Performance Simulation For Parabolic Trough Concentrating Solar Power Plants And Export Scenario Analysis For North Africa

Performance Simulation For Parabolic Trough Concentrating Solar Power Plants And Export Scenario Analysis For North Africa By Daniel Horst A Thesis Submitted to the Faculty of Engineering at Cairo University in Partial Fulfillment of the Requirement for the Degree of MASTER OF SCIENCE In MECHANICAL POWER ENGINEERING FACULTY OF ENGINEERING, CAIRO UNIVERSITY GIZA, EGYPT 2012

Performance Simulation For Parabolic Trough Concentrating Solar Power Plants And Export Scenario Analysis For North Africa By Daniel Horst A Thesis Submitted to the Faculty of Engineering at Cairo University in Partial Fulfillment of the Requirement for the Degree of MASTER OF SCIENCE In MECHANICAL POWER ENGINEERING Under Supervision of Dr. Adel Kahlil Professor In Mechanical Power Engineering Department Faculty of Engineering Cairo University Dr.-Ing. Jürgen Schmid Professor In Electrical Engineering Department Faculty of Engineering Kassel University Dr. Carsten Pape Frauenhofer-Institute for Wind Energy and Energy System Technology FACULTY OF ENGINEERING, CAIRO UNIVERSITY GIZA, EGYPT 2012

Performance Simulation For Parabolic Trough Concentrating Solar Power Plants And Export Scenario Analysis For North Africa By Daniel Horst A Thesis Submitted to the Faculty of Engineering at Cairo University in Partial Fulfillment of the Requirement for the Degree of MASTER OF SCIENCE In MECHANICAL POWER ENGINEERING Approved by the Examining Committee: Prof. Dr. Adel Khalil Prof. Dr.-Ing. Jürgen Schmid Dr. Hany Nokrashy FACULTY OF ENGINEERING, CAIRO UNIVERSITY GIZA, EGYPT 2012

I Table of Contents List of Figures List of Tables List of Abbreviations List of Symbols Abstract 1. Introduction...1 1.1 Renewable Energy and Climate Change...1 1.1.1 Solar technology...1 1.2 Parabolic trough technology...3 1.2.1 Reflectors...4 1.2.2 Absorber...4 1.2.3 Tracking and controlling...5 1.2.4 Heat transfer medium...5 1.2.5 Thermal storage...6 1.3 Thesis objectives and outlines...7 2. Theoretical performance calculation CSPP...8 2.1 Geometrical relations...8 2.1.1 Sun earth geometry...8 2.1.2 Sun collector geometry... 10 2.2 Solar filed... 12 2.2.1 Optical losses... 12 2.2.2 Heat losses... 14 2.3 Thermal storage... 15 2.3.1 Stratification Storage the Multi- Node- model... 16 2.3.1 The Plug- Flow model... 18 2.4 Power block... 19 2.4.1 Carnot Cycle... 19 2.4.2 Clausius- Rankine Cycle... 21 2.4.3 Steam Turbine... 26 3. Recourse assessment for CSPP...29 3.1 Land Recourse assessment... 29 3.1.1 Slope... 30

II 3.1.2 Land Cover... 31 3.1.3 Hydrology... 31 3.1.4 Geomorphologic features... 32 3.1.5 Protected areas... 33 3.1.6 Industry and Population... 33 3.1.7 Technical potential... 34 3.1.8 High voltage grid... 36 3.2 MED- CSP Scenario CG/HE... 37 3.2.1 Growth of population... 38 3.2.2 Growth of economy... 38 3.2.3 Electricity demand... 39 3.2.4 Scenario for energy security... 40 3.3 Weather data... 42 3.3.1 Irradiation data sets... 43 3.3.2 Ambient data sets... 45 4. Simulation Program...47 4.1 Program overview... 47 4.2 GIS database and scenario processing... 49 4.2.1 Scenario processing... 50 4.2.2 Side evaluation... 51 4.2.3 Side depositing... 54 4.3 CSP performance simulation... 56 4.3.1 Solar field simulation... 56 4.3.2 Thermal storage simulation... 60 4.3.3 Power block simulation... 64 4.4 Curve fitting and ELCC calculation... 74 4.4.1 Least square optimization... 74 4.4.3 ELCC calculation... 76 5. Export scenario 2050...80 5.1 Scenario description... 80 5.1.1 Installed capacity and plant distribution... 80 5.1.2 Residual load curve... 81 5.1.3 Power plant park... 84 5.2 Simulation results... 86 5.2.1 Simulation for SM1... 86

III 5.2.2 Simulation for SM2... 87 5.2.3 Simulation for SM3... 89 5.2.4 Simulation for SM4... 91 6. Conclusion...94 6.1 Simulation program... 94 6.2 Export Scenario... 96 References...100 Appendix A (Validation table CSP performance calculation)...103 Appendix B (Transmission grid maps North Africa)...105

IV List of Figures Figure 1: Functional principle of a parabolic trough [REB]..3 Figure 2: Absorber tube of a parabolic trough collector [REB]...4 Figure 3: Basic concept for the integration of thermal energy storage into a solar thermal parabolic through power plant [REB].6 Figure 4: Geometrical relation between sunbeam and tilted surface [EEG].8 Figure 5: Displacement of the sun image [SLP]...11 Figure 6: Three-node stratification liquid storage tank [SETP]..17 Figure 7: Plug and Flow model whit 4 layers [SLP].19 Figure 8: Carnot cycle p,v and T,s Diagram [TDK].20 Figure 9: Schematic drawing of a steam power plant and the T,s Diagram of a Clausius-Rankine cycle [TDK] 21 Figure 10: Real Clausius-Rankine Process T,s Diagram [KWT]..23 Figure 11: Increasing of the main steam parameters [KWT]...24 Figure 12: Water steam cycle with reheating [KWT].25 Figure 13: Regenerative feed water preheating [KWT] 26 Figure 14: Schematic drawing of a high-pressure turbine [KWT]...27 Figure 15: Areas whit a slope higher than 2.1% [SID]...30 Figure 16: The land cover in the Euro-Mediterranean Region [SID] 31 Figure 17: The Hydrology of the Euro-Mediterranean Region [SID] 31 Figure 18: Geomorphologic exclusion criteria in the Euro-Mediterranean region [SID].32 Figure 19: Protected areas of the Euro-Mediterranean Region [SID]..33 Figure 20: Industry and population of the Euro-Mediterranean region [SID]..34 Figure 21: Annual direct normal irradiation in kwh/m 2 /y on non-excluded areas in the areas in the Euro-Mediterranean Region [SID] 34 Figure 22: Annual direct normal irradiation on non-excluded areas in global scale [SID].35 Figure 23: Electrical Transmission System Network of North Africa [GENI] 36 Figure 24: Envisage Mediterranean interconnections with Europe [GENI]..37 Figure 25: Population growth in North Africa by countries [MCSP]..38 Figure 26: Average GDP growth rate between 2003 and 2050 for North African countries [MCSP].39 Figure 27: Energy consumption in North Africa until 2050 [MCSP]..40 Figure 28: Electricity production in the MENA region until 2050 [MCSP].41

V Figure 29: Area for the CSOMO-EU model displaying topographic height in meter [COE]...43 Figure 30: Global horizontal irradiation calculated out of the COSMO-EU data compared whit global horizontal irradiation data from HC3 database..44 Figure 31: I, b and DNI for 31 N 29 E and year 2007 simulated out of COSMO-EU data..45 Figure 32: Monthly average ambient data for Egypt June 2007 provided by the COSMO-EU database..46 Figure 33: General overview of the Matlab program for CSPP calculation.47 Figure 34: Basic program structure 48 Figure 35: Basic program structure scenario analyzes...50 Figure 36: Number of CSPP according to MED CSP Scenario.51 Figure 37: Black-white map for evaluating available areas 52 Figure 38: Schematic drawing of high voltage grid in North Africa...53 Figure 39: Possible installed gross capacity per country per country. Simulated with a grid distance of 25km and 200 MW el installed gross capacity per pixel.53 Figure 40: Irradiation map of North Africa, CSPP distribution for scenario 2030 54 Figure 41: Irradiation map of North Egypt, CSPP distribution for scenario 2030 Cairo area..55 Figure 42: Irradiation map of North Africa, CSPP whit cooling system distribution for the scenario 2020..55 Figure 43: Basic Program structure performance calculation 56 Figure 44: Incidence angle, tilt angle and solar height at 31 N 20 E for the 21 st of March 2007.57 Figure 45: Geometrical Collector losses in a location of 31 N 29 E at 21 st March 2007 one axis tracking system, collector orientation north south..59 Figure 46: Efficiency and thermal energy harvest by the solar field with a size of 188,000m 2 and a direct normal irradiation of 800 W/m 2 at 31 N 20 E for 21 st of March 2007 one axis tracking, collector orientation north south 60 Figure 47: Schematic drawing of storage arrangement in the simulation program 61 Figure 48: Schematic drawing of SM arrangement, storage unit is 6h 62 Figure 49: Full load operating hours /Simulation operating hours for solar-multiple 1-4 simulated for a CSP plant equipped with wet cooling system and back cooling system located at 31 N 29 E metrological year 2007.64 Figure 50: Schematic drawing of water steam cycle with basic design parameters..65 Figure 51: Schematic drawing of the arrangement for a wet cooling system equipped whit back cooling system and water turbine....67

VI Figure 52: Schematic drawing of evaporation cooling system with air-cooling tower 69 Figure 53: Schematic drawing of condenser connected to a dry cooling system...71 Figure 54: Plant performance calculation for SM1 equipped with dry cooling system at a location of 31 N 29 E for the meteorological year 2007. Collector orientation is north south..73 Figure 55: Plant performance calculation for SM4 equipped with evaporation cooling system at a location of 31 N 29 E for the year 2007. Collector orientation is north south...74 Figure 56: Feed in series direct calculated and optimised and for CSPP SM 4 with wet cooling and back cooling system located at 31 N 29 E. CSOMO-EU ambient data for 2007..76 Figure 57: Schemata for the ELLC calculation [UVE].77 Figure 58: Method of recursive convolution, example for 3 power plants [DENA1]..78 Figure 59: Distribution of exporting CSP plants for Germany, with installed gross capacity of 200 MW el for each plant, in North Africa for 2050 80 Figure 60: SIM-EE model IWES for simulation of the residual load curve [UBA]...82 Figure 61: Annual load duration curve for Germany in 2050 with an energy supply by 100% renewable energy sources [UBA] 84 Figure 62: Residual load curve from the UBA study 100% renewable electricity supply by 2050 [UBA]..84 Figure 63: SIM-EE model IWES for simulation of the residual load curve. Changes are under taken in the position of importing energy according to the simulation in this thesis. [UBA]...85 Figure 64: Weekly average energy production from CSP plants SM1 based on the weather data COSMO-EU 2007.87 Figure 65:Residual load curve without import energy and residual load curve including import energy form CSP with SM2. Irradiation and ambient data from COSMO-EU model 2007 88 Figure 66: Partial view of figure XX residual load curve without imported energy and residual load curve including imported energy from CSP with SM2. Irradiation and ambient data from COSMO-EU model for 2007.88 Figure 67: Annual load duration curve for the residual load with and whiteout CSP import SM2 89 Figure 69: Residual load curve without imported energy and residual load curve including imported energy from CSP with SM3. Irradiation and ambient data from COSMO-EU model 2007 90

VII Figure 69. Partial view part of figure XX residual load curve without import energy and residual load curve including import energy form CSP SM3 as well as energy production CSP. Irradiation and ambient data from COSMO-EU model for 2007..90 Figure 70: Annual load duration curve for the residual load with and whiteout CSP import SM3.91 Figure 71. Residual load curve without imported energy and residual load curve including imported energy form CSP SM4. Irradiation and ambient data from CSOMO-EU model 2007.92 Figure 72. Partial view part of figure XX residual load curve without imported energy and residual load curve including imported energy from CSP with SM4. Irradiation and ambient data from COSMO-EU model for 2007..92 Figure 73: Annual load duration curve for residual load with and without CSP import. Ambient data COSMO-EU 2007.93

VIII List of Tables Table 1: Compulsive and optional criteria for the exclusion of land used for CSP plants [SID].29 Table 2: Areas for CSP in km 2 available in the MENA countries for different DNI Classes [SID]..35 Table 3: Concentrated solar thermal potential in North Africa [MCSP]...42 Table 4: Basic input parameters for individual simulation of CSPP..49 Table 5: Input parameter for scenario analyze..50 Table 6: Design parameters for simulation of the solar field for CSP plant with gross capacity of 200 MW el [DLS].57 Table 7: Design parameter for storage simulation [DLS] 60 Table 8: Design parameters for simulation of the power block whit gross capacity of 200 MW el [DLS].64 Table 9: Design parameters for water steam calculation with a wet cooling system [DLS] 67 Table 10: Design parameters for water steam calculation evaporation cooling system [DLS 69 Table 11: Design parameters for water steam calculation dry cooling system [DLS] 71 Table 12: Installed exporting CSP plants and gross capacity separated by countries for North Africa 2050.81 Table 13: Unplanned outage probability [DENA1]...85 Table 14: PPN for ELCC simulation...86 Table 15: Summery of the results from export scenario analyze..97

IX List of Symbols A = Area C = Concentration ratio (chapter 2) / ventilator constant (chapter 4) c = specific heat capacity D = distance collector row F = area angle (chapter 2.2) / control function (chapter2.3) f = focal length (chapter 2.) / solar field coefficient (chapter 4) G = irradiation at absorber pipe (chapeter2.2) / Gibbs free energy (chapter2.3) H = height of collector h = specific enthalpy / probability density function (chapter 4.4) I = global horizontal irradiation I b = global horizontal beam radiation I d = global horizontal diffuse radiation L = length of the collector l = length of the not irradiated absorber part m = mass flow N = number of nodes n = number of the day P = power / black out probability (chapter 4.4) p = pressure Q = thermal energy q = specific thermal energy S = entropy s = specific entropy T = temperature in Kelvin t = time U = heat transfer coefficient (chapter2.2) / internal energy (chapter 2.3) V = volume (chapter 2) / availability collector field (chapter 4) v = specific volume (chapter 2) /wind speed at ground height (chapter 4) W = work w = specific work (chapter 2)/ speed (chapter 4) x = distance (chapter 2) /water contend (chapter 4.3) z = distance for water transportation

X Greek letters α s = surface azimuth α r = surface azimuth relative β = tilt angle β s = solar elevation χ= storage capacity δ = angle of declination ε = emissivity φ = latitude η= efficiency ϕ = relative humidity ρ= density θ= incidance angle θ Z = zenith angle ψ = solar field reduction factor

XI List of Abbreviations ALB_RAD: ASOB_S: CC: CSP: CSPP: DLR: DNI: DSMW: ELCC: GDP: GHG: GIS: GLCC: GNI: HC3: IDR: IEA: IWES: IPCC: LOLE: LOLP: MENA: NDVI: PPA: PPN: RE: SM: SEGS: UBA: WBGU: WCU: ground albedo short wave radiation at ground high capacity credit concentrated solar power concentrated solar power plant German aerospace center direct normal irradiation digital soil map of the world effective load carrying capacity gross domestic product green house gasses global information system global land cover characterization gross national income HelioClime3 Incidence direct radiation international energy agency Frauenhofer-institute for wind energy and energy system technology Intergovernmental Panel on Climate Change loss of load probability loss of load probability Middle East and North Africa normalized difference vegetation index purchasing power parity power plant network renewable energies solar multiple solar energy generation system German Federal Environmental Agency German Advisory Council on Global Change World conservation Union

XII Abstract Since the Intergovernmental Panel on Climate Change (IPCC) report 2007 established: it is very likely that global warming nowadays is man made [IPCC2007a], it becomes obviously that the emission of CO 2 have to be reduced drastically. The green house gas (GHG) concentration for 2010 was 39% above the preindustrial level. Therefore the warming trend has increased significantly over the last 50 years [IPCC2007a]. Still GHG emissions associated with the provision of energy services are the major cause of climate change. Nevertheless, cold-fired power plants are still the basis of electricity production all over the world. In order to work against this trend research and political influence is necessary to avoid the negative impact of the global warming. Based on the history of industrial development the industrialized countries have the duty to reduce their own emissions, as well as supporting developing countries by doing so. In this context, options like exporting solar energy form the deserts of North Africa to Europe would offer great possibilities and both sides might have a benefit. Because of the high solar irradiation form 2500kw/m 2 /y in some parts of North Africa and the high share of direct radiation concentrated solar power plants (CSP) are an excellent option for a sustainable electricity production in this region. The thesis is providing a model witch can calculate the electrical output of CSP parabolic trough plants for several locations in North Africa. Furthermore it is analyzed how these CSP plants can contribute to a 100% renewable energy supply in Germany for the year 2050. Therefore the first part of this thesis presents a model estimating suitable locations for CSP plants in North Africa and calculating the electrical output of CSP plants. Several criteria s for land use, annual irradiation and infrastructure, are processed, and rules for depositing CSP plants were specified. In alliance with weather prediction models an energy-balanced CSP model simulates the electrical output of exclusive solar driven CSP parabolic trough plants. By using different configurations the simulation can display various forms of storage sizes and different cooling system. The second part of the thesis is focused on how the CSP plants can support a 100% renewable energy system in Germany in the year 2050. Therefore, the effective load carrying capacity (ELCC) and the capacity credit (CC) for a number of CSP plants in relation to the residual load in Germany is evaluated. The focus is on the influence different storage sizes have on the possibility to supply energy on demand for Germany.

XIII The work shows that even a high capacity credit of around 53% for solar multiple four CSP plants is not necessarily helpful to reduce significant the high demand in Germany during winter times. This can be mainly related to the seasonality noticed in the CSP output even whit high storage capacities. The thesis is outlined by the MED-CSP study [MCSP] and the study 100% renewable electricity supply by 2050 [UBA2050] of the German federal environmental agency.

! 1 1. Introduction 1.1 Renewable energy and climate change The 2007 IPCC report states that Most of the observed increases in global average temperature since the mid-20 th century is very likely due to the observed increase in anthropogenic greenhouse gas concentration. [IPCC, 2007a] The concentration of CO 2 has been increased to about 390 ppm CO 2 at the end of 2010, 39% above the pre-industrial level. This lends to an average global temperature increase of 0.76 C. In order not to exceed a global temperature increase of 2 C with a probability of 67% the CO 2 concentration has to be limited at 450 ppm by 2050. This means that only 750 billion tons of GHG can be emitted until 2050. For that reason the 2050 GHG emissions must be bisecting in relation to the emissions of 1990. Taking into consideration historical reasonability as well as their economic strength, developed countries must reduce their emissions by about 80% to 95% by 2050. Countries in transition and developing countries have more time to reduce their emissions. However, taking into consideration the growth of population and the ongoing development in countries of transition and developing countries, a significant increase in energy consumption along with GHG emissions will be noticed. In 2011 the global energy consumption was around 510 EJ/year compared to 340 EJ/year in 1990. In parallel to this trend, the yearly amount of GHG emissions is increasing, reaching 30 Gt CO 2 in 2010. If this trend is ongoing the limit of 750 Gt GHG emissions will be reached before 2035 and it would cause the 2 C goal to not be met. Nevertheless all societies require energy services to meet basic human needs like lighting, cooking, mobility, communication, and to serve productive processes, but GHG emissions associated with the provision of energy services can be seen as the major cause of climate change [IPCC, 2007a]. Consequently, other ways of energy production have to be found. 1.1.1 Solar technology The use of renewable energies (RE) offers a great chance to reduce the GHG emissions in an economical way. The costs and the challenges for the integration of RE into an existing energy supply system is mainly dependent on the actual system characteristic, the current share of RE and the availability of RE resources. Based on these circumstances the WBGU recommends establishing representative projects for introducing RE on a large scale. In this way, incentives on a strategic level for a

! 2 global change in energy policies can be set. A strategic partnership between the European Union (EU) and the MENA region can be seen as a key element of such a policy. In such cooperation the EU can offer technologies and finances in order to address their national and international responsibility for climate protection. The MENA region can benefit from transferring their renewable resources as an export product along with benefiting from the accompanying economic growth. [DLR, MED- CSP] The MENA countries have the greatest potential for using direct solar energy on a global scale, with a minimum of 412 EJ/year and a maximum of 11,060 EJ/year. Solar energy especially offers a significant potential for the near-term (until 2020) and the long-term (until 2050) climate change mitigation. The wide range of technologies for harvesting the energy provided by the sun can be excellently implemented in North African countries and the Middle East. The costs of these technologies have considerably reduced over the last 30 years and political conditions as well as technical development will offer additional cost reduction in the future. Besides thermal energy for heating and cooling, PV systems and concentrating solar power electricity generation can play an important roll in the MENA countries. The majority of the world s electricity production nowadays derives from nuclear, coal, gas, oil, and biomass driven power plants. The CSP plants work on the same concept while simply providing an alternative heat source. Therefore CSP can benefit from not only the improvements made in the solar concentrator technologies but also from the ongoing advantages in steam and gas turbine cycles. Further advantages of this technology are that it does not need exotic materials and adding a thermal storage for operations on grid stability is possible. Also the CSP technologies that can be used are cheap. It can be installed in small-scaled applications with a few kwinstalled-capacities (dish/stirling systems) up to multiple MW s (tower and trough systems). [IPCC, SREEN] Parabolic trough plants can be operated as hybrid plants, together with gas turbines, that can be used as desalination plants or operating as pure solar plants with different storage sizes. Actualization of this technology already exists in the MENA region like the Kuraymat power station. The focus in this work is with parabolic trough technology installed in North African regions. The next part will give a rough overview about the parabolic trough technology.

