IMPACTS OF WIND POWER ON ENERGY BALANCE OF A HYDRO DOMINATED POWER SYSTEM Juha Kiviluoma and Hannele Holttinen, VTT Finland VTT Technical Research Centre of Finland, P.O. Box 1000, FIN-02044 VTT, Finland, Tel. +358-20-722 6671 Fax +358-9-722 7048, e-mail: email@example.com Summary The increasing penetration of wind power in the electricity networks changes the energy balance of the system and the operation of regulative capacity. In the Nordic system, there are large hydro reservoirs that can be used to compensate for the varying and partly unpredictable nature of wind. The results indicate that regulating wind power is not a problem for a hydro dominated system. Other issues are likely to arise before regulation. Since wind power has very low marginal costs, it pushes other forms of production out of the markets. This affects especially condensing power plants, but also CHP plants are operated less. Wind power lowers the price levels in the system and changes the way hydro power reservoirs are operated over the year. In a case where the hydro dominated Nordic system is injected with 30% wind power penetration, cost duration curve drops significantly. The marginal power is most of the time nuclear or CHP and during windy periods the price goes even down to zero. New and most of the existing investments would not be profitable in such a setting. A price drop caused by increase in wind production in the Nordic countries can be mitigated with new transmission capacity to the continental grid or with additional consumption possibly in the form of heat pumps, electric vehicles or hydrogen production. Also the functioning of the market mechanism can be questioned, when most of the power comes from sources with low marginal costs. Introduction With this study we are trying to get some answers for the questions: How hydro power regulates large amounts of wind power in hydro dominated system? How the energy balance of the hydro dominated system changes with large amounts of wind power? What types of power plants are affected? How the prices react? Wind power production is variable and partly unpredictable. It is capital intensive and has low marginal costs. Hydro power has also low marginal costs. In the Nordic system hydro power is mostly reservoir hydro with large reservoirs and good possibilities to regulate the output power. When these two power ingredients form the backbone of a power system, one can expect to get some problems. First, one has to ask if the regulative capacity of hydro power is enough to compensate for the variation in wind production. Second, since both have low marginal prices, it could mean that power market might get too low prices to cover up investments in new power production. To assess the dependencies between hydro and wind power, we use detailed energy system market model (WILMAR) of Nordic countries and Germany operating in hourly time-scale. It includes stochastic properties for wind to present the variability and partial unpredictability of this energy source and a long term model to present the scheduling of hydro power reservoirs according to value of water stored in reservoirs. Power system in the model is divided into 12 regions which have limited transmission capacity between each other. The model also handles large number of separate district heating areas, which is important in order to incorporate CHP production properly. This article continues the work of Holttinen  in analysing the impacts of large scale wind power production on the Nordic power system. Compared to previous work, this study uses a new model, which can represent the hourly changes in wind power and thus produces more accurate results on some issues. Second, the analysis has been extended to a case where the energy penetration of wind has increased to 30%. In the report “Long-term challenges in the electricity system” Elkraft  analysed effects of 21 GW of wind power in the Nordic system. The model used for the analysis was modified Balmorel, which was the basis for the model used in our study as well. The main differences between Elkraft’s and our Wilmar model are that in Wilmar there is stochastic water value model included and time-series for wind are made taking into account the geographical dispersion of large-scale wind power. Also there are more heating networks. The inclusion of Germany in the model makes our results more robust in regards to transmission outside the examined area. On the other hand Elkraft report makes assumptions on plants that are decommissioned beyond 2010. This results in more realistic and constrained capacity balance and higher prices during high consumption – low wind periods. However, it can be argued that by taking some capacity away between the cases one would also lose comparability between different wind cases and the base case. Research setting In order to see effects of increasing share of wind power, we have presumed a Nordic power system for the year 2010. Since we are analyzing