Energy Economics 33 (2011) 33–43
Contents lists available at ScienceDirect
Energy Economics j o u r n a l h o m e p a g e : w w w. e l s e v i e r. c o m / l o c a t e / e n e c o
Electricity tariff design for transition economies Application to the Libyan power system Javier Reneses a,⁎, Tomás Gómez a, Juan Rivier a, Jorge L. Angarita b a b
Instituto de Investigación Tecnológica, Universidad Pontificia Comillas, c/Alberto Aguilera 23, 28015 Madrid, Spain Europraxis Operations, Paseo de la Castellana 50, 28046 Madrid, Spain
a r t i c l e
i n f o
Article history: Received 8 May 2009 Received in revised form 8 April 2010 Accepted 9 April 2010 Available online 4 May 2010 Keywords: Electricity tariffs Libyan power system Cost causality Regulation Transition economies
a b s t r a c t This paper presents a general electricity tariff design methodology, especially applicable for transition economies. These countries are trying to modernize their power systems from a centralized environment (with normally, a public vertically integrated electric company) to a liberalized framework (unbundling electricity companies and, eventually, starting a privatization process). Two issues arise as crucial to achieving a successful transition: i) ensuring cost recovery for all future unbundled activities (generation, transmission, distribution and retailing), and ii) sending the right price signals to electricity customers, avoiding cross-subsidies between customer categories. The design of electricity tariffs plays a pivotal role in achieving both objectives. This paper proposes a new tariff design methodology that, complying with these two aforementioned criteria, requires a low amount of information regarding system data and customer load profiles. This is important since, typically, volume and quality of data are poor in those countries. The presented methodology is applied to computing tariffs for the Libyan power system in 2006, using real data. © 2010 Elsevier B.V. All rights reserved.
1. Introduction Tariff design is a key issue for regulatory authorities given that tariffs are the interface between electricity companies and final consumers. Properly designed tariffs are essential both for ensuring that the system is used to the best advantage in the short-term and for mapping out long-term demand trends. Tariffs must achieve two main objectives: first of all, they must generate the income required to cover all the costs of supplying electricity. Secondly, they must send the right economic signals to each customer to ensure that they use the service in the most efficient way, socio-economically speaking. Tariff design is a crucial part of every regulatory framework. Generally speaking, there are a number of regulatory principles that are essential for tariff design (Berg and Tschirhart, 1988; Pérez Arriaga and Smeers, 2002). Among them, three can be highlighted, especially in the first stages of the regulation processes: • Business sustainability. This is the basic principle of most regulatory schemes and is intended to guarantee that suppliers recover all the accredited costs incurred in producing the regulated good or service. Sustainability can guarantee an industry's viability, and the extent to which this principle is met is determined in two phases. The first involves determining the tariff level, which is directly related to how each activity is regulated and its accredited costs determined. The second involves the selected tariff design and its capacity to
⁎ Corresponding author. Tel.: + 34 915422800; fax: +34 915423176. E-mail address:
[email protected] (J. Reneses). 0140-9883/$ – see front matter © 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.eneco.2010.04.005
recover these accredited costs. In both phases, this principle may clash with the principle of efficiency, described below. Business sustainability is related to another regulatory principle: cost additivity. This principle implies that final tariffs must be designed by adding different tariffs, each of them linked to the cost of one activity (typically, generation, transmission, distribution and retailing or customer services). This cost additivity will allow the clear identification of the costs of the different activities and, hence, make it possible to establish adequate remuneration for the providers of those activities. • Economic efficiency. Two types of economic efficiency are of interest here: ○ Productive: involving the production of the good or service at the lowest possible cost, for a predefined level of quality. Productive efficiency is not really achieved through tariff design, but through efficient remuneration of each activity. ○ Allocative efficiency: in other words, allocating the resource to whoever uses it best or values it the most. In this case, instead of allocating the resource, a price signal is issued to encourage each consumer to use the amount of the resource that is most efficient for the system as a whole. This signal must serve both for the short and the long term. The way to ensure that economic efficiency is achieved is by using the cost-causality principle, that is to say, assigning the costs to whoever originates them. According to the very abundant economic literature on tariffs for regulated activities, short-term marginal costs/prices should be used whenever possible to achieve this goal. Those prices, based on
34
J. Reneses et al. / Energy Economics 33 (2011) 33–43
the cost-causality principle, give the right signals for the short-term operation of the system. Unfortunately, very often these signals clash with the principle of sustainability, having, in some cases, even more disadvantages than advantages, as in the case of distribution marginal costs.1 So other criteria must be used to “round out” the tariffs and ensure that revenues match accredited costs. These criteria have to be consistent with the mentioned cost-causality principle. Lastly, if there are any other costs that cannot be attributed to anyone (such as the costs related to the regulatory commission, the system operator and the market operator), other allocation criteria, usually called “second best”, must be applied. One such criterion, designed to distort consumer short-term behaviour as little as possible, is to apply Ramsey prices,2 taking due care to avoid clashing with the non-discrimination principle. • Equity or non-discrimination between different customers. To apply this principle, it is necessary to first define what discrimination is and what it is not, a definition that can be usually found in the power system legal framework. The general consensus of opinion regarding the application of this principle is that rates are held to be nondiscriminatory if consumers are charged the same amount for using the same good or service, regardless of the purpose for which it is used and the nature of the consumer. This does not contradict the cost-causality idea, according to which everyone must be allocated the part of the cost that they are responsible for, in terms of how they use the good or service. In fact, meeting this principle has much to do with the level of consumer information that can or must be used to allocate costs of service. In economic terms, more efficient tariffs can be designed if consumers' cost utility function is known. Yet that entails knowing the purpose for which the service is used and would lead to discrimination between different consumers. Care must also be taken when applying Ramsey or second-best tariffs, because under this system the users who are least elastic to price changes are charged the most. Once again, this allocation may be discriminatory if based on users' “private” characteristics (such as the level of income or the intended utilization of the electricity), and not strictly on external objective data, i.e., the contracted capacity, the volume of energy consumed or the utilization pattern. Current legislation usually stipulates the amount of information that can be used to determine tariffs. A subject directly linked with discrimination is the existence of cross-subsidies between tariffs. Usually, it is considered desirable not to have cross-subsidies, that is to say, each customer pays exclusively and completely his own costs. The most common definition of a tariff without any cross-subsidies is that the tariff has to be fixed between the marginal cost of the good or service, and the cost of providing such a good or service only to the customer whose tariff is being defined (stand alone cost). But there are situations in which, for a variety of reasons, cross-subsidies to certain type of customers are wanted, for example low-income customers. Such cross-subsidies are justified if there are objective criteria to identify the customers that have to be subsidized. These criteria for “positive” discrimination are consciously designed and have a social or economic aim. Sometimes, to prevent major tariff disruptions, the non-discrimination principle may have to be applied gradually before and after a new tariff methodology is enforced.
1 The use of marginal costs for setting tariffs in distribution systems has important drawbacks. For example, two similar users can be charged a very different price only depending on the configuration of the network (Rodríguez Ortega et al., 2008). 2 Ramsey pricing is based on assigning costs to customers inversely to their demand elasticity. That is to say, the users who are the least elastic to price changes are charged the most. This way, the impact of those costs (which cannot be directly attributed to any customer) on the short-term behaviour of customers is minimized, since elastic consumers (who could change their consumption pattern) are charged the least (Ramsey, 1927).