! 3 1.2 Parabolic trough technology The solar field produces thermal energy by using direct normal irradiation (DNI), and delivers this energy to a steam power plant. The solar field can be considered in a first approach as a solar steam generator. The glass mirror of the solar field has a parabolic shape and is reflecting the incoming direct radiation with a concentration value of around 80 to the absorber tube. One of the most modern Collectors nowadays is the type LS-3. One LS-3 collector consists of 224 mirror segments, where each segment has an area of 2.68 m 2. Taking into consideration the bending of the mirrors, an area of 545 m 2 is reached with one LS-3 collector. In addition the collector contains 24 absorber tubes. The complete construction is a lightweight metal structure, which normally is equipped with a single axis tracking system. Figure 1: Functional principle of a parabolic trough [REB] A complete solar field contains several parallel rows of solar collectors, which get connected in loops of normally 6 LS-3 collectors. The power block is located in the center of the solar field. The distance between the collector rows is planned according to minimizing the piping costs on the one hand and having a minimal shading effect between the rows on the other. In general the design of the solar field depends on plant and collector size, the temperature and pressure losses in the piping system and the specific ambient conditions. Parabolic trough fields can be erected in any direction, but erected in a north-south direction leads to the highest

! 4 possible energy yield over the year while an east-west orientation smoothes down the seasonal fluctuations. Some of the components like the metal structure, the tracking system, the controllers and other subsystems, which make up around 60% of the direct solar field costs, are standard components and can be ordered from several countries and in different forms. The reflectors and the absorber tube, however, are special components and have to be produced specifically for the parabolic trough solar field. 1.2.1 Reflectors The concentrators (see figure 1) consist of a heat-formed glass cake. It is carried by the metal structure of the collector. By using special production techniques, like the float-glass method, absolute evenness of the cake is guaranteed. Glass, which is used in solar applications, must have very low iron content for getting a transmissivity in the solar spectrum of around 91%. The iron content of a so-called White Glass is around 0.015% compared to normal glass with an iron content of around 0.13%. The binding of the reflectors is done under heat conditions. Several safety layer coatings are added, giving additional protection for the mirror. Finally the contour accuracy is tested using a laser beam. 1.2.2 Absorber For the LS-3 collectors the absorber pipe consists of a stainless steel tube with a length of 4 meters and a thickness of 70mm. A glass pipe surrounds the tube (see figure 2). The glass tube allows evacuating of the area between the absorber tube and the glass pipe in order to minimize convection and conduction heat losses. Figure 2: Absorber tube of a parabolic trough collector [REB] The vacuum also serves to protect the highly sensitive coating. Nowadays, such selective coatings remain stable in temperatures of 450 C upwards to 500 C. On average the solar absorption is currently above 95% and at an operational

! 5 temperature of around 400 C the emissivity is below 14%. This leads to an optical efficiency of around 80% for upcoming perpendicular radiation. Furthermore the hydrogen getter (see figure 2) absorbs the hydrogen, which is getting through the glass pipe and the stainless steel pipe by diffusion. A membrane finally pumps the hydrogen out of the vacuum. As a final point, glass/metal joints realize extension bellows compensating the thermal expansion of the pipe, and the connection between the glass pipe and the metal structure. 1.2.3 Tracking and controlling Solar fields in a CSP plant use single axel tracking systems. The tracking is according to the position of the sun and/or the requirement of the power block. Therefore a solar sensor is used to evaluate the sun position. Sensors consisting of a convex lens focus the sun light to a small photovoltaic cell, reaching a resolution of around 0.05%. These kinds of sensors are used in the so-called SEGS plants, where they prove most effective. The tracking system must have sufficient torque to operate the collectors even at higher wind speeds. For LS-3 collector s normally electrohydraulic drives are used. In the design specs the movement can take place with a speed of 9 m/s. For emergency reasons or for operation conditions, which are not requiring a high optical efficiency, the speed can be increased up to 20 m/s. In existing plants the controlling of the field takes place in two separate stages. The overall control is located in the central control room and the second stage is placed on each collector unit. The local units take care of the incoming irradiation; wind speed and mass flow of heat transport medium. In case of emergency the local units can shut down parts of the solar field. The overall unit operates the solar field according to the overall plant requirements, mainly the electrical output in relation to the actual solar radiation. 1.2.4 Heat transfer medium At the moment high-boiling synthetic thermal oil has been applied as the heat transfer medium in the absorber tubes. According to the thermal stability of this oil the actual operation temperature of the solar field is approximately 400 C. When operating at this temperature, the oil has to be pressurized at around 12 to 16 bar. The thermal oil is circulating in the collector tubes, where the driving forces are speed adjustable pumps. For the purpose of thermal expansion during its heat up an expansion vessel is installed.

! 6 1.2.5 Thermal storage The availability of thermal storages currently plays an important role for the economic success of solar thermal power plants. In parabolic trough plants sensible heat storages, operating with temperatures between 300 C and 400 C, are in use. The storage has a significant influence on the operating conditions of the solar thermal plant. Changes in the solar radiation availability lead, without proper storage, to a change in the electrical output. This not only leads to the plant having a reduced supply security, but also to a reduced lifetime of the steam turbine itself. Frequent changes to the load of the turbine lead to more thermal stresses, thus reducing the lifetime of the turbine. Larger storages are able to support load shift to non-day times, for example the evening hours when peak load demand is needed. In combination with an over dimension of the solar field the yearly operational hours can be extended significantly. Therefore the operation time of around 2,000 hours per year can be increased to 4,000 hours per year by doubling the solar field and storing the produced energy over the day. The solar multiple can extend up to 8,000 operating hours per year. This allows solar multiple (SM) plants to operate as base load plants. SM is used as an indicator of how much the solar field is oversized. SM 2 means an additional energy for around 2,000 operating hours per year, SM 3 for a sum of 6,000 operating hours per year and so on. The figure 3 displays one possible arrangement of a CSPP with thermal storage. Figure 3: Basic concept for the integration of thermal energy storage into a solar thermal parabolic trough power plant [REB]

! 7 1.3 Thesis objectives and outlines The objective of the thesis is to develop a physical model; witch can predict the electrical output of CSP parabolic thought plants in North Africa. Furthermore it have to be analyzed how these CSP plants can contribute to a 100% renewable energy supply in Germany for the year 2050. Consequently the model must be able to deal with data sets out of the SIM-EE model [UBA2050]. Using Matlab as programming software it is possibility to fulfill this requirement. So far no exclusive solar driven CSP plants installed in North Africa. Realistic predictions for the future have to be developed, and ways of distributing the CSP plants in North Africa must be evaluated. Therefore, the first work for the thesis is to find a way to estimating suitable locations for CSP plants in North Africa. Criteria s for land use, annual irradiation and infrastructure, can be used to do so. Furthermore weather prediction model must be added to the simulation for getting exact results. Also different configurations for the plant like various forms of storage sizes and different cooling must be considered. The second part of the thesis is implementing process criteria s from the SIM-EE model in order to compare the results from part one, with results from the SIM-EE simulation for 2050. The focus should be on the effective load carrying capacity witch will allow to make statement related to the influence import energy from North Africa has on the 100% renewable energy system in Germany for the year 2050. The thesis starts with a theoretical description of the physical processes, occurring in a parabolic trough CSPP. In the next chapter the recourses and the potential for CSP plants in North Africa, related to several criteria s are analyzed. The program development is described in the 4 th chapter. This chapter contains also the results of the simulation program and the added program parts from the SIM-EE model. The 5 th chapter finally, presents the outlines and the results for the export scenario analyze. Chapter six summarizes the results and is giving an outlook for possible further investigations.

! 8 2. Theoretical performance calculation CSPP 2.1 Geometrical relations The energy contained in the direct normal radiation is harvest by the concentration collectors of a solar field. By reflecting the received normal direct radiation to the absorber, thermal energy is produced and transported to the power block. Conventional technology is used to convert this energy into electrical energy. Therefore the incoming radiation can be seen as the fuel of a CSP plant. For a precise calculation of the CSPP performance it is essential to understand the geometrical relations between the sun and earth at any time of the day. In this chapter these theoretical relations are described. 2.1.1 Sun earth geometry The geometrical relationship between the incoming direct radiation and a plane of any orientation at every hour can be described in terms of several angles. The following schematic drawing gives an overview of these geometrical relations. Figure 4: Geometrical relation between sunbeam and tilted surface [REG]. The most important angel, in this arrangement is the so-called incidence angle! it can be calculated out of [SETP; S.14]: (2.1) Here " is the tilt angle of the plane, and # the surface azimuth angle, with 0 in the southern direction, then becoming negative when moving towards an eastern direction or positive in a western direction. Both angles can be either fixed or adjusted according to sun position by using a tracking system. The angle of

! 9 declination $ can be found by the approximate equation of Cooper (1969) [SETP; S.16]: # " = 23.45sin 360 284 + n & % ( (2.2) $ 365 ' This angle represents the angular position of the sun at solar noon in relation to the plane of the equator. The number of days in a year is represented by n. The hour angle % sets the angular displacement of the sun east or west of the local meridian. This is done due to rotation of the earth with a value of around 15 per hour. The angle will be negative in the morning and positive in the afternoon. The latitude in equation (2.1) is mentioned with & and reveres to a certain location. The zenith angle can be understood as the incidence angle of beam radiation on a horizontal surface. Therefore equation (2.1) can be simplified to [SETP; S.16]: cos" z = (cos# cos$ cos% + sin# sin$) (2.3) The sun height " s shown in figure 4 is the complementary angle of! z. Due to this reason it can simply be stated as: " s = 90 #$ z (2.4) Tracking systems for the collector are important in terms of harvesting the maximum amount of energy out of the sunlight. The standard tracking system for a parabolic trough collector is a single axis tracking system. The collector orientation is normally either in a north-south or east-west direction. For a fixed east-west direction the surface azimuth is defined as [SETP; S.21]: $ " = 0 if " s < 90 % & 180 if " s # 90 (2.5) Therefore the tilt angel of the surface can be calculated by [SETP; S.21]: tan" = tan# z cos$ s (2.6) The aim of the one axis tracking system is to minimize the angle of incidence. For tracking in north south direction the angle can be evaluated according to [SETP; S.21]: cos" = (1 # cos 2 $ sin 2 %) (2.7) The same calculations can be undertaken for a fixed north-south orientation of the collector surface. Here the surface azimuth is defined by [SETP; S.22]:

! 10 % " = 90 if " s > 0 & '#90 if " s $ 0 (2.8) Consequently, the collector tilt angle is evaluated according to the following equation [SETP; S.21]: tan" = tan# z cos($ % $ s ) (2.9) According to equation (2.8) the surface azimuth angle will be 90 or -90 depending on the sign of the solar azimuth angle. The incidence angle for a plane rotating about a horizontal north-south axis, with continuous tracking, is [SETP; S.21]: cos" = (cos 2 " z + cos 2 # sin 2 $ ) (2.10) Based on the angles described in this chapter it is possible to calculate the incoming beam radiation on a collector field, with one axis tracking system for a certain location and time. 2.1.2 Sun collector geometry For the operation of a CSPP temperatures of around 400 C are necessary. These high temperatures cannot be reached with a flat plate collector. Therefore concentrating collectors are used. One type of these concentrating collectors is the parabolic trough collector. The direct normal radiation reaching the collector is concentrated on the absorber tube located in the focal point of the parabolic collector. The most important characteristic factor therefore, is the concentration ratio. It is defined as the aperture area in relation to the absorber area [SETP; S.327]: C = A a A abs (2.11) Here A a is the aperture area and A abs the area of the absorber. For a three dimensional system like a parabolic dish system with a two axis tracking system, which is focusing on one point, the maximum concentration ratio is around 45,000. However, the parabolic trough is a two dimensional system, where a maximum concentration ratio of 200 can be reached. The most significant losses under some circumstances occurring in a solar field are the shading losses. This reduction is happening when one collector row reflects their shadow onto the next row. In well-designed CSP plants this effect only shows up in the morning or evening hours with shadowing due to low tilt angle. However at these times shading losses can reach a maximum of up to 100%, where the rest of the day

! 11 these losses are close to zero. The part of the collector area that is not in the shaded region, can be approximately calculated according to [DLS; S.10]: & )& ( + H sin$ sin% ) ( D + D " SL =1# ( 1# + ( tan$ ( & H cos$ + cos% 1# s + ) + ( & r ( ( sin$ + + H cos$ + cos% ) + r ( ( sin$ + L+ ' ' cos$ s **' ' cos$ s * * (2.12) If the result of the equation above is smaller or equal to zero the collector is completely in the shadow. If the result is greater or equal to one, no shading losses are occurring on the collector. It is obvious that the shading effect is dependent on the collector size and the distance between the collector rows. In equation (2.12) D represents the distance between the collector rows in meters, H the height of the collector in meters and L the length of a collector row also in meters. The relative azimuth angle # r can be calculated as the absolute value of the different between the solar and the surface azimuth angel. Another loss factor occurs because, the incoming radiation to the collector is not exact perpendicular and the absorber tube has a finite length. At the end of each collector a certain part of the absorber tube will be not be irradiated. This displacement of the sun image is shown in the following figure Figure 5: Displacement of the sun image [SLP]. In the northern hemisphere these effects can be noticed especially during wintertime. Normally these losses are under two percent in feasible areas for CSP plants. Nevertheless, the reduction factor can be calculated according to [SLP; S.22]: l = f tan" sin (# $ # s ) (2.13) In formula (2.13) l represents the length of the not irradiated part, displayed in figure 5. The focal length f depends on the collector design. Consequently, the losses in percentage can be calculated as:

! 12 " EV =1# l L (2.14) Further losses that occur depend on the finite earth sun distance where the beams reaching the collectors are also not exactly parallel. As a result the sun image on the absorbers is not precisely circular. The image can be seen in the form of an ellipse, which changes the frame, depending on the angle of incidence. Only for! is zero the image will be circular. By increasing the incidence angle the performance characteristic of the sun image becomes worse. This happens because the absorber is designed for a perfect circular sun image. This effect is called incidence angle modifier and can be predicted according to Marco (1995) as follows [SLP; S.23]: " IAM = cos#(1+ sin 3 #) (2.15) From formula (2.15) we can observe that the losses occurring are negligible. Nevertheless, they do increase with distance from the equator. Finally the irradiation reaching the collector can be separated in two parts. One is exactly perpendicular the other one is horizontal to the collector. The collector however only can reflect the perpendicular part of the radiation. This leads to the socalled cosin-effect. Here the amount of useful irradiation can be calculated by [DLS; S.9]: " COS = cos# (2.16) Finally the amount of energy, which can be reflected per square meter of collector area to the absorber, can be calculated [DLS; S.10]: IDR = " COS " IAM " SL " EV (2.17) IDR stand for Incident Direct Radiation and represents the useful part of energy provided under ideal conditions to the absorber tube. 2.2 Solar field Inside of the absorber tube a heat transport medium, mainly synthetic oil, is used to collect the thermal energy and transport it to heat exchangers for producing steam in order to operate a steam turbine. During the transportion of the oil thermal losses can be recognized, and also further losses during the concentration process have to be taken into account. 2.2.1 Optical losses Energy losses not only occur because of geometrical reasons like shading or unirradiated absorber parts, but also from the material properties of the mirror, the

! 13 hull pipe, and the absorber. Therefore, some operating figures will be defined. They are all dependent on the design quality of the collector elements being used [DLS; S.12]: Reflectivity of the mirror: A small part of the incidence radiation is not reflected to the absorber, because the reflectivity of the mirror being finite. The mirror absorbs a part of the incoming radiation. As an average the reflection coefficient of a mirror used in solar thermal abdications can be set to '=0.93 [DLS; S.12]. Contamination of the mirror: The part of irradiation, which is absorbed by the mirror, is increasing with ongoing contamination of the mirror surface. Taking into consideration frequently washing procedure the contamination factor can be estimated with $=0.98 [DLS; S.12]. Transmission factor of the mirror: The glass on the top of the mirror also partly absorbs the irradiation. The irradiation has to pass the glass cover two times, which leads to a transmission coefficient of around ( s =0.99 [DLS; S.12]. Quality factor of the mirror: The quality factors depending on the production processes as well as the erection on site. For example the absorber tube is not exactly mounted in the focal point of the mirror additional losses will occur. Also different focal length of the mirror plats will lead to additional losses. Nowadays a quality factor of )=0.90 is assumed. [DLS; S.12]. Transmission factor of the hull pipe: A small part of the reflected irradiation is again reflected by the glass pipe, which surrounds the absorber tube. This transmission coefficient can be set here to ( H =0.95 [DLS; S.12]. Absorption factor of the absorber pipe: At the absorber tube not all of the reflected radiation will be absorbed. Due to physical conditions a part of the radiation will always be reflected. The absorption factor can be estimated with #=0.95 [DLS; S.12]. All the quality and material factors mentioned above are factored for an LS-3 collector type. Taking into consideration all the additional factors above it is finally possible to calculate the amount of energy received per square meter absorber pipe:

! 14 G A = IDR"#$% H % 2 S = IDR& opt (2.18) As displayed in equation (2.18) the combination of all quality factors can be summarized as the optical efficiency of the mirror * opt. In the formula above G A represents the irradiation reaching the absorber pipe per square meter. Now the part of how the transformation into thermal energy and consequently, the losses occurring during transportation in the absorber pipe is described. 2.2.2 Heat losses Until now it has been described how the energy of the direct solar radiation reaches the surface of the absorber. Here the form of energy is now changed into thermal energy. This can be described by constructing an energy balance equation like [SLP; S.23]: G A A" = Q n+ Q (2.19) losses Here A is the absorber area and + is the absorbance of the absorber. Q n identifies the useful thermal energy collected in the thermal oil and the thermal losses are combined in Q Losses. Physically there are three different types of heat transportation, occurring naturally. These are heat transport according to convention, conduction and radiation. However these phenomena are the reason for the thermal losses in the solar field. The losses depending on conduction and convection can be set in a first approach in a linear relation with the difference between the average absorber temperature and the ambient temperature [SLP; S.24]: Q CC = U CC A(T a " T U ) (2.20) U CC is the heat transfer coefficient, which is adjusted by measurement results. The average absorber temperature is labeled T a, and T u is the ambient temperature. In addition to the convection and conduction losses, the thermal radiation losses must be taken into consideration. Therefore the heat flux between two surfaces by thermal radiation is described as [VDW; S.255]: Q "(T 4 rad = 2 # T 4 1 ) 1 #$ 1 + 1 + 1 #$ 2 $ 1 A 1 A 1 F 1,2 $ 2 A 2 (2.21) T represents the temperature for each of the two surfaces, F is the area angle between the surfaces, A is the area of the surfaces, + is the Stefan Boltzmann constant, and, is the emission coefficient for each surfaces. In case of a solar field it

