Large-scale desalination facilities are increasingly employed to supplement potable water supplies for many cities, where the demand for water is having a negative impact on the sustainability of natural water resources. A primary environmental concern with the establishment of these large-scale facilities is the effective disposal of the hyper-saline effluent brine, so that harmful effects on the marine environment are minimised. In countries with effective effluent discharge regulations, the negatively buoyant brine is typically released through submerged diffusers, where the ports are inclined towards the ocean surface to aid dispersion processes. Predictive models provide an inexpensive method of considering different sources and ambient design parameters. It has been demonstrated that predictive models developed primarily for positively buoyant discharges significantly underestimate dilution measured by physical studies, when applied to negatively buoyant desalination discharges. It has been shown recently that predicted dilution and geometric parameters can be improved if a reduction in the buoyancy flux of the main flow is incorporated into these models. Here, the reduced buoyancy flux (RBF) approach is modified through the use of a physically based buoyancy loss mechanism. Improvements in the predictive capabilities of the new model are demonstrated through comparisons with predictions from existing models and an extensive range of results-from physical studies for geometric, dilution, and velocity parameters. (C) 2015 Elsevier B.V. All rights reserved.