Report No FR0134

J C Ellis

Nov 1990



To make an initial assessment of the connections between the various statistical bases for consenting and monitoring sewage effluent quality and the maximum daily mean basis proposed in the draft EC Municipal Waste Water Treatment Directive.


In order to assess the implications of the forthcoming EC Municipal Waste Water Treatment Directive for the UK water industry, it is important to have a mechanism for translating existing percentile-based measures of effluent quality into their equivalents in terms of maximum daily mean quality.


The four final effluent quality data sets provided by Anglian Water have been subjected to a comprehensive year-by-year statistical analysis.

The numerical conclusions presented in the report are of course specific to those four effluents, but the approach developed is generally applicable and has been described in sufficient detail for the method:of analysis to be readily undertaken by others.

Both for SS and for BOD, typical conversion factors have been obtained linking 80%ile, 95%ile and maximum quality to annual mean quality. No overall bias was found between spot-based and composite-based annual mean quality, although appreciable differences could be seen in individual years. More surprising was the finding that the relationship between maximum and mean quality, though showing considerable scatter, vas not discernibly different whether spot sample or composite sample data was used. In other words, maximum daily mean quality seems to be no less 'spiky' than maximum instantaneous quality (at any rate, as seen through the historical sampling window).

An approximate method has been developed for assessing the implications of using 'mini-composite' samples based on five spot samples. This indicates that such a sample could, one time in ten, deviate from the full 24-hour composite result by up to 70%.

An assessment has been made of the SS:BOD ratio. This shows unexpectedly vide variations across the four data sets, with values ranging between 1 and 4 around a typical ratio of 2. In particular, there is little evidence in support of the commonly-assumed ratio of 1.5.


  1. The general statistical approach described and illustrated in the report should be applied to other final effluent quality data sets. This would be an effective way of quickly building up a more substantial base of representative effluent quality performance measures.
  2. Additional sampling over and above the utilities' current spot and composite sampling programmes is needed to gain a better understanding of the short-term behaviour of effluent quality over the 24-hour period. This could most usefully take the form of additional spot sampling at, say, two- or three-hourly intervals on certain selected days.


First, composite sample data was used to develop a link between maximum daily mean and annual mean. A graphical form of representation was used to highlight the fact that the maximum:mean ratio increases according to the number of samples. Next, spot sample data was used to develop links between annual mean quality and 80%ile, 95%ile and maximum quality. (Again, the maximum:mean ratio was recognised as being a function of sample size.) Date-matched composite and spot sample data was then used to estimate the potential bias between annual mean quality as determined (a) by composites and (b) by spot samples.

The end product of this three-stage analysis was a set of factors - for any particular determinand, year and data set - linking all the various statistical measures of quality relevant to the study.

Lastly, two subsidiary (and originally unplanned) tasks were undertaken. The first involved a more detailed study of the within-day variation in effluent quality; the second was to determine the values typically taken by the SS:BOD ratio.

Copies of the Report are available from FWR, price 25.00 less 20% to FWR Members