Portal de Periódicos da CAPES

Long‐term impact of variable generation and demand side flexibility on thermal power generation

2018; Institution of Engineering and Technology; Volume: 12; Issue: 6 Linguagem: Inglês

10.1049/iet-rpg.2017.0107

1752-1424

Niina Helistö, Juha Kiviluoma, Hannele Holttinen,

Integrated Energy Systems Optimization

IET Renewable Power GenerationVolume 12, Issue 6 p. 718-726 Research ArticleOpen Access Long-term impact of variable generation and demand side flexibility on thermal power generation Niina Helistö, Corresponding Author Niina Helistö niina.helisto@vtt.fi Smart Energy and Transport Solutions, VTT Technical Research Centre of Finland, Espoo, FinlandSearch for more papers by this authorJuha Kiviluoma, Juha Kiviluoma Smart Energy and Transport Solutions, VTT Technical Research Centre of Finland, Espoo, FinlandSearch for more papers by this authorHannele Holttinen, Hannele Holttinen Smart Energy and Transport Solutions, VTT Technical Research Centre of Finland, Espoo, FinlandSearch for more papers by this author Niina Helistö, Corresponding Author Niina Helistö niina.helisto@vtt.fi Smart Energy and Transport Solutions, VTT Technical Research Centre of Finland, Espoo, FinlandSearch for more papers by this authorJuha Kiviluoma, Juha Kiviluoma Smart Energy and Transport Solutions, VTT Technical Research Centre of Finland, Espoo, FinlandSearch for more papers by this authorHannele Holttinen, Hannele Holttinen Smart Energy and Transport Solutions, VTT Technical Research Centre of Finland, Espoo, FinlandSearch for more papers by this author First published: 22 February 2018 https://doi.org/10.1049/iet-rpg.2017.0107Citations: 32AboutSectionsPDF ToolsRequest permissionExport citationAdd to favoritesTrack citation ShareShare Give accessShare full text accessShare full-text accessPlease review our Terms and Conditions of Use and check box below to share full-text version of article.I have read and accept the Wiley Online Library Terms and Conditions of UseShareable LinkUse the link below to share a full-text version of this article with your friends and colleagues. Learn more.Copy URL Share a linkShare onFacebookTwitterLinkedInRedditWechat Abstract This study presents the potential role of thermal power generation in a future power system with high shares of variable generation while considering different sources of demand side flexibility such as heat pumps and heat storages in district heating, demand response from industries and electric vehicles. The study was carried out using a generation planning model combined with a unit commitment and economic dispatch model. The results from the planning model show a strong shift away from combined cycle gas turbines to open cycle gas turbines and gas engines as the share of wind power and solar photovoltaic increases. Demand side flexibility measures pushed this trend further. The results from the unit commitment and economic dispatch model demonstrate that the flexibility measures decrease the ramping frequency of thermal units, while the ramp rates of thermal units remain largely unchanged or increased. This indicates that the flexibility measures can cover smaller ramps in the net load more cost-effectively but that thermal power plants are still valuable for larger ramps. Impacts on emissions and electricity prices are also explored. 1 Introduction Thermal power plants tie up a large amount of capital for decades, and it is consequently important to understand their potential role in a future power system, where the increasing amount of variable generation (VG, e.g. wind power and solar photovoltaics (PV)) is changing the role of thermal generation from the main source of electricity towards a provider of flexibility. On the other hand, different sources of controllable electricity demand also need to be considered, as they compete with thermal power plants in the provision of flexibility. Potential sources of flexibility include: demand response from industry, demand response from smaller consumers, heat pumps and electric boilers combined with heat storages in district heating grids, smart charging of electric vehicles (EVs), batteries, power-to-gas facilities, reservoir and pumped-storage hydropower plants, cycling of thermal power plants, and curtailment of VG itself. Increased use of transmission lines to neighbouring areas will partially mitigate the impact of variability especially in the case of wind power where the smoothing effect is strong [1]. The need and value of additional flexibility to integrate high shares of VG have been addressed, e.g. in [2, 3], albeit with a focus on wind power (by 'share of VG' we denote VG's share of total annual electricity consumption, while we use 'instantaneous share of VG' to denote VG's share of power production at a given moment in time). The results demonstrated that flexibility in the generation and use of heat as well as smart charging of EVs can decrease costs