Other principles that should be born in mind when designing tariffs are transparency in tariff setting, the simplicity of the process, stability in the long term, and consistency with existing regulations. It is hard, and at times impossible, to comply with all of them at once, at least fully. At times, this is simply because of a lack of knowledge and at other times, because principles clash with each other. The final objective is to strike a reasonable balance between the aforementioned principles. Keeping in mind these general principles, this paper proposes a systematic and comprehensive methodology for designing electricity tariffs, with a special focus on the application of the methodology to transition economies. We refer to a transition economy as one which is changing from a centrally planned to a free market regime. The common objectives in transition economies are related to achieving economic liberalization for the main productive sectors (including restructuring and privatization), macroeconomic stabilization, and the creation of a private financial sector. In the electricity sector, these countries aim at modernizing their power systems by migrating from a centralized scheme organized typically around one state-owned vertically integrated electric company to a liberalized framework. This liberalization involves unbundling electricity companies and, eventually, starting a privatization process (Rothwell and Gomez, 2003). The design of a new tariff structure in accordance with the general principles previously enumerated and described is a necessary step of key importance to a successful transition process. The tariff design has to show that the different activities (generation, transmission, distribution and retailing) of the electricity sector which is to be unbundled will be profitable and their revenue flows stable in the medium and long term. This is a necessary condition if the sector is to be made attractive to possible future investors. As regards the previously mentioned tariff design principles, in transition economies special emphasis has to be placed on recovering all costs involved in each of the different electricity activities (cost recovery), and to allocate those costs among the different customer categories in an objective and transparent way, avoiding crosssubsidies between them (correct economic signals). The proposed methodology ensures cost recovery for generation, transmission, distribution and retailing as separated businesses and allocates their costs in a transparent and objective manner to the different customer categories. This will allow a clear determination of the way in which the costs of those activities are recovered, providing stability to the businesses and making them attractive to investors. Therefore, the final tariff is computed by adding the tariffs corresponding to each activity (cost additivity). The new tariff structure should also be kept as simple as possible (simplicity principle), allowing a smooth transition from the past situation, with its subsidized customer categories, to a new one in which explicit subsidies can be implemented for low-income customers (and are not generalized for specific customer categories). Although the tariff design principles mentioned above are also applicable to liberalized electricity systems with already unbundled businesses and implemented electricity markets, the focus of the paper is to show how they are implemented in detail for transition economies. Here some hypotheses regarding how to decompose the total utility cost between the different activities have to be formulated. In addition, in developed electricity sectors it is customary the use of marginal prices in trading between electricity production and consumption, while in some developing or transition countries the concept of economic dispatch based on merit order is not fully implemented. Another issue affects customer categories and subsidies they receive. In developed countries, in general, customer categories have been defined according to load profiles and voltage levels. However, in transition countries these are new concepts that necessarily imply a large-scale transformation of the final tariff structure. The proposed methodology is designed to test different final structures and identify current levels of subsidies between customer categories.
J. Reneses et al. / Energy Economics 33 (2011) 33–43
Finally, in developed countries the amount of available information is normally greater than in transition economies, allowing the design of more complex tariff structures, for instance with high time discrimination (i.e. more time blocks) and signals to customers (who should be better prepared to react to complex tariffs than customers in transition countries).3 Many publications have dealt with the problem of determining the revenues permitted for regulated activities (see, for example, Green and Rodríguez Pardina, 1999; Román et al., 1999; Rudnick and Donoso, 2000; Liston, 1993; Jamasb and Pollitt, 2000; Pollitt, 2005; Lowry and Getachew, 2009). However, there is an important lack of literature on the allocation of these revenue requirements between final customers. In addition, the proposed approaches make use of an important amount of information (Rodríguez Ortega et al., 2008; Larsson, 2003) that, usually, will not be available in transition countries. The tariff design methodology that is used in this paper makes use of basic information about the sector that normally is known to (or can be estimated by) regulators. Once the liberalization process is mature enough, a second-generation regulation can be implemented, normally making use of a larger amount of information. It is important to note that tariff design methodologies used worldwide are not usually made public in detail. In some cases, electricity tariffs have become a political instrument and the corresponding government is not interested in having a close and transparent methodology.4 In other cases, regulators are mainly concerned about the total cost recovered by companies, but the companies are free to propose the final tariff structure to the regulator. In general, scientific publications discuss the theoretical aspects of the problem: the use of short or long-term marginal costs, revenue reconciliation, etc. (Bonbright, 1961; Feldstein, 1972; Weston, 2000). However, the authors have not found in the literature a comprehensive and detailed method such as the one presented in this paper. The benefits of applying the methodology presented in this paper in a transition country can be considered to be very significant. First of all, the segregation of costs by activities will clearly reveal the economic situation of each part of the integrated company and will provide clues about its future sustainability and attractiveness for private investors. Second, appropriate customer categorization and the computation of the required tariffs will also clarify one important issue: the level of commercial losses and in which customer types they are concentrated. Finally, sending the right electricity prices to final consumers will achieve more efficiency in the use of electricity, even if some direct subsidies are implemented for low-income consumers belonging to specific customer categories. The paper presents the results obtained when applying the proposed methodology to the computation of electricity tariffs for the Libyan power system in 2006. This process is the result of a project promoted by the General Electric Company of Libya (GECOL), with the participation of IFS,5 Soluziona6 and Comillas University. The paper is organized as follows. Section 2 introduces some of the essential elements in tariff design. Section 3 describes the proposed methodology for cost allocation, while Section 4 presents a case study, 3 The authors know that these assumptions are not completely satisfied in some liberalized electricity markets, in which the proposal of this paper could also be very useful. However, we consider that, from an academic point of view, the proposed methodology is better applicable for transition economies, while liberalized economies should use a second-generation tariff design, sending more elaborated prices and options to customers and, consequently, making use of a higher amount of information. Actually, in many industrialized countries (like in the EU) electricity tariffs have been abolished, and only network access tariffs or default tariffs (for some consumer classes) exist. Still, the principles in this paper can be applied to these residual tariffs for the applicable components. 4 The situation differs considerably from country to country. While some of them have established a tariff structure with a high degree of complexity and options, others just have set a very simple tariff structure, setting one single price for each customer category without using any methodology. This is the case of the Libyan power system. 5 http://www.ifsworld.com/. 6 Nowadays, Soluziona belongs to Indra: http://www.indra.es.
