Theoretical efficiency limits for energy conversion devices

Energy 35 (2010) 2059e2069 Contents lists available at ScienceDirect Energy journal homepage: www.elsevier.com/locate/energy Theoretical efficiency limits for energy conversion devices Jonathan M. Cullen*, Julian M. Allwood Department of Engineering, University of Cambridge, Cambridge CB2 1PZ, UK a r t i c l e i n f o a b s t r a c t Article history: Received 14 July 2009 Received in revised form 6 January 2010 Accepted 19 January 2010 Available online 5 March 2010 Using energy more efficiently is a key strategy for reducing global carbon dioxide emissions. Due to limitations on time and resources, actions must be focused on the efficiency measures which will deliver the largest gains. Current surveys of energy efficiency measures assess only known technology options developed in response to current economic and technical drivers. However, this ignores opportunities to deliver long-term efficiency gains from yet to be discovered options. In response, this paper aims to calculate the absolute potential for reducing energy demand by improving efficiency, by finding the efficiency limits for individual conversion devices and overlaying these onto the global network of energy flow. The potential efficiency gains for each conversion device are found by contrasting current energy demand with theoretical minimum energy requirements. Further insight is gained by categorising conversion losses according to the underlying loss mechanisms. The result estimates the overall efficiency of global energy conversion to be only 11 per cent; global demand for energy could be reduced by almost 90 per cent if all energy conversion devices were operated at their theoretical maximum efficiency. Ó 2010 Elsevier Ltd. All rights reserved. Keywords: Energy efficiency Sankey diagram Prioritisation Exergy analysis Conversion loss 1. Introduction: the efficient use of energy The reasons for using energy more efficiently are clear: to relieve pressure on scarce energy resources, to reduce energy costs by avoiding wastefulness, and perhaps most pressing, to reduce energy related carbon dioxide (CO2) emissions which contribute to climate change. The well-known Kaya identity [1] expresses the generation of energy-based CO2 emissions as the product of four drivers: population, per capita wealth, energy intensity (energy per unit wealth) and carbon intensity (CO2 per unit energy). The first two drivers are socio-economic and are difficult to limit in practice. The third and fourth drivers are technical options which require energy to be used more efficiently (which lowers energy intensity) and the decarbonisation of energy supplies (which reduces carbon intensity). To date, emission reduction strategies have focused primarily on energy supply options: renewable energy technologies, nuclear power, carbon capture and storage (CCS) and fuel switching. Yet, the International Energy Agency (IEA) asserts that ‘energy efficiency improvements . represent the largest and least costly savings’ [[2], p. 4] available. There is further need to develop energy efficient technologies and understand the scope of efficiency measures to reduce CO2 emissions. * Corresponding author. Tel.: þ44 1223 760360; fax: þ44 1223 332662. E-mail address: jmc99@cam.ac.uk (J.M. Cullen). URL: http://www.lcmp.eng.cam.ac.uk 0360-5442/$ e see front matter Ó 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.energy.2010.01.024 In the 1975 conference Efficient use of energy, Ford et al. [3] stated that the primary objective of any technical energy study is to define a target ‘standard of performance’ against which current demand for energy can be compared. Such a target may be chosen from several options, for example, current best practice, the extrapolation of an historical trend, or the projected gains from a specific design innovation. The difference between today's energy demand and this target provides a measure of the improvement potential, or possible energy savings from energy efficiency measures. Finding the global improvement potential from energy efficiency requires tracing the scale of energy flow through technical devices in the energy network, and assessing the efficiency gains available in each device. Equation (1) is used to calculate these potential savings from efficiency, in conversion devices:  Target Scale of Potential for  ¼ efficiency savingenergy energyflow Current efficiency  (1) where the energy terms are measured in joules (J) and the efficiency terms in percentages (%). Assessing the scale of energy flow through the global energy network is the subject of a previous article by Cullen and Allwood [4]. Their research traces energy flow from fuel to final service, including the technical steps of fuel transformation, electricity generation, and end-use conversion, as shown in Fig. 1. The results are presented visually in Sankey diagram form, permitting identification of the technical areas with the largest energy flows. Particular attention is given to the technical components, rather 2060 J.M. Cullen, J.M. Allwood / Energy 35 (2010) 2059e2069 Fig. 2. Diagram of the