! 15 can be assumed that the absorber area is relatively small compared to the ambient area. These allows to simplify equation (2.21) to [SLP; S.24]: Q rad = " 1 A 1 #(T 2 4 $ T 1 4 ) (2.22) In the formula above the emissivity of the environment is considered with one. A 1 displays the area of the absorber. Now the energy used for heating up of the collector, for example in the morning hours can be calculated by [SLP; S.25]: Q C = AU C (T c " T a ) (2.23) U C shows the heat transfer coefficient of the collector and Tc is the collector temperature. Putting all these thermal losses together and combining it with formula (2.19) the amount of thermal energy produced by the solar field can be calculated out of : Q n = G"A # AU CC (T c # T a ) #$% c A(T 4 c # T 4 a ) # AU c (T c # T a ) (2.24) Consequently, the amount of energy collected in the absorber must be transported either into storage or to the power block. Here additional losses in the absorber pipe will be occurring. Finally these piping losses can be calculated according to [SLP; S.33]: Q P = U P (T F " T a ) (2.25) Now all necessary relations for the description of how the energy provided by the sun is harvested and distributed either into storage or to the power block are described. The point of interest is now the storage and power block arrangements. 2.3 Thermal Storage There are two general types of thermal storage mechanisms. The first one is based upon the use of sensible heat in various forms of solid and/or liquid materials. The other storage type involves the latent heat of phase change reactions. Sensible heat is added to a material simply by heating it up. Generally all energy that is involved in changing the temperature of a medium is called sensible heat, and it amounts simply to the product of the specific heat and the temperature change. [ES; S.22]: q = "c p V#T (2.26) In equation (2.26) ' is the density of the storage material, c p the specific heat capacity at constant pressure of the storage material and V the volume of the

! 16 storage. A different mechanism for storing thermal energy involves a phase transition with no change in the chemical composition of the storage material. A simple example therefore is water. At low temperatures under 0 C it is solid, at temperatures between 1 C and 99 C it becomes liquid and at temperatures above 100 C it converts into gas. Thus, it can undergo two transition phases, with associated changes in entropy and enthalpy. The Gibbs free energy, or chemical potential, for the two phases are in equilibrium with each other at the transition temperature. Therefore it can be evaluated according to [ES; S.23]"! "G = "H # T"S = 0 (2.27) At the temperature, the change in heat content -H at the transition temperature is equal to T-S. The slope of the temperature dependence of the Gibbs free energy and is proportional to the negative value of the entropy, which is different in diverse phases and materials. Considering that state-of-the-art sensible heat storages are used in CSP plants, here two different ways of calculating this kind of thermal storage will be described. 2.3.1 Stratification Storage the Multi-node-model Liquid storage tanks are operated at a significant degree of stratification. The degree of stratification in real operation is strongly dependent on the design of the tank and its location. In the multi-node model heat storage is modeled and divided into N nodes in order to display the stratification layers. In order to formulate the necessary equations, it will be appropriate to set some assumptions about how the liquid will be entering the tank and how the distribution to the several nodes takes place. The density of the liquid is dependent on the temperature. Therefore we can assume that the storage material will find its way to the node with the same temperature as the liquid. As it is displayed in figure 6 the mass flow coming from the collector field m c finds its way to a node according to its temperature between T s,1 and T s,3. The same physical effect can be discovered when the liquid enters the tank upstream.

! 17 Figure 6: Three-node stratification liquid storage tank [SETP]. In the three node-model displayed in graph 6 the nodes are counted from the top to the bottom. The liquid flow inside of the tank however, can run from the bottom to the top or the other way around depending on the thermodynamic conditions of the different layers. A collector control function F c i can be defined in order to determine, which layer receives the thermal energy coming from the solar field. [SETP; S.384]: $ 1 if i =1 and T c,0 > T s,1 & 1 if T F C s,i"1 # T c,0 > T s,i i = % & 0 if i = 0 or if i = N +1 '& 0 otherwise (2.28) During the operation of the solar field the control functions can be non-zero only for one node. The liquid returning from the power block can be controlled in the same way like the mass flow coming from the collector field. Therefore the control function F L i is established [SETP; S.384]: $ 1 if i = N and T L,r < T s,n & 1 if T F L s,i"1 # T L,r > T s,i i = % & 0 if i = 0 or if i = N +1 '& 0 otherwise (2.29) Equation (2.29) follows the same assumption as for equation (2.28). Therefore only one control function can be set 1 during an operation. The net flow between the nodes can now be either up or down. This depends on the magnitudes of the load flow rates and the values represented in the two control functions (2.28) and (2.29) at any particular instance. It is appropriate to define a mixed flow rate in order to describe the net flow rate into node I from node i-1 in the

! 18 multi-node model. The equations for calculating these flow rates can be presented as [SETP; S.385]: m m,1 = 0 (2.30) i"1 N m m,i = m c F c L # j " m L # F j (2.31) j =1 j =i+1 m m.n +1 = 0 (2.32) By using the control functions (2.28 and (2.29) it is now possible to describe the energy balance for each node i. This function can be expressed as [SETP; S.385]: m i dt s,i dt " = UA % $ # c ' p & (T a ( T s,i ) + F c i m (T c,0 ( T s,i ) + F L i m c(t L,r ( T s,i ) # % + m m,i(t s,i"1 " T s,i ) if m m,i > 0 $ m m,i+1(t s,i " T s,i+1 ) if m &% m,i+1 < 0 (2.33) Here the first term representing the heat losses to the to the environment with the temperature T a. By increasing the number of nodes the model can be used for describing processes in highly stratified tanks. Therefore equation (2.33) can be seen as the basic description of the energy balance. Furthermore the model allows adding a heating system in one or more nodes or going into more details for the description of the thermal losses. For solving equation (2.33) numerical integration can be performed by techniques like the explicit Euler or the Runge-Kutta methods. However, for this type of storage model very little experimental evidence for supporting the results are available. Nevertheless, the assumptions are all based on physical facts. 2.3.2 The Plug-Flow Model The plug flow model is another possibility for describing a stratification tank. Here different layers, flowing around in the tank are the focus of interest. When the heat transport fluid is entering the tank a new layer is simply added to the model. On the other hand when liquid is leaving the tank a layer will be removed from the model. The size of each segment varies depending on the flow rate and the time increments set for the calculation. Figure 7 presents a heat storage tank according to the plug and flow model with four layers. On the x-axis the temperature will be marked, while the y-axis represents the

! 19 height of the storage. V i presents the volume of each storage layer. Figure 7: Plug and Flow model whit 4 layers [SLP]. Here, losses on the surface of the tank occurring. The temperature change of each layer i can be detected with the following equation [SLP; S.30]: m i dt s,i dt " = UA % $ # c ' (T a ( T s,i ) (2.34) p & i The definition m i represents the mass at each layer, T s,i the Temperature of the layer, U represents the heat transfer coefficient, A is the outer area of the storage and T a stands for the ambient temperature. Both models have their advantages and disadvantages. The decision which model is used in the simulation has to be determined by considering the required accuracy, the available data sets, and the available computational capacity. 2.4 Power Block The heart of the CSP plant is the power block: here the thermal energy delivered either from storage or from the solar field is transferred into electrical energy. In previous chapters, the process has been described for how to harvest and transfer solar energy, as well as the transport and storage of thermal energy. Here a closer look of how the thermal energy is transformed into electrical energy is undertaken. For CSP plants, well-established techniques are used. These techniques are depending theoretically on the Clausius-Rankine cycle. 2.4.1 Carnot Cycle Cycle processes play a very important roll in the mean of transferring thermal energy into mechanical energy. The ideal thermodynamic cycle is the Carnot process. It has

! 20 the maximum thermodynamic efficiency, the Carnot efficiency * c. This is because all changers of states are reversible. In practical applications this efficiency can never be reached but is used for indicating the thermodynamic quality of a real process. Figure 8 shows the Carnot cycle in a p,v and a T,S diagram it can be seen that all parts of the cycle are reversible. Figure 8: Carnot cycle p,v and T,s Diagram [TDK]. As it is displayed in the graph above, the process consists of two isothermal and two reversible adiabatic changes of state. The isothermal expansion form state (1) takes place by adding thermal energy Q 12 and performing work W 12. A reversible adiabatic expansion takes place from state (2) to state (3). The isothermal compression between states (3) and (4) releases the thermal energy Q 34 and must be supplied by work W 34. Lastly, by reversible adiabatic compression the working gas is brought back to state (1). By using the first law of thermodynamics the process can be described as [TDK; S.66]: # du = # "W + # "Q (2.35) Taking into consideration, that the change of energy in a closed cycle is zero, the formula (2.35) can be simplified to [TDK; S.66]: $ "W = # $ "Q (2.36) Equation (2.36) shows that the sum of all added and removed work is equal to zero. Therefore, it can be defined as [TDK; S.66]: W 1,2 +W 2,3 +W 3,4 +W 4,1 + Q 1,2 + Q 3,4 = 0 (2.37) From equation (2.37) the Carnot efficiency can be explained only by the added and removed thermal energy in the cycle [TDK; S.66]:

! 21 " th =1# (#Q 3,4) Q 1,2 =1# T 3 T 1 (2.38) By using the entropy difference the thermodynamic efficiency can be also descript as the relation of temperature T 3 and T 1. This efficiency finally can be used for evaluating the quality of any other circular process. 2.4.2 Clausius-Rankine Cycle In the CSP plants the water steam cycle is operated on the basic concept of the ideal Clausius-Rankine cycle. The primary working medium here is water/steam.!"#$%&#'($)('*+%*+,-'./%.#$)01(#$$ Figure 9 represents all essential parts of a power block, and the T,s diagram for the related Clausius-Rankine cycle. Figure 9: Schematic drawing of a steam power plant and the T,s Diagram of a Clausius- Rankine cycle [TDK]. The cycle starts with a reversible adiabatic compression (1.2) by the feed water pump (P), the compression takes place in the liquid phase. Followed by an isobaric heat addition (2.3) in the evaporator. The heat source in case of a CSPP is the thermal energy provided by the solar field or the thermal storage. The heat switch over takes place in several heat exchangers (KE). When reaching the evaporation line (3) an isobaric isothermal heating up of the steam is taking place (3.4). Followed by another isobaric heat adding (4.5) in the superheated (Ü). After superheating the steam a reversible adiabatic expansion (5.6) takes place in the steam turbine (T), where the thermal energy is transformed into mechanical energy and again transformed by the generator (G) into electrical energy. Finally the water steam mix is isobaric condensates (6.1) in the condenser (Ko) until it reaches the

! 22 evaporation line (1). Bringing the thermal energy into the cycle between point (2) and (5) can be described as [TFI; S.188]: Q add = m (h 5 " h 2 ) (2.39) Where m represents the mass flow of the water/steam, and h 5 and h 2 are the specific enthalpy values at point (5) and (2) out of graph 8. The enthalpy values can be either taken from the water steam table or form the h,s diagram. The transferred energy in the turbine (5.6) can be calculated by [TFI; S.187]: P = m (h 6 " h 5 ) (2.40) In the formula above m stand for the mass flow of the steam, h 6 and h 5 are the enthalpy values in state (5) and (6). In the condenser the thermal energy of the exhaust steam will be removed according to the function [TFI; S.188]: Q 0 = m (h 1 " h 6 ) (2.41) The energy consumption of the feed water pump can be calculated by the water/steam mass flow m together with the specific volume of the condensate and the pressure increase between (1.2) [TFI; S.189]: P P = m v 1 (p 2 " p 1 ) (2.42) In formula (2.42) P p stands for the power consumption of the feed water pump, and v 1 is the specific volume of the condensate. From equations (2.30) through (2.42) it is possible to calculate the thermodynamic efficiency of the ideal Clausius-Rankine cycle. In most of the cases the power used by the feed water pump compared to the energy generated in the steam turbine is very small. Therefore this power is not taken into consideration when calculating the thermal efficiency [TDK; S.120]: " th = P Q =1# Q 0 Q (2.43) It becomes obvious that even in a reversible process cycle the thermal efficiency in the Clausius-Rankine cycle is lower the efficiency in the Carnot cycle (equation 2.38).!""#$#"%&'(#)*+,-.#)/0)%1,1#)&-)1+#)2(,3%&3%45,-6&-#)*7*(#!) So far all processes in the cycle were considered as being reversible, but in reality these processes are irreversible. The change of state in the working medium (water/steam) in the feed water pump, the steam generator, the turbine and the condenser are irreversible. This leads to differences in the operation points. Figure

! 23 10 displays this difference between a reversible and an irreversible cycle. Figure 10: Real Celsius Rankine Process T,s Diagram [KWT]. The dotted line presents the real process, where the continuous line shows the ideal process. The main differences, which can be acknowledged in figure 10, are: a: irreversible compression in the feed water pump b: pressure losses in the steam generator c: irreversible expansion in the steam turbine d: pressure losses in the condenser The increase of pressure in the feed water pump is irreversible. Because of this an increase of the entropy can be noticed. The change of state taking place in reality is not from (1.2) it is from (1.2 r ). Here the pressure increase from p 2 to p 2r is the same as the pressure reduction of the working medium in the steam generator, caused by friction in the pipe. In modern steam plants the parasitic load of the feed water pump can be assumed with around 2%-3% of the generated power. The irreversible behavior in the steam generator is mainly dependent on the temperature difference between the working medium and the heat transport fluid from the solar field. Furthermore, the movement of the working medium through the steam generator leads to a pressure drop. The lost energy of the working medium h 2r -h 2 has to be added also by the feed water pump. Depending on the irreversible expansion in the steam turbine only state (4 r ) is reached at the outlet of the steam turbine. The additional heat, which has to be removed in the condenser, logically is increasing according to: " Q = h 4 r # h 4 (2.44) Consequently, the efficiency of the real Clausius-Rankine cycle in reduced by [TFI; S.192]:

! 24 "# th = " Q Q add (2.45) Finally the irreversible behavior in the condenser is dependent on the technical necessary temperature difference and the pressure loss because of friction in the piping system.!89"/$&-.)/0)1+#)clausius-rankine)*7*(#)#00&*&#-*7)) The first possibility to increase the efficiency of the water steam cycle is increasing the temperature and pressure of the working medium before entering the turbine. Figure 11 shows the thermodynamic consequence of this procedure in a T,sdiagram. Figure 11. Increasing of the main steam parameters [KWT]. For the processes 1234 and 12 3 4 the amount of energy removed by the condenser is equal, but the yield energy in the process 12 3 4 is higher than in the process 1234. This is displayed in figure 11 by the greater area covered by the process 12 3 4. For the process 12 3 4 the removed heat in the condenser is less compared to the other process cycles in the graph above. The temperature at point (3 ) is mainly limited due to material reasons. However for solar thermal power plants this procedure is less interesting because of the temperature which can be reached by the collectors is around 400 C nowadays, where in conventional plants the steam temperature is already at around 600 C. Furthermore, a limit in the scenario shown above is the amount of wet steam allowed in the last stages of the steam turbine. Due to erosion on the steam turbine blades, the amount of wet steam must be reduced as much as possible. Another optimization of the Clausius-Rankine cycle is possible, the so-called reheating. This means that the working medium is reheated after a partial expansion

! 25 in the high-pressure stage of a steam turbine. The following T,s-diagram shows the thermodynamic consequences of this procedure. Figure 12: Water steam cycle with reheating [KWT]. The process (3.4) presents the expansion in the high-pressure stage of the steam turbine (4.5) is the reheating process in the steam generator and (5.6) is again the expansion in the intermediate and low-pressure stage of the turbine. The thermodynamic efficiency, not including the feed water pump, can now be defined as [KWT; S.70]: " th =1# T 6 (s 6 # s 1 ) h 3 # h 1 + h 5 # h 4 (2.46) Equation (2.46) shows that for increasing the thermal efficiency by using a reheat system the average temperature of the added heat in the reheater has to be higher than the average temperature of the cycle without a reheating system. The average temperature of the basic process can be calculated as [KWT; S.70]: T zu,1 = h 3 " h 1 s 3 " s 1 (2.47) Where the average temperature for the reheating process is calculated [KWT; S.71]: T zu,2 = h 5 " h 4 s 5 " s 4 # 0.5(T 4 + T 5 ) (2.48) In normal operation conditions the temperature for the reheating system is equal to the temperature of the main steam temperature (T 5 =T 3 ). Therefore, the efficiency will be increased if T zu2 >=T zu1. This lead to [KWT; S.72]: T 4 > 2T zu,1 " T 5 (2.49) Another advantage of the reheating stage can be seen in the reduced amount of wet steam in the low-pressure part of the steam turbine.

! 26 Increasing the cycle efficiency is also possible by adding a preheating system. It heats up the feed water above the condensation temperature, and therefore the efficiency of the Clausius-Rankine cycle gets closer to the efficiency of a Carnot cycle. In practical use the steam for heating up the feed water comes from several stages of the steam turbine. Figure 13: Regenerative feed water preheating [KWT]. Figure 13 gives a schematic overview of the preheating process. It is clearly shown that the heat is shifted from b to a by using the preheating system. In order to avoid evaporation effects in the feed water the water is pressurized. The thermal efficiency of the water steam cycle can now be calculated as [SLP; S.33]: " th =1# h # h ' 3 1 (2.50) T 1 (s 4 # s 1 ) Increase in efficiency mainly depends on the reduction of the mass flow through the condenser. This results in a reduction of the condenser losses. Another conclusion of the preheating system is that the steam mass flow to the highpressure part of the steam turbine is increasing where the steam flow in the low pressure and intermediate pressure part is decreasing. This leads, depending on sealing losses in the turbine, to an increase of the internal turbine efficiency. 2.4.3 Steam Turbine A steam turbine is an axial turbo engine, in which the thermal energy stored in the steam is converted into mechanical, mainly rotational energy. For the transition of the enthalpy into kinetic energy the working medium is seeded up in the nozzles. These nozzles are composed by the outlines of the guidance wheels. After this a switch of flow direction of the working medium takes place by using the rotating wheels. As a reaction of the impulse forces occurring now on the wheels, a torque is created and transferred to the turbine shaft. The set of a guidance wheel and a rotational wheel is

! 27 called turbine stage. Figure 14 gives an overview of the construction concept of a modern steam turbine. Figure 14. Schematic drawing of a high-pressure turbine [KWT]. The construction schema above shows the inner cover a, the outer cover b, the labyrinth sealing c and the turbine shaft d. Furthermore, figure 14 is giving an overview of how the thermal energy stored in the steam is transferred into the kinetic energy of the shaft. Simplified the power produced by a steam turbine can be calculated out of the enthalpy drop over the turbine, the steam mass flow, and the turbine efficiency [KWT; S.261]: P T = " T m #h (2.51) The turbine efficiency * T is depending on several losses occurring during the energy conversion in the turbine. For an easier understanding of the turbine efficiency we can calculate the efficiency like [KWT; S261]: " T = " i " mech (2.52) In formula (2.52) * i is considered as the inner efficiency. It is depending on losses occurring on the wheels, which are manly friction losses. Gap losses depending on the design of the turbine and ventilation losses are also recognized in this factor. Furthermore, losses depending on wet steam and losses occurring during the steam are leaving the turbine because of rearrangement in the flow direction. As a benchmark the inner efficiency for modern turbine can be assumed to be between 93% and 95%. The mechanical efficiency * mech includes steam losses in the labyrinth sealing as well as friction losses between shaft and bearings. Because of using modern hydraulic bearings for carrying the turbine shaft the efficiency here can be assumed whit 98% or 99%. All the assumptions above are related to the design point of the steam

! 28 turbine, which is normally at the maximum rated power of the turbine. The turbine power however is controlled by the steam mass flow. This is affecting the efficiency as well as the net power output of the generator. The relationship between the steam mass flows for the different operation modes (full load or part load) are described by the cone law of Stodola [KWT; S.261]: m T = p "T m 0 2 2 p " 0 2 # p $T 2 # p $ 0 T " 0 T "T (2.53) Here # stand for the entrance of the turbine and % stands for the turbine outlet, 0 characterizes the full load operation, where T represents part load behavior. Over the years, several possibilities like fixed pressure operation, sliding control or equivalent sliding pressure have been established for the control of the steam turbine. Based on equations (2.53), the operation mode of the steam turbine, and also the part load behavior of a CSPP can be assumed. This chapter has outlined theoretical relations used for the developed model explained in chapter 4. However not only were technical factors taken into consideration for the simulation but also a recourse assessment for CSP plants in North Africa is undertaken in the next.!.

