Abstract
This paper presents an industrial application of mixed-integer nonlinear programming (MINLP) models for the optimisation of a water network, which was initiated by an integral need for cleaner production within a brewery. Several mathematical models were developed in order to reduce the use of freshwater, whilst considering the specific requirements of each particular production section. These models are based on the design method developed by Kim and Smith [1]. The original formulation is modified to enable efficient integration of discontinuous and semi-continuous water-using processes in the packaging area. Semi-continuous processes are treated as water sources of limited capacity. The option of installing storage tanks for semi-continuous water streams is included in the model, in order to re-use these streams during the shutdown periods of semi-continuous operations. The original model is additionally extended with the option of installing local (on-site) wastewater treatment units operating either in batch or semi-continuous modes. This enables the analysing of opportunities for regeneration re-use within the production area, i.e. the brewhouse with a cellar, because of high contaminant concentrations at these sites. The scheduling of batch wastewater treatment units is performed simultaneously in order to adjust the treatment schedule to a fixed schedule of batch processes. © 2009 Elsevier Ltd. All rights reserved.