Low-cost greenhouse gas abatement via emission

Journal of International Council on Electrical Engineering

ISSN: (Print) 2234-8972 (Online) Journal homepage: http://www.tandfonline.com/loi/tjee20

Low-cost greenhouse gas abatement via emission intensities Aleksis K. Xenophon & Jin Zhong To cite this article: Aleksis K. Xenophon & Jin Zhong (2018) Low-cost greenhouse gas abatement via emission intensities, Journal of International Council on Electrical Engineering, 8:1, 206-213, DOI: 10.1080/22348972.2018.1539151 To link to this article: https://doi.org/10.1080/22348972.2018.1539151

© 2018 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group. Published online: 29 Oct 2018.

Submit your article to this journal

View Crossmark data

Full Terms & Conditions of access and use can be found at http://www.tandfonline.com/action/journalInformation?journalCode=tjee20

JOURNAL OF INTERNATIONAL COUNCIL ON ELECTRICAL ENGINEERING 2018, VOL. 8, NO. 1, 206–213 https://doi.org/10.1080/22348972.2018.1539151

ARTICLE

Low-cost greenhouse gas abatement via emission intensities Aleksis K. Xenophon and Jin Zhong Department of Electrical and Electronic Engineering, The University of Hong Kong, Hong Kong SAR, China ABSTRACT

ARTICLE HISTORY

Following recent international commitments to reduce greenhouse gas emissions, governments around the world are considering various emissions abatement strategies – with electricity sectors required to play a key role. Cap-and-trade schemes and carbon taxes are among the most frequently implemented and discussed policy approaches. However, these schemes can have negative economic impacts due to energy price increases, as well as their interactions with pre-existing taxes. An alternative approach, known as an emissions-intensity scheme (EIS), can mitigate the extent of these negative economic consequences. Rather than setting an absolute cap on emissions, an EIS targets the emissions intensities of individual generators by establishing a sector-wide emissions-intensity baseline. The current paper investigates the influence of EIS policy parameters on total emissions, electricity prices, and net scheme revenue. A simulationbased approach is adopted, utilizing a direct current optimal power flow (DC OPF) modelling framework. In doing so, the complementary roles of the emissions price and emissions-intensity baseline are illustrated. Data from the simulation based analysis are then used to demonstrate how, for some systems, an EIS may be calibrated to deliver emissions abatement without increasing electricity prices, and at no cost to the government.

Received 24 January 2018 Accepted 18 October 2018

1. Introduction Following the 2015 Paris Agreement on climate change, governments around the world are considering various policies to reduce greenhouse gas emissions. At present, most policy approaches place a price on emissions, either through the imposition of a tax, or the establishment of emissions trading schemes (ETS) [1]. While schemes with broad emissions coverage, such as the European Union Emissions Trading Scheme (EU ETS), have the advantage of exposing a large quantity of emissions to a carbon price, they can be difficult to manage effectively. This is particularly true for the EU ETS, where permit prices have collapsed in recent years [2]. In order to overcome these challenges, there has been increased interest in policies tailored to specific sectors, in particular those involved in electricity and heat production [3,4]. The abatement potential in these sectors alone is significant, with emissions from electricity and heat production constituting 42% of global CO2 emissions [5]. An approach designed to reduce electricity sector emissions, known as an emissions intensity scheme (EIS), has been recommended to the Australian Government by independent consultants [6], and more recently by an expert advisory body [3,7]. The scheme involves targeting the emissions intensity (emissions per unit of electrical output) of generators by establishing a CONTACT Aleksis K. Xenophon

[email protected]

