Benchmarking Economical Efficiency of Renewable Energy Power Plants: A Data Envelopment Analysis Approach

Advanced Materials Research Vol. 772 (2013) pp 699-704 Online available since 2013/Sep/04 at www.scientific.net © (2013) Trans Tech Publications, Switzerland doi:10.4028/www.scientific.net/AMR.772.699

Benchmarking Economical Efficiency of Renewable Energy Power Plants: A Data Envelopment Analysis Approach Corrado lo Storto*,a, Gabriella Ferruzzi*,b *University of Naples Federico II, Piazzale V. Tecchio n. 80, Naples Italy a

[email protected], [email protected]

Keywords: Data Envelopment Analysis, Benchmarking, efficiency, Renewable energy, Selection, ranking, Subsidies.

Abstract. This paper implements Data Envelopment Analysis (DEA) to calculate an efficiency measure index of 21 energy power plants that use different technologies, including both renewable and conventional types. Super-efficiency measurements are used to generate a ranking of plants based on their efficiency score that can be used to select among alternatives. It is also showed how DEA can also be adopted to estimate the amount of financial subsidies necessary to make a renewable energy plant as efficient as a conventional energy plant, by calculating the extent to which inefficient power plants over-utilize specific inputs or under-produce outputs. Introduction Energy demand has largely increased in recent years as a consequence of the rapid economy growth all over the world. Generally, developed countries have fulfilled the energy need by exploiting fossil fuels. But, the awareness of decreasing availability of fossil resources in the long term and a growing concern for the environment induced governments to stimulate the construction of power plants that generate clean energy from renewable sources (wind, geothermal, biomass, solar, photovoltaic and small hydropower plants). Governments in the OECD area stimulated the diffusion of renewable technologies by implementing a variety of policies, mostly providing financial support, as these technologies need a higher amount of investment capital per unit of energy power produced in comparison to more conventional technologies. That, together with the large portfolio of available technological alternatives, makes the measurement of the economical efficiency and selection of the renewable power plant typology and size an important multi-criteria decision-making problem. Multi-criteria methods allow to take into account in the same time conflicting and qualitative factors that become relevant when environmental issues besides financial issues are considered in the comparison of conventional and renewable sources energy plants. This paper implements a particular type of multi-criteria method, Data Envelopment Analysis (DEA), to compare 21 conventional and renewable energy plants. DEA provides a unique efficiency score for each plant, and more important, identifies improvement trajectories useful to assess the amount of financial subsidies that make the adoption of renewable energy plants as efficient as conventional plants. Using economical efficiency as a criterion to rank and select energy generating power plants Literature on the decision-making process relative to the implementation of energy power plants has focused on several issues, i.e. site selection, alternative technology selection, size optimization, capacity planning, etc. and has suggested several approaches and methodologies to deal with the associated problems, i.e. traditional financial or socio-economical analysis, AHP, ANP, MCDM, MOLP, case-based reasoning, and DEA [1, 2, 3, 4, 5, 6, 7]. In particular, literature shows that in the last decade multi-criteria methods have been widely dominant in various decision-making problems relative to many selection and, more generally, to energy planning issues in both academic and industry literature. Reference [8] claims that multi-criteria decision analysis (MCDA) methods have become very popular in supporting the decision-making process relative to sustainable energy as a consequence of the multi-dimensionality of the sustainability goal and the complexity of socio-economic and biophysical systems. The scholars provide an extensive review on multi-criteria All rights reserved. No part of contents of this paper may be reproduced or transmitted in any form or by any means without the written permission of TTP, www.ttp.net. (ID: 95.75.106.95, University of Naples Federico II, Naples, Italy-31/10/13,22:23:07)