! 29 3. Recourse Assessment for CSPP 3.1 Land Recourse Assessment Using numerous criteria such as ground structure, water bodies, slope, shifting sand, protected and/or restricted areas, forest, and agricultural covered areas allows for the detection of land resources, which would permit the placement of concentrating solar collector fields. For collecting these data sets from the DLR are used, witch finally all combined in order to yield a map of usable areas. Some of the used criteria can be seen as optional. For instance, tourist areas or agricultural areas can be transformed into potential sites for CSP plants. Other information like slope of the terrain or water availability can be understood as compulsory criteria. For example if the slope of the terrain is greater than 2.1% the placement of a CSP plant will be, considering the state of the current technologies, impossible. Table 1 shows the compulsive and the optional criteria an area must fulfill for being considered as a possible construction site for a CSP plant. [SI; S.39] Table 1: Compulsive and optional criteria for the exclusion of land used for CSP plants [SI. Exclusion Criteria Compulsive Optional Slope of Terrain >2.1% X Land Cover Sea Inland Water Forest Swamp Agriculture Rice Culture Hydrology Permanent Inland Water Non-Permanent Inlet Water Regularly Flooded Area Geomorphology Shifting Sand, Dunes Security Zone for Shifting Sands 10km X X X X X X X X X X X

! 30 Salt Pans Glaciers Security Zone for Glaciers Land Use Settlements Airport Oil or Gas Fields Mine, Quarry Protected Area Restricted Area X X X X X X X X X In this work, all criteria, whether compulsive or optional, are used for evaluating the capability of an area. Out of table 1 it can be noticed that the criteria can be summarized into five major topics, which will be described in the following subchapters. Furthermore a minimum direct irradiation as well as the existing grid system is discussed as possible site exclusion criteria. 3.1.1 Slope The digital elevation map shown in figure 15 presents the slope of the terrain. As it was mentioned previously a slope of more than 2.1% excludes a site for installation of a CSP plant. Providing elevations in a 1 x 1 km 2 resolution, the digital elevation model from GLOBE is used for determination of the slope. [SI; S41] Figure 15: Digital elevation map for determination of the slope [SI]. The figure 15 highlights a slope of greater than 2.1% in bright red. Smaller slopes are displayed in different tones from white, which shows flat terrain to dark blue, which stands for a slope of 2.1%. Beside of a few areas in Morocco and Egypt no restriction

! 31 by the slope of the terrain can be noticed in North Africa. 3.1.2 Land cover The information about the land cover is extracted from the land cover characterization (GLCC) database. This database uses the normalized difference vegetation index (NDVI). The NDVI divides the land cover into ten classes, which are comprehensive to the global ecosystem's classification. [SI; S43] Figure 16: The land cover in the Euro-Mediterranean Region [SI]. Figure 16 displays the 10 classes of land use for the EU-Mediterranean region. The area of interest in North Africa is largely covered by desert and semi desert. In relation to table 1 numerous sites can be see as accessible areas for a CSPP according to the land cover information. 3.1.3 Hydrology Data sets for rivers, lakes etc. are also extracted from the GLCC database. Figure 17: The Hydrology of the Euro-Mediterranean Region [SI]. Map 17 shows the most significant hydrological features in the Euro-Mediterranean

! 32 region. Small rivers are not taken into account simply because of the fact that shifting a plant site of a maximum of 500 meters in any direction is consider as suitable. [SI; S43] Large rivers, however, mostly near the sea-confluence are taken into account. Regions where no data sets are available are marked red. As a final point the areas prone to flooding are displayed in the satellite image. Likewise subtracted from the map above, the hydrology will not lead to a great exclusion of obtainable sites in North Africa. 3.1.4 Geomorphologic features Due to their properties and physical conditions some areas due to their soils are not suitable for erecting concentrating solar collectors. Dynamic structures like shifting sand dunes can be taken as prohibiting areas for CSP plants simply not providing a compound strong enough for the erection of the pylons needed for the CSPP. Taking into consideration the capability to provide additional safety zones around glaciers and sand dunes has to be accounted for. The movement of a sand dune can be assumed to be 200m/year. If the operation time of the CSP plant is judged with 50 years a security zone of at least 10 km around sand dunes must be maintained. The restricted area around the glaciers must be reviewed individually for each glacier and is not part of this work. Spatial information about sand dunes and salt areas are extracted from the Digital Soil Map of the World (DSMW). The spatial resolution of the map amounts to approximately 10 x 10 km 2. The DSMW includes 26 groups of soil types, 106 soil types, as well as showing some non-soil features, including the sand dunes and salt areas of interest. [SI; S44]. Figure 18: Geomorphologic exclusion criteria in the Euro-Mediterranean region [SI]. Figure 18 gives an impression of the geomorphology characteristic in the Euro- Mediterranean region. Shifting sand dunes are marked yellow. These regions are considered as exclusion areas for CSPP plants. Here for the first time a larger

! 33 amount of restricted area in the North African region can be noticed. 3.1.5 Protected areas According to the definition published by the World Conservation Union (WCU) a protected area is: An area of land and/or sea especially dedicated to the protection and maintenance of biological diversity, and of natural and associated cultural resources, and managed though legal or other effective means. [SI; S45]. Areas, which meet the universal guidelines contained in this definition, can be seen as eliminated areas for CSP plants and presented in figure 19. Figure 19: Protected areas of the Euro-Mediterranean Region [SI]. However in practice the precise purpose for witch protected areas are managed differs greatly. Therefore the WCU has defined a series of six protected area management categories, listed in the legend of map 19. The six area types are strict nature reserved area, wilderness area, national parks, natural monuments, habitat/species management areas, protected land/seascape, and managed resource protected areas [SI; S.46]. The restrictions for erecting the CSP plants in North Africa therefore will cause it to be limited to a few sites in Morocco, Algeria, Libya and Egypt. 3.1.6 Industry and Population Figure 20 displays data about industry and highly populated places. This map also includes oil or gas fields, mines, quarry airports and desalination plants. The data set is based on the Digital Chart of the World ASCII.

! 34 Figure 20: Industry and population of the Euro-Mediterranean region [SI]. Still a great restriction for available CSP sides in North Africa based on industry and population criteria s can be not noticed. [SI; S47] 3.1.7 Technical potential The data sets explained previously are used to develop maps showing possible restricted areas for concentrated solar power plants based respectively on land configuration. Another powerful criteria is the yearly average of direct normal irradiance at a certain location. This is displayed in the following EU-Mediterranean map. Figure 21: Annual direct normal irradiation in kwh/m 2 /y on non-excluded areas in the Euro- Mediterranean Region [SI]. A site is considered to have a technical potential for a CSP plant when the yearly average direct irradiation is at least 2000 kwh/m 2 /year. In general solar thermal electricity generation is also possible with a lower irradiation. Taking into consideration economical factors, CSP first start to become feasible with a higher

! 35 irradiation. However, the 2000 kwh/m 2 /year represents a suitable value for operating a CSP plant. Furthermore, the image 21 can be seen as the resultant map including all criteria mentioned previously. Based on this map it is possible to calculate the accessible area in km 2 for the countries of interest. In this thesis a closer look at five countries of North Africa, Morocco, Algeria, Libya, Tunisia and Egypt is under taken. In the following description only these countries and their potential are taken into consideration for any calculation. Table 2 presents the maximum available areas with a direct normal irradiation of greater than 2000 kwh/m 2 /year. [SI; S48] Table 2: Areas for CSP in km 2 available in the MENA countries for different DNI Classes [SI]. DNI Classes Morocco Algeria Tunisia Libya Egypt 2000-2099 6083 6237 9288 7773 206 2100-2199 5650 34142 6445 25331 1481 2200-2299 10875 29006 9864 109712 16846 2300-2399 17194 39462 19464 176659 40969 2400-2499 34348 222860 22823 152875 41347 2500-2599 30569 384570 11637 183342 44613 2600-2699 18930 428487 240 155513 98004 2700-2800+ 48074 277580 373665 354972 TOTAL 171724 1422344 79761 1184870 598439 All data sets presented in this chapter are also available on a global scale and collected be the German Aerospace Center. Figure 22: Annual direct normal irradiation on non-excluded areas in global scale [SI]. For further scenario analyses the North African parts out of map 22 are used (see chapter 4). It can be seen that sometimes areas whit even a high irradiation are not

! 36 available because of other excluding criteria s. 3.1.8 High voltage Grid In order to supply the electrical energy produced by a CSPP plant a connection point to the transmission grid must be within a certain distance. This DESERTEC scenario includes high voltage direct current transmission lines. In this work only the actual existing high voltage and extra high voltage grid systems are taken into consideration. The transmission grid not only consists of high voltage lines, but also of transformers, switchgear and other electrical equipment. If power plants must be connected to high voltage grids over a long distance enormous investment costs must be taken into account. But not only equipment like cables, transformers and switchyards are expensive: the transmission losses over long distances in AC grid systems have to be taken into consideration. With the idea to minimize the connection costs and minimizing the transmission losses, the distance from the power plant to the existing high voltage Grid in North Africa has to be very short. Therefore the existing high voltage grid like it is presented in figure 23 gives another possibility of excluding possible areas for erecting concentrated solar power plants. Figure 23: Electrical Transmission System Network of North Africa [GEMI] For a more detailed description of the national grid systems in Morocco, Algeria, Tunisia, Libya and Egypt see appendix B. The images presented in the appendix are used for the simulation explained in chapter 4.2. The existing grid offers already the possibility to export energy from North Africa to Europe. Syria as well as Morocco having connection points to European transmission systems. Figure 24 giving a schematic overview of the interconnection points between the MENA region and the EU.

! 37 Figure 24: Envisage Mediterranean interconnections with Europe [GEMI]. For this work it is assumed that the transmission capacity of the existing grid s sufficient for transporting the electrical energy produced by CSP plants between North Africa and Europe. The finale side availability is now based on the access to infrastructure, ground and land characteristic as well as a minimum availability of direct normal irradiation. 3.2 MED-CSP SCENARIO CG/HE The scenario of the MED-CSP study is used in this work for further calculations. The study analyzes the demand of electricity and water in the MENA countries until 2050. The center of attention in this work is the electrical energy production by CSP plants in the North Africa region. The water demand will be only briefly summarized and is used for computing of the cost potential for electrical production. The CG/HE scenario is based on the assumption that economic growth rate of the MENA countries is sufficiently high enough to close the gap with the USA per capital national income to 50% by 2050. In other words most of the MENA countries will reach an income per capital equivalent to most of the states in central Europe nowadays. Along with the high economic growth rate a sharp increase in the efficiency of the electrical sector is assumed. This leads to a slightly lower growth of the electrical demand in the next decades. This scenario is called in the MED-CSP study: Closing the gap, High efficiency gains (CG/HE) and is the most optimistic scenario within the MED-CSP study. Nevertheless, nowadays most of the Maghreb countries are well on this track. For example, Egypt only has to accelerate its economic growth a bit in order to reach the goals of the scenario. To make the assumption more realistic the economic growth rate in the CG/HE scenario is limited

! 38 to a maximum of 7% per year per country. [MCSP; S.71] The principal factors the CG/HE scenario is based on are explained in this chapter. 3.2.1 Growth of population The increase in the population until 2050 is projected based on the data sets World Population Prospect of the United Nations for intermediate growth published 2002. Furthermore data from the German federal statistical office are taken into account. Figure 25 displays the results of this revise for countries in North Africa.. Figure 25: Population growth in North Africa by countries [MCSP]. The population in the North African countries will be increased to 250 million people in the year 2050. As it is shown in picture 25 the major share with a bit more than 50% of the population is in Egypt. Also the ratio between rural and urban population is taken into consideration. Here Egypt has the highest amount of people living in rural areas compared to Libya where the ratio of rural population to urban population is very low. For all other North African countries the ratio is close to the MENA average with 0.3 in the year 2050. [MCSP; S.76] 3.2.2 Growth of Economy The MED-CSP study uses the gross national income (GNI) in purchasing power parity (PPA) US$-2001 per capital as an indicator for the economic growth of a country. Nevertheless the gross domestic product gives a better overview of the economic development within a country, where the GNI can be better used to indicate the income of the population. However, there is a close link between these two factors. As a benchmark for the calculated scenarios until 2050, the average annual growth rate of 1.2% of the gross domestic product (GDP) per capital (in 1995 US$(PPA)) for the US and Canada (2000-2030) according to the reference scenario

! 39 of the IEA (2002) is used. The restriction of the annual growth rate within 7% GDP in the CG scenario seems to be an upper boundary on long-term growth streaming possibility. All societies in North Africa should be able to master these scenario goals. The countries in North Africa also have differences in groups according to the income of the population. The first group at this time is the middle-income countries, containing Morocco, Tunisia and Egypt. The second group of interest is Countries that depend to a great extent on export of energy resources ; here Algeria and Libya are the two countries in North Africa. For the countries in the first group the GDP growth rate calculated in the MED-CSP study seems to be very optimistic, where the growth rate of the second group is mainly dependent on the world energy market. The optimistic assumption in the scenario can be seen in the fact that the GDP/capital in 2050 for a country in the second group will be well above the current GDP/capital of the USA. For countries in the first group the GDP/capita will be between the current level of France and the USA in 2050 [MEDC; S.83]. However the CG scenario has been chosen in order to show the maximum potential for CSP electrical generation in these countries. Figure 26: Average GDP growth rate between 2003 and 2050 for North African countries [MCSP]. Finally, graphic 26 shows the average GDP growth rate for countries in North Africa. Based on the economic growth the installed capacity respective to the electrical demand for each country can be evaluated. 3.2.3 Electricity Demand The growth of the economy nowadays is directly linked to a growth in the electrical demand. Furthermore, with a growth in the population a growth of the overall electrical demand can be assumed. The MED-CSP study calculates the electricity

! 40 demand until 2050 from the factors of GDP growth and the increase of population. [MEDC; S.93]. Figure 27: Energy consumption in North Africa until 2050 [MCSP]. Figures 27 plot the final results out of the analysis, in the MED-CSP study, for North Africa. The figure shows the power consumption per capita per year based on the CG/HE scenario. 3.2.4 Scenario for Energy Security Based on the data set explained in the previous chapters, a qualitative annex scenario for different renewable energy systems is developed in the MED-CSP study. The scenario is specific for each country, taking into account the differences in demand and supplies structures and has a time frame until 2050. The scenarios developed in the MED-CSP study highlights methods to bring demand and production together in an economical, ecological and social way. A cost optimizing calculation for the different kinds of technologies in a conventional way is not included, because of the difficulties to do so for an outlook of the next 40 years. The following parameters are taken into account as boundary conditions, for estimating the potential of the renewable energy sources in North Africa. First the availability of renewable energy resources, like wind, solar irradiation or biomass. The maximum growth rate for the production capacity of each renewable energy technology is considered. As mentioned above the population growth together with the economic growth of each country has a significant influence on the distribution between the different technologies and is considered as well. Furthermore, a scenario for the peak load and the resulted spinning reserve is evaluated. The amount of renewing/repowering of existing power plant capacities, within the financial possibilities and the cost of electric power production with renewable technologies

! 41 compared to conventional techniques is also part of the scenario analysis. Lastly political incentives and economical boundary conditions as well as the grid infrastructure and additional cost for establishing the necessary infrastructure are taken into account. All of these parameters cannot be seen as fixed values, they have to be understood in the scenario analysis as dynamic factors. [MEDC; S111] The factors collectively lead in the end to an assumption of how the electrical energy production in 2050 can look like, separated by different types of technology. Figure 28 presents these for all countries of the MENA region together. Figure 28: Electricity production in the MENA region until 2050 [MCSP]. It can be noticed that until 2030 oil and gas are playing an important role in the energy mix of the MENA countries. But after the depletion of these natural resources the importance of the conventional technologies is decreasing. At the same time CSP plants will become more significant. One reason for the strong increase in the electrical production capacity of CSP technology is the possibility to store parts of the thermal energy and operate the CSPP with certain modification within close to 8000 base load hours a year. Consequently, CSP can be used for the purposes of grid stability operating as a base load plant as well as peak load plant and shows a lot of the flexibility conventional plants are offering nowadays. Out of the results presented in the figure above the tentative CSP potential for 2050 is estimated. The MED-CSP study also includes scenario analysis for water demand in the countries of the MENA region. This part is only considered, in the thesis, as not available potential at the cost for CSP electrical production. Due to the fact that the MED-CSP study sees CSP plants also as an excellent possibility for sea water desalination plants. Finally the data sets, which get derived out of the CG/HE scenario, are summarized in the following table for the five countries of North Africa.

! 42 Table 3: Concentrated solar thermal potential in North Africa [MCSP]. Country Tentative CSP 2050 [TWh/a] Coastal potential [TWh/a] Water demand 2050 [TWh/a] Morocco 150 300 1.2 Algeria 165 57 2.8 Tunisia 43 352 1 Libya 22 498 25 Egypt 395 496 265 Only for Egypt the coast potential for electrical production is limited. The data summarized in Table 3 are used as input parameters for the simulation program presented in chapter four. 3.3 Weather data Further integral parts of the simulation program are the weather data sets. In this thesis the information is taken from the COSMO-EU model provided by the German meteorological service. The irradiation data is calculated out of the yearly time series data with an hourly resolution and a spatial resolution of around 7 x7km 2. But not only radiation data but also ambient condition information like ambient temperature, ambient pressure, wind speed and its direction are essential for the performance calculation of CSP plants. This information is also extracted out of the COSMO-EU database with the same resolution used for the irradiation data. The numerical weather prediction model covers the area of Europe and parts of North Africa. The following image shows that a complete area of North Africa is not part of the database. The limit of the map respective to the model in geo coordinates are for the left corner bottom 27.20 N 9.14 W for the right corner top 65.58 N 34.24 W, the right corner bottom is 26.12 N 34.24 E and the right side top is 62.40 N 63.47 E. [COE; S22] For all calculations about CSPP plants in North Africa the two bottom points are the limitation for the area taken into consideration. All data sets for the ambient conditions and the solar radiation are completely available for the year 2007.

! 43 Figure 29: Area for the CSOMO-EU model displaying topographic height in meter [COE]. 3.3.1 Irradiation data sets The data used for calculating the direct normal irradiation is getting out of the information delivered by the COSMO-EU model. Therefore the information of the short wave radiation at ground high (ASOB_S) and the albedo (ALB_RAD) also at ground level are used. The ASOB_S value is the timely average of the short wave radiation computed by a time horizon of 1 hour. For example if ASOB_S at the time t is the actual value then the information provided by the database is calculated like [COE; S.54]: ASOB _ S = 1 TA TA " ASOB _ S(t), TA =1h (3.1) 0 Here TA represents the time horizon of computation. The other value ALB_RAD is an actual magnitude takes in consideration the type of soil, the ground humidity, the snow and crop cover. Out of these values the global horizontal irradiation for the area of interest can be evaluated by [COE; S.54]: I = ASOB _ S (1" ALB _ RAD) (3.2) I represents the global horizontal irradiation, where ALB_RAD is the actual albedo value and ASOB_S is the timely average of one hour for the short wave radiation at ground. However putting together an average value with an actual value is not best practice. Therefore the so produced values will be judged by a sensitive analysis with a data set, provided by the Helioclime-3 (HC3) file.