large increases in wind power production, it would have made more sense to analyse a system further in the future. However, the changes farther into the future are more uncertain. There are several important uncertainty factors (including fuel prices, CO2 allowance prices, fossil and renewable subsidies, nuclear acceptance, transmission grid improvements and price development of renewables) that could change the system decisively. Our aim was not to predict the future. Rather we wanted to see how wind power changes hydro dominated system and that can be seen just as well with a 2010 system. Model description and limitations The study is conducted using a model developed in EU-project WILMAR (Wind Power Integration in Liberalised Electricity Markets). The model optimises power markets based on a description of generation, demand and transmission between defined model regions and derives electricity market prices from marginal system operation costs. The model is a stochastic linear programming model with wind power production as the stochastic input parameter, although stochasticity is not utilised in this study due to long solving time of whole year modelling. The model optimises unit commitment taking into account trading activities of different actors on different energy markets. The resulting prices are optimal and do not include factors like market power, so they indicate lower boundary for prices. The regulation market prices are low since there are no additional transaction costs. Three electricity markets and one market for heat are included in the planning model: 1. A day-ahead market for the planned delivery of electricity. 2. An intra-day market for handling deviations between expected production and consumption agreed upon the day- ahead market and the realised values of production and consumption in the actual operation hour. Regulating power can be traded up to one hour before delivery at the intra-day market. Currently the model does not have load forecast errors, so the demand for regulating power is caused only by wind forecast errors. In this study deterministic wind production was used instead of stochastic predictions in order to reduce computation time. This meant that intra-day market did not bring any changes to the dayahead market scheduling. 3. A day-ahead market for automatically activated reserve power. The demand for these ancillary services is determined exogenously to the model. However, maximum forecasted changes in wind power are taken into account when reserving ancillary services. 4. Due to the interactions of CHP plants with the dayahead and intra-day market, markets for district heating and process heat is included. Since the model has optimisation horizon of the spot market and hydro power has water reservoirs that can hold water for much longer periods of time, the model uses separate long term model to get water values for the hydro content. The hydro power presentation in the model is not currently detailed enough to make robust analysis of extreme regulation situations. River systems are sometimes complex due to flow restrictions and reservoir locations compared to downstream power stations. Also winter time presents additional restrictions to the utilisation of hydro power. These are not taken into account, which leads to an overestimation of the flexibility of hydro power. Also the long term model has only one reservoir for the whole model area, and therefore the model has only one water value for all the reservoirs in the model. Load flow restrictions are not present in the model, which may in some cases lead to overuse of transmission lines if the real system was run like the model suggests. Transmission limits are set only as capacity for each line. In addition to current transmission lines between model regions, the model has new transmission lines listed in Table 1. A more detailed description of the model is given in Brand et al . Transmission line Model regions Capacity Fennoskan 2 FI_R – SE_M 800 MW Nea-Järpströnnen NO_M – SE_N 750 MW Storebelt DK_W – DK_E 600 MW NorNed NO_S – DE_NW 700 MW Konti-Skan 1 update SE_M – DK_W +90 MW Ajaure - Rossagu NO_N – SE_N +100 MW Skagerrak IV NO_S – DK_W 600 MW Table 1. New transmission lines in the model. Case descriptions The research consists of four cases. The base case holds only planned changes to the 2010 system. 