induced by wind power variability and uncertainty, increase the cost-optimal share of wind power generation, and reduce CO2 emissions. In addition, energy storages and demand side flexibility were shown to have economic and environmental benefits, especially in power systems that had low amounts of flexible conventional generation coupled with a higher share of wind power. The value of different sources of flexibility has been compared in [4] for the Northern European power system with high shares of wind power. The study emphasised the advantages of new transmission links and flexibility in district heating grids as means to reduce total costs in systems with high amounts of VG. However, [2-4] have been limited to investigating the value of increased system flexibility, such as reduction in total costs, and do not show in detail the impacts on the operation of thermal generation as will be done in this study. The share of thermal power generation is inevitably diminishing as the VG share increases. Generation planning studies have addressed which power plant types are the most optimal for covering the remaining electricity demand. According to [5-8], there will be a reduced need for base load generation and an increased need for peak load power plants, and the capability for cycling has been recognised as an important characteristic for future thermal power plants. Cycling of thermal power plants refers to their operation at varying load levels, including on/off, load following, and minimum load operation, in response to changes in system load requirements [9]. Cycling is associated with technical limitations such as ramp rates and start-up times as well as costs. It causes wear and tear mainly due to increased thermal gradients in the turbine and boiler materials. According to [10], a 25% PV share can significantly increase the ramping of thermal fleet and the following thermal generator features will be helpful when the PV share is high: the ability to start and stop with short minimum downtimes, reliable and repeatable start-ups, low minimum generation levels, and a wide range of ramping capability. Troy et al. [11] demonstrated that an increasing share of wind power affects base load power plants differently depending on their characteristics: combined cycle gas turbine (CCGT) power plants saw rapid increases in the start-stop cycling and a plummeting capacity factor, whereas coal power plants saw an increase in part-load operation and ramping in the 2020 Irish system. Pereira et al. [12] carried out simulations with the Portuguese system. As they increased the amount of wind power, they observed an increase in the number of start-ups of gas units and a decrease in the electricity production of CCGTs. Although the case study included hydropower plants with reservoirs and pumping capacity, it did not specifically analyse the impact of flexibility measures on the role and operation of thermal power plants, which will be in focus in this study. The impact of storage, transmission and demand side flexibility on the operation of thermal power plants has been examined in [11, 13]. Dupont et al. [13] showed in a Belgian case study with 2012 and 2025 scenarios that demand response on average decreased the loading of mid-peak and peak power plants over the year as well as during peak load periods. This was also reflected in the reduced number of start-ups. Troy et al. [11] demonstrated within the Irish system that the presence of storage or interconnection on a power system can even exacerbate the cycling of base load units at lower shares of wind power (<23–34% of total demand). This behaviour could conceivably be different in a larger power system with more interconnections and if other new sources of flexibility were to be considered. The importance to model not only direct start-up costs but also other cycling costs in the unit commitment models has been emphasised in [14-16]. A detailed modelling approach was used by Deane et al. [17], who analysed the impact of sub-hourly modelling on the cycling of units with the All Island power system for the year 2020. With higher resolution, all units ramped more significantly over shorter periods. However, modelling sub-hourly time scales and cycling costs was not seen particularly important for generation planning models in [18, 19]. Instead, [18, 19] emphasised the chronological variability of load and VG, together with hourly resolution operating constraints, such as ramping limits, minimum up and down times and reserve requirements, which are all included in this study. In addition, this study takes into account start-up costs and decreased efficiency due to cycling but excludes, e.g. ramping costs. This study complements the existing literature by addressing the combined impacts of VG and demand side flexibility on the operation of thermal power generation based on optimised power plant and flexibility portfolios. The existing literature