KEYWORDS

Emissions abatement; emissions intensity schemes; climate policy; revenue recycling

sector-wide emissions-intensity baseline. Each generator is given an emissions allowance equal to the level set by the baseline for each MWh they produce. Generators with emissions intensities exceeding the baseline are obligated to purchase additional permits, while those generators with emissions intensities below the baseline can sell their surplus emissions allocations [3,6,8]. Over time, the baseline can be decreased, incentivising the early implementation of efficiency improvements to existing plant, and investment in renewable energy technologies. A key advantage of these schemes is that revenue is recycled among generators via a transparent rule-based mechanism. Internalisation of these transfers enables an EIS to achieve abatement targets while minimising its impact on other sectors of the economy [3]. While emissions abatement schemes involving output-based rebating have been extensively covered in the economic literature [9–12], they are less often discussed in the context of power systems. When the electricity sector has been considered, an EIS is shown to have less of an impact on electricity prices relative to other abatement mechanisms [3,6,8]. Additionally, schemes that allocate permits based on output, such as an EIS, are shown to have less of a distortionary impact on investment decisions [13]. Despite the demonstrated advantages of an EIS, little work has been done to investigate

LG201 Chow Yei Ching Building, The University of Hong Kong, Pokfulam, Hong Kong SAR, China

© 2018 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

JOURNAL OF INTERNATIONAL COUNCIL ON ELECTRICAL ENGINEERING

the optimal design of EIS scheme parameters. While [14] has considered the design of an optimal tax under an EIS in the context of the cement industry, it does not consider subsidies paid to those below the emissionsintensity baseline. Similarly, while [13] conducts an indepth analysis of the long-run implications of an EIS variant, little work is done to analyse the short-run effects of such a scheme. It is the objective of this paper to address these issues. The following analysis assumes that a policymaker is able to set an emissions-intensity baseline, along with an emissions price. These policy parameters are changed incrementally, with generator dispatch decisions recorded for each combination. Using these dispatch decisions, total emissions, electricity prices, and net scheme revenue (total penalties collected minus total credits paid) are observed. In doing so, the complementary functions of EIS scheme parameters are clearly illustrated – a key contribution of this paper. An additional contribution is the demonstration of a simulation-based approach that can help guide the selection of EIS policy parameters – facilitating the attainment of specific environmental and economic objectives. The paper also explores unique properties associated with an EIS. Specifically, it illustrates the scheme’s ability, in some instances, to be simultaneously price and revenue neutral. For the system investigated, it is shown that an EIS can be calibrated such that a 14% reduction in emissions is realised, without increasing the price of electricity, or coming at a cost to government. It should be noted that the formulation of the EIS previously described differs slightly from [3,4], where it is the trade of permits between generators that sets their price. The implications of these differences are discussed in Section 5. The paper is structured as follows. Section 2 formulates an EIS model, along with metrics used to assess its performance. Section 3 describes the data used to test the model formulated in Section 2. Section 4 presents the results, with Section 5 providing analysis and discussion of these results. Section 6 concludes the paper.

2. Model A direct current optimal power flow (DC OPF) modelling framework has been used to analyze the impact of an EIS on nodal electricity prices, total CO2 emissions, and scheme revenue (total penalties minus total credits). It should be noted that a DC OPF model is a linear approximation of the more accurate alternating current optimal power flow (AC OPF) modelling framework. Despite DC OPF models being unable to consider transmission losses, it is still considered an appropriate framework for the

207

current analysis. This is because it is the dispatch decisions of generators under different EIS policy scenarios that is the primary focus of this study – not issues related to network congestion or topology. These additional issues are likely to be the focus of future work. The formulation of a DC OPF model which incorporates an EIS is given by (1). min Pi

N X ½Ci ðPi Þ þ ðei ϕÞPi T

(1:1)

i¼1

s.t. Pi; min Pi Pi;max "i

Pi PiD ¼

N X

Bij θi θj "i ðλi Þ

(1:2)

(1:3)

j¼1

θREF ¼ 0

(1:4)