700

Material Researches and Energy Engineering

decision analysis methods in sustainable energy decision-making. They find that equal criteria weights are the most popular weighting method, while AHP is the most popular comprehensive MCDA method. However, the MCDA methods have a number of shortcomings: a) weights given to criteria have a direct influence on the outcome of decision-making, and, finally on the ranking and alternative selection; b) the normalization of attribute weights in MCDA various approaches has been identified as a procedural source of biases [9]. Weighting approaches may be thus a major concern of these methods. When a new decision alternative is added to an otherwise unchanged decision problem, it may cause a reordering of the previous rank order, determining rank reversal that is another cause of concern [10]. Furthermore, as reference [4] emphasizes, the measurement of power generating plant efficiency is a major concern in the electricity-generating industry as a consequence of high investment and operation costs. The measurement of the economical efficiency of a plant should henceforth be the final aim of the evaluation. The study setting Energy plant efficiency measurement. In this paper, the multiple criteria optimization problem is approached through Data Envelopment Analysis (DEA). While MCDA methods tend to focus on a small number of alternatives that are compared across a usually large number of criteria, DEA can be effectively used when there are many more options than criteria. Contrarily to what is done with MCDA, in DEA weights are endogenous because they are derived within the model and not by external decision makers. The DEA method determines an efficiency enveloping frontier that corresponds to a virtual efficient production technology set to which the real alternatives are compared, without assuming any a priori functional form. It is a non-parametric performance measurement technique that evaluates the relative efficiency of a number of homogeneous decision making units (DMUs), i.e., production plants, which transform inputs into outputs. The enveloping frontier is generated solving a sequence of linear programming (LP) problems, one for each DMU. For every DMU, the ratio between its weighted sum of outputs to its weighted sum of inputs measures the relative efficiency of that DMU. All DMUs that are placed on the efficient frontier are considered 100% efficient. In that case, there is neither linear combination of DMUs’ inputs and outputs, nor real DMU that could either decrease one input or increase one output without worsening the value of at least another one. When a DMU lies in the interior of the efficient frontier domain, it should be considered as inefficient. Efficient DMUs can be used as a performance benchmark for the inefficient DMUs. The relative efficiency score of a DMU is thus measured by the distance between the actual observation and the frontier obtained from all DMUs under examination, adopting the Farrell measure of technical efficiency as a measure for the DMU efficiency score. Here, the CCR DEA model is implemented to calculate plant efficiency [11], with the assumption of constant returns to scale and input orientation (minimization). As size of energy generating plants included in the study largely differs and these ones do not have comparable production levels, the adoption of a BCC DEA model with variable returns to scale was excluded to avoid any amplification of scale effects that would lead to a huge number of 100% efficient power generating plants. An input-oriented CCR DEA model is defined as: m Min Θ + ε  ∑ Si− +  i =1 n

s.t.

∑λ y j

s

∑S r =1

+ r

  

rj

- Sr + = y rj , r=1,...,s

ij

+ Si − = Θx ij , i=1,...,m

j =1 n

∑λ x j

j =1

λj ≥ 0 +

Sr , Si

j=1,...,n −

≥0

r=1,...,s , i=1,...,m

Advanced Materials Research Vol. 772

701

+ where Sr and Si- are slack variables, Θ is the efficiency measurement, n is the number of technologies, s and m are respectively the number of outputs and number of inputs. The outcome of the CCR DEA model is an efficiency measure which is between 0 and 1, but without any ranking of DMUs. In order to have a full not censored ranking of energy power plant efficiencies, the conceptualization of super-efficiency suggested by reference [12] is also adopted allowing the efficiency measurement of a plant j to be greater than 100% (AP DEA). Sample. Sample includes 21 power plants that use different technologies to generate energy, classified as to the [13] (see Table 1): 1) primary output type, electrical (E), thermal (T) and cogeneration technology (CT); 2) fuel burnt, fossil (F) or renewable (R).

