Fuzzy thinking makes management clearer

Posted on July 25, 2012

What’s the best way to manage a multipurpose reservoir?  An established method relies on minimising the deficit between realised and target levels for water storage and the release of water from the dam.  Recent years have seen growing interest in the possibility of improving on this approach by taking into account the fact that water management objectives are inherently ill-defined.  As a result, a new generation of “fuzzy” reservoir control models has emerged.  However, there have been few studies that compare the performance of standard optimisation models with their fuzzy alternatives.  Indian engineers have carried out such a comparison by focussing on the Hirakud reservoir system in the Mahanadi river basin.  In the case of Hirakud the main management objectives are to keep the reservoir as empty as possible during the flood season, satisfy the irrigation demand at different times of year and generate sufficient hydropower in each time period.  Opinions of experts and reservoir managers were used to describe the extent to which different levels of storage and release satisfied the three objectives, and linguistic variables were converted to numeric variables by assigning a value of 1 to the highest degree of satisfaction and 0 to the lowest.  Standard and fuzzy models were developed.  With respect to irrigation and hydropower, the results of simulations with the standard and fuzzy models were similar.  They showed that in most cases satisfaction was high because the difference between target and achievable levels was low.  However, in terms of the flood control objective the outputs from the two models were significantly different.  For four of 13 simulated seasonal time periods, levels of satisfaction were very low (4-9%) with the standard model but high (38-99%) with the fuzzy model, presumably because the flood control objective was defined explicitly only by the fuzzy model.  Overall satisfaction, defined as the product of the satisfaction levels for the three objectives, was higher for the fuzzy model (67%) than the standard model (47%).  While fuzzy models use imprecise language, they can be very descriptive and mirror the operations of human decision-making.  Their ability to utilise subjective information as well as precisely measured data makes them suitable for managing reservoirs and other  multipurpose systems.

Reference:  Chandramouli, S. & Nanduri, U.V.  2011.  Comparison of stochastic and fuzzy dynamic programming models for the operation of a multipurpose reservoir.  Water and Environment Journal 25, 547–554.  http://onlinelibrary.wiley.com/doi/10.1111/j.1747-6593.2011.00255.x/pdf


Posted in: flooding, hydrology, lakes