The DC OPF model given by (1) considers a network of N buses, each indexed by i. The load at each bus is PiD (MW), while Bij (S) is the value of the susceptance between nodes i and j. Voltage angles at each bus relative to the reference bus are given by θi (rad). The generalized cost function for a generator at bus i producing power at rate Pi (MW) is Ci ðPi Þ ($/h). Under an EIS, generators are exposed to an additional penalty or credit which is proportional to their output level, given by ðei ϕÞPi T. The magnitude of this penalty or credit per MWh for generator i is equal to the difference between its emissions intensity, ei (tCO2/MWh), the emissions-intensity baseline determined by the government, ϕ (tCO2/MWh), multiplied by the emissions price, T ($/tCO2). The objective of (1) is to minimize the total cost of generation (1.1), subject to power generation limits at each bus (1.2), power balance at each bus (1.3), and the voltage angle at the reference bus fixed to zero (1.4). A simulation-based approach was implemented by incrementally altering T and ϕ, running the model for each combination, and recording the generation level, Pi, at each bus. These different generation levels were then used to calculate total emissions, E (tCO2/h), using (2). E¼

N X

ei Pi

(2)

i¼1

Net scheme revenue, R ($/h), was calculated by summing the net penalties paid by each generator, and is given by (3), while the electricity price at each bus is given by the dual variable, λi , associated with constraints (1.3).

208

A. K. XENOPHON AND J. ZHONG

R¼

N X ðei ϕÞPi T

(3)

i¼1

There are several reasons why a simulation-based approach has been adopted. The most important is its ability to exposit the effect T and ϕ have on E, R, and λi . Computing these measures over the T – ϕ space allows the impact of the policy parameters to be easily visualized – clearly illustrating their complementary functions. Additionally, data obtained from the simulations can be used to identify parameter combinations that achieve specific economic objectives. Two objectives discussed in this paper are price and revenue neutrality. Price neutrality is achieved when the imposition of an EIS does not change electricity prices relative to the business-as-usual (BAU) case (e.g. electricity prices under an EIS are the same as under the scenario where T = 0). A T – ϕ combination is said to be revenue neutral when total penalties collected equals the total credits paid (e.g. combinations where R = 0).

3. Data An IEEE 118 bus system has been used to implement the model described in (1), with quadratic generator cost functions for coal, gas, and oil units taken from http://motor.ece.iit.edu/data/ltscuc/. It should be noted that these data do not include emission intensities. To deal with this, an emissions intensity allocation procedure was devised. First, emissions intensity ranges for coal, gas, and oil units, shown in Table 1, were calculated based on data published by the Australian Energy Market Operator (AEMO) [15]. For each fuel-type, generators with the lowest marginal costs (MC) were assigned the highest emissions intensity within their respective ranges, and vice versa. Using these high and low MC generators as reference points, emissions intensities were assigned to the remaining units via linear interpolation. In this way, the analysis considers a worst-case scenario, where prior to the imposition of an EIS, the cleanest units (for each fuel type) were the most expensive to operate. While this allocation method was possible for coal and oil units, it could not be used for gas generators, as the cost function parameters were identical for all units. In Table 1. Emission-intensity range by fuel-type [15]. Fuel Type Coal Gas Oil

Emission intensity range (tCO2/MWh) 0.891–1.558 0.423–1.012 0.900–1.014

this case, the 11 gas units were assumed to have emissions intensities at equally spaced intervals over the gas emissions intensity range shown in Table 1. From the set of generators, 54 were selected, yielding a final generation mix of 74% coal, 15% gas, and 11% oil. The minimum level of generation for each unit was assumed to be zero (Pi,min = 0). Demand at each bus was scaled such that total demand was equal to 60% of system capacity. It should be noted that this analysis does not consider renewable generators. If included, the implementation of an EIS would result in large profits for existing renewable plant, without leading to additional emissions abatement. This is because the MC for renewable plant (especially hydro, wind and solar) are typically lower than gas and coal units, meaning they would be dispatched first anyway. Omitting existing renewables from the implementation of an EIS is broadly consistent with recommendations found in [4]. While it is acknowledged that renewable plant constructed after the implementation of an EIS would be under the scheme’s remit, the current analysis is limited to the consideration of short-run impacts only. Also, given that the installed capacity of renewables is typically low compared to fossil-fuel fired plant, their exclusion is unlikely to have an appreciable effect on the main findings.