potential gains from energy efficiency. Fig. 1. The flow-path of energy. than economic sectors, within each chain of energy. The same technical categories are used in this paper to analyse potential efficiency gains. Cullen and Allwood make a novel distinction between ‘conversion devices’ (e.g. power stations, engines, and light bulbs) which upgrade energy into more useful forms, and ‘passive systems’ (e.g. houses, vehicles) where useful energy is finally ‘lost’ (referring to a loss of energy quality, or usefulness) as low-grade heat to the environment, in exchange for final services (e.g. transport, sustenance or thermal comfort). This becomes important when calculating efficiency targets, because in passive systems no conversion of energy takes place and therefore no meaningful theoretical efficiency target can be calculated. For example, a perfectly insulated building requires no heat input to maintain thermal comfort, giving an infinite efficiency limit for the heated space, which is nonsensical. Thus it is possible to set theoretical efficiency targets for conversion devices but not passive systems. The efficiency limits of passive systems depend instead on practical engineering design constraints and will be examined in future work. The current analysis aims to determine the efficiency terms from Equation (1) for conversion devices, and to overlay these onto the global map of energy flow. This will enable energy researchers and policy makers to:  compare the efficiencies of energy conversion devices on an equivalent basis;  determine which conversion devices have the greatest theoretical potential for reducing energy use and therefore CO2 emissions;  categorise avoidable losses according to the mechanisms which lead to the loss of energy quality. 2. Previous work: the limits of energy efficiency Estimates in the literature of possible energy savings from efficiency measures vary greatly depending on what target efficiency is chosen. For example, if the target efficiency is constrained by market forces an economic potential is defined, whereas a technical potential takes into account practical efficiency limitations, and a theoretical potential is based on thermodynamic efficiency limits. Fig. 2 summarises these approaches for calculating possible energy savings, and demonstrates with indicative values, how varied these estimates can be (adapted from Dyer et al. [5], p. 4437). Selecting a suitable target efficiency that is both objective and technically defensible is essential if the full potential for efficiency gains is to be gauged. Basing long-term targets on economic potentialsdby tracking historic efficiency indicators or surveying known technologiesdis risky because future economic drivers are difficult to forecast over long time periods. Economic drivers can vary and are in part pre-determined by governmental policies, whereas the technical and theoretical efficiency limits of energy devices do not change with time. To follow is a critical discussion of the current models used for predicting future efficiency gains, divided into three groups: top-down models, bottom-up models and theoretical models. Top-down models track historical trends in efficiency indicators and extrapolate these into the future, to determine energy efficiency targets. For example, in the World Energy Outlook reference scenario, the IEA predicts that the global average energy intensity (a measure of global energy efficiency) will fall on average by 1.7% per year from 2004 to 2030, based on the past 30 year trend [6]. However, there are two problems with such an approach. Firstly, the extrapolated efficiency target may be unachievable because it exceeds some theoretical or practical limit. An annual improvement in efficiency of 1.7% equates to a 35% saving by 2030 and an impressive 55% by 2050. For many technical devices, such gains may not be physically possible, leading to an exhaustion of the innovation potential if alternative solutions cannot be found, as discussed by Blok [7]. Secondly, these models assume that the underlying structural components of energy demand are stable and predictable over long periods, when in reality demand is affected by disruptive events, discontinuities and non-rational human behaviour. For example, Raupach et al. [8] show that the declining trend in global energy intensity from 1980 to 2000, has in recent years reversed, placing in doubt many predictions of future energy demand and associated CO2 emissions. In practice, future predictions based on extrapolation of energy trends are rarely accurate, prompting Vaclav Smil to comment that ‘long-range energy forecasts are no more than fairy tales’ [[9], p. 154]. Bottom-up models survey best practice efficiency technologies and estimate their combined potential for reducing energy demand. For example, a recent McKinsey Global Institute report found that global energy demand could be reduced by more than 20% in 2020 by diffusing currently available technologies throughout the world [10]. Efficiency options are typically ranked according to their abatement potential and cost of implementation, enabling cumulative graphs