Table 1 - List of power generating plant technologies T1 T2 T3 T4 T5 T6 T7 T8 T9 T10 T11 T12 T13 T14 T15 T16 T17 T18 T19 T20 T21

Output E E E E E E E E E CT E E E E E E E E, T E, T E, T E, T

Fuel F F F F F F F F F F R R R R R R F R R R F

Plant technology COAL FIRED STEAM TURBINE GAS FIRED TURBINE NATURAL GAS CYCLE COMBINATED (NGCC) NGCC + CCS INTEGRATED GASIFICATION COMBINED CYCLE (IGCC) IGCC + CCS PULVERIZED COAL (PC) PC + CCS NUCLEAR (EPR) COAL STEAM FIRED TURBIN COGENERATION SYSTEM SMALL HYDROELECTRIC PLANT MICRO HYDROELECTRIC PLANT MINI-TURBINE WIND GENERATOR LARGE TURBINE WIND GENERATOR PHOTOVOILTAIC ISOLATED SYSTEM PHOTOVOILTAIC GRID CONNECTED SYSTEM FUEL CELL (PAFC) WOOD CHIPS BIOMASS BIOMASS FROM WASTE BIOMASS FROM VEGETABLE OIL GEOTHERMIC

The CCR DEA model input and output variables. Two energy power plant efficiency models have been developed, Model 1 and Model 2. Both models include 2 inputs and 1 output, that are based on physical and economical data. Model 2 differs from Model 1 as it adds an estimation of the plant pollution cost associated to CO2 emission. An in-depth review of technical literature and limitation imposed by the availability of data addressed the selection of the input and output variables used [14]. Table 2 reports the list of variables. Input and output variables have been calculated as follows: n

CF YeCF = , n

n

∑ CVi YeCV =

1

n

∑ CV + t × ACO i

,

EN = PW × LF × 8,760,

YeCV * =

i

2i

1

n

where CF is total fixed investment cost to build the plant, CVi is the variable cost of plant operation at year i including cost of fuel and maintenance, n is the operational life of the plant, PW is the annual power generated by plant, LF is the factor load, 8,760 is the total amount of service hours per year, t is an estimated tax for pollution (€ 7/ton of CO2), and ACO2i is the total amount of CO2 produced by plant in year i. The following assumptions were made: 1) the decommissioning cost was considered only for nuclear power plants, and it was considered to be equal to 30% of the initial investment cost; 2) the annual operation and fuel cost increase was quantified according to the instructions provided by the IEA reports. In particular, the operation cost was increased by 1%, while fuel cost by 2%, except for

702

Material Researches and Energy Engineering

the vegetable oil cost that was increased by 2.5%; 3) for the CCS plants, capture cost is also included (equal to 75% of the total cost), but transport and storage costs have been neglected; 4) cost measures have been considered as independent of time, and therefore, they were not discounted. Table 2 - Input and Output variables Model 1 Input 1: Input 2: Output 1: Model 2 Input 1: Input 2: Output 1:

(YeCF) Average annual fixed cost (€) (YeCV) Average annual variable cost (€) (EN) Yearly amount of energy generated by plant k (kWh) (YeCF) Average annual fixed cost (€) (YeCV*) Average annual variable cost and pollution cost (€) (EN) Yearly amount of energy generated by plant k (kWh)