! 44 Figure 30: Global horizontal irradiation calculated out of the COSMO-EU data compared with global horizontal irradiation data from HC3 database. The green part in graph 30 shows the global horizontal irradiation calculated out of the COSMO-EU data ASOB_S and ALB_RAD for a location 31 N 29 E in the year 2007. The yellow part is the global horizontal radiation for the same location and year supplied by the HC3 database. Finally the red curve is the difference between the HC3 information and the data from the COSMO-EU model. At first a displacement of the data by two hours, was noticed. In order to get comparable results the time line for the COSMO-EU model was shifted 2 hours ahead. With this change the yearly global horizontal irradiation is calculated. When the HC3 data, with a value of 2,046kWh/m 2 per year, is subtracted the COSMO-EU information leads to 1,902 kwh/m 2 /per year. Here the difference is only 7.2% in relation to the HC3 value. For getting a better idea of the differences, the yearly average is evaluated with a higher irradiation of 17.6 W/m 2 in the HC3 model. Where the maximum difference is noticed with 817 W/m 2 higher irradiation in the COSMO-EU model on the other hand a day with 740 W/m 2 higher irradiation in the HC3 data are noticed. The differences are resulting in the different ways of evaluating the normal global horizontal data in the two models. For the ongoing processing the accuracy of the COSMO-EU data considered as sufficient. Out of global horizontal irradiation data sets, the direct normal irradiation used for further processing can be appointed. For that reason first the hourly clearance index is defined by [SETP; S72]: (3.3)

! 45 Where I represents the global horizontal irradiance at ground height and I 0 is the extraterrestrial radiation calculated out of the solar constant with 1,377 W/m 2. By using the hourly diffuse fraction model with correction for variability and surface albedo the relation between the global horizontal irradiance, and the diffuse part can be calculated: (3.4) By knowing the diffuse part I d of the global horizontal irradiation also the direct horizontal irradiation can be evaluated according to: I b = I " I d (3.5) Finally the direct normal irradiation (DNI) is defined by [SETP; S.24]: DNI = I b cos(" z ) (3.6) The following image displays the relation between I, I b and DNI calculated from the COSMO-EU data at the location of 31 N and 29 E for the year 2007. Figure 31: I, b and DNI for 31 N 29 E in the year 2007 simulated by using COSMO-EU data. Two cloudy days in the beginning of summer can be indentified in figure 31. The DNI values calculated in this way are used for the performance evaluation of CSPP in chapter 4.3 3.3.2 Ambient data sets The ambient data temperature, pressure, wind speed, and wind direction at ground for further calculations are also taken from the COSMO-EU model. For all these data

! 46 sets the spatial resolution is also 7 x 7km 2 and the timely resolution is one hour for the year 2007. The different data sets will be explained in the following by using the monthly averages for the data sets of Egypt. Consequently the next figure shows the monthly averages for June 2007 Figure 32: Monthly average ambient data for Egypt June 2007 provided by the COSMO-EU database. The averages shown in graph 33 above are calculated out of the hourly values provided by the weather prediction model. The resultant wind speed (W_10) is computed out of zonal patter (U_10) and the meridonal component (V_10) by using their geometrical relations. The index 10 refers to the elation height of 10 meter. As a further factor the actual wind direction is provided out of the database. The atmosphere pressure at ground height (PS) is calculated in the numerical model by extrapolation of the prediction of the overall pressure at the lowest modal pattern. At 2-meter elevation height the temperature (T_2M) information is provided. All the data sets presented in this chapter can be seen as input parameters for the simulation model presented in the next chapter. Sometimes theses data sets have to be adapted in order to fit for the following scenario and performance simulation of CSPP in North Africa. The adaptation is explained together with the structure of the simulation program in the next chapter.

! 47 4. Simulation Program 4.1 Program overview The intention of the simulation is an accurate reproduction of the thermodynamic processes occurring in any parabolic trough concentrated solar power plant located in North Africa. At the moment, not even one CSP plant, exclusively powered by solar energy is installed in this region. Consequently, predictions for the future have to be undertaken. In order to make these predictions reasonable, information about potential and available areas in the North African countries have to be emblazed and processed in a realistic way for a further performance calculation. The following flow chart gives a common overview of how this is put into execution. Figure 33: General overview of the Matlab program for CSPP calculation.

! 48 Three main program parts can be identified from the figure above. In the beginning, several classes of information about land use, weather etc. are being processed and prepared for further calculations. The second part is the computation of the thermal and resulting electrical output produced by a CSPP. In the last section it is possible to fit the CSPP operation condition to predefined load curves depending on the maximum storage capacity. Based on theses optimization results the effective load carrying capacity (ELCC) and the capacity factor (CC) are evaluated. Graph 34 gives a more comprehensive overview of the program structure, highlighting the three different parts of the program. The green part shows the CSP scenario analysis for North Africa based on the results of the MED-CSP study [MCSP] and the GIS input data. The orange part presents the final operation optimization and the blue part represents the thermodynamic performance simulation. Furthermore, the white parts in the sketch demonstrate the basic options for the program user. Figure 34: Basic program structure. The first initiation is the adjustment for a scenario simulation according to the MED- CSP study including the GIS database, or an individual performance calculation of a

! 49 CSPP at a certain location. Using the individual simulation modus the manual settings for the performance calculations are, the selection of a cooling system. The agenda offers here the three possibilities, wet cooling, evaporation cooling and dry cooling system. Furthermore, the use of a back cooling system has to be stated. Also the orientation of the solar collectors must be adjusted, by setting an angel between north-south and east-west orientation. Additionally, the storage size and the collector field area have to be adapted by the settings for solar multiple one to four. Finally, the load profile for the plants and the time horizon for the simulation have to be set as well as the geographical location. The number of geographical points set here will be the number of plants calculated by the simulation program. Table 4 summarizes these basic input parameters. Table 4: Basic input parameters for individual simulation of CSPP. Input parameters Options Input Cooling system Wet, Evaporation, Dry 1 2 3 Back cooling system Yes/no 1/0 Collector orientation Angle in rad Vector format Solar multiple From 1 to 4 1 2 3 4 Load profile 24 hours load profile Vector format Time frame Date including e.g. 12-March-2010 Location Latitude and longitude Vector format In the following program description it is assumed that using the GIS database and the input information from the MED-CSP scenario is adjusted. 4.2 GIS database and scenario processing The scenario analysis is the first automatic computation step in the simulation program. This subprogram is calculating according to input data such as the resource map explained in chapter 3 and economical outlines from the GC/HE scenario the number of plants in Morocco, Algeria, Tunisia, Libya and Egypt. In addition, available locations are evaluated and the CSP plants being allocated. If the location for a plant is set as engaged the program extracts out of the COSMO-EU database the required weather data sets. Figure 35 displays the flow diagram of the GIS based scenario simulation process (highlighted in flow chart 34 green). Due to the long computing time which is based on the high amount of processed data, it is possible to do the calculations for coast distance and grid distance only in the first time running the program and loading the

! 50 data for all available areas in the following runs. However, here all process steps will be explained. The yellow path in the following chart leads the way through these process steps. The input parameters displayed in white are explained in chapter 3. Figure 35: Basic program structure scenario analysis. Also for these computations some basic design parameters have to be adjusted manually. The setting of these input parameters is undertaken in the main script called CSPP and shown in the following table. Table 5: Input parameter for scenario analysis. Input parameters Options Input Collector orientation Angle in rad Vector format Solar multiple SM1 SM2 SM3 SM4 Number Load profile hourly load profile Vector format Year of interest Year e.g. 2015 4.2.1 Scenario Processing The first process step evaluates the installed capacity per country and its transformation in a number of CSP plants for the year of interest. The installed

! 51 capacity by CSP plants in the MENA region until 2050 is based on the MED-CSP Scenario CG/HE1 [MCSP]. The average plant size for the performance simulation is considered with 200 MW el gross capacity. By using the actual installed capacity in the years 2005 and 2010 as well as the predicted capacity in 2050, the annex curve displayed in figure 36 for the North African countries is generated. Figure 36: Number of CSPP according to MED CSP Scenario. For developing the curves the exponential curve fitting function out of Matlab statistic toolbox is used. As it is shown in figure 36 the annex scenarios are expected for all countries to be exponential. For Egypt and Morocco always one existing CSP plant in 2010 is set, which leads to a faster increase of the annex curve. In this case ISCCS are considered as pure CSP plants. Together with the year of interest, set in the input parameters, the numbers of plants for the scenario analysis is evaluated out of the annex graph. 4.2.2 Site evaluation The map for positioning the CSP plants is designed out of map 22 presented in chapter 3. The part representing North Africa was cut out and geo-referenced by using the software QGIS [QGIS]. The boarders of the countries Morocco, Algeria, Tunisia, Libya and Egypt are defined according to shape files of these countries. The finalized map was exported from QGIS to Matlab. Due to the fact that the yearly irradiation data presented in the original map is not of interest for the scenario analysis the map is transferred from a three-canal RGB picture into a gray-scaled image by using the Matlab mapping toolbox. All these procedures are resulting in the map displayed in Figure 37.

! 52 Figure 37: Black-white map for evaluating suitable areas for CSPP. The map above consists of 535*2199 pixel points, where each black point is presenting a suitable area for a CSPP. In geo-coordinates the map is stretch from longitude 16.78 W to 37.95 E and latitude 18.76 N to 39.47 N. The coastline in the figure above is designed by using the predefined Vector-layers named coast out of the Matlab mapping toolbox [MWS]. To indentify the best suitable areas for CSP plants in the next simulation steps the available areas close to the coastline get evaluated. This can be attractive because of advantages like unlimited water availability or a benefiting from existing grid systems, which are normally fully developed along the coastline in North Africa. As mentioned above the coastline in Matlab is displayed by means of a vector layer. The distance between the stating points of each vector and the available area pixel is calculated. The distance between two geographical points can be defined by dist = acos(sin(lat1)*sin(lat2) + cos(lat1) * cos(lat2) * cos(lon1 " lon2))*6370 (4.1) Here the value of 6370 is equivalent to the mean radius of the earth. The available areas with a minimum distance from the coastline are, taking into consideration that the coastline consist of connected vectors, all black points in a radius equivalent to the length of the coastline vector. All pixel in this circle are considered as available coast areas. Separated by country Morocco has 25019, Algeria 165993, Tunisia 3615, Libya 19378, and Egypt 10482 potential areas for CSP plants close to the coast. In the next step the available areas, within a certain distance to the existing high voltage grid in North Africa are selected. Many of the available areas, especially with a very high yearly average irradiation value, are located somewhere in the desert, without any connection to infrastructure for distributing the electrical energy. Due to this fact and the high investment costs for transmission grids the distance of all available areas form the existing grid system is evaluated. Based on the grid maps

! 53 presented in the appendix B a vector-layer for each country illustrating the high voltage grid is created in QGIS and exported to Matlab for the following computation. Figure 38 displays the North African map including the yearly average irradiation and the vector layers for the high voltage grid. Figure 38: Schematic drawing of high voltage grid in North Africa. The color code for the irradiation data is displayed in figure 22 in chapter 3. Here the colored map is only used to show the high number of well suitable areas without any access to the necessary infrastructure. The calculation of the distance between the electric grid and the available areas is designed in the same way like the distance evaluation for the coastal areas. In a first approach the distance between the potential CSPP areas and the electric grid is set to not more then 25 km. This leads to a theoretic potential of 36060 sides in Algeria 212433 in Egypt 341837 in Libya, 514686 in Morocco and 51090 in Tunisia. Therefore the maximum installed capacity per country is displayed in diagram 39.. Figure 39: Possible installed gross capacity per country per country. Simulated with a grid distance of 25km and 200 MW el installed gross capacity per pixel. The installed capacity shown in figure 39 is based on the assumption that on each

! 54 suitable pixel a CSPP with a gross capacity of 200 MW el is installed. 4.2.3 Side depositing In the following step, the simulation is depositing the CSPP plants for the scenario year of interest in the countries using the predefined criteria. First, the intersection between the available regions along the coastline and the areas fulfilling the criteria of minimum grid distance is determined. In the next calculation step, the installed capacity of CSP plants used for desalination according to the MED-CSP scenario CG/HE is removed from these theoretical coast potentials. In case there is sufficient potential of useable areas left, 80% of the CSP plants are distributed along the coastline. In case there are not enough available areas, like in Tunisia for the year 2050 only the amount of available areas left along the coast is used. The rest of the concentrated solar power plants are distributed around all other qualified areas taking only into consideration the minimum distance towards the high voltage grid. Therefore all plants are fulfilling the criteria, maximum distance of 25 km to the existing transmission grid and all other land use criteria s explained in chapter 3. Because of using the ambient data from the COSMO-EU model explained in chapter 3 the available areas have to be cut at the latitude of 27 N. The positioning itself is done by a random permutation according to [MFL; S.364]: " C w (n;k) = n % $ ' for(k ( n) (4.2) # k& In this formula, k are the available areas and n is the number of plants, here the assumption is that they are always more areas available per country than plants will be erected according to the scenario. Finally, a map consisting the grid the available areas and the CSP plants, is plotted for a visual control. Figure 40: Irradiation map of North Africa, CSPP distribution for scenario 2030. The pixel, where the CSPP are located are marked red. At this point of the simulation no separation between the different plants according to cooling system and other

! 55 operational conditions are undertaken, simply the geo coordinates for the plants are set. The next map gives a detailed view of the area around Cairo in the year 2030. Figure 41: Irradiation map of North Egypt, CSPP distribution for scenario 2030 (Egypt). For further calculations the distance between the plants and the coast as well as the Nile River is appointed. By knowing this distance it is possible to assign a specific cooling system for each CSP plant. This is of interest because of the significant performance influence by the cooling system. All plants along the coastline supposedly have a wet cooling system for optimal performance. At a distance of 1.5 times of the minimum coast distance an evaporation system is assumed. All other plants are simulated with dry cooling systems. The next map displays this distribution for the year 2020. Figure 42: Irradiation map of North Africa, CSPP with cooling system distribution for the scenario 2020. In picture 42 the blue marked plants are water cooled, the green ones are cooled by an evaporation cooling system and the red points represents plants equipped with dry cooling systems.

! 56 4.3 CSP Performance Simulation Based on the input data, either by scenario simulation or by manual input, the performance of all predefined CSP plants is calculated. Therefore, each performance evaluation is processed separately. After each loop the data of interest are stored for further processing. This devolution is shown in the following flow diagram. Figure 43: Basic Program structure performance calculation. The program path in the figure above is also highlighted in yellow. All steps have to be performed in order to calculate the electric power output of the CSPP. The two predefined process steps water steam table and Mollier h-x-diagram are only shortly summarized in this chapter. The extraction of the ambient data sets and the calculation for the specific DNI values at the predefined location is explained in chapter 3.3. 4.3.1 Solar field simulation Based on the input data the first program step evaluates, all necessary angles for predicting the sun position at a specific location and time. All geometrical calculations, necessary for describing the sun's position are explained in chapter 2.1

! 57 Solar geometry. In the simulation, 0 longitude refers to the Greenwich meridian. Furthermore, it is assumed that all solar fields are equipped with a one axis tracking system. For the case that the collector is erected in north-south direction the calculation of the collector slope at each hour is simplified by the definition [DLS; S.6] tan" = tan# z sin$ s (4.3) Here " represents the angle of slope. Figure 44 displays the solar height, the incidence angle and the tilt angle simulated in hourly resolution for the location 31 N 29 E at the 31 st of March 2007. The collector rows are erected in north south direction. Figure 44: Incidence angle, tilt angle and solar height at 31 N 20 E at 21 st, March 2007. It can be noticed that the tracking system first starts with sunrise and stops with sunset. The maximum allowed angle for the system is 90 in east west direction. To proceed, the design parameters for the solar field must define. These parameters are essential for predicting the thermal energy output of the solar field. The design parameters have to be set manually before starting the simulation. Values for the common LS-3 system are shown in table 4 Table 6: Design parameters for simulation of the solar field for CSP plant with gross capacity of 200 MW el. [DLS]. Parameter Value Unit Type Distance between collector row 17.3 m LS-3 Height of the collector 5.76 m LS-3 Length of the collector 94.6 m LS-3

! 58 Focal length of the collector 2.12 m LS-3 Reflection coefficient of the mirror 0.93 % LS-3 Contamination coefficient of the mirror 0.98 % LS-3 Correction coefficient of the mirror 0.90 % LS-3 Transmission coefficient of the mirror 0.99 % LS-3 Absorption coefficient of the mirror 0.95 % LS-3 Emission coefficient of the mirror 0.16 % LS-3 Heat loss coefficient of the absorber 2 W/(m 2 *K) Inlet temperature of the heat transport medium Outlet temperature of the heat transport medium 298 C LS-3 390.5 C LS-3 Concentration ratio of the collector 82 LS-3 Solar field coefficient 0.99 % LS-3 Transmission coefficient of the piping system 0.95 % LS-3 Area of the collector field (SM1) 884,450 m 2 Gross capacity 200 MW el Availability of the solar field 0.995 % Table 6 presents the design parameters for the solar field, as they are used for all further calculations in this work. The area of the collector field results from the predefined gross capacity of 200 MW el. Furthermore, the area is adjusted during processing according to the predefined solar multiple values. It can be recognized by the parameters for the temperature of the heat transfer fluid that the simulation presents a steady state operation. The transient processes like heating up during starting time and shutting down of the plant, are considered as not relevant for further analysis and mainly covered by the hourly resolution of the simulation. In the next two simulation steps, the losses in the solar field are calculated, based on the design parameters, ambient temperature, direct normal radiation and wind speed. The part Geometrical losses is focused on the calculation of losses depending on incidence angle, shading losses and the axial end losses following the description in chapter 2.2

! 59 Figure 45: Geometrical collector losses at a location of 31 N 29 E at March 21, 2007 for a one axis tracking system, collector orientation north south. Figure 45 shows that the axial ending losses as well as shading losses having nearly no effect during the daily operation. Shading losses only occurring in the morning and evening hours. The losses of the incidence angle modifier can be see as constant during the day, but these losses are always under 5%, in the areas of interest. Therefore the only significant geometrical losses are represented by the cosine losses. In the next simulation process the optical efficiency of the collectors as well as the heat losses in the solar field are simulated. In conjunction with the so called reduction factor the output of thermal energy by the solar field can be determined. Here, the ambient conditions temperature, wind speed, and wind direction for each location is taken into account. The reduction factor is calculated according to [DLS; S.13]: " = V f w f (4.4) In the formula mentioned above, f (solar field coefficient) represents the losses due to collector and piping connection, where V is the availability of the collector field. Both parameters are set as empirical values in the design parameters [DLS; S.13]. The factor f w describes the losses occurring according to the wind speed at twometer height. It can be calculated out of an empiric model like [DLS; S.13]: f w =1.125 " 0.0083v w (4.5) v w presents the wind speed at ground height, taken from the ambient data provided by the COSMO-EU database (see figure 32).

! 60 Figure 46: Efficiency and thermal energy harvest of a solar field with a size of 188,000 m 2 and a direct normal irradiation of 800 W/m 2 at 31 N 20 E for March 21st, 2007 with a one axis tracking and north-south collector orientation. The upper graph displays the overall efficiency of the solar field compared to the thermal energy harvested by a solar field with a direct normal irradiation of 800 W/m 2 throughout the day. A constant irradiation value was assumed for this presentation in order to highlight the effect occurring by the losses described above. 4.3.2 Thermal storage simulation In the following simulation part the thermal energy harvest, by the solar field is distributed between the thermal storage and the power block. This is done to follow a predefined load profile. The used design parameters for the thermal storage simulation are summarized in the following table: Table 7 : Design parameter for storage simulation [DLS] Parameter Value Unit Type Specific heat capacity 2.5 kj/(kg*k) Thermo oil Specific heat capacity 1.495 kj/(kg*k) Molted Salt Density 1899 kg/m 3 Molted Salt Height of the storage 2 m For the simulation processes the temperatures in the storages, hot and cold parts are assumed as fixed temperatures according to the operation condition of the solar field. Consequently, only energy transfer in the heat exchanger and between the hot part of the storage and environment is simulated. The schematic drawing 47 explains the storage arrangement including the basic operation parameters..

! 61 Figure 47: Schematic drawing of storage arrangement in the simulation program. It can be seen that the storage can be either in loading or in unloading operation according to the requirement of thermal energy at the power block and the energy provided by the solar field. Thermal energy transfer into the storage is assumed whenever the solar field provides more thermal energy then is needed at the power block. This means more energy is harvest by the solar field than required for electrical energy production. Consequently, the molted salt is pumped from the cold part via the heat exchanger into the hot part. The molted salt is heated up, hence the storage is in so-called loading operation. Another possibility is that the solar field cannot provide a sufficient amount of thermal energy and stored energy is available. Then the molted salt is pumped back from the hot part to the cold part, also via the heat exchanger. Here, stored thermal energy is delivered to the power block and the storage is in so-called unloading operation. If the complete storage capacity is used and there is no need for thermal energy in the power block no more energy is produced by the solar field. [PSK; S.45]. In practical operation this means that some collector rows are tracked manually out of the sun. In the simulation program thermal storage capacities are only available for a pre selection of solar multiple two or higher. The general arrangement for the SM configuration is shown in the following sketch.