of abatement potential to be created. Some wellknown examples of abatement curves include: the Global climate abatement map by Vattenfall [11]; the IPCC bottom-up analysis for sectoral mitigation in 2030 from Sims et al. [12]; the IEA marginal abatement cost curves for sectors [2]. Such studies provide a useful snapshot of current economic and technological drivers, and show where efficient technologies can be immediately applied. However, bottom-up models in their current form are incomplete for two reasons. Firstly, these models aggregate potential efficiency gains into broad economic sectors (e.g. transport, buildings, industry), often ignoring the complex chains of technical conversion devices and energy systems involved. Efficiency gains cannot simply be added together, because a saving in one device often reduces the potential for gain in a connected device. For example, a more efficient electric motor requires less electricity for the same load, reducing the demand for generation and therefore the absolute benefit of J.M. Cullen, J.M. Allwood / Energy 35 (2010) 2059e2069 efficiency gains in that upstream generation. Secondly, bottom-up models assess only known or emerging technologies that have evolved under today's economic drivers and technical conditions. Hence, surveys of current efficiency options identify mostly incremental gains to existing processes and tend to overlook opportunities from novel disruptive technologies or divergent development pathways, which are beyond the influence of industry. Attempting to assess all available technical options, including those that are yet to be discovered, is problematic, so instead researchers have found ways to calculate the theoretical improvement potential, independent from known efficiency measures. Theoretical models define an absolute target by calculating an upper efficiency limit based on thermodynamics. These models avoid setting a theoretically impossible target and are not constrained by currently known technologies or industrial practice. The thermodynamic property exergy (also known as availability) shows how far a device is operating from its thermodynamic ideal, allowing all energy conversion devices to be compared on an equivalent basis. Detailed exergy models exist for many individual conversion devices and include useful breakdowns of exergy losses. Rosen et al. [13] stress that exergy analysis has an important role to play in charting the increase of efficiency in society, because it clearly identifies possible efficiency improvements and reductions in thermodynamic loss. The first exergy analysis of an entire society was published by Reistad [14] and estimated the overall efficiency of the United States to be 21%. A review paper by Ertesvag [15] summarises a further 15 societal exergy studies, including coverage of numerous countries, regions, and one global study by Nakicenovic et al. [16]. The analysis by Nakicenovic et al. estimates the global efficiency of energy conversion in 1990 to be about 10% of the theoretical limit, but the paper is highly technical and difficult to comprehend for a non-expert reader. Although many exergy analyses have been performed on individual conversion devices, these are also technical in nature and appear only in specialist thermodynamic journals. Attempts to aggregate exergy information for conversion devices into an accessible global form are rare, and for this reason theoretical models are often overlooked when determining research priorities and creating energy policy. In this article, a theoretical model is used to assess energy conversion devices providing an absolute basis for identifying and ranking efficiency options. This requires comparing the current efficiency of conversion devices with their theoretical maximum, while considering the scale of energy flow through the technical devices in the global energy network. Inevitably, using a purely theoretical measure of efficiency promotes an ideal which may not be practically achievable, either economically or technically. However, such an approach provides a useful theoretical target from which to set practical limits, and an absolute basis from which to measure progress. 3. Constructing a map of global energy efficiency Previous efforts to assess the potential savings from efficiency measures are useful for identifying options and directing responses in the short term. Yet their reliance upon recent economic trends and known technical options makes them unlikely to be accurate over the times scales being negotiated in climate change policy. Using an absolute measure of efficiency, such an exergy analysis, avoids the uncertainty which results from the extrapolation of economic trends and captures the potential of yet to be discovered efficient designs. This paper aims to provide a visual map of global energy efficiency, which allows options to be identified and compared according to an absolute basis, independent of benchmarks based on economic or technical limitation. Three tasks must be completed 2061 to create such a map. The first is to determine the global scale of energy flow through conversion devices, which can be found in the previous analysis by Cullen and Allwood [4]. The second, requires determining the theoretical efficiency limit for each type of conversion device, and superimposing these onto the device energy flows. Finally, it is important to present the results in a visually accessible formatdsuch as a Sankey diagramdpermitting the maximum savings from efficiency measures to be visualised. 