10% case has additional wind power so that the wind energy penetration in Norway, Sweden and Finland is 10% in the analysed year. Also Germany and Denmark have bit more wind power than in the base case. The only difference between the 10% case and 20% and 30% cases is additional wind power (see Table 2). Wind power is added as zero price production time series, which are created from original wind or wind power production time series by using methods [1, 3] that take into account the smoothing factors caused by large scale production and geographical dispersal of capacity. base 10% 20% 30% Wind capacity NO+SE+FI 2.5 17.8 35.7 52.5 [GW] Germany 28.6 35.8 35.8 35.8 Denmark 4.1 4.6 4.6 4.6 Energy from wind 16 49 87 119 NO+SE+FI+DK [TWh] Table 2. Wind power capacity and production in different cases. The rest of the assumptions are same for all the cases. One of the decisive factors for future power systems is the price of CO2 allowances, which is set to 17 €/CO2 ton. Another important factor are fuel prices, these are shown in Table 3 and Table 4. In hindsight peat and oil prices are too low and natural gas prices probably too high compared to the general price level. Current Model prices IEA IEA prices (Finland) (2010) (2030) Light oil 53 113 - - $/barrel Fuel oil 46 66 - - $/barrel Crude oil - 60 22 29 $/barrel Coal 74 69 40 44 $/t Natural gas (Europe) 8.4 5.9 3.3 4.3 $/MBtu Table 3. Study fuel prices compared to current prices in Finland and IEA scenarios. Model [€2002/GJ] Wood waste 4 Wood 4.3 Straw 4.4 Waste 0 Peat 1.5 Light oil 7.19 Fuel oil 6.16 Coal 2.25 Natural gas 6.16 Nuclear 0.35 Table 4. Fuel prices used in the study. The modelled year uses 2001 profiles for hydro inflow, wind, load and heat demand. Demand time series have been adjusted to presumed level of 2010 consumption. Hydro inflow in the year 2001 was average. In the start of the year reservoir were higher than normal. Spring floods were quite small, but autumn rains were large. The year was not particularly windy. The model has import time series for electricity from Russia and Poland, which are also from year 2001. Both transmission lines are fully utilized for most of the time. In the highest wind cases imports should cease for part of the year since Nordic prices go so low. However, this does not happen in the model. If it would, there would be less hydro spillage. There is around 6000 MW of new power capacity other than wind added to the system during 2004-2010. About half of this is nuclear. During the same time around 2200 MW of power capacity is expected to be decommissioned, most of it is coal. Figure 1 shows the capacity balance of the model during the hour of highest consumption. There would still be more capacity available, but the model has not opted to bring them online, so this is not the same thing as ordinary capacity balance. The model actually has over 93 GW of non-wind capacity for the year 2010, so the modelled system is not stressed with this consumption level. The model does not include restrictions for the usage of hydro power capacity and therefore the hydro capacity is too high especially during winter. 85 Capacity balance in Nordic countries [GW] 80 Primary reserves 75 Non sp inning reserves 70 Realised wind p roduction 65 Production (not wind) without wind 60 ts e n lin io en pt on itm um y m c it ns m pa Co Co Ca Figure 1. Capacity balance of the model in the Nordic countries during the hour of highest consumption Results First of all, production forms with high marginal costs are often forced out of the market when wind penetration increases. Their full load hours and the number of hours they are in use go down to such extent that some of them will likely not be profitable any more. In particular, the model results indicate that other condensing than nuclear suffers a lot. In the 30% wind case the number of hours when one of these sets the marginal price is less than 300. Also CHP based on fuel oil, natural gas, wood or straw experience markedly reduced amount of production hours (see Figure 2). If there were more heat boilers, then CHP might be producing even less. Nuclear is used all the time, but usually not with full output. Figure 2. Weekly average electricity production of the Nordic CHP plants. Since the spring is somewhat dry, hydro power is not used that much in the first half of the year. The additional wind power is partly used to help to fill the reservoirs. During the winter prices stay high enough for CHP and some of the condensing plants. When the summer - autumn proves to be rainy, most of the reservoirs fill up to their maximum levels and hydro spillage is inevitable. Hydro power becomes very cheap, and since it’s the time of low consumption, prices plummet. They start to come back up in the end of the year when the hydro inflow diminishes and consumption increases. Although there are clear differences between the wind cases, the trend toward much lower prices in the rainy season is clear. Price levels in the model regions are depicted in Figure 3. Prices in Germany are not much affected by the low levels of prices in the Nordic countries. The amount of cheap power from Nordic countries is small compared to the size of the German market and therefore does not have much effect. Figure 3. Price levels in the model regions. w10, w20 and w30 refer to wind cases 10%, 20% and 30%. DE_CS, DE_NE, DE_NW are German regions; DK_E, DK_W Danish; FI_R Finland; NO_M, NO_N, NO_S Norwegian; SE_M, SE_N, SE_S are Swedish. Figures 4, 5, 6 and 7 show the weekly averages of power production, consumption and transmission as well as hydro spillage, wind