4. Results The model outlined by (1) was run for combinations of T and ϕ, with T 2 f0; 1; . . . ; 100g, and ϕ 2 f0; 0:05; . . . ; 1:3g. A total of 2,727 T – ϕ combinations were tested, with the least-cost dispatch found for each combination using GAMS. Nodal electricity prices were observed for each combination, with (2) and (3) used to calculate total emissions and net scheme revenue respectively. Total emissions were plotted over the T – ϕ space, as can be seen in Figures 1. and 2. It can be seen from Figures 1 and 2 that only T influences total emissions. This follows from the fact that ϕ is fixed for all generators, meaning that all units receive the same gross output subsidy per MWh. Consequently, ϕ cannot change the relative cost between generators. This is simply demonstrated by considering two units, each with the following generalized total cost (TC) and marginal cost (MC) functions: TC ¼ Ci ðPi Þ þ ðei ϕÞPi T 0

MC ¼ Ci ðPi Þ þ ðei ϕÞT

(4) (5)

Consider the case where, for given values of T and ϕ, the difference in marginal costs between Units 1 and 2 is given by γ, as is shown in (6).


209

Figure 1. Total emissions surface.

Figure 2. Total emissions contours.

C1 0 ðP1 Þ þ ðe1 ϕÞT ðC2 0 ðP2 Þ þ ðe2 ϕÞT Þ ¼ γ (6) Expanding terms, we see that: C1 0 ðP1 Þ C2 0 ðP2 Þ þ ðe1 e2 ÞT ¼ γ

(7)

Evidently ϕ has no impact on the relative cost of production between units 1 and 2. In this way, one can consider T and ϕ as having complementary roles. T serves to bring about a change in the merit order of production, changing dispatch decisions, and with it total emissions. This contrasts with ϕ, which determines the level at which to subsidize output per MWh, influencing net revenue generated by the scheme, and the price of electricity. The impact that ϕ and T have on electricity prices is shown in Figures 3 and 4. First, consider the case where ϕ = 0, and T is incrementally increased. This scenario corresponds to a carbon tax where the government collects all revenue. As the emissions price is increased, electricity prices also increase. Note that for any given T, if ϕ increases, the electricity price decreases. This is because increasing ϕ amounts to increasing the output subsidy paid to generators, decreasing their MC, which in turn decreases the price of electricity.

It is important to be cognizant of the competing objectives of mitigating electricity price increases, and having a scheme that is costly to run. This is illustrated by the surface and contour plots of Figures 5 and 6 respectively, which show how T and ϕ alter net revenue generated by the scheme. The shape of the net revenue surface in Figure 5 is similar to the price surface in Figure 6. However, while a policymaker may find high electricity prices undesirable, a large amount of net revenue is desirable. This is because revenue from the scheme could be used to achieve other policy objectives. It could also be passed on to households and businesses, through tax cuts or other concessions, to compensate for rising energy prices. However, also note how these concessions become less necessary if electricity prices are not substantially affected relative to the BAU case. Therefore, a trade-off exists. Policy makers can choose to earn large revenues at the cost of increased electricity prices. Alternatively, revenue can be forgone in favor of increasing the emissions-intensity baseline – effectively providing generators with a larger output subsidy per MWh, decreasing generators’ production costs, resulting in lower electricity prices. As the simulations have demonstrated, an EIS can give policymakers considerable flexibility when

210


Figure 3. Average nodal electricity price.

Figure 4. Average nodal price contours.

Figure 5. Net revenue surface.

Figure 6. Net revenue contours.