3.1. Selecting a consistent measure of efficiency To calculate the theoretical efficiency limit for each conversion device an appropriate measure of energy efficiency is required. Conventional energy efficiency, which is based on the first-law of thermodynamics, is typically defined for a conversion device as: h¼ energy output ðusefulÞ energy input (2) A natural gas power plant operating at 40% efficiency, an electric motor that is 95% efficient, and an air conditioner with a Coefficient of Performance (COP) of 1.8, are all typical examples of reported first-law efficiencies. However, this measure of efficiency is of limited use when comparing different types of conversion devices, because it is possible to have a maximum efficiency greater than 100%, and the quality of energy is not considered. For example, in space heating applications, a typical “high-efficiency” gas burning furnace has a first-law conversion efficiency of 95%, and an electric heating system is 100% efficient. Based on these figures, it could be assumed that space heating devices are already approaching their maximum efficiency limits. However, a typical heat pump has a COP of 3 (equivalent to an efficiency of 300%) and under ideal conditions can approach 10 (or 1000%). Such large variances in efficiency result from the failure of conventional efficiency definitions to consider the quality of energyd electricity and mechanical work are more valuable energy carriers than low temperature heat. Conventional energy efficiency (based on the first-law of thermodynamics) does not take into account this difference in quality and hence is not an objective basis for evaluating energy conversion devices. In contrast, exergy efficiency (based on both the first and second laws of thermodynamics, and similar in concept to effectiveness or availability) provides a more equitable measure of conversion efficiency. It uses mechanical work rather than energy as the basis for comparing devices with each other and their thermodynamic ideal. Exergy efficiency is defined for a device as: e¼ exergy output work output ¼ exergy input maximum possible work output (3) By definition, the theoretical limit of exergy efficiency for an individual device or a chain of multiple conversion devices, is always unity. Mechanical work is chosen as a basis for comparison because it is a high quality, low entropy form of energy. Electricity, which can be perfectly converted into mechanical work, is another high quality form of energy. Thus for a device which converts one form of mechanical energy to another (e.g. gearbox), or electrical energy to mechanical energy (e.g. electric motor), exergy efficiency and energy efficiency are almost the same. However, when the input or output of the device is heat (e.g. space-heater) the energy value of the heat must be downgraded into equivalent units of mechanical work. The importance of using an absolute measure of efficiency is explained using an example of lighting devices. It is sometimes argued that replacing incandescent light bulbs with more efficient compact fluorescent bulbs saves little energy, because the building's 2062 J.M. Cullen, J.M. Allwood / Energy 35 (2010) 2059e2069 space heating requirements are offset by the bulb's waste heat production. Ignoring the fact that in many climates space heating is not required in summer, and that waste heat from the bulb may compete with air-conditioning systems, the argument is flawed because it ignores the ‘quality’ of the energy. According to the firstlaw of thermodynamics, 100% of the electricity input to the bulb is converted to either light or waste heat. Yet, from a second law perspective the electricity is high quality energy (it can be converted into work almost completely), whereas the bulb's waste heat is a low quality form of energy (it is at low temperature, so is difficult to convert to mechanical work). If a more efficient lighting device was installed, the electricity saved could be used to run a high efficiency device like a heat pump that could deliver 3 times more of the same low quality heat than the light bulb (assuming a typical COP of 3). Clearly, not all forms of energy are equal in quality or usefulness, and therefore a consistent measure such as exergy is needed to equate device efficiencies. Exergy efficiency can be calculated directly, by finding the ratio of the output to input exergy flows through the device, but in practice this is complicated. Instead, if the conventional energy efficiency (h) is known, then the exergy efficiency (e) can be estimated using: e ¼ hn Table 1 Energy and exergy efficiencies for upstream conversion devices. Device Description h n e % % % Electricity generation from: Oil Crude oil and petroleum products 37a 