! 62 Figure 48: Schematic drawing of SM arrangement, thermal storage unit is designed for 6 hours base load operation [MCSP]. According to this drawing, the area for the solar field is simulated depending on the predefinition of SM by: " A design for SM1 $ 2A design for SM2 A solarfield = # $ 3A design for SM 3 % $ 4 A design for SM 4 (4.6) After setting the thermal energy harvest by the solar field the energy transfer during loading operation is calculated according to the energy balance equation: " HC (m salt c p,salt #T) = m Oil c p,oil #T (4.7) Where the energy transfer from the storage to the power block can be simulated by: " HC (m Oil c p,oil #T) = m sat c p,salt #T (4.8) Here the efficiency for the heat exchanger * HC is assumed with 0.98%. This high efficiency is based on two different mass flows in the heat exchanger respective a flow controlled pump in the storage arrangement. All other parameters are predefined as input values or calculated as part of the solar field simulation. In addition the losses occurring in the hot part of the storage are defined in the simulation program as [VDIW; S.207] Q losses = "#A dt (4.9) dx It can be recognized from formula (4.9) that the heat losses in the storage are simplified by using only the convection losses in the storage. The area A represents the storage surface. It is calculated by assuming a cylindrical storage with a certain

! 63 height, set in the input parameter. The value dx is the filling level of the hot storage part according to the process requirements and temperature difference dt. $ 6hours if solarmultiple2 q " storage = nom & 12hours if solarmultiple3 *% # powerblock & 18hours if solarmultiple4 '& 0hours if solarmultiple0 (4.10) According to the listing shown above it is obviously that the storage capacity is simulated in order to supply the power block under design conditions depending on the SM sketch with sufficient thermal energy for either 12, 18 or 24 full-load-hours a day. The annual full load hours, can be approximately calculated according to [FCS; S.29]: Flh = (2.5171DNI + 694)("0.0371SM 2 + 0.4171SM " 0.0744) (4.11) By using the input parameters for the simulation with an annual direct irradiation of 2,038 kwh/m 2 per year at a location of 31 N 29 E in 2007 the yearly full load hours for SM2 would be 3,139h. However, the simulation program return only 2,482 full load hours, which is, corresponds to a difference of -20%. This difference between the theoretical value and the results of the program can be explained by the fact that the program is using a more detailed calculation for the power block taking into consideration additional ambient parameters such as wind speed and temperature. On the other hand, the storage model is simplified compared to the stratification models descript in chapter 2.2. However, the theoretical full load hours for SM3 are 4,325 hours and for SM4 5,132 hours. The simulation is calculating for SM3 3,788 full load hours and for SM4 5,188 hours. The difference here amounts to -12.4% and +1.3%, respectively. Taking into account that the formula mentioned above must be understood as an approximation the accuracy of the simulation is considered as sufficient. The following picture shows the annual full load hours simulated for a CSPP equipped with a wet cooling system and a back cooling system (see chapter 4.4.3) at a location of 31 N and 29 E in the meteorological year 2007. The results are summarized in figure 49.

! 64 Figure 49: Full load operating hours /Simulation operating hours for solar-multiple 1-4 simulated for a CSP plant equipped with wet cooling system and back cooling system located at 31 N 29 E metrological year 2007. 4.3.3 Power block simulation The thermal energy delivered to the power block is the most significant input value for the performance calculation presented in this chapter. The theoretical basis for this simulation is the ideal Clausius-Rankine cycle. However, the efficiency in the ideal process is around 10-15% higher than it is in the irreversible process. This difference described in chapter 2.3 is compensated based on the assumption that in modern steam plants installations like pre-heaters and re-heaters are used to reduce the losses occurring by irreversible incident in a real process. This efficiency increase is expected to be in the same range as the difference in efficiency between real and ideal Clausius-Rankine cycle [DLS; S 35]. For the computation processes the following design parameters have to be set manually by the operator. Table 8: Design parameters for simulation of the power block with a gross capacity of 200MW el [DLS]. Parameter Value Unit Type Efficiency nominal 0.97 % Generator Efficiency nominal 0.71 % Pumps Efficiency nominal 0.60 % Ventilation Efficiency nominal 0.97 % Steam turbine Mass flow 40 kg/s Steam Nominal thermal load 500 MW th Steam generator Pressure 100 bar g Feed water pump Condenser pressure See wet/evaporation/dry cooling system -T Boiler 19 C Steam generator

! 65 In table 8 all values are taken from the DLR Sokrates model, only the nominal thermal load was adjusted according to an installed gross capacity of 200 MW el. The following drawing, gives a schematic overview of the water-steam cycle and how it is simulated in the program. Figure 50: Schematic drawing of water-steam cycle with basic design parameters according to simulation program. The cycle contains the four main parts, heat generator, steam turbine, condenser and feed water pump. The silver marked condenser in figure 50 and its operational performance is mainly depending on the related cooling system. The necessary thermodynamic parameters like temperatures, pressures and enthalpies are being simulated for the four stages shown figure 50. For processing the enthalpies out of the water steam table the software X Steam is used. This program, developed by Excel engineering and provided as open source offers calculation of steam and water properties based on the "International Association for Properties of Water and Steam Industrial Formulation 1997 (IAPWS IF-97). A full implementation of the IF-97 standard provides very accurate calculation of steam and water properties in ranges from 0-1000 bar and 0-2000 C. Available thermodynamic properties are the temperature, pressure, enthalpy, specific volume, density, specific entropy, specific internal energy, specific isobaric heat capacity, specific isochoric heat capacity, speed of sound, viscosity, and vapour fraction. All properties can be calculated with two input parameters like, pressure and temperature known, pressure and enthalpy known, enthalpy and entropy known and some only with pressure and density known. Therefore, X Steam can be used to calculate all necessary values in the Clausius-Rankine cycle. Furthermore, the code is speed optimized with pressure and enthalpy as inputs for dynamic simulations [EXS].

! 66 First off all enthalpy before the turbine can be evaluated out of the design parameters by the X Steam software. For a more precise simulation different types of the cooling systems and their influence on the thermodynamic efficiency of the power block can be simulated. Consequently, the temperature and in some extent the pressure in point 1 and 4 is depending on the type of cooling system, selected at the beginning of the computation. The interaction between the different system configurations and the water-steam part will be explained in the following chapter. However, the condenser itself is always assumed to be a surface condenser with a certain -T between the steam cycle and the cooling part. Furthermore, the parasitic load also strongly depends on the selected cooling system. Only the electrical load of the feed water pump, the tracking system, of the collectors, and the heat fluid pump installed in the solar field are independent of the cooling system and vary only with the load status of the power block. For the tracking system a consumption of 1 W/m 2 is assumed in the simulation. The load of the solar field pump can be calculated by [DLS; S.34]: " P Pump,SF = f $ Pump $ # Q HTM Q HTM,no % ' ' & 2 (4.12) Here Q HTM represents the actual energy flow in the heat carrier medium and Q HTMno is the energy flow in the design point. The field factor f for the solar field is set to 8.3 W/m 2 [DLS; S.16]. Finally, the electrical consumption for the feed water pump can be estimated from [DLS; S.32]: P FE = m steam(h 2 " h 1 ) # Pump (4.13) The efficiency for all aggregates during partial load operation is calculated in a first approach [DLS, S.36]: $ $ & " i = " i,nom #& m & & % % m nom ' ) ) ( 2 $ + 2& & % m m nom '' )) ) ( ) ( (4.14) In the equation above m describes the actual mass flow of the transported medium, and m nom the mass flow at design point. Further electric consumers are described separately for each cooling system.

! 67 WET COOLING SYSTEM The wet cooling system is the most efficient cooling option for a power plant in relation to the thermodynamic cycle efficiency. On the other hand the parasitic loads can become high if a back cooling system has to be used for transportation of cooling water over a long distance. Another disadvantage, especially in dry areas, like in some regions of North Africa, is the high water consumption of this condenser type. The following graph shows the design for the cooling cycle. Figure 51: Schematic drawing of the design for a wet cooling system equipped with back cooling system and water turbine. Here the operational condenser pressure is preset to 0.056 bar absolute [DLS; S.90], considering a fixed cooling water temperature as shown above. Nevertheless, for a more precise condenser temperature calculation further design values have to be set. Table 9: Design parameters for water steam calculation with a wet cooling system [DLS]. Parameter Value Unit Type Terminal temperature difference 4 C Wet cooling system -T cooling water 10 C Wet cooling system Depending on the terminal temperature difference the temperature inside the condenser must always be higher than the temperature at the cooling water outlet. This can be explained by the normal operation behaviour of a heat exchanger. Furthermore, the temperature increase of the cooling water is taken into account. Using these relations of the design parameters the temperature inside the condenser can be calculated according to [DLS; S.38]:

! 68 T 4 = T inlet + "T coolingwater + "T termitemperature (4.15) Based on the temperature inside the condenser and the pressure, the enthalpy at point 4 (figure 50) using the water steam table X Steam is calculated. Now it is possible to evaluate the necessary physical values at all other points of the water steam cycle. The electrical output of the generator is therefore the enthalpy drop over the steam turbine taking into account the generator and turbine efficiency. As it can be seen in figure 51 different kinds of pumps respective water turbine for a wet cooling system are used. Furthermore the electrical consumption depending on the load of the power block is processed. The water turbine in the back cooling system is used to provide a part of the electrical energy required by the back cooling pump and reducing in this way the parasitic loads. Therefore, the driving force is the height difference between the cooling system and the water recourse. The electrical consumption of the cooling pumps can be evaluated by applying the energy balance from Bernoulli [TMD; S.345]: $ P Pump = m water "zg + w 2 2 + "p # + "p ' V 1 & ) (4.16) % # (* Pump Considering the water turbine in the back cooling system only the part -z*g have to be chanced to (-z-1)*g*(1-* WT ) for predicting the parasitic load of the back cooling pump. Here * WT represents the water turbine efficiency [DLS, S.40]. Finally this energy together with the consumption of the feed water pumps the tracking system and the heat flow pumps represent the parasitic load in the CSPP performance simulation with wet cooling system. EVAPORATION COOLING The evaporation cooling system, compared to the wet cooling system, results in a lower thermal efficiency of the Clausius-Rankine cycle, because of the higher pressures and temperatures occurring in the condenser. Nevertheless, the advantage of this cooling system is the lower water consumption in relation the wet cooling system. In addition, its efficiency is higher in comparison to a dry cooling system. Figure 52 gives a schematic overview of the arrangement for an evaporation cooling system.

! 69 Figure 52: Schematic drawing of evaporation cooling system with air-cooled cooling tower. The operational condenser pressure for an evaporation system is assumed to be 0.087 bar absolute. Furthermore, the preset design parameters for the simulation are: Table 10: Design parameters for water steam calculation evaporation cooling system [DLS]. Parameter Value Unit Type Terminal temperature difference 4 C Evaporation cooling -T cooling water 11 C Evaporation cooling As it can be seen in figure XX, the ambient conditions temperature and pressure from the COSMO-EU model are part of the calculation. According to the concept of evaporation cooling the air in the cooling tower can catch a certain amount of water vapour depending on their temperature and pressure where the enthalpy of the air remains constant while the saturation temperature decreases. The enthalpy for the inlet air can be stated as [DLS; S.43]: h A,1 = c p,a T A,1 + x A,1 (r + c p,s T A,1 ) (4.17) Here T A,1 is the ambient temperature when entering the cooling tower, c p,a is the heat capacity of dry air, c p,s is the heat capacity depending on the water vapour, x A,1 is the water content of the air in relation to the dry air mass, and r is the evaporation enthalpy of water. The water content is evaluated in the simulation according to: x A,1 = 0.622p s,1 p " A,1 # p s,1 (4.18) In the equation above p is the ambient pressure, p s is the saturation partial pressure

! 70 of the air when entering the cooling evaporator and / is the relative humidity of the ambient air. The saturation partial pressure is evaluated by using a linear interpolation in the table for wet air based on the ambient temperature T A,1. [TMD; S.XX]. Based on the enthalpy evaluated for the inlet air the saturation temperature of the air can be appointed. The tables of wet air, used for this assignment, are based on the relation T s1 = f(h 1 ) [TMD; S.241]. It is assumed that an equilibrium between the water sprayed in the air-cooled condenser and the raising air in the cooling tower is reached at the air outlet. Therefore, the enthalpy at the outlet of the cooling tower can be defined as [DLS; S.44]: h A,2 = c p,a T A,1 + x A,1 (r + c p,s T A1 ) + c p,w "T Hx (4.19) C p,w is the specific heat capacity of the cooling water where -T Hx presents the temperature increase in the cooling water cycle. Based on this, the temperature T s,2 can be evaluated in the same way like T s,1. The final temperature at point 4 (figure XX) can now be calculated by including the design parameters from table 10 as [DLS; S.46]: T 4 = T S,1 + (T s,2 " T s,1 ) + #T Hx + #T condenser (4.20) By knowing the condenser temperature and pressure all other points of interest in the water steam cycle can be calculated and so the electric production of the plant is evaluated. By knowing the condenser temperature and pressure, all other points of interest in the water steam cycle can be calculated and thereby the electric production of the plant is evaluated. To calculate the parasitic loads, the sketch 52 highlights the two main pumps of interest for the cooling cycle. Furthermore, the cooling tower is assumed to be a forced draft tower operated by a ventilator. The power consumption of the main cooling pump is stated in equation (4.15). In the same way the electrical intake for the make up water pump is calculated. Hence, the water mass flow is equivalent to the evaporated water mass in the cooling tower. The parasitic load caused by the ventilator can be estimated from [DLS, S.45]: 3 1 P V = m A C V (4.21) " V The ventilator constant C V is calculated from the pressure losses in the cooling tower and the design air mass flow. The air mass flow is directly proportional to the relative humidity of the inlet air and the evaporation losses. Finally, the ventilator efficiency is

! 71 part of the design input parameters. By adding the consumption of all aggregates the parasitic load is simulated. DRY COOLING The dry cooling system is the cooling system with the lowest water consumption, but also with the lowest thermodynamic efficiency in relation to the water-steam cycle. Normally, dry cooling systems are used in desert or semi-desert areas where the water availability is very low and/or the costs for water are very high. The following graph gives a schematic overview of the air-cooled condenser arrangement. Figure 53: Schematic drawing of a condenser connected to a dry cooling system. As shown in figure 53, the ambient air directly cools the outlet steam of the turbine. The condenser pressure is therefore set to 0.143 bar absolute. The necessary input parameters for the further simulation in this case are displayed in the following table. Table 11: Design parameters for water steam calculation dry cooling system [DLS]. Parameter Value Unit Type Terminal temperature difference 17 C Dry cooling -T cooling air 16 C Dry cooling By using these input parameters and the ambient temperature provided by the COSMO-EU model the temperature at point 4 (graphic 50) can be calculated as [DLS; S.50]: T 4 = T ambient + "T coolingair + T termteperaturedif (4.22) Based on the temperature and the pressure in the condenser, the enthalpy at point 4 2 and 1 (figure 50) is simulated. For the dry cooling system also a forced draft arrangement is assumed. Therefore, the additional parasitic load depends on the ventilator in the wet cooling system. It is calculated according to the electrical

! 72 consumption of the ventilator in the evaporation cooling system (formula 4.19). The difference can be seen in the air mass flow. Here, the air mass flow must be able to add the energy difference between the cycle point 4 and point 1. A weak point in the calculation of the cooling systems is the assumption of a fixed condenser pressure. In real operation there is a close link between the condenser temperature T 4 and the pressure. With a temperature increase also the pressure is increasing, this can lead to worse performance conditions. Trying to compensate this fact, the maximum allowed condenser temperature in the program is 47 C. If there would be a higher temperature occurring it is assumed that suitable retaliatory actions like increasing make up water supply in the evaporation system or operating a spray water system in the wet cooling cycle are undertaken. However after simulation of the necessary parameters at all 4 points in the Clausius- Rankine cycle the enthalpy drop between point 3 and 4 (figure 50) is used for the computation of the thermal energy provided to the steam turbine [DLS; S.32]: P th = m S (h 3 " h 4 ) (4.23) m S represent the steam mass flow, and h 3 and h 4 are the enthalpies before, respectively behind the turbine. Furthermore the efficiency of the power block can be simulated by the equation: " th = P th Q SF = h 3 # h 4 h 3 # h 2 (4.24) The formula above stated the enthalpy drop over the steam turbine and the enthalpy increase in the steam generator from h 2 to h 3 as the thermodynamic efficiency of the water-steam cycle. The gross efficiency of the Clausius-Rankine cycle is based on equation 4.24 is [DLS; S.38]: " gross = " th " G " ST (4.25) Here * G is the efficiency of the electrical generator, which depends on the operation condition as follow [DLS; S.37]: $ P ' & th # (1 #" P G,nom )) % th,nom ( " G = $ P ' & th ) % P th,nom ( (4.26) Finally, the net efficiency including accounting the parasitic loads is calculated by [DLS; S.37]:

! 73 " Net = P th" G # P Parasitics (4.27) Q SF This allows, according to the design parameters and the ambient condition, a prediction of the electrical output produced by a CSP plant at a certain location and time frame. Figure 54: Plant performance calculation for SM1 equipped with dry cooling system at a location of 31 N 29 E for the meteorological year 2007. Collector orientation is north south. The simulation displayed in figure 54 was undertaken with the presettings for a dry cooling system and solar multiple one for a plant located at 31 N 29 E in Egypt. The annual electrical output of the CSPP is around 221.95 GWh/a for the meteorological year 2007. The installed gross capacity is set to 200 MW el. For comparison, a CSPP with dry cooling system and solar multiple four at the same location and time frame is simulated with an annual output of 892.58 GWh/a. The influence of the cooling system is significant. By changing the cooling system to wet cooling including a back cooling system, the annual electrical output for the same CSPP is estimated to 241.86 GWh/a operating with solar multiple one and 971.18 GWh/y with solar multiple four. The influence of the back cooling system with a transportation distance of 500 m can be seen as not significant for the yearly electrical production. The overall efficiency has an annual average of 25.79% for wet cooling where the efficiency of the CSPP with dry cooling has only 24.18%. For a SM1 wet cooled it is 12.25% and SM1 dry cooled has 11.40%. This low efficiency depends according to figure 54 on the high amount of partial load operation hours during the year 2007. The CSPP cooled by evaporation cooling is settled in between the dry and wet cooled CSP plant. Operating as solar multiple four, the annual output is 911.44

! 74 GWh/a as solar multiple four with an average efficiency of 17.81%. The differences in the performance according to the cooling systems are related to the differences in the parasitic load for each system as well as the different performance of the condenser in the water steam cycle. The following picture shows the electrical output of a solar multiple four CSP plant using an evaporation cooling system, at the location of Kuraymat (31 N 29 E) for the meteorological year 2007. Figure 55: Plant performance calculation for SM4 equipped with evaporation cooling system at a location of 31 N 29 E for the year 2007. Collector orientation is north south. The bigger size of the solar field in a solar multiple four arrangements allows the plant can operate in the summer months nearly 2,500 hours in base load. Also during the winter months the CSP can work close to its design point of 200 MW el gross capacities. Therefore in months with a low irradiation nearly no energy is available for storing. For all calculations the yearly average direct normal irradiation is 2,038 kwh/m 2. 4.4 Curve fitting and ELCC calculation The processed feed in the time series shown in figure 53 and 54 can be fitted to a predefined demand. This makes calculation for supplying energy on demand possible. Using these results the effective load carrying capacity is chronologically evaluated. Both simulation steps explained in this chapter are predefined processes, supplied by the Frauenhofer-institute for wind energy and energy system technology (IWES). 4.4.1 Least square optimisation An objective method to evaluate the feed in the time series according to the criteria