35
in which tariffs are computed with the actual data of the Libyan system in 2006. Finally, Section 4.2 summarizes the paper. 2. Tariff design framework As it has been stated above, the first step involved in a tariff design process is to clearly define the allowed total income which has to be recovered through tariffs. This total income is the result of adding the allowed incomes for the different activities involved in electricity supply: mainly, generation, transmission, distribution, and retailing (customer services). The second step is dividing these costs into the different cost drivers. For the tariff design proposed in this paper, three cost drivers have been considered: peak demand (kW), energy consumption (kWh) and number of customers. Thus, three different charges will be treated throughout the paper: a demand charge7 (in monetary units, m.u., per kW), an energy charge (in m.u. per kWh), and a fixed charge (in m.u. per customer). After the allowed costs are split between the different cost drivers, the next step consists of allocating them to the different customers' categories previously defined (customers clustering is not within the scope of this paper; for the Libyan system the current categories have been used). This allocation is carried out by using the cost-causality principle, i.e., by taking into account the specific contribution to the costs of each activity that can be attributed to the energy consumption and the peak demand of the different customers. Finally, the tariff structure can be computed using the additivity principle for tariffs. This is to say, for each customer category, time block and cost driver, all the unit costs are added in order to obtain the final tariff. The next subsections describe some key aspects that must be defined in order to achieve tariff design: network model, loss factors, time-block definition and customer categories. The terminology used in the paper is detailed in Appendix 1. 2.1. Network model In order to design a tariff scheme, a network model has to be defined to calculate energy flows between different voltage levels. This network model has to be made as simple as possible and with a degree of aggregation so that similar unit cost networks are grouped together, and no information is lost when a specific network is assigned to a common category. Fig. 1 shows a simple network model which is used in the case study for the Libyan system and could be suitable for other power systems. It is important to note that this simple network model considers that energy flows from higher to lower voltage levels. This simplification should be acceptable for most power systems, although the growth of distributed generation might change the current situation. All the network costs of the system have to be distributed between voltage categories, as will be shown in Section 3. The network model makes it possible to compute, for example, the total energy generated in the system EVLg, adding the energy demand from all customers: h i E EVLg = ∑ EVLg ðck Þ = ∑ EVLk ðck Þ⋅ 1 + fk;g : Ck
Ck
ð1Þ
Note that EVLg(ck) denotes the energy demand in voltage level VLg due to the customers ck located at VLk level, while EVLk(ck) denotes the actual energy demand of customers ck located at VLk level. The energy demanded at VLg is affected by the energy loss factors f E. 7 Through the paper, this charge to the peak demand (in m.u./kW) will be referred as a demand charge, although it could also be called a capacity charge.
36
J. Reneses et al. / Energy Economics 33 (2011) 33–43
losses will be covered will be necessary, but this point is beyond the scope of this paper. The significance and the treatment of these commercial losses will be briefly presented in the case study in Section 4. 2.3. Time-block definition
Fig. 1. Network model used for representing the Libyan power system.
The same computation can be made for the total peak demand PVLg at the generation level: h i P PVLg = ∑ PVLg ðck Þ = ∑ PVLk ðck Þ⋅ 1 + fk; g : Ck
Ck
ð2Þ
2.2. Loss factors Energy and power loss factors have to be taken into account if the previously defined network model is to be applied. In this model, energy loss factors can be defined to represent the losses caused by energy flows between different voltage levels. These energy loss factors fe;E j can be defined in order to represent how energy demand at a voltage level VLe affects a higher voltage level VLj (j N e). For instance, when allocating generation costs, it is necessary to know the total energy demand EVLg at the generation level VLg, taking into account all the energy loss factors. Similar concepts can be applied to demand. Hence, at a certain moment, the total demand PVLg at voltage level VLg can be computed using the power loss factors fe;P j and the demand caused at that level * by the customers located at all the voltage levels. Since there are systems in which only one of these loss factors is known, there is an empirical equation that relates energy and power loss factors (Flinn et al., 1983): E
P
fr;k = fr;k ⋅ðc + ð1−cÞ⋅lf Þ:
ð3Þ
In this relationship, c is a coefficient which depends on the shape of the load curve and lf represents the load factor of the system, computed as a function of the total energy demand EVLg divided by the Max , both of these terms referring to the generation peak demand PVLg level VLg: lf =
EVLg Max 8760⋅PVLg
:
ð4Þ
It is important to note that these coefficients refer only to technical losses. Some power systems in transition countries may also have an important amount of commercial losses (i.e., electricity that is consumed but not paid for), mainly located at the low-voltage distribution level, VL0. Arrangements between the electricity supplier and the regulatory authorities to determine how the cost of these
According to the principle of economic efficiency, it is very important that customers receive a signal about the costs of the system with time differentiation. Customers should pay for their energy consumption, taking into account the moment they consume. The way to implement this in a tariff design is through the definition of different time blocks. Usually, these time blocks are defined by assessing the annual load curves for different voltage levels, assuming that there is a high correlation between the demand profile and electricity production costs. In general, both profiles do not necessarily have to coincide; they might not do so, for instance, due to seasonal hydro conditions or in the case of renewable generation. However, in the case of a firstgeneration tariff design for transition economies such as is presented in this paper, the use of the annual load duration curve to define the time-differentiated blocks is considered to be adequate for the definition of time-differentiated economic signals. Another aspect to bear in mind when specifying the tariff time blocks is that, for practical reasons, they are defined by using daily, weekly, monthly or seasonal patterns. For example, peak hours could be defined as being from 5 p.m. to 9 p.m. on winter working days. Therefore, the time blocks are not completely linked to the load duration curve. For instance, in some cases, demand at some specific times or on certain days in the flat period can be higher than demand during periods that were classified as ‘peak’. This fact will have an important impact in the division of the costs between the different time blocks, as will be shown later in Section 3.1.1. 2.4. Customer categories Customer category definition is crucial in a properly functioning tariff scheme. Each customer category groups a number of customers with a similar profile and responsibility in the cost. All the customers in a category should be connected to the same voltage level and have a similar load profile. It is very important that all customers receive the correct economic signal and this can only be possible if customers are classified into homogeneous categories. The methodology proposed in this paper does not consider the definition of customer categories (see, for example, Chicco et al., 2003; Figueiredo et al., 2005). It will involve carrying out a metering plan, which allows updating (and, perhaps, redefining) customer categories with current information. Thus, tariff design will be performed for existing categories. For each one of these customer categories, the following data have to be collected: energy consumption by time block, peak demand by time block, and number of customers classified within that category. 2.5. Tariff design principles in transition economies As has been stated above, economic signals in tariff design have to be based on the cost-causality principle. However, when designing tariffs for transition economies, a tradeoff has to be made between this principle and the desirable simplicity of the tariff structure. In the end, the desired objective is the allocation of the cost of each activity to the different customer categories, time periods and cost drivers. For transition economies, the two key issues that have to be addressed are: 1) the correct allocation of the costs of each company's activity to the different customer categories, in order to avoid crosssubsidies between them, and 2) the introduction of differentiated electricity prices between time periods for those consumers who are prepared to react to them (typically, industrial consumers).
J. Reneses et al. / Energy Economics 33 (2011) 33–43
Finally, regarding the selection of the cost drivers and the allocation of the different costs to them, the main recommendation for transition economies is to achieve greater economic efficiency without introducing a high degree of complexity. When determining which part of the generation costs or/and the network costs should be allocated to the energy charge and which part to the demand charge, ad-hoc procedures can constitute an acceptable and relatively uncomplicated first step for tariff design. A more complex method can be subject to controversy and, normally, would require an amount of information that might not be available in transition economies. 3. Tariff design methodology Once the allowed income to be recovered for each activity — Generation, Transmission, Distribution, and Customer Services — has been determined, the second step is to allocate the cost of each activity to the different cost drivers. The design proposed in this paper considers three cost drivers: peak demand for each time block, energy consumption for each time block, and the number of customers. Generation, transmission and distribution costs will be allocated to demand (kW) and energy (kWh), while customer services costs will be allocated to the number of customers. Hence, for each customer category and each time block, two different charges will be defined: a demand charge (in m.u. per kW), and an energy charge (in m.u. per kWh). In addition, a fixed charge (in m.u. per customer) will be computed for each customer category. 3.1. Generation cost allocation Under ideal conditions, an optimal planned generation system will recover all its costs (fixed and variable), by charging the short-term marginal production costs to the energy consumed, for instance, in each hour. The functioning of current electricity markets is based on this principle and economic efficiency in tariff design is based on applying these generation marginal costs to the consumers as closely as possible. However in some of these markets, there is a supplementary capacity mechanism, resulting in a capacity payment (or long-term guarantee of supply) to all generators. The amount is equal to the compensation that is estimated to be necessary to make whole the recovery of the fixed cost of the adapted peak generation technology (i.e. the technology which produces energy and provides the required level of operational reserves at peak hours). In these markets, this capacity payment to generators should be collected as a demand charge to the consumers. Correspondingly, in the case of transition economies (where a wholesale market has not been implemented), the energy charge should be calculated by means of computing the short-term generation marginal costs. The demand charge could be estimated by computing the capacity payments that would correspond to a market with the same reserve margin. Since the sum of these two charges will not correspond (in general) to the actual generation costs, a final adjustment should be made. This adjustment could be based on applying a proportional coefficient to both charges (in order that their sum is equal to the total generation costs), or even on applying a second-best criterion, such as Ramsey prices. Although this solution could send the best possible economic signals, it may not be appropriate for transition economies. As mentioned above, such a complicated method can be subject to controversy and, normally, requires an amount of information that might be not available in transition economies. Hence, the proposal here is to adopt an ad-hoc procedure, which can constitute an acceptable first step for tariff design in transition countries. This procedure will divide generation costs into an energy charge and a demand charge, taking into account the particular characteristics of the corresponding power system.