94 35 Biomass Combustible plant/animal products 25b 90 23 and municipal/industrial waste Gas Natural gas and gas works 40a 96 38 Coal Hard coal, lignite and derived fuels 34a 94 32 (e.g. coke, blast furnace gas) Nuclear Nuclear fission (heat equivalent 33c 100 33 of electricity) Renewable Hydro, geothermal, solar, wind, tide, 80b 100 80 and wave energy Fuel transformation In petroleum refineries, gas works, 93d 100 93 coal preparation, liquefaction, distribution and own-use CHP Combined heat and power plants (all fuels) 56d 62 35 Heat Utility heat plants (all fuels) 85d 24 20 Notes: h ¼ energy efficiency, n ¼ quality factor, e ¼ exergy efficiency. a IEA (p. 73) [25]. b Estimated. c IEA (p. 138) [26]. d Calculated from IEA [24]. (4) where a dimensionless quality factor (n) is used to correct for the loss of energy quality in the conversion process, resulting from two sources. Firstly, the chemical exergy in a fuel is marginally higher than the standard enthalpy of combustion, due to the additional contribution of the post-combustion water vapour (lower heating value) and the flue-gas components. Ertesvag and Mielnik [[17], p. 959] give values called ‘exergy factors’ which vary by between 4 and 11% across typical fuel sources. Secondly, where energy is converted into heat, the heat output must be downgraded to be measured as mechanical work, using the thermal efficiency for a reversible Carnot engine (defined as j(T T0)/Tj, where T is the heat carrier and T0 is the ambient temperature, both in Kelvin). Formal definitions of energy and exergy efficiency are found in standard thermodynamic textbooks such as Çengel and Boles [18], while more detailed explanations of the exergy method are provided by Ahern [19], Hammond [20], Kotas [21], Szargut et al. [22] and Szargut [23]. 3.2. Calculating efficiency limits in conversion devices Creating a map of global energy efficiency requires assigning average efficiencies to each conversion device in the energy network, including fuel transformation, different modes of electricity generation and end-use applications. It is important to select efficiency values that are representative of the global device average, calculated in a consistent way, and are from credible sources. The input and output energy flows for the upstream conversionsdfuel transformation and electricity generation are well defined in the energy literature, allowing efficiencies to be deduced. However, global energy flow data is not available for end-use conversion devices, and instead the efficiency values must be found by a survey of literature. The conversion efficiencies for fuel transformation and electricity generation are calculated from the 2005 Balance Table for the World, produced by the IEA [24]. This table provides values for the global energy supply broken down by fuel type, and the ‘final’ energy delivered to consumers in the form of refined fuels and electricity. Thus the average energy and exergy efficiencies for fuel transformation, electricity generation, and heat production, can be inferred from these flows and other literature sources, as shown in Table 1. Some minor differences are found between energy and exergy efficiency for combustion based electricity generation (n < 100%). The input to these devices is increased when it is changed from energy to chemical exergy, while the electricity output remains unchanged. Thus the ratio of electricity output to chemical exergy input is reduced, by the factor n, and the exergy efficiency is lower. For Combined Heat and Power plants, and Heat plants, the difference between energy and exergy efficiency is larger because the heat output of these devices must be downgraded to mechanical work. In contrast, no difference is found for fuel transformation (n ¼ 100%) because the inefficiency relates to material losses during processing, nor for nuclear and renewable sources, for which the device input remain unchanged. Finding representative efficiency values for the global stock of end-use conversion devices is difficult. Efficiencies cannot be inferred from statistical studies of global energy flows, as this data is not available for end-use devices. Instead published values for energy and exergy efficiency must be used, but these vary considerably depending on the technology and vintage of the equipment surveyed, the chosen system boundary for each device, and the geographical scope of the study. Table 2 presents a review of 10 studies, covering the last 40 years, that list conversion device efficiencies. The review indicates whether the study uses energy efficiency or exergy efficiency (or equivalent second law measure), the number of devices categories given, and describes the scope of the study. The study by Nakicenovic et al. [16] is easily the most comprehensive and consistent analysis of global energy and exergy efficiency values. Unlike other studies, which use device case studies to estimate best practice values, Nakicenovic et al. aggregate data from 11 sub-regions, and across 6 fuels types, to create average global values of energy efficiency (h) [in their Table 3.4] and exergyquality factors (n) [in their Table 3.5]. The analysis also allocates global energy flows for 1990 to the selected devices, which helps to verify the efficiency values. The list of end-use conversion devices includes: Residential/commercial sector cooking, washer/dishwasher, space heating, hot tap water, space cooling, refrigeration, mechanical energy, lighting, electronic data processing (EDP)/ television (TV), other household appliances. 