pursuing economic and environmental objectives. For the purposes of this analysis, the case of revenue neutrality (total credits equal total penalties), and price neutrality (an EIS does not change the price of electricity relative to BAU) is considered. This scenario is important because it directly deals with concerns governments are likely to have when implementing an EIS. The first is that an emissions abatement policy will lead to an increase in the price of electricity, which will have negative flow-on effects throughout the rest of the economy. Secondly, governments may be reluctant to devote significant financial resources to greenhouse gas abatement policies. By considering a revenue-neutral scheme, an EIS can be implemented at no cost (or very low cost if one considers the resources required to administer the scheme) to government. Consequently, consideration of these scheme configurations is useful, as it deals with the major economic and political barriers facing emissions reduction policies. The combinations of T and ϕ that result in revenue and price neutrality are shown in Figure 7. The level of emissions for each combination is also shown for reference. In Figure 7 we can see that up until an emissions price of 21.5 $/tCO2, the price-neutral baseline is less than the revenue-neutral baseline. It is also known that increasing the baseline leads to a decrease in the price of electricity, as shown in Figure 4 (because the output subsidy for all generators increases). Therefore, points above the blue curve in Figure 7 result in lower electricity prices relative to the BAU scenario. Similarly, from Figure 6, for a given emissions price, decreasing the baseline leads to an increase in net revenue. This follows from the fact that decreasing the baseline results in more penalties being collected, and fewer credits paid out. Consequently, points below the red curve result in net revenue increasing relative to the BAU case. This implies that setting T and ϕ in the orange area of Figure 7 will result in electricity prices decreasing and positive net revenue, while simultaneously placing a price on emissions – leading to

Figure 7. Revenue and price-neutral T – ϕ combinations.

211

abatement. If revenue and price-neutrality is desired, it can be achieved by setting the T and ϕ at the intersection of the price-neutral and revenue-neutral contours. In this case the emissions price is 21.5 $/tCO2, and the emissions-intensity baseline is 0.99 tCO2/MWh. This policy parameter combination amounts to a 14% reduction in emissions – solely brought about through changing the relative costs of generators.

5. Analysis and discussion The preceding analysis illustrates the mechanics of an EIS, and demonstrates the capability of such a scheme to have a moderating impact on electricity prices when placing a price on emissions. This is due to the recycling of penalties and credits among generators. A subtle but important differences exists between the revenue recycling discussed here, and that used in countries such as Canada, which purports to have implemented a ‘revenueneutral carbon tax’ [16]. Unlike the Canadian scheme, which relies on tax cuts and offsets to minimize the impact of a carbon tax on households, an EIS internalizes transfers within the electricity sector. This obviates the need for complex re-distribution arrangements, and is one reason an EIS has been shown to reduce emissions at a relatively lower economic cost when compared to other schemes [3,8]. Another important distinction between an EIS and widely implemented abatement methods, such as an ETS, is the consistency of the price signal sent to electricity market participants. The EIS discussed in this paper considers the case where the emissions price is exogenously determined. This contrasts with an ETS where the demand and supply of permits sets the price. Permit price volatility may dissuade potential investors, especially when returns on renewable energy projects are contingent on the emissions price. Therefore, the relative stability of the emissions price under an EIS with an exogenously determined

212


emissions price may be beneficial to investors. However, the disadvantage is that scheme revenues may fluctuate with a fixed emissions price. If the emissions intensity-baseline is incorrectly calibrated, total credits may exceed total penalties, causing an EIS to run at a deficit. It is therefore important to not only consider the static short-run implications of an EIS, but also its medium and long-run implications. For this reason, considering varying levels of demand over multiple periods is an important extension, and constitutes part of the scope for future work. It must be mentioned that the EIS investigated in this paper differs slightly from that presented in [3]. In [3], demand and supply determine the permit price, whereas here it is set exogenously. The benefit of the EIS in [3] is that it will always be revenue neutral (the government only sets the baseline, and allows generators to trade amongst themselves). However, the disadvantage is that permit prices may fluctuate, sending a less clear signal to investors. The advantage of fixing the permit price, while beneficial for investors, may come at the expense of fluctuating scheme revenues. Further analysis is required to determine the magnitude of these differences, and is also part of the scope for future work.