37
Therefore, the total generation cost GC, is divided into a cost GCE, which is assigned to energy, and a cost GCP, which is assigned to peak demand. These two costs are assigned to customer categories and time periods in the next sections. 3.1.1. Generation cost allocated to peak demand Theoretically, the generation cost GCP to be allocated to peak demand should be wholly imputed to the peak demand at the generation level; this is to say, to the power consumption at the time when there is the highest demand in the system. Nevertheless, this methodology is not applicable in practice, since tariffs are computed ahead of time. That is, when computing tariffs there are many periods which, a priori, could be deemed to be likely to have the highest demand of the year. Furthermore, the assignation of all the costs to only a few hours would result in unstable tariffs and similar customers would pay significantly different electricity prices. Hence, the proposed methodology is based on splitting the cost into the H hours with the highest demand of the system. The number H of hours should be determined according to the examination of the corresponding load–duration curve.8 When computing tariffs, these hours are determined by using the most recent load–duration curve and, of course, the experience of the regulators. The next step is to compute how many h(ti) of these H hours belong to the different time blocks. In an ideal time-block definition, all the hours with the highest demand would belong to the peak block and the generation cost should be completely assigned to this time block. However, this will not be the case in practice and the value of h(ti) will have to be determined by using the most recent available data. Once these hours are computed for each time block, the cost GCP is allocated to each time block by using the corresponding coefficient pc(ti): pcðti Þ =
hðti Þ H
ð5Þ
GCP ðti Þ = GCP ⋅pcðti Þ:
ð6Þ
When this cost has been split into the different time blocks, it is necessary to allocate these quantities to the different customer categories. The cost GCP(ti,ck) to be paid for customer category ck is related to its maximum demand in the time block considered.9 As previously mentioned, these demands have to be calculated at the generation level, VLg. GCP ðti ; ck Þ =
GCP ðti Þ⋅PVLg ðti ; ck Þ ∑ PVLg ðti ; ck Þ
ð7Þ
k
Finally, the unit costs UGCP(ti,ck) (in monetary units/kW) associated to peak demand generation costs are computed. For customer category ck and time block ti, the unit cost is calculated by dividing the total cost by the actual peak demand of these customers; that is to say, the demand at the corresponding voltage level VLj: GCP ðti ; ck Þ PVLj ðti ; ck Þ ck customers are connected to VLj voltage level:
UGCP ðti ; ck Þ =
ð8Þ
3.1.2. Generation cost allocated to energy As in the case of the cost allocated to demand, generation costs allocated to energy have to be split into the different time blocks and, 8
Recommended values for H can range between 1% and 10% of the total. In power systems in which detailed data about the profiles of the consumers are available, this allocation in each block could be made using the coincident peak demand instead of the highest customer demand. Again, in a first-generation tariff design, it is considered that the use of the highest demand constitutes a correct signal. 9
38
J. Reneses et al. / Energy Economics 33 (2011) 33–43
then, into the different customer categories. In the case of transition economies, depending on data availability, marginal costs can be calculated hour by hour (whether an economic dispatch is already implemented), or they can only be approximated by average values, taking into account the variable costs of each generation technology. The generation cost GCE to be allocated to energy is split into the different time blocks by using the average generation marginal costs MC(ti) and the duration of the time blocks l(ti): GCE ðti Þ =
GCE ⋅MC ðti Þ⋅lðti Þ : ∑ ½MC ðti Þ⋅lðti Þ
ð9Þ
i
The economic signals provided to consumers by the marginal costs are crucial. Hence, even if marginal costs are not used to remunerate generators (which could happen in transition or developing countries), they should be computed or estimated in order to carry out the tariff design process. Again, it is necessary to divide these costs between the different customer categories. The cost GCE(ti,ck) to be paid for customer category ck is related to its energy consumption in the time block considered. As for the peak demand, these consumptions are referred to the generation level, VLg. GCE ðti ; ck Þ =
GCE ðti Þ⋅EVLg ðti ; ck Þ ∑ EVLg ðti ; ck Þ
ð10Þ
k
The last step is calculating the unit costs UGCE(ti,ck) (in monetary units/kWh) associated to energy generation costs. For customer category ck and time block ti, unit cost is calculated by dividing the total cost by the actual energy consumption of the customers; this is to say, the energy demand at the corresponding voltage level VLj: GCE ðti ; ck Þ UGCE ðti ; ck Þ = EVLj ðti ; ck Þ ck customers are connected to VLj voltage level:
economies it is proposed that the use is made of ad-hoc coefficients for splitting the total network cost at voltage level VLj, NCVLj, into a cost NCE,VLj, assigned to energy, and a cost NCP,VLj, assigned to peak demand. These coefficients can be determined on the basis of past experience in similar power systems (the values assigned to the Libyan systems are detailed in Section 4).The authors of this paper consider that the use of estimated coefficients will provide a better signal than allocating all the cost to a demand charge. The allocation of these two costs NCE,VLj and NCP,VLj to cost drivers is explained in the next subsections. 3.2.1. Network cost allocated to peak demand The network cost NCP to be allocated to peak demand is the main component of the total network cost (especially at low-voltage levels). Similar to what was done with the generation cost, network costs are split into the different time blocks using the coefficients pc(ti). Although from a theoretical point of view these coefficients could be defined by voltage level (since load curves are not equal at different voltage levels11), the authors consider again that for transition economies the signals provided by a single coefficient are good enough. Thus, the network cost NCP(ti) for the time block ti is computed as: NCP;VLj ðti Þ = NCP;VLj ⋅pcðti Þ:
Now, these quantities have to be allocated to the different customers. The cost NCP,VLj(ti,ck) to be paid for customer category ck is related to its maximum demand in the time block considered (see footnote 9). These peak demands have to be calculated at the corresponding voltage level, VLj. NCP;VLj ðti ; ck Þ =
NCP;VLj ðti Þ⋅PVLj ðti ; ck Þ
ð11Þ
3.2. Network cost allocation Network costs include the costs associated to investments (including asset depreciation and investment return), as well as operation and maintenance costs. Some tariff design methodologies allocate all the network (transmission and distribution) costs to the peak demand, with a charge per kW. Nevertheless, the proposed tariff structure also considers the allocation of part of these costs to energy consumption. The main justification for this is the inclusion of reductions in energy losses and of quality-of-service requirements in network-design criteria. Energy losses costs and quality-of-service penalties and incentives are directly related to the supplied energy.10 Therefore, the first step is to split the total network costs into an energy component and a demand component. The main difference between developed energy sectors and transition economies would be the way this computation can be made. It will depend on the available data and tools in the corresponding power system. In countries with a highly detailed available data, the procedure can be based on the use of a network-reference model (Rodríguez Ortega et al., 2008). Nevertheless, this kind of model will not normally be of use in transition countries (normally, due to the lack of data and skilled personnel). Thus, as in the case of the generation costs, in transition 10 Although ohmic energy losses depend on the squared energy flow, they are represented as it has been commented in the Loss Factors Section by linear loss factor coefficients that relate total energy losses with total energy consumption. A more advanced representation could be differentiating loss factors among time blocks. This latter approach would require many more data than the one usually available in transition countries.