! 75 supply on demand is offered by the least square optimization. Here the optimum for each simulation point can be found if the sum of the square distance between the load and the feed in curve is minimal. [MIN; S.697] Matlab offers some predefined algorithms dealing with the method of least square optimizations. Here the solver lsqlin is used. Lsqlin solves linear least square curve fitting problems according to [MWS] min x = 1 2 Cx " d 2 such that 2 $ Ax # b & % A eq x = b eq & ' lb # x # ub (4.28) It can be seen that the algorithm accepts inequalities and equalities in linear form, as well as upper and lower bounds. The boundary conditions lb and ub are used to define the operational area for the CSP power plant. This can be between no operation with 0 MW el output and an operation at maximum load. A disadvantage here is that no lower limit for energy production is given. Theoretically the power plant can operate with an electrical output less then 5 MW el, what is in reality depending on technical criteria s of the power block not possible [SIS; S50]. Due to the fact that this occurs very raw in the optimization, it is seen as acceptable. Furthermore, a simulation of more than one CSPP creates more flexibility for sharing the electrical production, which presents another possible option to eliminate, this problem. The criteria for inequalities are used to define the possibility of energy shifting depending on the storage size. The storage size is predefined according to the SM configuration (see chapter 4.3) where the feed in the time series of the CSPP represents the available energy at a certain time for shifting. The variable A is a matrix used for the allocation of the variables according to the time line. However another mismatch between the optimization and the CSPP performance calculation can be noticed. The thermal losses for the storages are simulated depending on the actual charge of the storage during the performance calculation; by shifting the storage charge during the optimization process the amount of thermal losses in the storage will be different. However, the thermal losses can be assumed to be relatively small compared to the amount of thermal energy produced by the solar field for a CSPP with a gross capacity of 200MW el. Therefore, the displacement of storage capacities during charge shifting and the related inaccuracy in thermal losses calculation are understood as acceptable for this work. Finally the predefined load curve where the CSPP output should be fitted to is defined by the variable d in equation (4.28). So the solver finds the minimum

! 76 electrical output the plant has to produce in order to reduce the demand of the predefined load curve in an optimum way. The yearly time series are optimized for a preset time horizon and updated for a predefined time step. The following figure shows the curve fitting process according to a residual load curve for a CSPP plant located at 31 N 29 E in Egypt. Figure 56: Feed in series direct calculated and optimised and for CSPP SM 4 with wet cooling and back cooling system located at 31 N 29 E. CSOMO-EU ambient data for 2007. For the simulation above the time horizon was set to 72 hours and the update frequency is 24 hours. The upper part shows the energy production according to the performance calculation, where the second plot presents the production after the optimisation process. In order to highlight the optimisation effect a time between 8,000 an 8,760 hours is plotted. The graphic shows how the output is fitted to the residual load (lower plot). A strong displacement during the wintertime can be noticed. For the SM 4 configuration a thermal storage capacity of 22.5 GWh th is available for load shifting. The figure above is only used for giving a graphical example of how the curve fitting process is working. 4.4.2 ELCC calculation Each electrical production technology can be characterized by the capacity credit (CC), that is to say, their contribution to secured power capacity. One possible classification for the secured power can be the additional load that can be covered by a production unit without a change in the loss of load probability (LOLP). This amount of energy is called effective load carrying capacity (ELCC). Furthermore the security of supply can be defined by the loss of load expectation (LOLE). Where the LOLP can be understood as a value of a possible load loss for each hour, the LOLE value

! 77 defines the statistical time of load loss for one year. As a maximum for the LOLE a value of 2.4 hours per year are used for all further calculations. The 2.4 hours per year are related to an expectable load loss for one day in 10 years. [UVE; S.26] Different methods, for calculating the ELCC, are available. In this thesis the so-called ELCC-chronologic is used. For this method the CSP energy production time series are considered as negative load. Therefore the actual load is reduced at each hour by an amount of energy that is supplied by the CSPP. Due to yearly fluctuations in the solar irradiation, normally production time series for more than one year are used in the calculation. However due to the limitation of available solar radiation data, only time series based on the metrological year 2007 are used during this work. The next schema shows the basic calculation concept for the ELCC-chorological method. Figure 57: Schemata for the ELLC calculation [UVE]. According to the schemata above an input parameter containing a power plant network (PPN) with the installed capacity and the blackout out probability for each plant has to be designed. Furthermore the load curve must be predefined and fed to the program. Based on these input parameters the LOLP without CSP production for the system of interest is evaluated. Therefore the probability of each plant defined in the PPN of supplying energy at base load condition is evaluated. The simulation considers only the two possibilities on and off which consequently places each power plant supply on nominal load with a probability of 1-P. P represents the blackout possibility for the power plant. By considering only these two options the cumulative density allocation for the complete PPN can be evaluated by the method of recursive

! 78 convolution. The method is summarized in the following figure for a power plant park consisting of 3 plants with different installed capacity. Figure 58: Method of recursive convolution, example for 3 power plants [DENA1] The flow chart shows that the probability density function for two plants can be evaluated out of the convolution of the density function for power plant one 1 (f) and power plant 2 (g). In mathematical expression it will be: h = f " g (4.29) Based on the commutative law for the convolution an additional plant (x) can be added to the function by: x " f " g = h " x (4.30) Using this calculation method the probability function for the complete PPN can be calculated [UVE; S.22]. Finally out of the cumulative density allocation the LOLP for each hour in relation to the predefined load curve is evaluated. The gray part in sketch 57 starts now with an iterative scaling until the LOLE reaches the predefined value of 2.4 hours per year. The relation between the LOLP and the LOLE can be stated therefore as [UVE; S.26]: T LOLE = " LOLP(Lt) (4.31) t =1 The load profile is multiplied with these load factors and the CSP production time series is subtracted from the new load profile. At the end the LOLP is calculated out of the relation between the new load profile and the cumulative density allocation of the PPN. In the blue part, now the load curve is scaled down until the original LOLE of 2.4 hours per year is reached. The new factor is multiplied with the load curve and subtracted from the original input curve. The difference between these two curves

! 79 can be seen as the ELCC. Dividing the ELCC value by the installed capacity of CSP plants results in the capacity credit (CC).

! 80 5. Export scenario 2050 5.1 Scenario description The following calculations are based on input parameters from the study 100% renewable electricity supply by 2050 published by the German federal environmental agency (UBA) and the MED-CSP study supplied by the DLR. The aim is to evaluate the capacity credit, respectively the firm capacity, for the energy exported by CSP plants to Germany in 2050. 5.1.1 Installed capacity and plant distribution The installed capacity of CSP plants in North Africa for the year 2050 is taken from the MED-CSP study (see chapter 3.2). It is assumed that the installed CSP plants are exclusive producing electricity for exporting it to Europe. To estimate the share of installed CSP capacity exclusively for Germany, the share of the country's electricity consumption in Europe is evaluated# Germany had in 2008 a share of 19.1% of the primary energy consumption of the EU-25 [SBA; S.73]. Consequently, a share of CSP plants in the same height is considered to be reserved for Germany for the following theoretical study. Each plant is simulated with an installed gross capacity of 200 MW el. The resulting number of plants and its distribution can be seen in the following map. Figure 59: Distribution of exporting CSP plants for Germany, with installed gross capacity of 200 MW el for each plant, in North Africa for 2050. The blue dots display the location for each plant. The ambient conditions and the irradiation data corresponding to the geo-coordinates of the CSSPs and taken from the COSMO-EU database. Having a closer look to the plant configuration, Morocco has 17 CSP plants installed for exporting of solar energy out of which 5 plants are equipped with a wet cooling system, 6 plants operating with evaporation cooling and

! 81 another 6 plants with dry cooling system. This leads to an installed cross capacity of 3.4 GW el. In Algeria 19 CSP plants are located, 3 of them with wet cooling system, 6 with evaporation cooling and 10 are equipped with a dry cooling system resulting in a total installed gross capacity of 3.8GW el. In Tunisia 5 CSP plants simulated with wet cooling system are installed, the overall gross capacity here is 1 GW el. Libya is hosting 1 evaporation-cooled plant and 2 dry cooled plants. Here the gross export capacity is 0.6 GW el. Finally, 24 wet cooled, 15 evaporation cooled and 5 dry cooled CSP plants are installed in Egypt, which equals an installed gross capacity of 8.8 GW el by a total of 44 CSP plants. The following table is summarizing these results. Table 12: Installed exporting CSP plants and gross capacity, separated by countries for Germany in 2050. Country Number of CSP plants Installed gross capacity [GW] Morocco 17 3.4 Algeria 19 3.8 Tunisia 5 1 Libya 3 0.6 Egypt 44 8.8 Sum 88 17.6 5.1.2 Residual load curve The residual load curve is taken from the study 100% renewable electricity supply by 2050 (UBA2050). This study described a scenario where the energy demand in Germany is supplied to 100% by renewable energies in the year 2050 The residual load curve has a significant influence on the further calculations and will be described more closely. This residual load curve is the result of the SIM-EE model developed by the Fraunhofer IWES. The following illustration presents the program parts of the SIM-EE model leading to the residual load curve used in this work.

! 82 Figure 60: SIM-EE model IWES for simulation of the residual load curve [UBA2050]. It can be seen that the residual load is a product, out of the basic load, the feed in of renewable energy must run units and additional consumers for E-mobility heating and climatization. The basic load curve includes all consumers participating nowadays in the grid. Furthermore, the transmission losses are part of this load curve. In sum the basic load for one year is 401 TWh. Due to the renewable energy law in Germany the production units always feeding into the grid when energy is produced. Therefore, they are simulated as must run units. For PV an installed capacity of 120 GW for the year 2050 is assumed. 80% of the PV systems are located on the roofs of residential buildings and the rest is installed on suitable fronts. The efficiency for the modules is calculated with 17% and the availability is 98%. The distribution is undertaken in relation to the distribution of settlement areas in Germany. The calculation is outlined by the meteorological data sets from the COSMO-EU model for the ambient temperature and the irradiation data from the HelioClim3 database. The wind energy production is divided into wind energy harvest onshore and offshore. For this simulation part, an installed capacity of 60 GW el onshore is assumed. The metrological data sets are also form the COSMO-EU model whit a spatial resolution of 14 x 14 km!. The distribution for the wind turbines is done between all available areas for an average wind speed of more the 5 m/s. The availability is assumed with 98.5%. The feed in time series are also simulated in hourly resolution. The offshore wind potential is set with 45 GW el in the year 2050. The availability is 97% and energy losses of 5% for the energy transmission to land is assume. In all other point the model follows the simulation for on-shore windmills.

! 83 Hydropower is simulated in hourly resolution taking into account 5.2 GW el installed capacity. Finally, for the geothermal plants an installed net capacity of 6.4 GW el is assumed in 2050. The availability is assumed to be in the same range like coal-fired plants with 90%. All the production units above are simulated as must-run-units. This means whenever energy is produced it will be fed into the electrical grid system and therefore must be subtracted at any time from the basic load profile [UBA2050]. For smoothing the load curve E-mobility, electric heat pumps and climate control is integrated via a load management. The potential of heat pumps supported by thermal energy is related to a heating demand of 100.5 TWh/year in residential buildings where in commercial and industrial buildings a consumption of 40.2 TWh/year is assumed. Due to thermal storage integration it is possible to uncouple heat demand from electricity demand allowing shifting the heat pump operation to times where electric overproduction is noticed. Consequently, a reduction of electrical energy demand for heating in times with low renewable energy availability takes place. Based on this concept, the residual load curve can be smoothed. The basis for the E-mobility simulation is a statistic evaluation of the driving habits in Germany. Furthermore it is assumed that all cars not in use are connected to the grid. This leads to an additional storage capacity of 180 GWh for the demand side management. The climatization is only simulated for commercial and industrial buildings. An additional annual requirement of 10 TWh for the year 2050 is assumed. Storages also offering the possibility to shift these loads by approximately three hours. Figure 61 shows the residual load curve for the year 2050 based on the metrological data sets form the year 2007. A positive residual load means that energy has to be added to the system either from stored energy, by backup power plants or the requirements have to be satisfied by importing renewable energy from abroad. The negative part represents the overproduction of renewable electrical energy calculated by the UBA2050.

! 84 Figure 61: Annual load duration curve for Germany in 2050 with an energy supply by 100% renewable energy sources [UBA2050]. The highest leak of energy in this scenario is 53.9 GW el. On the other hand an overproduction with a maximum of 63.7 GW el can be noticed. The annual deficit of 63.04 TWh is compared to an overproduction of 113.80 TWh. Figure 62 shows the residual load in hourly resolution for the year 2050 based on the weather data from 2007. Figure 62: Residual load curve from the UBA study 100% renewable electricity supply by 2050 [UBA2050]. The positive part of the load curve in the figure above is the input for the storage optimization and also for the calculation of the firm capacity. 5.1.2 Power plant network For the calculation of the firm capacity it is necessary that a power plant network (PPN) be designed. In this scenario simulation the PPN is set up according to the

! 85 long-term storage option and the amount of required backup power plants in the UBA2050 study. The study assumes that the positive residual load is covered by a PPN consisting of 2.5 GW el net installed capacity of gas turbines operating with biomethane, 30.4 GW el net capacity of gas and steam turbine plants using hydrogen as fuel and 17.5 GW el net capacity backup gas turbines also operating with biomethane. This can be shown in relation to the SIM-EE model as follow. Figure 63: SIM-EE model from Fraunhofer IWES for simulation of the residual load curve. Changes are undertaken in the position of importing energy according to the simulation in this thesis [UBA2050]. It can be seen that using excess renewable electricity produces hydrogen for operating combined cycle power plants. The amount of imported energy for the year 2050 based on the weather data from 2007 is assumed with 19.7 TWh. This value is related to a maximum import capacity of 6.9 GW el. However, in the UBA2050 study the imported energy is not considered as secured energy. Therefore, an additional net capacity of 14.7 GW el provided by gas turbines is assumed as backup capacity in order to guarantee the actual security of supply of 99%. Furthermore, the specific non-availability for each power plant must be defined for simulation of the ELCC. The necessary data therefore are provided by the DENA- Grid study [DENA1] and summarized for the assumed types of power plants in table 13. Table 13: Unplanned outage probability [DENA1] Type of power plant Unplanned outage probability Gas and steam turbine plant 1.8% Gas turbine plant 3.0%

! 86 The unplanned outage probability does not include the normal (projectable) outage times. It is assumed that the disposable outages can be shifted to times with low energy demand mainly during summer. Based on this assumption the unplanned outage probability is sufficient for the further calculation. The PPN designed for the scenario simulation is summarized in the following table Table 14: PPP for ELCC simulation Type of plants Number of plants Average net capacity per plant Installed net capacity Unplanned outage probability Gas and steam turbine plants Gas turbine plants 132 231 30,492 1.8% 239 145 34,665 3.0% Sum 371 65,157 Based on the input parameters described above, the maximum amount and the ELCC of imported energy from CSP plants in North Africa are calculated. 5.2 Simulation results The simulation is done according to the predefined definitions in chapter 5.1 for configuration SM1 to SM4. The design parameters for the collector field are adjusted for a LS-3 collector type with an irradiation of 800 W/m 2 at the design point. The collector orientation is always north south in order to harvest the maximum amount of energy. For irradiation data and data about the ambient condition the information from the COSMO-EU database 2007 are used. The curve fitting is done with a forecast horizon of 72 hours and an update frequency of 24 hours. For the energy transmission between North Africa and Germany no losses are assumed. 5.2.1 Simulation for SM1 The SM1 configuration is simulated without storage. Therefore, no energy shifting according to the demand of the residual load curve is possible. In sum all CSP plants producing an amount of energy from 15.3 TWh in the year 2050. The following figure shows the weekly average energy production for all 88 CSP plants in North Africa.

! 87 Figure 64: Weekly average energy production CSP plants SM1 based on the weather data COSMO-EU 2007. Figure 64 show that the energy production follows directly the solar irradiation for an SM1 configuration. Due to that fact a very low capacity credit of 0.38% according to the demand of the residual load curve is calculated. This value is corresponds to an ELCC of 67 MW el with a LOLE of < 2.4 hours per year. 5.2.1 Simulation for SM2 For the 88 CSP plants in SM2 configuration an electric output of 30.5 TWh based on the weather data from 2007 is simulated. For the displacement of the energy a storage capacity in SM2 configuration of 7.5 GWh th is calculated. For further calculation the feed in time series after the optimization is used. The next figure presents the result of the optimization in hourly resolution. Here also the high-energy demand during the winter months can be noticed.

! 88 Figure 65: Residual load curve without import energy and residual load curve including import energy form CSP with SM2. Irradiation and ambient data from COSMO-EU model 2007. The yellow area above displays the amount of energy produced by all CSP plants following the optimization according to the residual load curve. The green part presents the residual load curve after it is smoothed by the imported solar energy and the blue part shows the original residual load curve. For getting a better overview of the curve a part of 2,000 hours in spring and summer is sliced out of the figure above. Therefore, the next graph plots the same arrangement as above for the period between hour 2,800 and hour 4,800. Figure 66: Partial view of figure 65 residual load curve without imported energy and residual load curve including imported energy from CSP with SM2. Irradiation and ambient data from COSMO-EU model for 2007. In sum, only an amount of 23.6 TWh out of the production are useable for the reduction of the positive residual load. This leads to an amount of 6.9 TWh per year

! 89 remaining either for consumption in North Africa or to be used for hydrogen generation by electrolyzers. Figure 67 shows the resulting annual load duration curve. Figure 67: Annual load duration curve for the residual load with and whiteout CSP import SM2. The figure shows that importing solar energy from CSP pants in SM2 configuration reduces the peak load from 63.7 GW el to 49.5 GW el and in sum the yearly deficit of energy is reduced from 63.0 TWh to 39.4 TWh. In case that all the produced energy is transmitted to Germany the overproduction is increased to 120.7 TWh with a peak of 64.1 GW el for 2050. For making a more qualified statement about the imported solar energy the effective load carry capacity is calculated with 4.49 GW el. This leads to a capacity credit of 25.5% for the CSP plants with SM2 configuration. 5.2.2 Simulation for SM3 The simulation was repeated with a storage arrangement for the CSP plants of SM3. This means a storage capacity of 15 GWh th is available for each plant. An energy production of 47.1 TWh for the year 2050 is simulated. Due to the limitation according to the maximum production capacity for each plant with 200 MW el the load shifting during summer time is less distinctive than it is for the SM2 configuration. The changes mainly take place during the winter months. Due to the higher storage capacity also the amount of energy produced by the CSP plants is increased compared to solar multiple 2. The energy used for smoothing the demand presented by the residual load is calculated with 28.3 TWh for the year 2050. This reduces the deficit of energy to 34.7 TWh. The next picture displays the original and the

! 90 smoothed residual load curve for Germany as well as the energy produced by the CSP plants in the year 2050. Figure 68: Residual load curve without imported energy and residual load curve including imported energy from CSP with SM3. Irradiation and ambient data from COSMO-EU model 2007. Figure 68 displays the energy production of the CSP plants after optimization in yellow. The blue part presents the original residual load and the green part shows the residual load after smoothing by solar imports from CSP plants. For a more detailed view overview the following picture presents a time line of around 2,000 hours sliced from the figure above. Figure 69. Partial view part of figure 68 residual load curve without import energy and residual load curve including import energy form CSP SM3 as well as energy production CSP. Irradiation and ambient data from COSMO-EU model for 2007.

! 91 The peaks during summer months are more reduced in SM3 configuration than for SM2. Also the high demand in winter could be reduced more. An additional amount of 18.8 TWh (yellow part in the negative residual load) could either remain in the North African countries or be used for hydrogen generation in Germany. The following figure shows the energy distribution in the annual load duration curve including the imported solar energy from North Africa in 2050. Figure 70: Annual load duration curve for the residual load with and whiteout CSP import SM3. According to figure 70 the maximum value of the positive residual load is reduced to 46.6 GW el. If the negative part of the residual load curve contains the additional energy produced by the CSP plants an overproduction of 132.3 TWh with a peak of 68.1 GW el is calculated. Finally, the ELCC for SM3 configuration amounts to 7.64 GW el corresponding to a capacity credit of 43.5% of the installed capacity. 5.5.3 Simulation for SM4 Finally, the simulation is repeated with SM4 configuration. This allows an optimization with a maximum storage capacity of 22.5 GWh th. Using the feed in time series according to the optimization configurations a maximum amount of 64.1 TWh for exporting is produced by all CSP plants in the year 2050. Out of these 64.1 TWh only 31.8 TWh are used for lowering the positive part of the residual load curve. Figure XX shows the original and smoothed residual load curve as well as the energy for Germany provided by CSP in North Africa.

