An integrated water-electricity market design for multi reservoir, mixed operation.
Thesis DisciplineManagement Science
Degree GrantorUniversity of Canterbury
Degree NameDoctor of Philosophy
Water markets are often regarded as the most promising method of managing this increasingly important natural resource, but the literature on water market concepts is only emerging. Most of the focus is on physical trading arrangements, but financial property rights appear both conceptually and practically appealing, as a way to develop commercial and organizational arrangements to improve liquidity and ultimately increase efficient resource use. This thesis focuses on market arrangements to manage hydrology dependent surface water supplies, where consumptive and/or non-consumptive use occurs in a network with storage. Binding resource constraints create temporal and locational price differences. Moreover, the uncertainty about price differentials creates barriers to trade. Participant bids, reflecting their marginal use values, are assumed to be cleared by a benefit-maximising optimisation, such as Stochastic Linear Programming. This also creates price differences between locations, and time periods, and causes the market to accumulate a “settlement surplus” of rents associated with resource constraints. This thesis draws on the Financial Transmission Right (FTR) concepts developed for electricity markets to outline a general structure of financial hedging instruments that could be used to deploy this settlement surplus to hedge against price risks, across space and time. We also consider a swing option based approach, which bundles the above rights to create a virtual “slice of system” model that could be practically and conceptually appealing to both aggregated and disaggregated hydro reservoir systems. While only preliminary, our discussion of these options suggests that developments along these lines may be important in creating a water market environment that is acceptable to potential consumptive and non-consumptive participants.
The remainder of this thesis is about the problem of intra-period consumptive and non-consumptive water allocation in a mixed-use catchment. We develop a deterministic nodal Constructive Dual Dynamic Programming (CDDP) procedure which implicitly clears a market determining both consumptive and non-consumptive water allocations, across all nodes in a catchment with a single reservoir. Consumptive users extract water from the system, so each unit of water flow can only be used for a single consumptive use. A non-consumptive user transfers water from one node to another, extracting some benefit, or incurring some cost. Arc flow bounds may limit the opportunities for using water at the nodes. Costs can be associated with arc flow bounds and distributary demands to represent in-stream and environmental reserve flows enforced using penalty costs. The algorithm constructs the intra-period demand curve for release by sequentially forming marginal water value curves at each node, passing these curves towards the reservoir.
This approach can generate net demand curves representing all possible market-clearing solutions at nodal and user levels. It can also be used to construct net demand curves for water release from the reservoir, in each period, which could then be used in a stochastic inter-temporal CDDP model to construct marginal water value curves stored in the reservoir over an appropriate time horizon. Several variants on this approach are explored.
We discuss extending the procedure to assess the marginal value of water stored in two inter-connected reservoirs in a mixed-use catchment. A “lower level” intra-period CDDP is applied to construct a two dimensional “demand surface” for transfer, representing the marginal benefit from net release into either end of the inter-reservoir chain between the two reservoirs. Then a higher level inter-period CDDP demand-curve-adding method could be deployed to strike the optimal trade-off between the current release demands for the inter-reservoir chain and other sub-trees leading from the two reservoirs and the future storage demands.