Case Studies, wastewater treatment

Part II: What Constitutes Consent Failure?

This is the continuation of the Analysis of the Risk of Consent Failure of UK Wastewater Treatment Plants series from Alan Bland. I hope you enjoy:

In the early 1990s, when this project was undertaken, the established sewage treatment works discharge consents were those originally issued under the 1963 Water Resources Act, referred to as WRA consents. For very small discharges the consent would be “descriptive”, for example requiring that the visual impact of the discharge on the environment be negligible, but for larger discharges the consent would be numeric. Numeric consents covered BOD5, SS and sometimes NH4. In catchments containing industrial premises, the consent determinands might also include heavy metals (Pb, Zn, Cu, Hg etc.). In all cases, these WRA numeric consents were framed to require 95% compliance with the consent value, which in many cases was the so-called “Royal Commission Standard” of 30mg/l SS and 20mg/l BOD.

Compliance sampling was carried out initially by the NRA and later by the EA. In principle, the dates chosen for sampling were random, to avoid any systematic bias, but in practice weekends and public holidays were avoided. The number of samples taken per year varied according to the size of the discharge. This created a statistical problem. How many samples could be allowed to fail the consent value before the discharge was found to be non-compliant? At first inspection the answer might be thought to be obvious; 95% compliance means only 5 out of every 100 samples should be allowed to fail, so if 20 samples are taken then only 1 sample failure should be allowed. But this frequency of sampling at thousands of discharges across the UK would be very expensive and, furthermore, a single sample failure could be unrepresentative. The consent is a legal requirement; if non-compliance was contested in court, could it be proven that a single failure showed the discharge was non-compliant? Generalising this, there are four cases to consider:

  1. The discharge is actually compliant, but the sampling indicates non-compliance
  2. The discharge is compliant and the sampling also indicates compliance
  3. The discharge is non-compliant, but the sampling indicates compliance
  4. The discharge is non-compliant and the sampling indicates non-compliance

What was needed was a method of analysing samples that reliably identifies case 4 but ignores the other three cases. This gives the benefit of any doubt to the discharger, whose discharge may actually be non-compliant, as in case 3. To meet the need, a device called the Look Up Table (or LUT) was introduced.

The LUT utilises the Binomial probability distribution, which deals with the likelihood of successes and failures. The probability distribution formula is b(x: n,P) = nCx*Px*(1-P)n-x , where P = probability of success (which in this context is a failed sample), x is the number of successes (failed samples), n is the number of trials (in this context the number of samples per year), nCx is the number of combinations of n samples taken x at a time and b is the probability of achieving exactly x successes (failed samples).  For our purposes we need the cumulative probability distribution, which adds together the probabilities of achieving up to and including x successes.

Using this distribution, the number of samples required to demonstrate non-compliance is indicated in the LUT below:

Number of Samples Taken Per Year         Number of Failed Samples Allowed

4 to 7                                                                      1

8 to 16                                                                      2

17 to 28                                                                      3

29 to 40                                                                      4

41 to 53                                                                      5     (and so  on)

These numbers are selected from the cumulative Binomial probability distribution such that it is 95% probable that, for example, more than 4 sample failures in 35 samples taken in a year would indicate the true level of compliance with the consent is less than 95%. Note that the LUT even caters for very small numbers of samples taken in the year (between 4 and 7 samples), but still provides assurance that more than one failed sample could confidently be declared as indicating non-compliance of the discharge.

The next posting will examine the problems the LUT presents for a water company trying to achieve compliance within the limitations of its fixed assets.



About noahmorgenstern

Entrepreneurial Warlock, mCouponing evangelist, NFC Rabbi, Innovation and Business Intelligence Imam, Secular World Shaker, and General All Around Good Guy


4 thoughts on “Part II: What Constitutes Consent Failure?

  1. For the benefits of those non-UK readers, would you mind spelling out the names of the organizations referenced the 1st time they’re used? I’ve been trying to compare how the UK compliance processes compare to US processes but finding it hard because I don’t know the acronyms. Thanks! Laura from DC

    Posted by Laura | January 31, 2012, 3:42 pm
    • Laura, I did provide the full titles in Part 1 of the blog, which also gives some explanation of the evolution of UK water and wastewater management. The abbreviation NRA stands for National Rivers Authority, which was merged with HM Pollution Inspectorate (HMIP) in 1995 to form the current Environment Agency (EA), which I guess is a similar body to the EPA, with similar responsibilities to monitor and control pollution of the air, land, and aquatic environment. For further information see the EA web site

      Posted by Alan Bland | February 1, 2012, 1:22 pm


  1. Pingback: Online Statistics Course - Variance of a Population - January 31, 2012

  2. Pingback: Part 3: How can consent failure be predicted before it happens? « Water Quality and Security - February 13, 2012

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: