Use of Reservoir Operation Optimization Methods in Practice: Insights from a Survey of Water Resource Managers

The use of mathematical models to guide reservoir operations has a long history. The first reviews of the scientific literature on the topic already appeared in the 1980s (e.g. Yeh 1985), while the number of papers introducing new methods and applications has steadily grown in recent years (see e.g. Fig. 2 in Dobson et al., 2019a). Over time, algorithmic advances have enabled the application of reservoir operation optimization to increasingly complex simulation models and to larger number of objectives (e.g. Reed et al. 2013). Given the renewed interest in dam construction, particularly for hydropower development (Zarfl et al. 2014), and the pressure to expand the range of interests considered in dam operation, particularly towards environment conservation targets (e.g. Poff and Schmidt 2016; Chen and Olden, 2017), (multi-objective) optimisation would be expected to play a growing role in informing reservoir operations.

Despite this potential, however, there is a shared perception among researchers that optimisation methods have seen limited uptake by practitioners. For example, in a state-of-art review of the Water Resource System Analysis (WRSA) field, Brown et al. (2015) concluded that, while simulation models are widely used for what-if analyses and manual appraisal of options, optimisation methods are rarely used outside academia (with the notable exception of hydropower applications, see e.g. Ibanez et al. (2014)). Perhaps surprisingly, attempts at formally surveying practitioners to assess the validity of this perception have been quite limited so far. To our knowledge, the first study of this type dates back to the survey of US practitioners by Rogers and Fiering (1986), who reported a very limited uptake of WRSA methods at the time. More recently, Rosenberg et al. (2017) interviewed some practitioners in the US and Asia and found that “all practitioners mentioned use of simulation modeling” whereas most “indicated that they never implemented formal optimization algorithms”, and “were more inclined to either manually generate scenarios or use simple search algorithms”. The apparent disconnect between research and practice communities is a recurrent theme in commentary papers in the WRSA field, and further efforts have been advocated to provide more stringent evidence of the contribution of WRSA to society (Kasprzyk et al., 2018).

In this paper, we contribute to this ongoing discussion by presenting the results of a survey of practitioners of water companies in England and Wales, aimed at assessing specifically the use of reservoir simulation and optimisation tools. We complement the survey results with interviews of practitioners in consultancy companies and our own experience of interacting with the UK water industry. Finally, we suggest some directions for future research that we think may be interesting for researchers while also helping to make the field more relevant for practice.