• Print
  • Decrease text size
  • Reset text size
  • Larger text size
28/08/2017

THE MISSING MONEY PROBLEM IN THE BELGIAN POWER SECTOR

QUANTIFYING THE MISSING MONEY OF GAS-FIRED POWER PLANTS BY 2030

The large-scale deployment of renewable energy sources (RES) tends to push conventional gas-fired power plants out of the merit order, some of which are shutting down due to profitability issues. This is commonly referred as the Missing Money Problem: the investments in gas-fired power plants require important upfront expenditures whereas the declining power prices and the limited utilization rate of the aforementioned power stations do not ensure return on investments.

As long as Belgium can rely on its nuclear capacity, the missing money is not much of a threat as the closing of unprofitable capacity does not directly impact the security of supply. However, by 2025, the seven nuclear power plants in the Belgian territory are foreseen to shut down. This phase-out of 6 GW nuclear capacity will require massive investments in baseload generation.

In order to cope with the missing money problem and incentivize investments in the power sector, the Belgian Government is considering introducing capacity remuneration mechanisms[6].

More than relying solely on the revenues made from the generated energy, market players would also be remunerated for their available capacity.

The goal of this paper is to quantify the missing money of gas-fired power plants in the Belgian power sector by the horizon of 2030.

Determining the Belgian generation portfolio in 2030

Introduction to the model

The first step in the quantification of the missing money is the determination of the Belgian generation portfolio by the year 2030. The elaboration of the most optimal electricity mix in 2030 is calculated by minimizing the total cost to society [1] [2]:

where i represents each generation technology namely: Onshore wind, Offshore wind, Solar PV, CCGTs, OCGTs, biomass-fired generation, coal-fired generation with carbon storage and finally the interconnectors, and h represents each hour of the year.

Several constraints have been taken into account during the optimization exercise, notably:

  • The match between supply and demand, at each hour of the year. As proposed by reference [1], wind power and sun power are treated as negative demand. To this purpose, typical hourly wind profiles and sun’s radiation profiles of Belgium have been incorporated into the model.
  • The ramping rates of the N different technologies.
  • A periodic maintenance factor which downscales the installed capacity to the available capacity.
  • The renewable targets foreseen by the National Renewable Energy Action Plan [3], as lower boundary condition of the installed RES capacity. Furthermore, with the deployment of Norther and Rentel, offshore wind will likely reach a minimum capacity of 2200 MW [7].
  • A conservative maximal import capacity, estimated at 4000 MW for 2030, as upper boundary condition. 4000 MW represents Elia's current investment effort to bring the current interconnection capacity to 6500 MW around 2020 (approximately 50% of our peak demand [8]), with an availability handled as a probabilistic variable with an expected value of 66%. [8].

 

Input parameters

In order to compute the most cost-optimal electricity mix in 2030, several assumptions regarding the future electricity supply and demand have been made.

Regarding the demand, two opposite trends suggest an annual electricity consumption in 2030 roughly similar to the consumption in 2010 (~90 TWh). On the one hand, Belgium is moving towards a full electrification of the society. One factor being the increasing popularity of electric vehicles, another one being the ever-growing demand for electric appliances in our daily lives. On the other hand however, Belgium is aiming for serious efficiency targets, public awareness campaigns on rational use of energy, smart metering as well as strict residential insulation measures. These would suggest a drop in the electricity demand. More precisely, the Federal Planning Bureau [3] estimates the total Belgian electricity consumption to lie between 91 TWh and 93 TWh in 2030.

To this end, the demand used in this study consists of a total energy consumption similar to the Belgian 2010’s demand curve, but with peaks slightly higher and off-peaks slightly lower than the original curve.

Regarding the supply, the different generation technologies inputs – namely the ramping rates and the investment, O&M, fuel, CO2 costs – have been incorporated into the model, based on references [1] and [2]. No nuclear technology is allowed since the Belgium nuclear phase-out will be completed in 2025. Additional flexibility mechanisms such as demand response and hydro-storage have been integrated into the model, as fixed parameters, accounting for respectively 889 MW and 1300 MW[2].

For the sake of clarity, only two scenarios have been retained for the constrained optimization. As proposed by reference [2], the first scenario considers a targeted CO2 allowance price of 35,4 €/MWh with objective to incentivize investments in green generation. The second one considers a more conservative price of 6,5 €/MWh.