ð12Þ
∑ PVLj ðti ; ck Þ
ð13Þ
ck
Note that this cost is only applied to customers located at levels with a voltage lower than VLj. This is to say, PVLj(ti,ck) = 0 for customers ck located at VLe, e N i. This means that, for example, customers located at transmission level will not pay for distribution network costs. Finally, the unit costs UNCP,VLj(ti,ck) (in monetary units/kW) associated to peak demand network costs have to be computed. As for generation costs, for customer category ck and time block ti, the unit cost is calculated by dividing the total cost by the actual peak demand of the customers; this is to say, the demand at the corresponding voltage level VLe: NCP;VLj ðti ; ck Þ PVLe ðti ; ck Þ ck customers are connected to VLe voltage level:
UNCP;VL j ðti ; ck Þ =
ð14Þ
3.2.2. Network cost allocated to energy The network cost NCE to be allocated to energy could also be split into the different time blocks (Rodríguez Ortega et al., 2008). An important part of this cost is related to network investments required to ensure the reliability of supply at any moment of the year. Therefore, these costs should be charged to energy consumption. Another part of these network costs is related to energy losses. In distribution networks, energy losses directly affect the choice of the optimal size of power line conductors. Hence, this part of the network cost should also be charged to energy consumption. 11 For instance, peak demand in a certain voltage level will not coincide, in general, with peak demand in other level.
J. Reneses et al. / Energy Economics 33 (2011) 33–43
However, these costs allocated to energy consumption cannot be straightforwardly allocated to the different time blocks (as in the case of the peak demand), since the cause of the costs cannot be directly determined. For the sake of simplicity, in the approach presented here no differentiation has been made between time blocks. Thus, all time blocks have the same unit energy costs. Then, what is necessary is to divide these costs between the different customer categories. The cost NCE,VLj(ck) to be paid by each customer category ck is proportional to its energy consumption. These consumptions must be calculated at the corresponding voltage level, VLj. NCE;VLj ðck Þ =
NCE;VLj ⋅EVLj ðck Þ ∑ EVLj ðck Þ
ð15Þ
ck
Then, the unit cost UNCE,VLj(ck) (in monetary units/kWh) for customer category ck can be calculated by dividing the total cost by the actual energy consumption of the customers; this is to say, the energy demand at the corresponding voltage level VLe: NCE;VLj ðck Þ EVLe ðck Þ ck customers are connected to VLe voltage level:
UNCE;VLj ðck Þ =
ð16Þ
The customer services cost CC corresponds to the costs associated to customer management, such as metering, billing or customer attention. This cost is assumed to be driven by the number of customers and therefore allocated on a per consumer basis. Hence, the total cost should be previously divided into the different customer categories. In developed systems, this information can be obtained from the company's cost accounting reports. In transition countries, this accounting division might not be available. In that case, the total cost could be split between the different categories using some ad-hoc criteria, such as the number of bills issued per year, the number of phone calls, or other required services per year and per customer for each category. Unit costs (in monetary units per year and customer) for customer category ck can be computed by dividing the cost CC(ck) by the number of customers nc(ck) in the customer category. CC ðck Þ ncðck Þ
proportionality per activity to allocate those structure costs to each activity. The application of “second-best” criteria for allocating structure costs will not improve significantly the economic efficiency and may constitute a source of controversy. Thus, if GCS is the structure cost associated to the generation activity, the final unit cost associated to power generation costs can be recalculated as: UGCP ðti ; ck Þ =
GCP ðti ; ck Þ GC + GCS ⋅ : GC PVLj ðti ; ck Þ
ð18Þ
The first factor was the one obtained in Section 3.1.1 and the second factor is the correction in order to take into account structure costs. That is, if the structure costs GCS for the generation activity are 5% of the allocated costs GC, then the unit cost for generation will rise 5% for all the customer categories and time blocks. The same treatment is applied to the rest of the unit costs described in this section. Finally, if some of the structure costs cannot be assigned to any of the activities (such as the regulatory commission cost), a global coefficient has to be applied to the complete tariff. 3.5. Tariff structure Once the different unit costs have been computed for all the activities, customers' categories and time blocks, the complete tariff structure can be computed by using the additivity principle. The resulting tariff structure consists of:
3.3. Customer services cost allocation
UCC ðck Þ =
39
ð17Þ
• An energy charge (in monetary units/kWh) for each voltage level and time period. • A demand charge (in monetary units/kW) for each voltage level and time period. • A charge per customer (in monetary units/customer-year) for each customer category. Usually, all the billing variables defined in the mentioned tariff structure are not defined for some of the existing customer categories. For example, in transition economies (and even nowadays in advanced economies), domestic consumers may not have timeblock differentiation. Furthermore, in many countries, domestic customers do not have a peak demand meter or maximum power
Table 1 Costs in 2006 for the Libyan system (MLD). Generation costs Peak demand Energy Network costs Transmission 400 kV Transmission 220 kV HV — distribution MV — distribution LV — distribution
3.4. Structure cost Besides the costs mentioned above, there are other costs that cannot be allocated to any cost driver. Some examples are the costs associated to the functioning of the system operator or the regulatory commission. Other common situations in several countries are stranded generation costs, subsidies to renewable energy and support to domestic fuels. These costs have been called ‘structure’ costs. The allocation of structure costs may constitute an important problem in some energy systems if the volume of these costs is a large fraction of the total costs. In these cases, the allocation of structure costs may distort the economic signals provided by the final tariffs. However, the amount of structure costs is usually small compared to the total costs of the main activities.12 Hence, the approach proposed in this paper is to directly calculate a coefficient of 12 If this is not the case, an exhaustive analysis should be carried out about the particular costs considered as structure costs.
