12th International Symposium on District Heating and Cooling

optimal investments in new power plants and heatstorages.The study has been restricted to residential andindustrial district heating systems. Buildings notconnected to district heating systems were notconsidered, although these also require heat. Coolingdemand could also offer similar possibilities, but theproblem was not addressed here. Industrial heatdemand and water heating do not usually have strongseasonal variation and can therefore be more valuabletowards the integration of variable power.METHODS AND DATAThe model and assumptions used for the analysis aredescribed in more detail in [2]. For convenience, mostimportant sections are referenced below. The heatsector of the model is described more thoroughly here.The Balmorel model is a linear optimization model of apower system including district heating systems. Itcalculates investments in storage, production andtransmission capacity and the operation of the units inthe system while satisfying the demand for power anddistrict heating in every time period. Investments andoperation will be optimal under the input dataassumptions covering e.g. fuel prices, CO2 emissionpermit prices, electricity and district heating demand,technology costs and technical characteristics (eq. 1).The model was developed by (Ravn et al. [1]) and hasbeen extended in several projects, e.g. (Jensen &Meibom [10], Karlsson & Meibom [11], Kiviluoma &Meibom [2]).min iIExOperation Ci CiwtciPi, t, QitInvFixciCi ci,iItTiIThe optimization period in the model is one yeardivided into time periods. This work uses 26 selectedweeks, each divided into 168 hours. The yearlyoptimization period implies that an investment is carriedout if it reduces system costs including the annualizedinvestment cost of the unit.The geographical resolution is countries divided intoregions that are in turn subdivided into areas. Eachcountry is divided into several regions to represent itsmain transmission grid constraints. Each region hastime series of electricity demand and wind powerproduction. The transmission grid within a region isonly represented as an average transmission anddistribution loss. Areas are used to represent districtheating grids, with each area having a time series ofheat demand. There is no exchange of heat betweenareas. In this article, Finland is used as the source formost of the input data.The hourly heat demand has to be fulfilled with the heatgeneration units, including heat storages (eq. 2).The <strong>12th</strong> <strong>International</strong> <strong>Symposium</strong> on District Heating and Cooling,September 5 th to September 7 th , 2010, Tallinn, Estonia(1)194Loading of heat storage adds to the heat demand. Lossduring the heat storage process is not considered. Thedynamics of heat networks were not taken intoaccount.iIQi t hrt Z t Ta A, ,i,tHeatStoiIa ; (2)Analysis is done for the year 2035. By this time, largeportion of the existing power plants are retired. Threedistrict heating areas were considered. These have arather different existing heat generation portfolio by2035. This helps to uncover some interesting dynamicsin the results section.In this paper, scenarios without new nuclear power arecompared (scenarios ‗Base NoNuc‘ and ‗OnlyHeatNoNuc‘ in article [2]). This meant that wind power had avery high share of electricity production. Accordingly,there was more demand for flexibility in the system.‗Urban‘ area presents the heat demand in the capitalregion of Finland. The existing power plants in 2035cover over half of the required heat capacity. Largestshare comes from natural gas, which is a relativelyexpensive fuel in these model runs. The annual heatdemand is smallest of the considered areas: 6.2 TWh.‗Industry‘ area aggregates the known industrial districtheating demand from several different locations. This isa necessary simplification, since Finland has overhundred separate DH areas and the model would notbe able to optimise all of these simultaneously. Theindustrial heat demand in Finland is driven by paperand pulp industry, which produces waste that can beused as energy input. This capacity is assumed to beavailable in 2035 and as a consequence the modeldoes not need more industrial heat capacity. Theannual heat demand is 46.8 TWh.‗Rural‘ area aggregates non-industrial heat demandexcluding the capital region considered in ‗Urban‘. Thisis probably the most interesting example, as theexisting capacity covers only 20% of the heat capacitydemand. Therefore, the model has to optimise almostthe whole heat generation portfolio. There are woodresources (limited amount of forest residues and moreexpensive solid wood) available unlike in the urbanarea. The annual heat demand is 21.0 TWh.RESULTSFigures 1–3 give an example how heat productionmeets heat demand in the different areas during thesame 4.5 days in January. Negative productionindicates charging of heat storage. Electricity price ison separate axis together with the cumulative contentof heat storage. When electricity price is low, storage isloaded with electricity using heat boilers and heatpumps. When electricity price is high, CHP units