 

Results of the model

The results of the cost optimization are shown in Figure 1. Both scenarios result in similar electricity mixes in terms of installed capacity. However, the scenario with a high CO2 cost relies more on biomass-fired power plants as they are less carbon intensive than gas-fired generation. Consequently, less gas-fired capacity is installed compared to the scenario with low CO2 price, as it is partly substituted by biomass.

Similarly, the scenario with a high CO2 price relies less on OCGT generation because this technology is responsible for considerable amount of carbon emissions[1].

It is also interesting to note that coal-fired generation with Carbon Capture and Storage is not present in any of the scenarios’ results, due to its excessive costs.

Figure 1: Projected installed capacity in Belgium in 2030

In terms of total investments costs, the model computes a total amount of 22 Billion € for the high CO2 price scenario against 20 Billion €[2] for the low CO2 price scenario, this difference mainly originates from the higher investments costs of biomass technology which is more present in the first scenario.

Figure 2 gives an overview of weekly load curve in January with the optimal mix as calculated by the model. On the one hand, technologies such as interconnectors and biomass-fired generation both act like true base load. On the other hand, gas-fired generation together with hydro turbines act rather like intermediate or peak load, to cope with the intermittency of the renewables.

Figure 2: Power generation of installed capacities - low CO2 price scenario

Validation of the model

Another common way to determine the optimal generation mix is by means of the Load Duration Curve [1].  This method will be used to validate the model presented before. The load duration curve represents the number of hours of the year during which the demand exceeds a certain level. By plotting the costs of each technology on top of the load duration curve, the optimal electricity mix can be deduced, in terms of minimum social welfare cost [1]. Using the same input parameters as for the model, the 2030 curve is shown in Figure 3.

Figure 3: Load duration curve and marginal costs of conventional units

Note that it is purely static method; the technique does not account for dynamic limitations such as the intermittency of renewable generation, sudden load peaks nor the ramping rates of different technologies. Therefore, the cost-minimization with operational constraints is more accurate than the load duration curve technique. Nevertheless, the latter is still a useful tool to approximate and validate the results.

The residual load duration curve is obtained by subtracting from the original curve the intermittent generation of renewables, the imported electricity and the generation or consumption of flexible mechanisms such as storage and demand response (Figure 3). The residual load duration curve represents the demand that needs to be fulfilled by conventional power plants.

It should be noted that the already existing biomass plants, accounting today for 420 MW, constitute one of the two only technologies that acts as a true base-load, the other one being the interconnectors. The former are assumed to receive green certificates up until 2030, moving them to the beginning of the merit order, whereas new plants are assumed to do without. These 420 MW have therefore been deducted from the residual load duration curve.

For operating hours exceeding ~800 hours, CCGTs become more profitable than OCGTs (Figure 3).  Moreover, the residual load duration curve exceeds the level of 7000 MW approximately 800 hours during the year. This means that, in order to reach a minimal social welfare cost, 7000 MW should be filled by CCGTs and the remaining capacity - during couple of hours - should be filled by OCGTs, as they would operate less than 800 hours during the year. This remaining capacity accounts for a range between 1000 and maximum 2500 MW.

In conclusion, the load duration curve technique suggests a maximal mix of approximately 7000 MW in CCGTS, 2500 MW in OCGTs, and no extra biomass capacity on top of the initial 420 MW already in place. This result is consistent with the results of the cost minimization with operational constraints, shown in Figure 1, that resulted in 7500 MW in CCGTs, 2000 MW in OCGTs and 420 MW in biomass power plants.

 

Determining the missing money in 2030

Missing money of gas-fired generation

The large-scale deployment of Renewable Energy Sources tends to push power prices down, as these technologies operate with almost zero marginal cost. As a consequence, conventional power plants are being pushed out of the merit order and some of them are shutting down due to profitability issues. Paradoxically, renewables are intermittent and require the support of dispatchable and flexible power plants, especially in times of low load factors to balance the lack of renewable generation.

This leads to an intricate situation where gas-fired generation is non-profitable yet crucial for balancing purposes in the power system. This issue is commonly referred to as the missing money problem.

