Input Data

In this subsection, we describe the input data that we use for our fundamental models. We consider the German electricity market. Foreign markets are not explicitly modeled, but included with time-series of expected cross-border elec- tricity exchanges. For model calibration, data from the years 2006 and 2007 is

2.4. COMPARISON WITH OBSERVED PRICES 93 Unit Nuclear Lignite Coal Gas CC Gas GT Oil

Availability % 94.0 84.7 82.1 88.7 90.7 90.7 Capacity GW 20.3 21.5 26.2 14.6 4.6 4.4 Minimum Up-Time h 12 8 8 8 0 0 Minimum Down- Time h 8 8 8 4 0 0 Operating costs e/MWh 0.5 1.6 2.0 1.2 1.2 1.2 Start-up costs e/MW 2.0 5.0 5.0 8.0 8.0 8.0 Start-up fuel usage MWth/MWel 16.7 6.2 6.2 3.5 1.1 1.1 Table 2.2: Technical parameters of the considered units. Sources: Bagemihl [Bag02], Swider and Weber [SW07], Swider et al. [Swi+07].

used. In the following, we describe how electricity supply and electricity demand are modeled, and we give an overview of the fuel and carbon prices used for the comparison.

Electricity Supply:

The power plants considered in our model correspond to the existing units in Ger- many. The data is combined from different publicly available sources like the EEX, data provided by generation companies, and a list of power plants of the German Federal Environmental Agency [Umw09]. We adjusted the data by adding a coal unit, an oil-fired unit and a lignite unit such that the total capacity of each fuel type corresponds approximately to the capacities provided by the German Fed- eral Statistical Office [Des08].

For each unit, we multiply the capacity with an availability factor. This factor is used to account for non-availabilities of plants due to maintenance or forced outages. The values for the availabilities are taken from Swider et al. [Swi+07, p. 67]. In the LDC model, we are constrained to use a constant factor for each unit for the whole year. As most planned maintenance is scheduled in the summer months when demand is low, a seasonable availability factor is more realistic than a constant one. In the LP model, we consider such seasonal availabilities.

Table 2.2 shows the constant availability factors used in the LDC model as well as other parameters of the considered units. As far as the efficiencies of the plants are

publicly available, we use these efficiencies. For those plants where the efficiencies are unknown, we approximate the efficiency based on the plant type, the fuel and the age of the plant, as described by Schröter [Sch04].

Electricity Demand:

For the electricity demand, hourly load profiles provided by the European net- work of transmission system operators for electricity (ENTSO-E) are used.23 As there is an obligation to feed-in electricity produced by renewable sources, we directly subtracted the production of hydro-units (run-of-river plants and hydro reservoirs with a natural inflow) and wind power units from the demand. For the calculation of the (expected) hydro-production, we use the time-series of hydro- production from the publicly available database of the EU project Wind Power In-

tegration in Liberalised Electricity Markets (WILMAR). 24 Concerning the electric-

ity produced by wind power, we use two different data sources. For the elec- tricity produced by onshore wind units, we use the (hourly) wind power feed- in provided by the four TSOs (EnBW, Amprion, transpower, 50Hertz). As the offshore wind speed characteristics are different from the onshore ones, we use the (offshore) wind speed time-series of the FINO (Forschungsplattformen in Nord-

und Ostsee)project25 to derive the expected offshore wind production. The wind

speed forecasts are transformed to power using a power transform curve (see, e.g., McLean [McL08]).

The increased amount of wind power leads to an increasing demand of control power. Therefore, we model the control power demand depending on the in- stalled wind power capacity as described by DEWI et al. [DEW+05]. In the LDC model, we add the control power demand to the total electricity demand, as pro- posed by Burger et al. [Bur+07, p. 174].

Additionally, electricity demand is adjusted by electricity exchanges with neigh- boring countries. This is based on hourly time-series of cross-border physical load flows published by the TSOs.26

23_{These hourly load profiles are available in the “Country Package - Production, Consumption,}

Exchange” at http://www.entsoe.eu/, last time accessed December 01, 2009

24_{For a documentation, see Kiviluoma and Meibom [KM06]. The WILMAR database is available}

at http://www.wilmar.risoe.dk/Results.htm, last time accessed December 01, 2009.

25_{Data provided by the Bundesministerium für Umwelt, Naturschutz und Reaktorsicherheit (BMU),}

the Projektträger Jülich (PTJ) and the Deutsches Windenergie Institut (DEWI). The data is available (af- ter registration) at http://www.bsh.de/de/Meeresdaten/Beobachtungen/Projekte/FINO/, last time accessed December 1, 2009

26_{As there are no transmission values to Sweden published by transpower, we took this data}

2.4. COMPARISON WITH OBSERVED PRICES 95

Coal Gas Carbon

2006 2.11 e/GJ 5.93 e/GJ 17.40 e/t 2007 2.33 e/GJ 5.55 e/GJ 0.68 e/t 2008 3.77 e/GJ 7.45 e/GJ 22.45 e/t 200927 _{2.35 e/GJ 3.80 e/GJ 12.68 e/t}

Table 2.3: Fuel and carbon prices. Sources: BMWi [BMW09], EEX.

Fuel Prices:

The fuel and carbon prices that we use for the comparison are shown in Table 2.3. While electricity demand and wind power production are based on the values of 2006 and 2007, we use the observed fuel prices of 2008 and 2009 for the following comparison. This choice is motivated by the objective to test how well our models perform under the assumption of pertinent fuel prices. For the power generation expansion problem, fuel price uncertainty is reflected by considering different fuel price scenarios.