Customer services costs MMR Large agriculture Heavy industry Light industry (I) Desalination Light industry (II) Domestic Small agriculture Commercial State offices Street lighting Structure costs Generation Transmission Distribution Customer services
346.70 47.26 299.44 175.99 1.86 48.76 34.72 42.88 47.78 58.69 0.00 0.11 0.01 0.04 0.00 1.34 45.34 5.15 5.66 0.82 0.23 113.14 13.58 12.82 72.35 14.39
40
J. Reneses et al. / Energy Economics 33 (2011) 33–43
Table 2 Customer categories characteristics in 2006 for the Libyan system. Customer categories
Transmission (VL3) [220 kV] Distribution HV (VL2) [66–30 kV] Distribution MV (VL1) [11 kV] Distribution LV (VL0) [0.4 kV]
Energy (GWh)
MMR Large agriculture Heavy industry Light industry (I) Desalination Light industry (II) Domestic Small agriculture Commercial State offices Street lighting
Intermediate
Base
Peak
Intermediate
Base
69.7 124.7 203.7 9.0 26.2 83.0 1373.7 148.2 329.0 252.2 431.1
167.5 347.4 509.8 22.3 72.1 205.8 2243.9 566.7 792.9 1117.7 172.4
95.4 65.9 308.8 8.7 42.6 80.9 1134.0 112.7 372.6 352.9 603.5
49.9 127.1 152.8 7.0 23.1 64.5 833.5 210.3 246.9 207.9 304.0
43.3 109.2 151.5 6.8 19.9 63.2 756.5 189.3 227.8 366.0 261.6
44.1 37.3 156.5 4.0 20.3 37.2 708.6 53.8 176.2 302.1 266.5
limitation device, making it impossible to define a demand charge per kW (metered or contracted). In these cases, the computed tariff structure has to be adapted to the actual billing variables for each customer category, by using the estimated energy consumption and peak demand for each time block. For instance, if domestic customers have only an energy charge, the rest of the charges (demand charge and fixed charge per customer) have to be “energized” by computing the total amount to be recovered with domestic tariffs and dividing it by the total energy consumed. 4. Case study: Libyan system in 2006 This section includes a numerical application of the proposed tariff structure, with the actual data of the Libyan system in 2006. The total generation of the Libyan system in 2006 was 23,992 GWh, with a peak demand of 4005 MW. About 40% of the total energy is generated by using natural gas (in gas turbines); the rest of the energy is generated in steam turbines using heavy fuel oil (33%) and light fuel oil (27%). The total number of customers was almost 1,200,000 (the electrification rate was over 99%), with a per capita consumption of about 4000 kWh. The voltage categories into which the power network has been divided are transmission (two levels have been considered: 400 kV, referred to as VL4 in the paper, and 220 kV, referred to as VL3); highvoltage distribution (30–66 kV, VL2 in the paper), medium-voltage distribution (11 kV, VL1 in the paper), and low-voltage distribution (lower than 1 kV, VL0 in the paper). The network model used in the methodology is represented in Fig. 1. It constitutes a good approximation to the actual Libyan electricity system and portrays all the features and information needed for calculation purposes. With respect to the time differentiation, the proposal for the Libyan system is to consider three time blocks: peak (5 h with the highest demand of every day, from 19.00 to 0.00); base (the 8 h with the lowest demand of every day, from 1.00 to 9.00); and intermediate (the period when the demand is neither high nor low, with the rest of the 11 h of the day). No season differentiation has been considered. Tariffs have been calculated for the existing customer categories: MMR (pumping hydro units) for VL3 voltage level; Large agriculture and Heavy industry for VL2 level; Desalination and Light industry for VL1 level; and Light industry, Domestic, Small agriculture, Commercial, State offices and Street lightning for VL0 level. 4.1. Data description Table 1 shows the costs (in millions of Libyan Dinars13 — MLD) used to apply the described methodology, while Table 2 contains the information for the different customer categories. 13
Maximum demand (MW)
Peak
For the year 2006, the exchange rate was about 1.3 Libyan Dinars per US dollar.
Number of customers 1 3 30 213 1 20871 703537 79853 86060 11548 3533
Table 3 Unit generation costs. Unit generation costs
Transmission (VL3) [220 kV] Distribution HV (VL2) [66–30 kV] Distribution MV (VL1) [11 kV] Distribution LV (VL0) [0.4 kV]
Energy (LD/MWh)
Peak demand (LD/kW)
Peak Intermediate Base
Peak Intermediate Base
MMR
16.28 14.16
11.33
8.43 2.64
1.59
Large agriculture Heavy industry Light industry (I) Desalination Light industry (II) Domestic Small agriculture Commercial State offices Street lighting
16.89 14.69
11.75
8.85 2.77
1.67
16.89 14.69
11.75
8.85 2.77
1.67
17.52 15.24
12.19
9.29 2.91
1.75
17.52 15.24 17.85 15.52
12.19 12.42
9.29 2.91 9.52 2.98
1.75 1.79
17.85 15.52 17.85 15.52
12.42 12.42
9.52 2.98 9.52 2.98
1.79 1.79
17.85 15.52 17.85 15.52 17.85 15.52
12.42 12.42 12.42
9.52 2.98 9.52 2.98 9.52 2.98
1.79 1.79 1.79
The criterion used to divide generation costs is to allocate the variable costs (mainly fuel costs) to the energy charge and the rest of the costs (mainly depreciation and fixed costs) to the demand charge. This ad-hoc criterion was adopted once it had been checked that the percentage of the cost allocated to energy (about 85%) was much higher than the percentage allocated to the demand, which represents capacity payments. With respect to the coefficients for dividing network costs into an energy component and a demand component, they have been estimated by means of the experience of other tariff design processes (e.g., Rodríguez Ortega et al., 2008). For transmission networks (VL4 and VL3), 70% of the cost has been assigned to demand, for highvoltage distribution (VL2) this coefficient is 75%, while for mediumvoltage distribution (VL1) and low-voltage distribution (VL0) they are, respectively, 80% and 90%. Finally, the parameters pc(ti), which indicate how the demand charges are allocated to different time blocks have been computed by means of the load curve for the Libyan System, as indicated in Section 3.1.1. According to the obtained results, the parameters have been set to 0.7 for peak block, 0.2 for intermediate block and 0.1 for base block.14
14 The results obtained have been rounded to these values, which the authors (according to their experience in similar tariff design processes) consider to be reasonable.
J. Reneses et al. / Energy Economics 33 (2011) 33–43 Table 4 Unit transmission costs. Unit transmission costs
Table 6 Unit MV distribution costs. Energy (LD/MWh)
Peak demand (LD/kW)
Unit MV distribution costs
Peak Intermediate Base Peak Intermediate Base Transmission (VL3) [220 kV] Distribution HV (VL2) [66–30 kV] Distribution MV (VL1) [11 kV] Distribution LV (VL0) [0.4 kV]
MMR
0.71
0.71
0.71
6.31
1.98
1.19
Large agriculture Heavy industry Light industry (I) Desalination Light industry (II) Domestic Small agriculture Commercial State offices Street lighting
0.73
0.73
0.73
6.63
2.08
1.25
0.73
0.73
0.73
6.63
2.08
1.25
0.76
0.76
0.76
6.96
2.18
1.31
0.76 0.78
0.76 0.78
0.76 0.78
6.96 7.14
2.18 2.23
1.31 1.34
0.78 0.78
0.78 0.78
0.78 0.78
7.14 7.14
2.23 2.23
1.34 1.34
0.78 0.78
0.78 0.78
0.78 0.78
7.14 7.14
2.23 2.23
1.34 1.34
0.78
0.78
0.78
7.14
2.23
1.34
Distribution MV (VL1) [11 kV] Distribution LV (VL0) [0.4 kV]
Light industry (I) Desalination Light industry (II) Domestic Small agriculture Commercial State offices Street lighting
Energy (LD/MWh)
Peak demand (LD/kW)
Peak
Intermediate
Base
Peak
Intermediate
Base
0.47
0.47
0.47
7.39
2.32
1.39
0.47 0.48
0.47 0.48
0.47 0.48
7.39 7.57
2.32 2.37
1.39 1.42
0.48 0.48
0.48 0.48
0.48 0.48
7.57 7.57
2.37 2.37
1.42 1.42
0.48 0.48 0.48
0.48 0.48 0.48
0.48 0.48 0.48
7.57 7.57 7.57
2.37 2.37 2.37
1.42 1.42 1.42
Table 7 Unit LV distribution costs. Unit LV distribution costs
Energy (LD/MWh)
Peak demand (LD/kW)
Peak Intermediate Base Peak Intermediate Base
4.2. Results Using the described data, the different unit costs can be computed for each one of the activities, customer categories and time blocks. Tables 3–8 show these unit costs, while Table 9 shows the total unit cost following the application of the additivity principle. Note that unit HV distribution costs are not shown for customers belonging to the VL3 voltage level, since these customers do not pay for the network costs of the rest of levels. The same consideration can be done. The same applies to the rest of network costs (Tables 6 and 7). Finally, Table 10 presents the alternative tariff structure in which only an energy charge is considered (peak demand and customer charges are included in this energy cost for each customer category). This is the current situation of Libyan tariffs. Note that, in this case, different customers located in the same voltage level pay different tariffs, due to the use of load curves in the energy tariff setting. The table also shows the actual tariffs applied in the Libyan system for year 2006.