Based on the model’s optimal electricity mix, the bottom line (revenues – costs) of each technology can be analyzed quantitatively. Whenever it is negative, the term missing money is used. The power price is assumed to be equal to the marginal cost of the last (i.e. most expensive) running unit in the merit order. The model does not account for scarcity-rent on top of the quasi-rent in times of scarcity, as the optimal portfolio foresees enough installed capacity to cover the demand every hour of the year. Moreover, the model does not take into consideration additional revenue streams for ancillary services that gas-fired generation could potentially benefit from.

Using the calculation logic explained above, the missing money for the CCGTs comes down to 39 500 €/MW/year, while OCGTs have a bottom line of 27 300 €/MW/year.

CCGTs and OCGTs are the units with the most expensive marginal costs, hence the last technologies in the merit order. They are the most affected by the missing money problem. Indeed, the 420 MW of biomass-fired generation foreruns gas-fired generation as the former benefits from subsidies. All the remaining technologies have an annual positive bottom line and are not affected by the missing money issue. Their investment costs can easily be recovered for two main reasons: first of all, these technologies are activated almost all year long, while gas-fired generation is only activated to balance out the intermittency of renewables; secondly, their marginal cost is significantly lower than the power price, resulting in a higher margin in €/MWh compared to gas-fired generation.

 

Capacity remuneration mechanisms

For the above reasons, under the hypothesis of a purely competitive market, CCGTs and OCGTs would theoretically have left the market by 2030. As a matter of fact, this behavior is already observable in the energy-only market: gas-fired plants are decommissioning since they are no longer profitable. Yet, this technology is crucial to balance out the mismatches between supply and demand and more importantly to ensure the security of supply, especially after the phase-out of the Belgian nuclear sector that accounts today for almost 6 GW of base load power.

During winter 2015 to 2017, Belgium has introduced Strategic Reserves [4] to tackle this issue on a short-term. However, on a longer-term, Belgium is thinking of introducing other types of capacity remuneration mechanisms that would allow market players to be remunerated not only for the energy they produce but also for the capacity they provide. This is a common way to keep flexible power plants into the market.

A thorough analysis of the different capacity remuneration mechanisms that would suit the Belgian power landscape is outside the scope of this paper. However, for the sake of completeness, the most common mechanisms are listed below [4]:

  • Capacity Payments, a price-based mechanism in which the price in €/MW is set by an independent party. This mechanism is currently deployed in Spain, Portugal and Ireland. It is also seen as the most expensive CRM.
  • Strategic Reserves, a volume-based mechanism in which a limited capacity is tendered from offline power plants and activated in times of scarcity. Such a mechanism is currently deployed in Belgium, Finland and Sweden
  • Capacity Auctions, a volume-based mechanism where the capacity is tendered centrally, typically by the TSO, to cover the peak demand in the years to come. Such a mechanism is deployed in the United States (PJM grid) and is to be deployed as well in Great Britain.
  • Capacity Obligations, a volume-based scheme in which large consumers are entitled to reserve a volume of capacity in line with their power offtake, either by relying on their own means (power plants, demand response) or by buying capacity certificates in the market. This system is deployed in the United States (CAISO grid) and in France.

 

Case study on Capacity Payments

Depending on the selected mechanism, the adequate remuneration can be calculated. As an example, for the mechanism “Capacity Payments”, the remuneration per MW is calculated in such a way that gas-fired generation is no longer facing profitability issues, meaning the power stations do not have to leave the market. According to the model, CCGTs are the units that will be facing the largest amount of missing money, about 39 500 €/MW/year. It is thought to be the most expensive scenario as all technologies are financed by the missing money of CCGTs.

According to Figure 3, the residual load duration curve reaches its peak at a range of 8 000 and max 10 000 MW (for couple of hours). This corresponds to the minimal required installed capacity of flexible power plants. The introduction of such expensive Capacity Payments would therefore require 395 million euros per year to be distributed among market players, proportionally to their installed capacity in flexible power plants. This amount would be borne by the end consumers in return for offering the security of supply, resulting in an increase of the electricity bill estimated at 4,36 €/MWh according to the model, or 13 €/year for an average household in Belgium. The cost of the current short-term solution to the issue, the Strategic Reserve, attains approximately 0,67 €/MWh or 61 million euros [4], which is significantly less, although not future-proof for Belgium’s socio-economic developments.