2063 J.M. Cullen, J.M. Allwood / Energy 35 (2010) 2059e2069 Table 2 Survey of conversion device efficiencies. Reference Energy Summers [[41], p.151] U Reistad [[14], p. 431] Ford et al. [[3], p. 50] O'Callaghan [[42], p. 108] Culp [[43], p. 33] Gilli et al. [[44], p. 11] Nakicenovic et al. [[16], p. 228] Hammond and Stapleton [[45], pp. 152e157] U USDOE [[27], p. 1] Warr et al. [[46], pp. 34e35] U U U U U Exergy No. of devices Scope 25 Chart of best available technology, by the type of energy conversion Table of US data, including electricity generation at 38% Table, system boundary includes upstream electricity generation Chart, system boundary includes upstream electricity generation Chart of typical operational efficiencies Chart of end-use devices, with range of efficiencies shown Global values for energy and exergy efficiency, categorised by fuel Charts and tables, by domestic, commercial, industrial and transport applications Table of US power generation and industrial equipment Charts of UK devices U U U 23 17 38 28 17 20 11 U 14 10 U U U Industry process heat (low and medium temperature), high temperature heat/electrolysis, mechanical energy, other industrial uses. Transport bus/truck (Diesel), car/truck (Otto), airplanes, internal navigation (by water), rail, other. These minor adjustments made to the Nakicenovic et al. [16] data result in the list of 14 end-use conversion devices, with efficiency values, shown in Table 3. However, three adjustments are required to bring the efficiency values reported by Nakicenovic et al. into a form which is suitable for directing technical priorities today. Firstly, some individual device efficiencies are updated to reflect changes in system boundaries. The energy efficiencies for ‘mechanical energy’ devices (relating to electrical motors) are reported as 70% in industry and 54% in residential, by Nakicenovic et al. However, USDOE [27] calculate an average efficiency of 45% for the entire motor driven system, including the pump or compressor. This lower value more accurately describes the boundary system for an end-use conversion device used in this paperdthe device output is measured in its final useful form, in this case fluid motion. Ayres et al. [[28], p. 1117] provides a breakdown of industrial electricity use in the United States, which is used to separate refrigeration (6% of total net demand) from the broader category of ‘mechanical energy’ used by Nakicenovic et al. The efficiency for internal navigation (by water) is also applied to transport by international marine vessels, and biofuel powered engines are assumed to have an efficiency of 10%. Secondly, the reported efficiency values represent 1990 technology and are therefore outdated. Therefore the efficiencies are scaled to match historical improvements in global energy intensity using the IEA's reported sector improvements (cumulative from 1990 to 2004, updated for 2005): buildings 13.3%, industry 22.7% and transport 8.2% [29]. Applying a uniform efficiency improvement across the devices in each sector might lead to device efficiencies greater than 100%. Instead the historical scale factors are applied uniformly to the non-useful output of energy from each conversion device, grouped by economic sectors. Thirdly, the devices presented by Nakicenovic et al. are grouped by economic sectors instead of individual technologies. For example, the electrical motor could be listed as a distinct technology, but instead is included in four different categories: ‘mechanical energy’ (both industry and residential/commercial), ‘other household appliances’ and ‘other industrial uses’. The diesel engine is hidden in these same four categories, and in four additional transport categories: ‘bus/truck’, ‘internal navigation’, ‘rail’ and ‘other’. The selected device categories also vary considerable in scale of energy flow, from as low as 0.1% of global energy demand for ‘washer/dishwashers’ to greater than 16% for ‘space heating’. It is preferable to organise the efficiency data into technically discrete categories of conversion devices, of approximately equal scale of energy flow. For any real process (and thus energy conversions) there are always thermodynamic irreversibilities present. To provide further insight into how energy is degraded, and therefore what strategies could make better use of energy, the calculated exergy losses from each conversion device are aggregated into ten engineering loss mechanisms, as described in Table 4. There is no single study which provides a breakdown of global exergy loss across the range of conversion devices considered. Instead a number of exergy analyses of individual conversion devices are consulted: Dunbar and Lior [30] and Prins and Ptasinski [31] for generic combustion processes (applicable to engines, heaters and fossil fuel based electricity generation); Dunbar et al. 3.3. Grouping losses by engineering mechanisms Table 3 Energy and exergy efficiencies of end-use conversion devices. End-use device Description h n e % % % Motion average Diesel engine Compression ignition diesel engine: truck, car, ship, train, generator Petrol engine Spark ignition otto engine: car, generator, garden machinery Aircraft engine Turbofan, turboprop engine Other engine Steam or natural gas powered engine Electric motor AC/DC induction motor (excl. refrigeration) 26 90 24 22 95 21 28 99 27 47 53 25 60 93 56 Heat average Oil burner 58 24 14 61 25 15 13 99 12 Oil combustion device: boiler, petrochemical cracker, chemical reactor Biomass burner Wood/biomass combustion device: open fire/stove, boiler Gas burner Gas combustion device: open fire/stove, boiler, chemical reactor Coal burner Coal combustion device: open fire/stove, boiler, blast furnace, chemical reactor Electric heater Electric resistance heater, electric arc furnace Heat exchanger Direct heat application: district heat, heat from CHP Other average Cooler Light device Electronic All devices Refrigeration, air con.: industry, commercial, residential Lighting: tungsten, fluorescent, halogen Computers, televisions, portable devices 34 20 7 64 21 13 59 31 19 80 30 24 87 15 24 60 14 104 6 Notes: h ¼ energy efficiency, n ¼ quality factor, e ¼ exergy efficiency. 8 7 13 90 12 20 30 6 51 50 25 2064 J.M. Cullen, J.M. Allwood / Energy 35 (2010) 2059e2069 Table 4 Loss mechanisms. Mechanism Combustion Internal heat exchange Description Heat transfer between product molecules leaving the reaction site (with kinetic and photon energy) and neighbouring unreacted molecules, leads to unrecoverable exergy loss. Internal heat exchange can be avoided if the reactant and product streams are separated. Oxidation Chemical interactions (intra-molecular, radiation, thermo-mechanical) result from the reaction of oxygen and fuel, producing irreversible changes of energy. Conversion of chemical energy to a useful form without combustion, for example in fuel cells, can prevent some of this loss. Mixing Spontaneous mixing of reactants in the pre-combustion stage, and products in the post-combustion stage, cannot be reversed without additional energy input. It is difficult to avoid mixing in combustion processes. Heat transfer Heat exchange Heat transfer through a finite temperature produces irreversibilities (e.g. from combustion gases to steam). Minimising the temperature difference reduces losses, but increases the heat exchanger costs. Avoiding the use of high temperature fuel combustion for low quality applications (space and water heating), and cascading heat can reduce losses. Exhaust Thermal and chemical potential of stack and tailpipe emissions. Extracting heat from water vapour (condensing boilers) and completely oxidising fuel can prevent some loss. Heat loss Heat transfer from equipment to the environmental reference state. Losses can be minimised using insulation, preventing leaks of hot gas and liquids, and ensuring reactants and products leave the system at the surrounding temperature. Other Electrical resistance Resistivity (I2R), eddy currents and magnetic hysteresis losses in devices (e.g. power distribution, electric motor, light bulb, electronic). Can be minimised by selecting superior materials/metals for electrical components, and by reducing the length of electrical wires through miniaturisation of electronics and localisation of electricity supply. Friction Friction (sliding and fluid flow), inelastic deformation and unrestrained compression/expansion leads to non-recoverable exergy loss (e.g. in motors, turbine, engine, pump and pipe). Losses are reduced by using lubricants, reducing fluid flow velocities, and resisting expanding gases. Fission (nuclear) Highly irreversible fission and heat transfer processes result in losses. Can be partially reduced by using fossil fuel fired superheat and reheat units in the downstream steam system. Fuel losses Transformation, own-use, distribution and transmission of primary fuels results in physical losses (e.g. oil and gas leaks from pipelines). These can be reduced with good design and maintenance, or by using a more localised energy source. [32] and Durmayaz and Yavuz [33] for electricity generation using nuclear fission; Ertesvag and Mielnik [17] for hydroelectricity; Rakopoulos and Giakoumis [34] for diesel engines; Ford et al. [3] for petrol engines; Turgut et al. [35] for aircraft engines; Mecrow and Jack [36] and USDOE [27] for electric motor drives; Kotas [21] for refrigeration. The exergy breakdowns do not always correlate directly with the conversion device categories used in this study; in these cases scale factors, interpolation and estimation were used to complete the data. The exergy efficiencies for these individual studies were compared with the global averages presented by Nakicenovic et al. [16] and found to be broadly consistent. 