6. Conclusions In summary, the careful calibration and design of an EIS can achieve both environmental and economic objectives. One of the main advantages of this scheme is its ability to recycle collected revenues among generators using a transparent rule-based mechanism. The analysis illustrates the complementary roles that the emissions price and emissions-intensity baseline play within an EIS. The interaction and effect of these parameters on net revenue, electricity prices, and total emissions can be clearly seen through the simulationbased approach employed. This approach also identified T – ϕ combinations that reduce electricity prices and deliver net revenue to government. These outcomes are likely to be viewed favourably by policymakers wishing to reduce emissions, but unwilling to bear negative economic shocks. These advantages make an EIS a good policy option for leaders wishing to establish a clear and consistent signal to encourage renewable energy investment, as well as efficiency improvements, while at the same time minimising the economic impact of an emissions abatement policy.

Disclosure statement No potential conflict of interest was reported by the authors.

Funding This work was supported by the Key R&D Project of China [Grant 2016YFB0901903].

Notes on contributors Aleksis K. Xenophon received a Bachelor of Mechanical Engineering and a Bachelor of Economics (Hons) from the University of Adelaide. He is currently undertaking a PhD in electrical engineering at the University of Hong Kong. His research interests include the design of emissions abatement mechanisms for power systems, the development of open-grid models, and the use of machine learning techniques to analyse electricity markets. Jin Zhong received the B.Sc. degree in electrical engineering from Tsinghua University, Beijing China, and the Ph.D. degree from Chalmers University of Technology, Gothenburg, Sweden. She is currently an Associate Professor in the Department of Electrical and Electronic Engineering at the University of Hong Kong. Her research interests include power system operation, electricity sector deregulation, ancillary service pricing, and smartgrid.

References [1] World Bank, “State and Trends of Carbon Pricing 2016,” 2016. [2] Koch N, Fuss S, Grosjean G, et al. Causes of the EU ETS price drop: recession, CDM, renewable policies or a bit of everything?—new evidence. Energy Policy. 2014 Oct;73:676–685. [3] CCA, “Policy options for Australia’s electricity supply sector - Special Review research report,” Climate Change Authority, Australia, Special review research report, Aug 2016. [4] AEMC, “Integration of energy and emissions reduction policy,” Australian Energy Market Commission, Sydney, Dec 2016. [5] IEA, “CO2 emissions from fuel combustion Highlights,” International Energy Agency, Paris, France, 2016. [6] Frontier Economics, “Options for the design of emissions trading schemes in Australia,” Frontier Economics, Melbourne, Apr. 2008. [7] CCA, “Towards a climate policy toolkit: special Review on Australia’s climate goals and policies,” Climate Change Authority, Australia, Aug 2016. [8] Frontier Economics, “Emissions reduction options - A report prepared for the Australian Energy Market Commission (AEMC),” Nov. 2016. [9] Fischer C, Fox AK, “Combining rebates with carbon taxes: optimal strategies for coping with emissions leakage and tax interactions,” Social Science Research Network, Rochester, NY, SSRN Scholarly Paper ID 1408692, May 2009. [10] Bernard AL, Fischer C, Fox AK. Is there a rationale for output-based rebating of environmental levies? Resource and Energy Economics. 2007 May;29(2):83–101. [11] Fischer C, “Rebating environmental policy revenues: output-based allocations and tradable performance standards,” Resources for the Future, Washington, D.C, 2001.


[12] Burtraw D, Fraas AG, Richardson ND, “Tradable standards for Clean Air Act carbon policy,” 2012. [13] Zhao J, Hobbs BF, Pang J-S. Long-run equilibrium modeling of emissions allowance allocation systems in electric power markets. Operations Research. 2010 Feb;58(3):529– 548. [14] Almutairi H, Elhedhli S. Carbon tax based on the emission factor: a bilevel programming approach. Journal of Global Optimization. 2014 Apr;58(4):795–815.

213

[15] AEMO, “Emissions intensities of available generators,” 2017. [Online]. Available from: https://www.aemo. com.au/Electricity/National-Electricity-Market-NEM/ Settlements-and-payments/Settlements/CarbonDioxide-Equivalent-Intensity-Index. [16] Murray B, Rivers N. British Columbia’s revenue-neutral carbon tax: A review of the latest ‘grand experiment’ in environmental policy. Energy Policy. 2015 Nov;86:674–683.