Table 5 Unit HV distribution costs. Unit HV distribution costs
41
Energy (LD/MWh)
Distribution Light LV (VL0) industry [0.4 kV] (II) Domestic Small agriculture Commercial State offices Street lighting
0.27
0.27
0.27
9.58
3.00
1.80
0.27 0.27
0.27 0.27
0.27 0.27
9.58 9.58
3.00 3.00
1.80 1.80
0.27 0.27 0.27
0.27 0.27 0.27
0.27 0.27 0.27
9.58 9.58 9.58
3.00 3.00 3.00
1.80 1.80 1.80
As it can be seen, there are important differences between the calculated and the current tariffs. Domestic and small agriculture consumers actually have a lower tariff than they should, while the rest of the customers are currently paying more than the calculated tariffs. The current tariff for domestic consumers is actually charged in three segments (depending on the energy consumption), which range from 20 to 50 LD/MWh. The value presented in the table is the average tariff for these consumers. These differences can be explained as an existing cross-subsidy between customers' categories: some customers (especially domestic and small agriculture) are being subsidized by others (commercial, street lighting and state offices). The proposed tariff scheme will provide the different consumers with the right economic signals. Of course, the implementation of this new scheme has to be gradual, with a transitory period of a few years. The government can also decide to subsidize some customers, depending on their economic situation. The key aspect of the proposed scheme is that it basically provides correct economic signals
Peak demand (LD/kW)
Peak Intermediate Base Peak Intermediate Base Distribution Large HV (VL2) agriculture [66–30 kV] Heavy industry Distribution Light MV (VL1) industry (I) [11 kV] Desalination Distribution Light LV (VL0) industry (II) [0.4 kV] Domestic Small agriculture Commercial State offices Street lighting
0.43
0.43
0.43
4.94
1.54
0.93
0.43
0.43
0.43
4.94
1.54
0.93
0.44
0.44
0.44
5.18
1.62
0.98
0.44 0.45
0.44 0.45
0.44 0.45
5.18 5.31
1.62 1.66
0.98 1.00
0.45 0.45
0.45 0.45
0.45 0.45
5.31 5.31
1.66 1.66
1.00 1.00
0.45 0.45 0.45
0.45 0.45 0.45
0.45 0.45 0.45
5.31 5.31 5.31
1.66 1.66 1.66
1.00 1.00 1.00
Table 8 Unit customer services costs. Unit customer services costs Transmission (VL3) [220 kV] Distribution HV (VL2) [66–30 kV] Distribution MV (VL1) [11 kV] Distribution LV (VL0) [0.4 kV]
Customer (LD/customer-year) MMR Large agriculture Heavy industry Light industry (I) Desalination Light industry (II) Domestic Small agriculture Commercial State offices Street lighting
0.0 0.0 193.3 193.3 0.0 64.4 64.4 64.4 64.4 64.4 64.4
42
J. Reneses et al. / Energy Economics 33 (2011) 33–43
Table 9 Tariff structure for 2006. Total unit costs
Transmission (VL3) [220 kV] Distribution HV (VL2) [66–30 kV] Distribution MV (VL1) [11 kV] Distribution LV (VL0) [0.4 kV]
Energy (LD/MWh)
MMR Large agriculture Heavy industry Light industry (I) Desalination Light industry (II) Domestic Small agriculture Commercial State offices Street lighting
Intermediate
Base
Peak
Intermediate
Base
17.0 18.1 18.1 19.2 19.2 19.8 19.8 19.8 19.8 19.8 19.8
14.9 15.8 15.8 16.9 16.9 17.5 17.5 17.5 17.5 17.5 17.5
12.0 12.9 12.9 13.9 13.9 14.4 14.4 14.4 14.4 14.4 14.4
14.7 20.4 20.4 28.8 28.8 39.1 39.1 39.1 39.1 39.1 39.1
4.6 6.4 6.4 9.0 9.0 12.3 12.3 12.3 12.3 12.3 12.3
2.8 3.8 3.8 5.4 5.4 7.4 7.4 7.4 7.4 7.4 7.4
Transmission (VL3) [220 kV] Distribution HV (VL2) [66–30 kV] Distribution MV (VL1) [11 kV]
Distribution LV (VL0) [0.4 kV]
MMR Large agriculture Heavy industry Light industry (I) Desalination Light industry (II) Domestic Small agriculture Commercial State offices Street lighting
Customer (LD/customeryear) 0.0 0.0 193.3 193.3 0.0 64.4 64.4 64.4 64.4 64.4 64.4
5. Summary
Table 10 Simple tariff structure for 2006 compared to actual tariffs. Customer categories
Peak demand (LD/kW)
Peak
Energy (LD/MWh)
Actual tariffs
18.82 24.93 22.08 28.79
32.00 31.00 42.00
26.71 36.78
42.00
44.56 45.09 36.11 31.18 38.10
22.10 30.00 68.00 68.00 68.00
and, even if some subsidies are implemented, they are explicitly defined and they do not directly affect other customers or the profitability of the different activities. Like many other power systems in transition countries, the Libyan power system has a very high level of commercial losses (more than 30% in 2006). Table 11 shows the impact of charging these commercial losses to the LV customers. The government has already established a plan to reduce these losses. In the meantime the regulatory authorities have to decide who should shoulder the cost of these losses, as it seems reasonable not to penalize the rest of LV customers.
This paper has presented a general electricity tariff design methodology, especially focused on its application to transition countries. The proposed methodology basically complies with all the key criteria that should be taken into account in a tariff design process, while trying to be as simple as possible and to use only data that are usually available in transition countries. Special emphasis has been placed on designing tariffs that recover all costs disaggregated in the different electricity activities, and to allocate those costs to the different customer categories while avoiding cross-subsidies between them. The paper also has presented a case study in which tariffs are computed with the actual data of the Libyan system in 2006. The methodology is divided into several steps. Once the total revenue has been divided into the different activities (generation, transmission, distribution and customer services), these recognized costs are allocated to three cost drivers: energy, peak demand and fixed cost per customer. Then, each one of these costs is assigned to time blocks (peak, intermediate and base, for the Libyan system). Finally, costs are allocated to customer categories, and the unit cost for each activity is obtained. The additivity principle allows computing the final tariffs with different tariff structures. The computed tariffs have been compared to actual tariffs, and significant differences have been identified. This new tariff scheme allows the identification of current cross-subsidies between customer categories. The significance of commercial losses has been also highlighted by computing the tariffs with and without them.
