Home > Water Mode > Theory > Variable Peak Factors
It is a known fact that peak factors are higher when demands are lower. When designing internal water supply consultants implement some of the empirical equations to cater for that. But, when designing water reticulations, they tend to ignore this fact, or, at the most, they implement higher overall factor to entire network when the network is rather small. This does not fully address the problem, as edges of even the largest networks have very high peak factors.
Reason for ignoring a problem is assumption that looped network cannot be analysed employing variable peak factors, and water reticulations are generally looped.
That assumption is in fact false, looped network can be analysed with variable peak factors if we employ probability theory and repeat analysis sufficient number of times. In each iteration, node demand will be calculated from probability theory, using Gumbel distribution. Therefore, in each iteration node demand will be different, but at least once during the analysis it will reach maximum as determined by specified confidence factor.
Result of analysis is summary envelope of all executed iterations, that is maximum node discharges and minimum pressures. Maximum pipe flows will also, by necessity, follow Gumbel distributions with specified confidence factor.
Three peak demand curves are built in and users can create additional curves.
Press Curve button to view or append curves
Select confidence factor as required.
Notes on confidence factor:
Default confidence factor is 99.9% and we recommend using it, unless there are strong reasons to alter.
Lower factor, 99% should be used only when occasional disruption or insufficient pressure could be tolerated. Some would say when developing rural or low-cost solutions, but I am strongly against low-income / high-income separation in design criteria. All people have equal constitutional rights, including rights to proper water supply. Or, if you want pragmatic approach, look at the number of riots caused by poor service delivery.
99.99% is probably too high and should be used in research domain, not for design.
Unit on horizontal axis is shown as kl/day, which is the same as Equivalent Erven used in SA. I prefer using something that is possible to understand internationally.
Selected curve is shown with the thick line.
Change confidence level to see its influence on curves.
Click Create if you want to create new curve.
Dashed line is a curve proposed in Guidelines for human settlements, and is only for display purposes. There are reasons for not using it:
Curve is arbitrary, and there is no possible statistic or theoretical method which would create a curve (or rather two straight lines) of that shape. Two points on the curve could be correct, see selected curve touching it at 2 and 2000 EE, but shape is entirely wrong. There is no explanation how it was created, but I suppose values where assessed at two points and connected with straight lines.
It does not provide for confidence factor, nor any explanation regarding confidence factor is given.
Numerous articles comparing measured peak factors and this curve point to large discrepancies, particularly in mid-range where straight line is more distant from statistical curves. I can provide a list of articles if needed.
To create Peak demand curves, we use method proposed by Zhang in 2005. Particular advantage of this method is that indoor water demand can be separated from outdoor demand, or rather garden watering.
Curve name must be unique.
Description should point to important parameters, and is displayed at analysis window as a reminder.
Base peak factor is peak factor for unlimited number of users, far to the right on the curve (minimal peak factor). Please note that it applies only to indoor use, outdoor use has its own rules.
Specify percentage of water used for garden. Zero is acceptable.
Watering in some parts of the country is highly seasonal, and number of moths where watering is used significantly influence peak factor.
Press refresh to update summary table showing peak factors for selected consumption.
Change confidence factor to see how it influence peak factors.