has not fully considered the impact of increasing demand side flexibility and this can be important in future. Optimising the power plant and flexibility portfolios is essential as it can have a large impact on the operation of thermal power plants. Costs and benefits of the analysed scenarios were published in [20] and are referenced in this study. While the study by Kiviluoma et al. [4] focused on comparing the system value of various flexibility options, this study considers how the flexibility options affect thermal power plants in particular. The study considers different scenarios for the Northern European power system in 2050 – this gives room for a large share of VG but is still relevant for potential new thermal power plants. Section 2 describes the approach and the modelling tools used in the study. Section 3 defines the scenarios as well as important input data and assumptions. Results are presented in Section 4. Finally, Sections 5 and 6 provide discussion and conclusions, respectively. 2 Methods 2.1 Approach The modelling and simulation procedure was as follows: Define scenarios with different wind power and solar PV investment costs and CO2 prices. Define flexibility cases by changing the availability of flexibility technologies. Run investment planning model for each scenario combined with each flexibility case. Run operational planning model for each scenario/flexibility case combination while considering the investments created in step iii. In this approach, the planning model is free to choose the share of wind power and PV, given the costs and the availability of flexibility technologies. Another way to consider investments would be to fix the generation capacities or the shares of wind power and PV in each scenario and use the same capacities or shares in all flexibility cases under each scenario. The chosen method gives a comparison for thermal unit operation in a fully optimised system where thermal units fulfil a role that is in line with their capabilities and system characteristics. 2.2 Models The study was carried out using two optimisation models: a generation planning model Balmorel and a unit commitment and economic dispatch model WILMAR Joint Market Model (JMM). Both models are formulated in GAMS and were solved using CPLEX. Only the basic principles of the models and the most essential thermal power plant constraints are described here. More detailed description of the models can be found in [21, 22]. Balmorel [21] is a generation planning model that minimises the total investment and operational costs. In this study, one future year was considered, which was represented by three selected weeks, each consisting of 168 hourly time steps. Balmorel balances the electricity demand and production as well as the heat demand and production at every time step in the hourly time series. It takes into account the chronological aspect of load, wind power, solar power and hydro inflow. Balmorel includes constraints for combined heat and power (CHP) and other thermal plants, storage levels, required reserves and required minimum generation capacity in each region. Balmorel is a regional model with transmission constraints modelled using net transfer capacity limits between price regions. JMM [22] is a unit commitment and economic dispatch model that minimises the total operational costs (1) The total costs (1) consist of fuel costs (), variable operation and maintenance costs (), fixed and fuel-based start-up costs (), revenue of online capacity and storage content in the end of the optimisation period (), changes in consumers' utility (cost of downward demand response, , and revenue of upward demand response, ), emission costs (), transmission and distribution costs (), as well as infeasibility penalties (). JMM uses hourly time resolution with an optimisation horizon of 36 h. The model simulates day-ahead and intraday markets, and takes into account reserve requirements. JMM takes the investment results from Balmorel and runs an 8760-h rolling optimisation with a more detailed representation of the operational constraints as well as uncertainties related to load and wind power forecasting. Each country is divided into one or more price regions. Transmissions between price regions are constrained by net transfer capacities similar to Balmorel. Each price region can include several heat areas, and in each heat area, heat production needs to cover heat consumption in each hour. In systems with large amounts of reservoir hydropower, water values need to be estimated. JMM has a separate water value estimation model to perform this. Furthermore, JMM was run in mixed-integer linear programming mode in order to properly optimise the unit commitment. JMM contains several constraints for thermal power plants. As these are relevant for the present study, they are presented in the following. Multiple same-size units are represented with integer variables indicating the number of