Findings Table 3 presents plant productivity rates. Plant productivity was calculated by running three DEA models as described the previous section. Data show that plants differ so much as to their productivity index. The average productivity rate of plants in sample is at 64.96% in Model 1 and 64.34% in Model 2, while it increases to 69.94% when super-efficiency is calculated. Standard deviation is about 27% in Model 1 and Model 2. Both in Model 1 and Model 2 there are 5 100% efficient plants that use the following technologies: Coal fired steam turbine, Gas fired turbine, Natural Gas Combined Cycle, Nuclear power plant and Large wind turbine. In Model 1, two plants based on the Fuel cell and biomass technology (Wood Chips Biomass) have the lowest productivity, respectively at 18.66% and 18.22%. The same energy power plants have the lowest productivity rates in Model 2, even though a little different. Contrarily to what expected, the addition of the CO2 cost in Model 2 has an almost not influential impact on the productivity rates of power plants. Indeed, the Coal Fired Steam Turbine plant which is characterized by the highest amount of CO2 emission remains 100% efficient in Model 2. The same is for the NGCC technology plant. The power plants T11, T12, T13, T15, and T16 even though have no CO2 emissions do not change their productivity score moving from Model 1 to Model 2. There are plants that have small amounts of CO2 emissions that even improve their productivity scores in Model 2, such as T4 and T10. The last column of Table 3 shows the super-efficiency scores of plants calculated using inputs and output of Model 1. Data indicate that the power plant which is based on the Large Wind Turbine technology is the most productive one among the efficient plants, being at 138.51% efficiency. Table 4 provides information relative to potential improvements of inefficient power generating plants in Model 1. Inefficient plants over-utilize specific inputs either in terms of fixed investment costs or operational variable costs, or under-produce energy. According to the input orientation approach, the performance of an inefficient plant could be increased by decreasing the amount of input consumed. In Table 4, percentage measures in columns “YeCF” and “YeCV” indicate the extent to which operational and fixed costs should be reduced to make a power plant 100% efficient. On average, a YeCV reduction of about 49% and a YeCF reduction of about 46% are necessary to make efficient a plant belonging to the inefficient group of sample. The minimum improvement required to make a plant efficient is 21.63% for both inputs. Thus, a Coal Steam Fired Turbine Cogeneration System plant may become 100% efficient by either decreasing annual operational costs or fixed investment by 21.63%. For most plants, the necessary input reduction is all but the same for both inputs. However, there are two plants in which reduction relative to input 1 slightly differs from reduction relative to input 2, T15 and T21. This difference is very strong for plant T16 (PV Grid Connected), in which a relative reduction of the annual operational cost may be more effective to increase plan efficiency in comparison to a greater fixed cost reduction. This information relative to potential trajectories leading to efficiency improvement can be particularly useful to identify more effective and customized financial subsidies (i.e., type and amount) to be assigned to a power plant whose production technology uses a certain renewable energy in order to make it more economically efficient moving on the efficiency frontier and, henceforth, competitive in comparison to more

Advanced Materials Research Vol. 772

703

conventional technologies. Indeed, it can be assumed that a cost reduction is equivalent to an increase of the value generated by that plant in the perspective of the owner or operator of that plant. For instance, providing a Small Wind Turbine plant with a financial contribution of about 61.49% of its fixed annual cost, that plant may become as efficient as a power plant based on a more conventional energy production technology (i.e., Coal Fired Steam Turbine or Gas Fired Turbine). Table 3 – Power plant efficiency scores T1 T2 T3 T4 T5 T6 T7 T8 T9 T10 T11 T12 T13 T14 T15 T16 T17 T18 T19 T20 T21

CCR DEA Model 1 100.00% 100.00% 100.00% 65.29% 74.49% 56.59% 77.01% 52.94% 100.00% 78.37% 52.72% 77.31% 38.88% 100.00% 24.39% 52.96% 18.66% 18.22% 60.76% 41.52% 73.99%

CCR DEA Model 2 100.00% 100.00% 100.00% 67.97% 71.42% 56.28% 73.67% 51.80% 100.00% 79.47% 52.65% 77.32% 38.78% 100.00% 24.71% 53.02% 17.71% 19.09% 51.70% 41.58% 74.07%

AP DEA Model 1 112.98% 101.67% 113.76% 65.29% 74.49% 56.59% 77.01% 52.94% 137.72% 78.37% 52.72% 77.31% 38.88% 138.51% 24.39% 52.96% 18.66% 18.22% 60.76% 41.52% 73.99%

Table 4 - Improvement trajectories in Model 1 T4 T5 T6 T7 T8 T10 T11 T12 T13 T15 T16 T17 T18 T19 T20 T21

Efficiency 65.29% 74.49% 56.59% 77.01% 52.94% 78.37% 52.72% 77.31% 38.88% 24.39% 52.96% 18.66% 18.22% 60.76% 41.52% 73.99%

YeCF -34.71% -25.51% -43.41% -22.99% -47.06% -21.63% -47.28% -22.69% -61.49% -89.85% -83.04% -81.34% -81.78% -39.24% -58.48% -33.80%

YeCV -34.71% -25.51% -43.41% -22.99% -47.06% -21.63% -47.28% -22.69% -61.12% -75.61% -47.04% -81.34% -81.78% -39.24% -58.48% -26.01%

Conclusion This paper has performed Data Envelopment Analysis to calculate an efficiency index of 21 energy power plants which use different production technologies, including both renewable and conventional types. As expected, results show that conventional power energy plants are generally more efficient than renewable plants, while only large wind turbine plants are comparable to traditional plants as to the productivity measure, or even apparently better. Data also show that the effect of a pollution tax in the case of conventional plants does not substantially improve the productivity of renewable energy plants.

