Water Behind Capacity
Master thesis
Permanent lenke
http://hdl.handle.net/11250/2404702Utgivelsesdato
2016Metadata
Vis full innførselSamlinger
- Institutt for elkraftteknikk [2503]
Sammendrag
When the power system becomes more dependent on intermittent renewable energy, the needfor ancillary services becomes larger. Therefore, the power producers ability to deliver reserveshas been analysed in this thesis. Additionally, the costs of delivering such services have beenevaluated, as these costs determine if it is profitable for hydro power producers to supply balancingpower instead of energy. Since hydro power is a well suited source for delivering balancingpower, and a major part of the Nordic power system, the amounts of reserves which can bedelivered from hydro power units have been the main topic of this thesis.
The thesis is based on the decision support tool SHOP, which is a short termoptimization modelfor hydro power producers developed at SINTEF Energy Research, in addition to its newly developedsimulation funcionality. The model is one of the industry s most used tools for findingthe optimal amount of energy to be sold in the power market in every hour for the days to come.However, the model is not adequate for deciding the amount of reserves to be be delivered, anddoes not consider hydrological constraints when the reserves are activated. Therefore, the firstobjective in this thesis has been to develop a methodology for assessing the amounts of reserveswhich can be delivered. In order to do this, the methodology has been based on the excisingmodels, and the model functionalities have been utilized in innovative and non-standard ways.Next, the second objective in this work has been to validate the methodology on illustrative andrealistic case studies.
Fromthe results in the analysis, it has been shown that the developed methodology successfullyassesses the amounts of reserves which can be delivered, and produces realistic results. Hence,the results are not only verifying the methodology, but are also illustrating concepts which arevery valuable for market participants who are delivering ancillary services. Furthermore, it hasbeen shown that the costs and amounts of reserves which can be delivered froma hydro systemin the regulating power option market are strongly related to the amount of available water inthe system s reservoirs. Additionally, the hydro system s flexibility also plays an important role,as more reserves can be delivered from a less constrained system. As more flexibility is availablewhen reserves are delivered from a hydro systemwhich consists of several cascaded plants, more reserves can typically be delivered from such systems, and at lower costs.
The analysis has been carried out in several steps. First, the expected number of hours whichthe reserved capacity is activated has been found. This estimate is based on available historicaldata of the Nordic power system, and statistical methods have been used to find the expectednumber. Next, the maximum amount of capacity which can be activated from a hydro systemduring the expected time period without emptying the system s reservoirs has been found. Finally,the costs of delivering different amounts of reserves have been found. The analysis hasbeen carried out on two different hydro systems, where different amounts of generation capacityhave been reserved for the regulating power option market, and the amounts of availablewater in the reservoirs have been varied.
At last, it has been shown that the amounts of reserves which actually can be delivered are limited,even though the system easily can withhold more generation capacity from the generatingunits. This is both due towater unavailability and violation of hydrological constraintswhen thereserved capacity is activated. Hence, in order to obtain decision support tools which are wellsuited for production scheduling in several markets, the excising models should be extendedto include algorithms which assess the volumes that can be delivered in the balancing marketswhile considering the risk of violating relevant constraints.