! 92 Figure 71. Residual load curve without imported energy and residual load curve including imported energy form CSP SM4. Irradiation and ambient data from CSOMO-EU model 2007. A partial view for 2,000 hours is displayed for more details. The following picture therefore shows the maximum possible reduction during summer time, based on the meteorological data of 2007. Figure 72. Partial view part of figure 71 residual load curve without imported energy and residual load curve including imported energy from CSP with SM4. Irradiation and ambient data from COSMO-EU model for 2007. The maximum reduction of the deficit energy for the year 2050 can be noticed. The total amount of electricity deficits is 31.3 TWh. The next figure displays the annual load duration curve for Germany with and without the imported energy form CSP with SM4 configuration.

! 93 Figure 73: Annual load duration curve for residual load with and without CSP import. Ambient data COSMO-EU 2007. According to the simulation an amount of 32.3 TWh can either remain in the countries of North Africa or be exported for the purpose of hydrogen generation. The peak demand in the residual load curve including the CSP imports is 44 GW el. The overproduction including the imported energy from CSP in North Africa amounts to 145 TWh with a peak of 70 GW el. The ELCC for SM4 configuration is calculated with 10.13 GW el and a capacity credit of 57.5%.

! 94 6. Conclusion 6.1 Simulation Program The aim of the thesis was to develop a simulation program that can evaluate the electrical output of CSP plants in North Africa. Such programs are already available for example Grennius or SAM. The advantage in this program can be clearly seen in the program part GIS database and scenario processing. This program part described in chapter 4.2 gives the possibility not only to fit the plant position to several land use criteria but also to the actual existing infrastructure for energy distribution. The simulation offers here already a basic approach for the plant siting in North Africa, mainly based on data out of the MED-CSP study. Nevertheless, this part can be made more precise by using informations from country reports for the MENA countries published for example by RECREE. Furthermore, informations about developing plans for a direct current high voltage grid as interconnection between Europe and North Africa could be collected and implemented to the program. Here the TRANS-CSP study published by the DLR could be a good starting point. Another possibility can be the implementation of more countries in the MENA region, therefore also the DLR publishes a lot of useful informations. Adding these data sets to the existing program can be done easily, due to the modular program structure and straightforward programming language offered by Matlab. A second advantage of the program is the ambient database in the background. The performance of a CSP plant can be simulated for every region in North Africa (resolution of 7 x 7 km!) taking into consideration the limits of the COSMO-EU model. By adding not only the irradiation data but also ambient temperature, ambient pressure wind speed near ground height and wind direction the results of the performance simulation getting more precise. Due to time reasons the relative humidity was not implemented in the CSP performance calculation. Considering the significant influence this parameter can have especially for an evaporation cooling systems, an adaption of this parameter should be undertaken during further work. Furthermore, a comparison of the irradiation data of the COSMO-EU model and the HelioClim 3 data from Soda would be advisable. The HC3 data sets using an interpolation of the irradiation data for a zenith angle of lower than 15 compared to the COSMO-EU model, which provides straightforward modeled values. By calculating the DNI a difference in the two data sets was noticed. Finally the program part scenario analysis can be understood as the link between two different studies,

! 95 the MED-CSP and the UBA2050 outlining this work. The amount of plants is calculated according to the scenario planning in the CG/HE of the MED-CSP while the final results of the export scenario are calculated by including outcomes of the UBA2050. The CG/HE scenario is the most ambitious scenario of the MED-CSP study regarding the installed capacity of CSP plants in North Africa. The intention was to show what could be possible, if the right incentives on both sides of the Mediterranean Sea will be set. In this context, also the UBA2050 can also be seen as an ambitious scenario for Germany until 2050. For the performance calculation the most reliable cooling system for each plant at a certain location depending on the water availability gets chosen. In this thesis, the availability of water is related only to the coast and the Nile River. Plants along the coastline or the Nile are equipped with a wet cooling system, close to this they are using evaporation cooling and where a lower water availability is assumed the plants operating with dry cooling system. Additional informations of the hydrology out of the DLR Solar import study can be used to make these selection criteria and its influents on the performance calculation more precise. The core of the program is definitely the performance calculation of a CSP plant. As described above, the program can calculate the performance in relation to the cooling systems and SM adjustment. However, the aim of the simulation is not to design a CSP plant. The electrical output of the plant in relation to several input parameters should be calculated based on physical criteria and giving an accurate value of the electrical output depending on the incoming irradiation. Therefore, a database containing design data of an LS-3 collector type together with the necessary basic parameters for the storages, the power block and the cooling systems where created and used. The simulation displays only steady-state operation conditions. In this work it is assumed that the transient processes are mainly covered by the hourly resolution of the results. The DLR SOKRATES model does similar assumptions for the same temporal resolution. If a higher time resolution like 15 min time series are realized, the influence of transient processes on the results have to be taken into account, especially for the solar field simulation. To obtain more precise values, the simulation is working with real ambient data during the performance calculation. The most significant influence of the ambient date not considering the irradiation could be noticed by the performance calculation of the conventional part. This is related to the condenser performance in relation to the ambient conditions. By mischance it was not possible to get any usable condenser performance curves for related power plants. Due to the complexity of calculating these curves it was impossible to indicate the condenser pressure in relation to the

! 96 ambient conditions. That limits the influence of the ambient data by the performance calculation. However, accepted values where set for the condenser pressure and the ambient data are considered for the temperature calculation. In this way the influence of the ambient data is given to a certain extend for evaluating the condenser parameters. As an outlook, it would be possible to implement these curves for getting more precise results of the condenser behavior. The storage arrangement is simulated for different solar multiple operation. Here also more advanced modeling techniques as described in chapter 2 could be used in further work. However, the performance simulation offers the opportunity to simulate for different arrangements, feed in time series in hourly resolution at a high accuracy. Finally, the two predefined program parts load optimization and effective load carrying capacity calculations where adapted to the simulation program. Some mismatching point between the written program in this thesis and the predefined steps during the export scenario simulation where noticed. Where most of them could be solved still a mismatching by the optimization process at the end of the feed in time series is noticed. Furthermore, an inaccuracy during the optimization in the negative part of the residual load curve can be seen. Solving this problem must be undertaken in a further work. However, the accuracy of the results from optimization simulation are considered as high enough for the ELCC_C calculation. Including these two simulation parts provided by the IWES the export scenario for 2050 is calculated. 6.2 Export Scenario The export scenario can be seen as a combination of results out of the MED-CSP and the UBA2050 study. Therefore, the amount of energy produced by a number of CSP plants according to the MED-CSP in North Africa, is put into relation to the residual load of Germany calculated in the UBA2050. The simulation is undertaken using weather data from 2007 for the year 2050. The focus of interest is on the effective load carrying capacity and capacity credit for the CSP plants providing solar energy to Germany. For the further discussion, the results from chapter 5 are summarized in table XX.

! 97 Table 16: Summery of results from export scenario analyze ELCC [GW el ] Capacity credit Sum export energy [TWh/a] Export energy positive [TWh/a] Gross installed capacity [GW el ] SM1 0.067 0.38% 15.3 17.6 SM2 4.49 25.48% 30.5 23.6 17.6 SM3 7.64 43.46% 47.1 28.3 17.6 SM4 10.13 57.5% 64.1 31.8 17.6 Due to the fact that these values are simulated, it is necessary to take a closer look at the used input parameters. First of all, the ELCC_C calculation normally requires input parameter for at least for four to five years. This is needed to minimize effects related to extreme weather conditions occurring infrequently [UVE; S.31]. In this thesis however, only the feed time series for 2007 where simulated, due to the unavailability of irradiation data sets for more years. Consequently, only the residual load for 2007 is used for further processing. Finally, for the installed capacity only one very optimistic scenario out of the MED-CSP study was studied. The comparison between different scenarios would be desirable. This effect gets even more boosted by the assumption that all CSP plants strictly used for exporting energy to Germany. Due to already existing shortages in most of the countries, this assumption has to be removed in a future work and a more precise evaluation of the local demand in the producing countries and the available energy for export have to be conducted. However, to compensate this effect at least the amount of energy produced during overproduction times in Germany could be seen as available energy for the production countries. Finally, the storage losses simulated in the program part performance calculation can be displaced during the dynamic optimization processes. Because of the program structure, no recalculations for the thermal losses taking place during the optimization process. This can lead to a different amount of losses depending on the storage time. However, related to the quantity of thermal energy produced in a CSP plant with a gross installed capacity of 200 MW el the different in the losses can be assumed as not significant for the ELCC and CC calculation. Furthermore, only a power plant scheduling can be used for calculating a coast-optimized operation of the CSPP, to do so will also be part of a future work. Knowing the accuracies of the input parameter the results of table XX will be discussed. First off all it can be noticed that for an SM1 configuration the capacity

! 98 credit is in the same range as the capacity credit for PV systems [UVE; S.31]. Also the ELCC with 0.067 GW el is very low and will not help to reduce backup capacities in Germany by using solar imports. It can be stated, that using a CSP plants without thermal storage in this way, it is only more expensive than PV systems. Looking at the results of CSP plants including storage capacities ones again the importance of storage arrangement in relation with CSP technologies can be highlighted. All results for the CC are in the range of CC s for parabolic through plans described in the MED- CSP study. Only plants using additional fossil fuel achieve higher CC values. However, the idea of the thesis is a 100% renewable scenario. The input data for the power plant park and the ELCC calculation are related to the hydrogen scenario of the UBA2050. For this approach, an energy import to Germany of 19.7 TWh in 2050 was calculated. This evaluation is also done with weather data from 2007 [UBA; S.91]. At a first look all three CSP configurations with storages are able to fulfill this criteria. However, an import capacity of 6.9 GW el is also required for all the times when energy is needed according to the residual load curve. Only the configuration SM3 and SM4 are providing an ELCC high enough to perform this requirement. Consequently the imported energy based on the criteria mentioned above can be considered as guaranteed for by using SM3 or SM4 configuration. As a result the addition backup capitates installed by the postulation that the import energy can be not considered as guaranteed in the UBA2050 can be removed. However, it can be noticed, that the energy production by CSP plants even with storages is depending on the seasonality of the irradiation. Consequently, the possible contribution of CSP plants reducing the residual load during winter times, with the highest demand is very small. Therefore, the use of long-term storage options like hydrogen storages cannot be reduced significantly by importing energy from CSP plants. In the summer time however, the CSP plants equipped with SM2 or higher can nearly deliver energy on demand, witch can be noticed observing the reduction of the demand during summer times. During this times the advantages of CSPP with storage compare to PV systems can be noticed. As an outlook simulations with different design points for the CSPP should be undertaken for getting more comparable results. During the thesis a program was developed which can evaluate the amount of energy produced by CSP plants in North Africa on a secure level. Further investigation in the program structure and calculating different scenario configurations would be helpful for getting more precise results.

! 99 To the best of my knowledge I do hereby declare that this thesis is my own work. It has not been submitted in any form of another degree or diploma to any other university or other institution of education. Information derived from the published or unpublished work of others has been acknowledged in the text and a list of references is given. -------------------------------- Daniel Horst Kassel, 05.03.2012,

! 100 References [VDW]: Association of German engineers (2006): VDI-Wärmeatlas. 10th, Berlin, Heidelberg: Springer. [DENA1]: Deutsche Netz Agentur (2005): DENA Grid study 1. Planning of the Grid Integration of Wind Energy in Germany Onshore and Offshore up to the Year 2020. In cooperation with: DEWI / E.ON Netz / EWI / RWE Transportnetz Strom / VE Transmission. Berlin: Deutsche Netz Agentur (DENA). [SETP]: Duffie, J / Beckman, W (2006): Solar engineering of thermal processes. 3rd, Hoboken, New Jersey: John Wiley & Sons. [IPCC, SREEN]: Edenhofer, O / Madruga, R / Sokona, Y (2012): Renewable Energy Sources and Climate Change Mitigation. Special Report of the Intergovernmental Panel on Climate Change. Cambridge: University Press. [EXS]:Excel Engineering: ULR:http://x-eng.com. Last access 05.01.2012 [SBA]: Federal Statistical office (2006): Deutschland in der EU. Wiesbaden: Federal Statistical office. [MCSP]: German Aerospace Center (2005): Concentrating Solar Power for the Mediterranean Region. In cooperation with the Federal Ministry for the Environment, Nature Conservation and Nuclear Safety Germany. Stuttgart: German Aerospace Center (DLR). [GENI]: Global Energy Network Institute (GENI): National Energy Grid Maps. URL: http://geni.org/globalenergy/library/national_energy_grid. Last access: 05.01.2012. [TMD]: Grote, K / Feldhusen, J (2007): Taschenbuch des Maschienenbaus. 21st, Berlin, Heidelberg: Springer. [SLP]: Hoyer, C (1998): Simulation der Leistungsabgabe von Parabolrinnekraftwerken. Master thesis, University Oldenburg. [ES]: Huggins, R (2010): Energy Storage. Dordrecht, Heidelberg, London, New York: Springer. [IPCC 2007]: Intergovernmental Panel on Climate Change (2007): Climate Change 2007 Synthesis Report Summary for Policymakers. Geneva: Intergovernmental Panel on Climate Change (IPCC).

! 101 [REB]: Kaltschmitt, M / Streicher, W / Wiese, A (2007): Renewable Energy/Technology Economics and Environment. Berlin, Heidelberg: Springer. [UBA2050]: Klaus, T / Vollmer, C / Werner, K / Lehmann, H / Müschen, K (2010): Energieziel 2050. 100% Strom aus erneuerbaren Quellen. Dessau: Research of the federal environmental agency. [DLSC]: Kearney, D / Kelly, B / Cable, R (2003): Overview on the use of a Molten Salt HTF in a Trough Solar field. NREL Parabolic Trough Thermal Energy Storage Workshop. Golden: National Renewable Energy Laboratory. [STS]: Laing, D / Tamme, R (2008): Speichertechnik nicht nur für Solarenergie. Stuttgart: German Aerospace center (DLR). [TFI]: Langeheinecke, K / Jany, P / Thieleke, G (2006): Thermodynamic for Engineers. 6th, Wiesbaden: Vieweg & Sohn GWV. [MWS]: MathWorks: Accelerating the pace of engineering and science/product documentation. URL: http://www.mathworks.de/products. Last access: 02.03.2012. [PSK]: Mohr, M / Svoboda, P / Unger, H (1999): Praxis solarthermischer Kraftwerke. Berlin, Heidelberg: Springer. [UVE]: Pape, C (2011): Untersuchung zur Versorgungssicherheit in Energiesystemen mit hohem Anteil erneuerbarer Energien. Master thesis, University Kassel. [MFL]: Papula, L (2003): Mathematische Formelsammlung. 8th, Wiesbaden: Vieweg & Sohn GWV. [MIN]: Papula, L (2011): Mathematik für Ingenieure und Naturwissenschaftler. Vektoranalysis, Wahrscheinlichkeitsberechnung, Mathematische Statistik, Fehler und Ausgleichsrechnungen. 6th, Wiesbaden: Vieweg & Sohn GWV. [QGIS]: Quantum GIS (2011): Benutzerhandbuch. Quantum GIS Development Team. URL: http://www.qgis.org. Last access: 15.12.2011. [COE]: Schulz, J / Schättler, U (2009): Kurze Beschreibung des Lokal-Modells Europa COSMO-EU (LME) und seiner Datenbanken auf dem Datenserver des DWD. Offenbach: German Weather Service/Research and Development. [SIS]: Siemens (2008): Technology Curse Steam Turbine Design. Power Plant design personal training. Erlangen: Siemens AG. [KWT]: Strauß, K (2009): Kraftwerkstechnik. 6th, Berlin, Heidelberg: Springer. [FCS]: Trieb, F (2011): Concentrating Solar Power 2/ Fundamentals of Concentrating

! 102 Solar Power. Stuttgart: German Aerospace center (DLR), REMENA Kassel. [DLS]: Trieb, F / Kornshage, S / Quaschning, V / Dersch, J (2004): SOKRATES- Projekt Solarthermische Kraftwerkstechnologie für den Schutz des Erdklimas. Freiburg, Stuttgart: Fraunhofer Institute for solar Energy systems. Technology database AP2 /AP2.1, German Aerospace center (DLR). [SI]: Trieb, F / Marlene, O / Thomas, P (2009): Characterization of Solar Electricity Import Corridors from MENA to Europe. Potential Infrastructure and Cost. Stuttgart: German Aerospace Center (DLR). [TDK]: von Wolfersdorf, J / Weigand, B / Köhler, J (2010): Thermodynamik kompakt. 2nd, Dordrecht,Heidelberg, London, New York: Springer. [REG]: Wesselak, V / Schabbach, T (2009): Regenerative Energietechnik. Dordrecht, Heidelberg, London, New York: Springer.!

! 103 Appendix A (Validation Table CSP performance calculation) The simulation of the CSP performance is validated by the results of the DLR SOKRATS MODEL. In the following tables the simulation results for Matlab and SOKRATS program will be shown. The validation is done for the solar field and for the three different modifications of the power block. The grey marked parts are preset input parameters for the validation. All design parameters were set according to the description of the DLR SOKRATES model. For closer information see model description of DLR SOKRATES. Validation table solar field. Matlab Simulation DLR SOKRATES DNI [W/m 2 ] 800 800 IDR [W/m 2 ] 490 490 Thermal energy solar field [MW] 49.02 49.06 Efficiency solar field [%] 0.32 0.33 Validation table power block with wet cooling system Matlab Simulation DLR SOKRATES Heat from solar field [MW] 91.63 91.63 Steam turbine power [MW] 36.81 36.74 Gross energy output [MW 35.53 35.46 Net energy output CSP [MW] 32.68 32.64 Consumption feed water pump [MW] 0.48 0.45 Consumption back up pump 0.05 0.03 Consumption cooling water pump 0.57 0.50 Consumption solar field [MW 1.75 1.75 Parasitic sum [MW] 2.85 2.75 Efficiency generator 0.97 0.97 Efficiency all pumps 0.71 0.71 Efficiency water turbine 0.80 0.80 Thermal efficiency 0.406 0.401 Gross efficiency 0.391 0.389 Net efficiency 0.362 0.359

! 104 Validation table power block with evaporation cooling system. Matlab Simulation DLR SOKRATES Heat from solar field [MW] 91.63 91.63 Steam turbine power [MW] 36.10 35.72 Gross energy output [MW 34.78 34.65 Net energy output CSP [MW] 31.12 31.21 Consumption ventilator [MW] 0.52 0.47 Consumption feed water pump [MW] 0.49 0.46 Consumption make up water pump 0.02 0 Consumption cooling water pump 0.88 0.74 Consumption solar field [MW 1.75 1.75 Parasitic sum [MW] 3.66 3.44 Efficiency generator 0.97 0.97 Efficiency all pumps 0.71 0.71 Efficiency ventilator 0.60 0.60 Thermal efficiency 0.395 0.390 Gross efficiency 0.381 0.378 Net efficiency 0.339 0.341 Validation table for power block with dry cooling system. Matlab Simulation DLR SOKRATES Heat from solar field [MW] 91.63 91.63 Steam turbine power [MW] 34.87 34.50 Gross energy output [MW 33.89 33.46 Net energy output CSP [MW] 30.07 29.87 Consumption ventilator [MW] 1.52 1.36 Consumption feed water pump [MW] 0.55 0.47 Consumption solar field [MW 1.75 1.75 Parasitic sum [MW] 3.82 3.60 Efficiency generator 0.97 0.97 Efficiency feed water pump 0.71 0.71 Efficiency ventilator 0.60 0.60 Thermal efficiency 0.381 0.376 Gross efficiency 0.368 0.365 Net efficiency 0.334 0.326

! 105 Appendix B (Transmission grid maps North Africa) The following transmission grid maps are used in the simulation program (Chapter 4.2) Transmission grid Egypt [GENI] Transmission grid Libya [GENI]

! 106 Transmission grid Tunisia [GENI] Transmission grid Morocco [GENI] For the transmission grid in Algeria no reasonable map was found. Therefore, the information of the transmission grid in Algeria is shaped out of figure 23.