shedding and average Nordic price level. There are several things worth noticing. There is lot less wind shedding than hydro spillage. Value of shedded wind is equal to the market price of the shedding hour. Hydro, on the other hand, is valued according to the future market price of the hour in which the water would have been used. Water in the reserves has thus potential to have high value. During winter there is no need to either shed or spill, since it is not yet known whether there will be so much water inflow that hydro reservoirs will fill up. Later during spring, when it becomes certain that the reservoirs will overflow, the water value in the reservoirs goes to zero, but the value of wind stays at the market price. So wind is shed only when the market price goes down to zero, or when wind shedding can help the system to regulate and avoid costly start-ups of thermal plants. The larger the area where wind power is located, the smaller the short time-scale variations become, since the weather patterns do not hit the wind power parks simultaneously. However, in the weekly time scale, there is considerable variation from week to week, since pressure areas can affect winds in large areas similar fashion for few days. This can be seen especially well in the graphs showing Swedish production. Variation in weekly average wind production increases with the increasing wind penetration. In a hydro dominated system bulk of the variation is smoothed out by hydro power. However, when the penetration gets higher also thermal plants start to participate. Since Swedish production is dominated by hydro and nuclear, nuclear also reduces output during windy weeks. This should not be technically problematic, since the changes are not fast and nuclear is able to ramp at those rates. However, it is problematic to cover the high capital costs of nuclear if utilisation ratio drops too much. Danish system is highly thermal apart from the wind power. Therefore it is not surprising that with increasing wind penetration biggest decrease in production happens in Denmark. Danish production is replaced by imports from Sweden and Norway. In Finland overall production increases although the system is quite thermal as well. One reason is that in our cases wind does not increase much in Denmark and it does in Finland (see Table 2). Another reason is that Finnish nuclear has low marginal cost and does not decrease as much as Danish condensing power. Figure 4. Weekly averages of power production, consumption and transmission in the base case. Figure 5. Weekly averages of power production, consumption and transmission in the 10% wind case. Figure 6. Weekly averages of power production, consumption and transmission in the 20% wind case. Figure 7. Weekly averages of power production, consumption and transmission in the 30% wind case. So far we have not really shown how the system regulates large amount of wind power in the hourly time-scale. As previously mentioned, the hydro representation is not detailed enough to give a proper picture of the available regulation capacity. Stochastic presentation of wind power would also make the case more realistic. However, keeping these in mind, one can still have a look at Figure 8, which shows a high consumption – low wind situation in the 30% wind case. Demand in the Nordic countries is little below 70 GW. Wind blows only for 5.2 GW worth during the most difficult hour from 9:00 to 10:00. Hydro produces at 41 GW and nuclear full power at 14 GW. CHP produces 11 GW. However, coal condensing is producing only 500 MW, peat 300 MW and natural gas 50 MW. On top of this Nordic countries are exporting to Germany at full power. However, this situation is probably not realistic. During winter time part of hydro power capacity is not really available as it has been in the model. There should probably be few gigawatts of condensing power in production to make up the difference or the transmission to Germany should be reversed. The difference could also be covered by demand side measures, if they were available. One should also remember that although that 5.2 GW of wind production is not much when the wind capacity is 53 GW, it is still quite a lot compared to the current non-nuclear condensing capacity in the Nordic system, which is less than 10 GW. Wind lends to capacity credit even in difficult situations. During high consumption periods the prices are high and accordingly hydro power is producing as much as it can (this is the period where it can make profits). During those times it cannot provide positive reserves unless it holds back in order to be able to contribute to those reserves. In the described hour about 35 % of the ancillary reserve need of 6.5 GW is provided by hydro power. Biggest part of the reserved reserves is condensing power. Figure 8. 