Abstract
An amount of water leakage in water supply of large city reaches to 17-20% of total supply quantity, and it becomes one of the most serious social problems. A region having much leakage quantity should be sought beforehand in order to use effectively an old listening noize method (i.e. to listen to the leakage noize on ground by specialists), and it is necessary to throw a few specialists and equipments into this region. A distribution pipeline network is divided into appropriate blocks, and then used quantity of water in each block is estimated by models. A new method is proposed, which estimates leakage quantity by subtracting this estimated used quantity of water from inflow measured quantity into the block.
The feature of the models are described as follows;
(1) The state of faucets used in each block is formulated as a queuing problem, which is expressed as a Poisson arrival exponential service system by considering number of household (=number of faucets used at midnight) as number of server and number of people in the block as number of customer. Under this assumption, a model estimating vacancy time at midnight (i.e. time of all faucets being closed) is developed by generating quasi-random number.
(2) Futhermore, a model is developed to estimate used quantity of water by considering the pattern of used water quantity in addition to the above model.
The vacancy time estimation modes is verified to be practically effective in experimental blocks in a large city. The water usage quantity estimation model will be expected to be used practically in the near future.
