Wadiso 6 User Guide > Tank Cost Optimization to be included in future Wadiso version > Storage Tank Optimization Technique and General Concepts > Critical Period Tank Sizing Technique > Accuracy of the Critical Period Technique

The critical period technique included into Wadiso has been tested extensively. It was used to estimate the tank balancing volumes in a large number of hypothetical water distribution systems. Four distinctly different sets of consumption information were used. The estimated tank balancing volumes were compared to the actual volumes as computed by time simulation. Very good comparisons were obtained. The average "error" made by the critical period technique was +1,5%, with a standard deviation of 5%. In certain multi-tank systems, however, errors for individual tank balancing volumes of up to 20% were recorded. The large errors in the testing of the method were recorded when rather unrealistic tank locations and elevations were used. The range and standard deviation of the errors generally decreased when more realistic networks were considered. For simple type networks such as the one in the Simple Pipe Network Layout, the method gives exact results. In this paragraph the errors made by the critical period technique are put into perspective.

The purpose of the critical period technique can be seen as an efficient estimator of storage volumes (and costs) which is used in an optimization procedure. The issue at stake is the trade-off between pipe costs and storage costs. Seen in this light, accurate estimation of the precise storage cost is not of prime importance. A more important requirement of the technique would be to describe accurately the gradient of storage costs with respect to a change in pipe costs. This gradient was always maintained by the technique, even in the cases where it made the relatively large errors in the estimation of tank size.

Uncertainties in the input parameters for pipe network analysis are considerable. Consumption characteristics such as average demand rate, peak day factor and the 24-hour demand pattern are at best only reasonable estimates of the expected situation. Pipe characteristics (e.g. roughness), pump curves and the distribution of consumption among the nodes are further examples of uncertainties which may be encountered. Seen in the light of these uncertainties in the system parameters, and the purpose for which it is to be used, the errors made by the critical period technique can be regarded as of limited significance.

Although the method is accurate enough to be included in an optimization procedure, it should not be used for final design purposes. The result of an optimization procedure employing the critical period technique would be an optimal combination of pipe sizes and estimated storage tank volumes. These volumes should be fine-tuned by the time simulation technique for final design volumes.