704

Material Researches and Energy Engineering

This paper has also illustrated how DEA can be purposefully implemented to identify most effective types and estimate amounts of financial incentives necessary to make a renewable energy plant as efficient as a conventional energy plant. Indeed, in the input-oriented model adopted, DEA allows to calculate the extent to which an inefficient power plant over-utilize specific cost inputs to produce energy. This over-consumption of financial resources could be balanced by an equivalent amount of financial benefit assigned to the owner or operator of the plant. As a further development of the proposed method, the implementation of an output-oriented DEA model might be also useful to determine the amount of financial subsidies necessary to increase plant efficiency as a tariff increase paid to the plant operator. Corresponding Author Corrado lo Storto, email: [email protected], phone: +39.081.7682932 References [1] T. Melchiora and R. Madlener, Economic evaluation of IGCC plants with hot gas cleaning, Applied Energy, Vol. 97 (2012), pp.170–184. [2] J.A. Matelli, E. Bazzo and J. C. da Silva, Development of a case-based reasoning prototype for cogeneration plant design, Applied Energy, Vol. 88 (2011), pp. 3030–3041. [3] H.A. Naim and M.G. Carvalho, Multi-criteria assessment of new and renewable energy power plants, Energy, Vol. 27 (2002), pp.739–755. [4] J.R. San Cristóbal, A multi criteria data envelopment analysis model to evaluate the efficiency of the Renewable Energy technologies, Renewable Energy, Vol.36 (2011), pp. 2742-2746. [5] R. Madlener, C.H. Antunes and L. C. Dias, Assessing the performance of biogas plants with multi-criteria and data envelopment analysis, European Journal of Operational Research, Vol. 197 (2009), pp. 1084–1094. [6] B.A. Akash, R. Mamlook and M. S. Mohsen, Multi-criteria selection of electric power plants using analytical hierarchy process, Electric Power Systems Research, Vol. 52 (1999), pp. 29–35. [7] A. Datta, A. Ray, G. Bhattacharya and H. Saha, Green energy sources (GES) selection based on multi-criteria decision analysis (MCDA), International Journal of Energy Sector Management, Vol. 5 (2007), pp. 271 – 286. [8] J.J. Wang, Y.Y Jing, C.F. Zhang and J.H. Zhao, Review on multi-criteria decision analysis aid in sustainable energy decision-making, Renewable and Sustainable Energy Reviews, Vol. 13 (2009), pp. 2263–2278. [9] A.A. Salo and R.P. Hamalainen, On the measurement of preferences in the analytic hierarchy process, Journal of Multi-Criteria Decision Analysis, Vol. 6 (1997), pp. 309-319. [10] W. De Keyser and P. Peeters, A note on the use of PROMETHEE multicriteria methods, European Journal of Operational Research, Vol. 89 (1996), pp. 457-461. [11] A. Charnes, W.W. Cooper and E. Rhodes, Measuring the efficiency of decision making units, European Journal of Operational Research, Vol. 2 (1978), pp.429-444. [12] P. Andersen and N.C. Petersen, A Procedure for Ranking Efficient Units in Data Envelopment Analysis, Management Science, Vol. 39 (1993), pp. 1261–1264. [13] A. L’Abbate, G. Fulli., F. Starr. and S.D. Peteves, Distributed Power Generation in Europe, Tech. Rep. JRC41239 (EC Joint Research Centre Institute for Energy, 2008). [14] F. Rossi, Gestione dei sistemi elettrici nei mercati liberalizzati (ESI, Italy 2007).

Material Researches and Energy Engineering 10.4028/www.scientific.net/AMR.772

Benchmarking Economical Efficiency of Renewable Energy Power Plants: A Data Envelopment Analysis Approach 10.4028/www.scientific.net/AMR.772.699