Remote communities are typically isolated from the main electricity grid and require local generation, most often relying on expensive resources (especially diesel) to supply their load. The economic and logistical aspects are such that there is value in exploring different approaches to reduce the consumption of diesel and to increase the energy input from local renewables. In this paper, we focus on using the virtual storage capabilities of the demand side to perform diesel generation optimization. Specifically, we revisit the economic dispatch problem by adapting it to the presence of flexible demand. We compare three different formulation approaches: one where demand is reallocated heuristically to shave the peak, a second where the reallocation is itself optimized and a third that attempts to optimize the overall efficiency of the diesel units. A case study based upon a remote microgrid shows that flexible demand can decrease diesel fuel consumption in an economically meaningful manner, which is still significant given the costs associated with the purchase and haulage of diesel to these small communities.