online units in a particular unit group i at time period t as follows: (2) (3) In (2) and (3), represents the number of start-ups and shut-downs, represents the number of online and offline units, and represents the minimum up and down times of units in the unit group. Start-up fuel consumption of a unit group i at time period t is defined as (4) In (4), represents the start-up fuel consumption per unit of started capacity, is the available electricity generation capacity of the unit group and is the number of units in the unit group. In addition, the following constraints are limiting the operation of thermal power plants: (5) (6) (7) Here, represents electricity and heat production, represents primary, secondary and tertiary reserves, of which primary and secondary reserve are divided into positive and negative parts, is the minimum load factor of the units, is the iso-fuel coefficient, and is the backpressure coefficient. The terms including only apply to extraction CHP plants. Equations (5) and (6) represent the maximum and minimum generation limits, respectively. 3 Scenarios, flexibility cases and assumptions The studied system was the Northern European power system in 2050. Year 2020 was simulated as a reference. The countries included in the modelling are Denmark, Norway, Sweden, Finland, Estonia, Latvia, Lithuania, Poland and Germany. This section first describes the scenarios, including wind power and solar PV investment costs and CO2 prices. Then, flexibility cases are presented, listing the available flexibility sources in each case. These are followed by the fuel price assumptions, initial generation capacities and the characteristics of thermal power plant investment options. Finally, other important region-specific parameters and their main sources are given, including the sources for the hourly load and wind power production and forecast data. 3.1 Scenarios Three scenarios were created with different wind power cost, solar PV cost and CO2 price assumptions: Traditional, Windy and Sunny. Traditional represents a relatively conservative business-as-usual case. Windy represents wind power and PV cost reductions and a higher CO2 price resulting in a higher VG share. Sunny represents a more profound cost reduction for PV. Table 1 presents the wind power and PV cost assumptions in the scenarios. The costs were iterated until the target shares of wind power and PV tabulated in Table 1 were approximately reached in the HeatFlex flexibility cases (see Section 3.2). For wind power, however, it was assumed that only at maximum 53,000 MW of new capacity can be built with the costs tabulated in Table 1. In addition, it was possible to build at maximum 88,100 MW of additional wind power with a cost that was €100/kW higher than the costs in Table 1. The rest of the wind power plants had to be built with a cost that was €200/kW higher than the costs in Table 1. This represents building higher towers and larger rotors in less windy sites to achieve the same capacity factors that could be achieved with lower costs in better sites. Fixed annual operation and maintenance costs were €36/kW/a for wind power and €12/kW/a for PV in all scenarios. Table 1 also tabulates CO2 price assumptions for 2050. In the 2020 reference case, CO2 price was €17/t. Table 1. Investment costs of wind power and solar PV and CO2 prices Scenario VG share Wind, €/kW PV, €/kW CO2 price, €/t Traditional 30% wind and 9% PV 1600 550 12 Windy 51% wind and 10% PV 1310 520 49 Sunny 41% wind and 19% PV 1340 270 49 3.2 Flexibility cases For each of the three scenarios, the optimisation procedure was repeated five times assuming different available flexibility options, as presented in Table 2. After first running the HeatFlex flexibility case for each of the scenarios, the investment costs in each scenario were fixed, and the capacities and shares of wind and PV were allowed to change in the other flexibility cases. The baseline investment options included transmission lines and the following production units: wind turbine generators, PV panels, CCGT power plants, gas turbine (GT) power plants, gas engine (ICE) power plants, biomass power plants, nuclear power plants, natural gas-fired heat boilers (€100/kW) and biomass-fired heat boilers (€400/kW). The additional flexibility options included heat pumps, heat storages and electric boilers in district heating grids, batteries, industrial demand response, EVs, hydropower, power-to-gas facilities, and thermal power plants with smaller minimum load. Table 2 tabulates how the investment options were included in the five flexibility cases. The table also presents the investment costs of the additional flexibility options. The characteristics of thermal power plants are presented later in Section 3.5. Table 2. Investment options in each flexibility case Case Description Base baseline investment options HeatFlex base + heat pumps (€575/kWheat), heat storages (€4/kWh) and electric