Table 11 Tariff structure for 2006 recovering commercial losses. Total unit costs
Transmission (VL3) [220 kV] Distribution HV (VL2) [66–30 kV] Distribution MV (VL1) [11 kV] Distribution LV (VL0) [0.4 kV]
Energy (LD/MWh)
MMR Large agriculture Heavy industry Light industry (I) Desalination Light industry (II) Domestic Small agriculture Commercial State offices Street lighting
Peak demand (LD/kW)
Peak
Intermediate
Base
Peak
Intermediate
Base
17.0 18.1 18.1 19.2 19.2 32.9 32.9 32.9 32.9 32.9 32.9
14.9 15.8 15.8 16.9 16.9 30.4 30.4 30.4 30.4 30.4 30.4
12.0 12.9 12.9 13.9 13.9 23.8 23.8 23.8 23.8 23.8 23.8
14.7 20.4 20.4 28.8 28.8 65.8 65.8 65.8 65.8 65.8 65.8
4.6 6.4 6.4 9.0 9.0 18.8 18.8 18.8 18.8 18.8 18.8
2.8 3.8 3.8 5.4 5.4 11.4 11.4 11.4 11.4 11.4 11.4
Customer (LD/customeryear) 0.0 0.0 193.3 193.3 0.0 64.4 64.4 64.4 64.4 64.4 64.4
J. Reneses et al. / Energy Economics 33 (2011) 33–43
Acknowledgements The authors would like to thank the valuable and insightful comments and suggestions received from the two anonymous reviewers of the paper. Appendix 1. Terminology This annex includes a detailed description of the terminology that has been used throughout the cost allocation procedure described in the paper.
Indexes VLj ti ck
Voltage level j Time block i Customer category k
Energy and demand PVLj(ti,ck) Demand at voltage level VLj and time block ti, due to the demand of customer category ck (in MW) EVLj(ti,ck) Energy consumption at voltage level VLj and time block ti, due to the demand of customer category ck (in MWh) PVLj(ck) Demand at voltage level VLj due to the demand of customer category ck EVLj(ck) Energy at voltage level VLj due to the demand of customer category ck PVLj(ck) Total demand at voltage level VLj EVLj(ck) Total energy at voltage level VLj
Costs GC GCP GCP(ti) GCP(ti,ck)
Total generation cost Generation cost assigned to peak demand Generation cost assigned to peak demand for time block ti Generation cost assigned to peak demand for time block ti, allocated to customer category ck UGCP(ti,ck) Unit generation cost assigned to peak demand for time block ti and customer category ck GCE Generation cost assigned to energy consumption GCE(ti) Generation cost assigned to energy for time block ti UGCE(ti,ck) Unit generation cost assigned to energy for time block ti MC(ti) Average generation marginal costs for time block ti NCVLj Network cost at voltage level VLj Network cost assigned to peak demand at voltage level VLj NCP,VLj NCP,VLj(ti) Network cost assigned to peak demand at voltage level VLj for time block ti NCP,VLj(ti,ck) Network cost assigned to peak demand at voltage level VLj for time block ti, allocated to customer category ck UNCP,VLj(ti,ck) Unit network cost assigned to peak demand at voltage level VLj for time block ti, allocated to customer category ck NCE,VLj Network cost assigned to energy consumption at voltage level VLj NCE,VLj(ck) Network cost assigned to energy at voltage level VLj, allocated to customer category ck UNCE,VLj(ck) Unit network cost assigned to power at voltage level VLj, allocated to customer category ck
43
CC Total customer services cost Customer services cost allocated to customer category ck CC(ck) UCC(ck) Unit customer services cost allocated to customer category ck Loss factors Loss factor corresponding to the energy losses caused by a fe,E j demand at a voltage level VLe when considering the energy flow at a higher voltage level VLj. Loss factor corresponding to the power demand losses fe,P j caused by a demand at a voltage level VLe when considering the power demand at a higher voltage level VLj.
Other symbols Percentage of costs allocated to demand assigned to time pc(ti) block ti l(ti) Duration of time block ti nc(ck) Number of customers for customer category ck
References Berg, S.V., Tschirhart, J., 1988. Natural monopoly regulation: principles and practice. Cambridge Surveys of Economic Literature Series Cambridge. Cambridge University Press, New York and Melbourne. Bonbright, J., 1961. Principles of Public Utility Rates. Public Utilities Reports, Inc. New York: Columbia University Press. Chicco, G., Napoli, R., Postolache, P., Scutariu, M., Toader, C., 2003. Customer characterization options for improving the tariff offer. IEEE Transactions on Power Systems 18 (1), 381–387. Feldstein, M.S., 1972. Distributional equity and the optimal structure of public prices. The American Economic Review 62 (1/2), 32–36. Figueiredo, V., Rodrigues, F., Vale, Z., Gouveia, J.B., 2005. An electric energy consumer characterization framework based on data mining techniques. IEEE Transactions on Power Systems 20 (2), 596–602. Flinn, D.G., Hall, S., Morris, J., Cummings, A.B., Northcote-Greene, J.E.D., 1983. Improved Methods for Distribution Loss Evaluation. Electric Power Research Institute (EPRI). Green, R., Rodríguez Pardina, M., 1999. Resetting Price Controls for Privatized Utilities. A Manual for Regulators, EDI Development Studies. . Jamasb, T., Pollitt, M., 2000. Benchmarking and regulation: international electricity experience. Utilities Policy 9 (3), 107–130. Larsson, M.B.O., 2003. The network performance assessment model. Regulation with a reference network. Proceedings of the Elforsk Market Design Conference, Stockholm, Sweden. Liston, C., 1993. Price-cap versus rate-of-return regulation. Journal of Regulatory Economics 5 (1), 25–48. Lowry, M.N., Getachew, L., 2009. Statistical benchmarking in utility regulation: role, standards and methods. Energy Policy 37, 1323–1330. Pérez Arriaga, J.I., Smeers, Y., 2002. Guidelines on tariff setting. In: Lévêque, Francois (Ed.), Chapter 7 in Transport Pricing of Electricity Networks. Kluwer Academic Publishers. Pollitt, M., 2005. The role of efficiency estimates in regulatory price reviews: Ofgem's approach to benchmarking electricity networks. Utilities Policy 13, 279–288. Ramsey, F., 1927. Contribution to the theory of taxation. The Economic Journal 37, 47–61. Rodríguez Ortega, M.P., Perez-Arriaga, I.J., Rivier Abbad, J., Peco, J., 2008. Distribution network tariffs: a closed question? Energy Policy 36 (5), 1712–1725. Román, J., Gómez, T., Muñoz, A., Peco, J., 1999. Regulation of distribution network business. IEEE Transactions on Power Delivery 14 (2), 662–669. Rothwell, G., Gomez, T., 2003. Electricity economics. Regulation and Deregulation. Ed. Wiley-IEEE Press, New York, USA. Rudnick, H., Donoso, J.A., 2000. Integration of price cap and yardstick competition schemes in electrical distribution regulation. IEEE Transactions on Power Systems 15 (4), 1428–1433. Weston, F., 2000. Charging for Distribution Utility Services: Issues in Rate Design. Vermont: Regulatory Assistance Project.