24 hours of high consumption – low wind power during February 26th. Conclusions 20-30% penetration levels change the energy balances a lot. As a cautious conclusion one can say that regulation is not likely to be a major problem for a wind-hydro system. However, this needs to be studied more carefully with better representation of hydro systems than used in this study. In the 30% case regulation by other condensing than nuclear is needed only for few hours and can be provided cost-effectively by demand side measures or, if necessary, by gas turbines. More pressing concerns for wind power are site availability, investment costs, fossil fuel prices, local grid issues and acceptability. Our study highlights how 2010 system would change with high amounts of wind. In reality high wind would change feasibility of future power plants. Wind power will not leave many operating hours for condensing plants and therefore additional condensing plants in wind and hydro dominated system should probably have low capital costs compared to operational costs. This would raise the prices during high consumption – low wind periods. However this will only happen once the old plants with lower marginal costs are decommissioned, but this could take time since it might be more cost-effective to leave the old plants as reserves than to build new ones. In the 30% case amount of high price hours would be minimal, since wind plus hydro could cover almost all of the situations. The addition of demand side response advocated by Nordel  would be one way to help to keep the prices down. If one assumes that the amounts of wind power analysed in this study could become reality at some point in time, it raises couple of serious issues for the future of the hydro dominated power systems. The results show that utilisation rates of some plants drop a lot. This is obviously an issue for the companies owning those plants as well as for possible future investments in those types of plants. A serious issue for wind power is the general drop in the price level. The prices go so low that investments in wind power or in any other form of production are not interesting unless there is some form of subvention, for example fixed prices. The drop in price level can be compensated by adding new connections to the continental grid, which will have a great demand for CO2 free electricity if the long range CO2 targets are to be met. Another possibility is to increase electricity consumption. There are strong pressures to replace oil in the traffic sector and electricity can help either with plug-in vehicles or hydrogen production. Also heating sector could use electricity with heat pumps or electric boilers, although the latter is waste of exergy of electricity. If the power prices are low it could also attract energy intensive industry to increase consumption. However, since all power production has negative effects, it is problematic to have low prices which do not give incentives for saving energy. Wind penetration in a liberalised hydro dominated system cannot go high without experiencing problems of too low prices. The way markets currently operate, condensing plants with low capital costs and high operational costs are needed as marginal plants to guarantee a price level at the spot market that is profitable for new investments. Currently Nordic hydro power gets its water value from the expectation that it can produce during profitable periods when condensing plants are price setters. Additional wind power pushes those price setters away from the market and this leads to a dramatic drop in the water values as well as in the spot prices. Situation is made more difficult by the fact that both hydro and wind have stochastic inflows of energy with substantial variation between different years. We think that the issue of low prices could indicate that price mechanism in hydro-wind system is not fully functional. It could well be that in the future hydro-wind system might be the cheapest option to power the Nordic system from the socio-economic perspective. If this is the case, then the market mechanism would not give incentive to reach this cheapest option. Nobody, disregarding consumers, would build wind power capacity up to 30% penetration if the prices would get as low as indicated by the results of this study. This should hold true even if one takes into account that in this study no capacity has been taken away when the wind was added, because peak capacity was used only during few hours. References 1. Holttinen H. The impact of large scale wind power production on the Nordic electricity system; VTT Publications 554, PhD dissertation; Otamedia: Espoo, 2004. 2. Elkraft System. Long-term challenges in the electricity system – Wind power and natural gas. Elkraft System: 2004. 3. Brand H, Weber C, Meibom P, Barth R, Swider DJ. A stochastic energy market model for evaluating the integration of wind energy, 6 th IAEE European Conferenc; Zurich , 2004. 4. Nordel. Developing demand response on the Nordic electricity market. Nordel: 2005.
Pages to are hidden for
"IMPACTS OF WIND POWER ON ENERGY BALANCE OF A"Please download to view full document