boilers (€60/kWheat) Flex HeatFlex + smaller minimum load of thermal power plants (no additional cost) All Flex + batteries (€150/kWh), demand response from industry (no investment cost, total 10,340 MW), hydropower (€329/kW, max. 6094 MW), power-to-gas (€1386/kWh) All + EV All + EVs (60% of passenger cars → 18% increase in electricity consumption, smart charging, grid-to-vehicle only) 3.3 Fuel prices Fuel prices in the study, tabulated in Table 3, reflect IEA New Policies scenario [23]. Table 3. Fuel prices Fuel price CO2 content (€/GJ) (kg/GJ) 2020 2050 coal 2.7 — 95 fuel oil 14 — 78 lignite 2.2 — 101 municipal waste 0 0 19 natural gas 8 10 56.9 nuclear 1 1 0 peat 3.5 — 107 shale 1.5 — 106 straw 4.5 4.5 0 wood 5 5 0 wood waste 2.5 2.5 0 3.4 Initial generation capacities The initial generation capacities in each region were mainly based on Platts power plant database and publications of national energy authorities [24, 25]. Part of the current power plant capacity was assumed to be retired by 2050 based on their estimated technical lifetimes. Initial wind power and solar PV capacities in each country were primarily based on the national 2020 targets [26], which resulted in a VG share of 24% of annual electricity demand for the Northern European case. 3.5 Investment costs and characteristics of thermal units Table 4 tabulates characteristics of thermal power plants that were available as investment options. The data is based on multiple sources, e.g. [9, 27-29]. Investment costs were converted to annual costs using an annuity payment factor of 0.094. It is possible to arrive at this annuity payment factor, for example, using 20 annual payments and a 7% discount rate. Table 4. Characteristics of thermal power plants (2050) Plant typea Investment cost, €/kW Fixed O&M costb, €/kW/a Variable O&M costb, €/MWh Efficiency, % Min. load c, % Unit size, MW Min. up time, h Min. down time, h CB CV Start-up costs, €/MW Start-up fuel consumption, MWh/MW CCGT 1300 40 0.8 60 30/50d 500 1 1 1.6 0.14 58 0.1 GT 550 17 0.5 40 5/10 60 1 1 0 0 25 0.4 ICE-C 670 17 3.5 46 10 20 1 1 0 0 0 0 ICE-E 1000 24 3.5 46 20 20 1 1 0.9 0.15 0 0 ST large 2000 60 2 36 20/35d 150 4 3 0.7 0.15 85 4.1 ST small 2800 85 2.1 36 25/40d 50 4 3 0.7 0.15 114 2.7 NU 4800 97 2.3 33 50 800 15 10 0 0 99 4.1 a CCGT: combined cycle gas turbine power plant; GT: gas turbine power plant; ICE-C: gas engine power plant (only electricity); ICE-E: gas engine power plant (electricity and heat); ST: biomass-fired steam power plant; NU: nuclear power plant. b O&M: operation and maintenance. c Minimum output of the unit in relation to the unit size. d For CCGT, GT, and ST units, the plant type with a smaller minimum load is used in the flexibility cases Flex, All and All + EV, and the plant type with a larger minimum load is used in the cases Base and HeatFlex. 3.6 Other region-specific data In addition to initial generation capacities, several other input parameters were defined for each region. These include annual electricity and heat demands, hourly electricity and heat demand profiles, minimum installed electricity and heat production capacities, annual biomass potential, maximum wind power capacities at different investment cost levels and hourly profiles for wind power, solar PV and hydro inflow. Moreover, hourly day-ahead forecasts were defined for electricity demand and wind power. The time series were from the year 2011. Historical wind power profiles were scaled in order to better match them to the capacity factors of state-of-the-art wind turbine generators; Table 5 presents the resulting capacity factors and also the annual electricity demands. The main data sources for these region-specific data were Nord Pool Spot, TSOs and Eurostat [30-37]. Initial net transfer capacities between price regions were based on ENTSO-E's plans [38]. Table 5. Region-specific data Region Wind power capacity factor, %a Solar PV capacity factors, %a Annual electricity demand 2020, TWh/a Annual electricity demand 2050, TWh/a Germany 30.8 11.4 557 600 Denmark East 39.4 10.4 16 18 Denmark West 40.5 9.9 24 25 Estonia 36.5 9.4 8 12 Finland 34.8 8.9 94 105 Lithuania 36.0 9.5 10 16 Latvia 37.1 9.8 7 10 Norway Middle 41.7 8.2 28 30 Norway North 42.8 7.4 8 9 Norway South 42.2 9.5 98 100 Poland 33.1 11.1 170 170 Sweden Middle 34.2 10.2 93 95 Sweden North 36.0 8.9 27 30 Sweden South 37.7 10.4 27 28 a Capacity factor: the ratio of average power generation over a year to the rated capacity. 4 Results The simulation results show the impact of VG and demand side flexibility on thermal power generation. Some results are shown for only some of the cases to keep figures more compact. The first subsection presents the results based on the generation planning model Balmorel. Then, annual production and emission results are presented, followed by economic results. Finally, cycling results are presented. 