Plumbing Africa January 2019 | Page 35

HEALTH AND SANITATION << Continued from page 31 Mitchel (2007) used a 6-min time step over a 50-year simulation period, and concluded that the selection of either the YAS or YBS operating rule has negligible impact on the estimation of yield and volumetric reliability. There are however other algorithms rather than YAS and YBS; Ghisi (2010) estimated rainwater tank sizing and potential for potable water savings using an algorithm of the Neptune computer program which is neither YAS nor YBS. MASS CURVE METHOD The mass curve methods use a mass balance of the worst drought recorded, thereby assuming that a more severe drought will not occur in the future (Ndiritu et al., 2014). The method further assumes that the tank is initially full at the beginning of the flow record and hence will be full at the start of the critical period (McMahon and Adeyole, 2005). Restriction control curves that are a function of storage content cannot be handled by the method (McMahon and Adeloye, 2005) and no estimate of the expected reliability is provided (Ndiritu et al., 2014). Moreover, the algorithm does not account for storage losses. www.plumbingafrica.co.za PROCEDURES BASED ON STOCHASTIC DATA GENERATION Storage estimates are made based on stochastic or synthetic streamflow data in conjunction with one of the critical period methods (McMahon and Adeyole, 2005). One of the methods based on this procedure is the behavioural analysis method discussed earlier. PROBABILITY MATRIX METHODS The probability matrix method is based on analytical derivation of probability distribution functions of design parameters; probabilistic modelling of the storage process is possible without the need of continuous simulations (Raimondi and Becciu, 2014). The method considers a maximum of two isolated rainfall events rather than the entire time series and it assumes that the tank is full at the end of the first rainfall event (Raimondi and Becciu, 2014). The limitation in this method is that each year of the record is simulated separately, thereby ignoring serial correlation of the hydrological variables involved (Ndiritu et al., 2011a). PA Next month we take an in-depth look at RWH models. November 2018 Volume 24 I Number 9 33