Coronavirus (Covid-19): Latest updates and information
Skip to main content Skip to navigation

About the Rainwater Tank Performance Calculator

The basic tank programme models a DRWH system over 10 years using daily steps. A daily step is used because longer steps (such as 1 week or 1 month) give misleading results whenever a small tank is used. Each day (any) roof run-off water is assumed to be added to the quantity of water in the tank and the user’s demand is subtracted. It is assumed that 85% of rainfall hitting a roof reaches the downpipe to a RWH tank. The tank is checked daily for overflow and for becoming empty. Average performance measures (reliability, satisfaction and efficiency) are computed for the 10-year period.

Thus to drive the tank programme requires rainfall data for 3650 days. As few enquirers have such data, and as transferring such big data sets would be inconvenient for users of this service, we have chosen to use ‘pseudo daily data’ instead of ‘actual daily data’. Thus for each month we use the relevant monthly total rainfall to generate a pseudo daily sequence. We employ an algorithm that gives a daily sequence with similar statistics to actual rainfall records. This we have tested both by comparing amplitude distributions and by comparing the real rainfall and pseudo rainfall sequences via the tank performance programme. Provided the real and pseudo daily sequences have the same monthly totals, the measures of reliability etc. agree within about 2%.

We therefore ask for 10 years of monthly data (i.e. for 120 values). If we had chosen more than 10 years, it would have made difficulties for many potential users of our computation service. Moreover in meteorology (where climate is always drifting as well as fluctuating) using a longer record gives better averaging but also makes such averages more ‘out of date’. Since we are here using past data to predict future performance, we want it to be as recent as possible. Unless only very extreme events are being forecast, 7 to 10 years is recommended as the best record length to use for future predictions.

Unfortunately many users don’t have access even to ‘actual monthly’ rainfall data. In such cases we must use their ‘mean monthly’ data, such as appears in for example atlases and national yearbooks, and treat every year is if it were an average year. This procedure will slightly (e.g. 2%) over-estimate such performance measures as ‘reliability’.

Within the programme as it now stands, the 10-year period is started with the tank half full. Under the two adaptive demand strategies the daily demand D is varied about the nominal demand DNOM according to the following rules:

Adapt to tank content

Tank content

Daily demand

>2/3 full

1.2 x DNOM

1/3 to 2/3 full

1.0 x DNOM

< 1/3 full

0.8 x DNOM

Adapt to rainfall

Rain in last 14 days?

Daily demand D


1.2 x DNOM


0.8 x DNOM

Components of the programmes underlying this computation service were developed by Terry Thomas during 1999-2001, in part fulfilment of contracts placed on Warwick University by first the European Union’s INCODEV programme and then by the (UK) Department of International Development’s KAR programme. The support of both of these organisations is therefore gratefully acknowledged. A decision was made that it would be more useful to most RWH practitioners to offer a design service rather than to disseminate a complex and large programme for them to mount on their own computers. The components were therefore turned into an on-line service by a Warwick student, Claire Brown, as her 2001-2 “3rd year Project” under a degree programme in Engineering Design and Appropriate Technology’.

Input Data