In comparison, Ireland tendered a market-wide capacity of 7 046 MW in 2015 at a price of 81 750 €/MW/year, resulting in an annual capacity payment sum of 576 million euros, even though the island has a lower demand in peak and volume [5].

 

Sensitivity analysis

The missing money issue being largely associated with the large-scale deployment of renewable energy sources, it is interesting to study its evolution in function of the penetration of renewables in the power system.

The deployed flexible capacity –mainly composed of gas-fired generation– is almost independent of the share of renewables installed, as a fixed amount of dispatchable capacity is crucial to cover the peak demand in times of scarcity, when the meteorological conditions are not suitable for intermittent generation. This is shown in Figure 4. As a result, the higher the penetration of renewables, the larger the installed capacity required to cover the same demand curve.

Surprisingly, although the annual operating hours of CCGTs are highly impacted by the penetration of renewables, the missing money is only slightly affected by the increase in climate target. This can be explained by the fact that the margin of the last technology in the merit order is assumed to be equal to zero according to the model. Consequently, its fixed costs cannot be covered regardless of whether those units are activated or offline. This is shown in Figure 4.

Quantitatively, the sensitivity study has shown that for an increase in wind target of 70 % compared to the initial target, the missing money only increases by 5 %. This slight increase is simply due to the decrease in OCGTs operating hours, only hours of the year during which CCGTs are able to make a margin as they forerun OCGTs in the merit order.

Figure 4: Sensitivity analysis of the missing money issue

Conclusion

Massive investments will be needed during the next decade in the power sector, about 20 to 22 Billion euros depending on the CO2 price. The projected installed capacity appears to rely mainly on gas-fired technology together with intermittent renewables, as well as biomass-fired generation in the high CO2-price scenario.

However, in order for the gas-fired generation to be profitable, the energy market will need to be coupled with a capacity remuneration mechanism. Indeed, CCGTs were identified to lack 39 500 €/MW/year in order to be profitable. If capacity payments would be introduced, a range between 61 and maximum 395 million € per year would be needed considering the aforementioned cost-optimal production park.

The missing money is only slightly impacted by an increase in the penetration of renewables, about 5 % for an increase in wind target of 70 %. The rationale behind this is that, even though CCGTs operating hours will decrease with the rise of renewables, they operate with almost zero margin. Consequently, their fixed costs cannot be covered regardless of whether gas-fired generation are running or offline.

Authors: Arthur Talpaert and Jean Trzcinski

 

Sources :

[1]: DETERMINING OPTIMAL ELECTRICITY TECHNOLOGY MIX WITH HIGH LEVEL OF WIND POWER PENETRATION, Cedric De Jonghe, Erik Delarue, Ronnie Belmans, William D’haeseleer

[2]: THE FUTURE ELECTRICITY MIX IN BELGIUM, Sia Partners, 2015

[3]: CLIMATE AND ENERGY FRAMEWORK FOR BELGIUM, 2015

[4]: ASSESSING THE COST OF THE STRATEGIC RESERVES, Sia Partners, 2015

[5] Capacity Requirement and Annual Capacity Payment Sum for Calendar Year 2015, Commission for Energy Regulation, 2014

[6] http://www.lecho.be/r/t/1/id/9784635, retrieved on 2 July 2016

[7] http://www.lecho.be/r/t/1/id/9776295, retrieved on 10 June 2016

[8] ÉTUDE DE L’ADÉQUATION ET ESTIMATION DU BESOIN DE FLEXIBILITE DU SYSTÈME ÉLECTRIQUE BELGE, Elia, April 2016

 

Copyright © 2017 Sia Partners. Any use of this material without specific permission of Sia Partners is strictly prohibited.


[1] Relative to the quantity of MWhe produced.

[2] Our first study from 2015 [2] depicted an amount of respectively 14 B€ and 30 B€ respectively for the high and low CO2 prices, for the horizon of 2025. The difference between the 2015 and 2017 model originates mainly from the hypotheses (e.g. the marginal prices, the availability of import capacity at 60% [Note [8]: Elia foresees a capacity of maximal 6500 MW], the foreseen wind capacity, the demand curve) but also from the optimization toolbox in Matlab that has been adapted.

0 comment
Post a comment

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.
Image CAPTCHA
Enter the characters shown in the image.
Back to Top