4.1 Future generation mix Fig. 1 presents the initial electricity production capacities and the electricity production capacities as invested by Balmorel. A significant amount of coal power was assumed to be in operation in 2020, but new investments in coal power were not allowed and the existing coal power plants were assumed to be phased out by 2050. In the Traditional scenario, large investments were made in CCGTs, whereas in the Windy and Sunny scenarios, new gas power plant capacity was more dominantly based on GTs and ICEs. Fig. 1Open in figure viewerPowerPoint Initial (striped) and invested (filled) electricity generation capacities. Investments in power-to-gas facilities were also allowed in 'All' and 'All + EV' but the model did not invest in them (a) Total generation capacity, (b) Thermal generation capacity split into categories (ST: steam turbines; CCGT: combined cycle gas turbines; GT: gas turbines; ICE: gas engines) Differences in optimal generation mix can be explained by changes in the net load. Net load is calculated by subtracting wind power production, PV production and run-of-river hydropower production from the load time series. Increasing the share of wind power and PV changes the net load in a way that indicates a transition from base load units to intermediate and peak load units [20]. Introducing district heating related flexibilities in the system decreased the investments in CCGTs and increased the investments in ICEs. A price sensitive demand response was able to replace a portion of GTs. EVs increased the electricity demand and consequently the total generation capacity needed to be larger. Table 6 presents the additional transmission capacities between price regions invested by Balmorel. In the Windy scenario, the model invested in larger transmission capacities than in the Traditional and Sunny scenarios. Windy2050_Base resulted in smaller transmission capacity investments than the next three flexibility cases, which can be explained by the smaller VG share in Windy2050_Base. When EVs were included, the model resulted in smaller transmission capacity investments compared with the three previous cases, although the share of VG was higher. The result indicates that smart charging of EVs displaces the flexibility provided by transmission links. Table 6. Invested transmission capacity between price regions Invested transmission capacity, GW 2020_reference 0 — — Sunny2050_HeatFlex 29 Windy2050_HeatFlex 32 Traditional2050_HeatFlex 10 — — Windy2050_All + EV 25 Windy2050_All 31 Windy2050_Flex 32 Windy2050_HeatFlex 32 Windy2050_Base 27 4.2 Annual production Fig. 2 shows the annual electricity production by fuel. In the Traditional scenario, district heating related flexibilities were not utilised significantly. In the different Windy cases, district heating related flexibilities increased the share of wind power while EVs increased predominantly the share of PV. Fig. 2Open in figure viewerPowerPoint Annual electricity productions. 'Electric' category represents the consumption of heat pumps and electric boilers in district heating grids, consumption of EVs, as well as battery losses Fig. 3 shows the resulting role of electricity (heat pumps and electric boilers) in the annual district heat production. Heat pumps, electric boilers and heat storages can reduce the amount of heat produced by natural gas by almost 70%, as can be seen by comparing Windy2050_Base and Windy2050_HeatFlex results. Fig. 3Open in figure viewerPowerPoint Annual heat production Compared with the 2020 results, CO2 emissions were reduced by 55% in Traditional2050_HeatFlex and by ∼70% in Windy2050_HeatFlex and Sunny2050_HeatFlex. From the different flexibility options, district heating related flexibilities and EVs played a key role in reducing the emissions. Wind power and solar PV curtailments were generally low in the studied scenarios: <1% in 2020_reference and Traditional2050_HeatFlex, 1.5% in Windy2050_HeatFlex, and 3.4% in Sunny2050_HeatFlex. 4.3 Economic aspects The results demonstrated that it is possible to achieve a large system benefit (1.5–2.5 billion euros per year) with flexibilities in district heating when the share of VG is high. Furthermore, it was possible to double the system benefit when additional flexibility options were included [20]. Higher system flexibility increased the share of wind power and PV (see Fig. 2). Due to this, electricity price volatility was higher in Windy2050_All + EV, which included flexibility from smart charging of EVs, compared with other Windy cases [20], although increased flexibility would by itself even out fluctuations in the electricity prices. The results highlight that electricity prices are sensitive to the combined changes in, e.g. the level of energy system flexibility and the share of different generation technologies. 4.4 Cycling of thermal power plants Fig. 4 shows the hourly ramps in the net load of Northern Europe. The 2050 cases in the figure are Traditional2050_HeatFlex (30% wind + 9% PV), Windy2050_HeatFlex