Specifying Feed Limits

[Bee5]

I need to specify parameter limits to a water treatment plant supplier, but I only have ten data points. How is this usually approached? Do I double the mean of my data set, or do I specify that the design should be to accommodate the mean plus three standard deviations from the mean?

 

[bimr]

It is customary to set a parameter limit (to the equipment supplier) such that the equipment will meet the limit at all times. The equipment supplier would consider this when he is designing the equipment.

Particular pieces of equipment have typical performance standards as well.

What parameters are you considering and what is the purpose of the equipment.

 

[Bee5]

I have a mean value for conductivity, and a minimum and maximum value, but I only have ten data points. I am specifying conductivity limits to a water treatment technology supplier. I am not confident that one can supply limits based on ten data points. How is this usually done in practice when little data is available?

 

[bimr]

Are you talking about data points for the raw water? What type of water treatment plant are you installing?

 

[Bee5]

Industrial wastewater treatment plant. The effluent needs to be treated, but very little data is available.

 

[thorium1]

What is the purpose of the treatment? To reduce conductivity only? You only have 10 data points for conductivity?

 

[thorium1]

Bee, its common to have limited data. In fact, there can be situations where no data are available, forcing engineers to resort to predictions, common sense, simulations etc. What one does is to give all the info, as much as possible and either make do with it or gather more. That is how it is done in the industry. You are troubling yourself with an unnecessary problem that has a trivial solution. You should be considering things like what type of process to use, the economics etc and not which of the 10 pieces of data to provide to your vendor or whether you should do some statistics on the data first

 

[Latexman]

In God we trust, all others bring data - that's what I say! If you NEED more data, and only you know this, look around for more data. Put out an APB on the data! What does the water permit(s) say? Does the lab that does the testing of the water have some data you don't? If you can't find more data, set up a sampling and testing program, and get more data. What is the time frame covered by these ten data points? Days, weeks, months, years? Are there seasonality changes to worry about? Does the data cover this adequately? No, be sure your sampling and testing program addresses this. Sometimes you just have to wait, and collect enough data to be able to characterize the problem. Ten data points to characterize the feed to a WWTU seems inadequate/risky to me. Obviously, you have the same concern. Also, do you need to provide for an upcoming expansion? Is a new unit coming to the plant soon? What are it's needs? Should you fold these future demands into your project? Ask the project sponsor.



Good luck,
Latexman

Need help writing a question or understanding a reply? forum1529

 

[Bee5]

Thank you Latexman and thorium1. I will stipulate to the suppliers that there is a lack of data, but I will give them the data we have available. If they require more data, we will need to set up a sampling and testing program until enough data is available.

 

[bimr]

For wastewater projects, the effluent guarantee will always be based on the raw wastewater parameters. These parameters are generally specified as an operational range. Wastewater tneds to have highly variable parameters and there is little value in doing a statistical evaluation because in the end, the wastewater treatment system has to achieve the effluent guarantees at all times.

Conductivity is rarely used in wastewater treatment unless one is doing some type of desalination process.

 

[thorium1]

bimr is right on this point. BOD, COD is an important parameter in wastewater treatment, but as the OP is unwilling to share the purpose of the treatment plant, the required effluent parameters as well as the selected treatment types, there is not much to help in that aspect. Damping the fluctuations using appropriately sized equalization tanks will be important, therefore mean values or his 3 standard deviations are not going to be helpful. In fact, ALL the data will be needed to help size the equalization tanks as tanks are integrating systems.