can explain abit more. Like what factor are you talking about? A mean value of what? What data do you have?

[Bee5]

For example, if I have a mean conductivity value of 1900 mS/m, but I only have 10 data points, where the maximum is 4500 mS/m and the minimum 270 mS/m. The data points are therefore fairly variable.

I would like to specify the range for the technology supplier of the water treatment plant. Do I simply multiply 1900 by an acceptable conservative factor?

[thorium90]

You shouldnt make up your own average numbers. Inform your supplier of the maximum and minimum. What process are they supplying you with? Biological treatments processes? It is always advisable to provide as much data as possible so that they may have enough information to provide you with an adequate process. If you make up some mean value, they might provide with enough just to handle that and your plant will have problems at maximum loads.

[Bee5]

The mean value is the actual mean calculated from the data. My concern is with how to specify the upper limit that the plant should be able to handle. How does one calculate the upper limit based on ten data points?

[markymaark]

Look on a risk assessment.

 

If your current maximum is 2800 mS higher than your average, then what damage would this cause.  What would the final product quality be if the feed had this high conductivity.  Usually I'd say design for the max.  If your 4500 data point is >3 standard deviations away then I would say it's the rare scenario that does not need to be designed for.  However, you need to know what the consequences would be for a feed with that conductivity.

 

If you design for the max, then all lower conductivties will be fine.  If you design for the average, then ~50% of the time you will have an insufficient reduction in conductivity (presumption).

 

Mark

[thorium90]

From what you have posted so far, I infer that your plant doesnt exists yet right? Thats why you have engaged some company to provide you with the technology to perform a certain function, which is to treat wastewater. You have thus gathered 10 samples of the wastewater and based on this data, you want your contractor to build you a new plant? So now you are puzzling how to organise the limited data to do just that?

 

Mark, I also specified the need to include the minimum values and queried if it is a biological process because if the plant is a biological process, it cannot run on too low amounts of wastewater too. The microbes will starve.... Such plants cant be designed for just the maximum, it needs to be designed for the minimum too. Although the OP hasnt quite clarified these. It would appear the OP is only concerned with how to calculate the max and min of data points...

 

Although, it could be industrial wastewater since the OP specified conductivity values and not BOD or COD values. In that case, what kind technology is the supplier going to provide you with?

Edited by thorium90, 13 March 2013 - 10:29 AM.

[Bee5]

@markymaark

Thank you, I was wondering about the designing on three standard deviations from the mean, but I did not know whether this approach is commonly used.

 

@thorium90

Correct, I am puzzling around how to specify the limits to the contractor on some scientific basis, with reasonable margins, and I do not know how to approach it considering my limited data set.

[thorium90]

Bee5, why cant you give them all your data? If you knew enough about wtp, you would be designing it yourself already. So evidently you contracted someone to do it but you are worrying about what data to give the people who know how do design one? Thats odd isnt it?

 

Give them all the data. Discuss with them what you want out of the plant. That will be alot more useful than banging on the statistical calculator

Edited by thorium90, 13 March 2013 - 10:34 AM.

[Bee5]

@thorium90
There are only ten data points in total per parameter. The waste water used to be treated by a third party, and analyses were not necessary. Now we would like to treat and have started analysing. But we have very little data due to the expense associated with lab testing.

We could give them all the data we have, but they would be in the same position as me, with trying to set limits on a small data set.

Edited by Bee5, 13 March 2013 - 10:36 AM.

[Bee5]

@thorium90

You are correct that I could give them the data, but I am interested to know what they would do with it. I would also like to learn about design and conceptual design, and am trying to solve the puzzle myself.

[markymaark]

Another Note. 

 

Bee5, you need to inform them of the data set.  If you give just an average and say "design for this" then if anything goes wrong due to high or low conductivity then only you are liable.  If you inform them then their design needs to accomodate all the information you give them.

Edited by markymaark, 13 March 2013 - 10:44 AM.

[Bee5]

Thank you both for your assistance.

[thorium90]

The contractors will have to make do with what you give or try to gather more data themselves. If you want to know more about the design and conceptual design, you will need to provide alot more data than just conductivity...