3.4. Results The global map of conversion efficiency is presented in Fig. 3. Energy flow is traced from primary energy sources (left), through fuel transformation, electricity generation, and end-use device conversion, to useful energy (top-right). The vertical lines show where energy is converted to a new form, with any non-useful energy output (exergy loss) being separated from the main flow and collated in the bottom-right corner. (The labels used to distinguish the energy flows are defined in: Table 1 for energy sources, Table 3 for conversion devices, and Table 4 for loss mechanisms). The thickness of each line represents the scale of energy flow, with the use of colour to help distinguish different energy flows. Useful energy, in the form of heat, motion, light, sound, and cooling, is collected in the top-right corner and indicates the energy required if the current conversion devices were all to operate at their theoretical maximum exergetic efficiency. Energy values are reported in exajoules (EJ ¼ 1018 J) and direct CO2 emissions associated with fossil fuels are shown in the red circles in billion tonnes of CO2 (GtCO2 ¼ 109 t CO2) (based on 2005 data from the IEA Key World Energy Statistics [[37], p. 44]). 3.5. Data accuracy Rigorous data for estimating conversion device efficiencies and allocating losses, is not readily available. Energy allocation varies considerably between countries and energy efficiency differs between devices depending on the age, operation, and type of device. Although some exergy loss breakdowns are available for specific devices, the various methodologies employed make it difficult to translate this data into a consistent global analysis. To minimise the influence of these variables, data was selected from only two sources with global coveragedIEA [24] for energy flow and Nakicenovic et al. [16] for conversion device efficiencies. These sources do not include a quantitative error analysis, and therefore a formal assessment of the data accuracy cannot be made. However, each data set has been prepared in a consistent manner, having been collated from many smaller regional energy studies and surveys. All energy values reported in this analysis are rounded to the nearest EJ. One simplifying factor in this analysis is the allocation of energy use directly to the physical devices which convert energy. There is no need to embed the energy associated with upstream conversion processes such as electricity generation, or non-direct energy inputs such as transport and capital equipment. These energy inputs are allocated directly to the conversion device. This avoids the complex boundary issues associated with other energy analysis methods, such as Life Cycle Assessment, where the allocation of non-direct impacts is subject to truncation and double-counting errors, as discussed by Cullen and Allwood [38]. Despite these known imperfections in data accuracy, the use of best available energy data provides a much needed basis for prioritising action in the area of energy efficiency. It is anticipated that over time new studies will provide more accurate efficiency data for energy conversion devices, which can be used to build further upon this research. 4. Discussion Having mapped the theoretical efficiency limits for conversion devices onto the global energy network, what can now be inferred about the efficiency with which society uses energy? How do the efficiencies of different conversion devices compare? How can we 2065 J.M. Cullen, J.M. Allwood / Energy 35 (2010) 2059e2069 Fig. 3. The global map of energy conversion efficiency. interpret the map of energy efficiency, in order to direct priorities for researchers, designers and engineers working in the field of efficiency? 4.1. How efficient are current conversion devices? Individual device efficiencies from different parts of the diagram cannot be compared directly with each other. To state that an electric motor is more efficient than a diesel engine, ignores the larger upstream losses from electricity generation and distribution that are linked with the electric motor. Instead, a compound efficiency (ec) can be calculated for each energy chain, by multiplying consecutive device efficiencies together along the entire chain length: ec ¼ ef  ee  ed (5) The subscripts used to indicate the type of conversion device are taken from the map of energy flow shown in Fig. 1: c ¼ compound efficiency, f ¼ fuel transformation; e ¼ electricity generation and distribution; d ¼ device conversion (end-use). The resulting compound efficiencies for energy chains are shown in Table 5, organised by the end-use conversion devices. These indicate the theoretical efficiency limit for each chain, from fuel to useful energy, irrespective of any particular combination of conversion devices. Table 5 demonstrates that the conversion of fuels to useful energy is typically inefficient, averaging only 11% across all devices. The efficiency of conversion devices has improved only marginally over the last 15 years, when compared with the 10% calculated by Nakicenovic et al. [16]. This small absolute improvement in average device efficiency places into sharp contrast the reported and acclaimed 15% relative improvement in global energy efficiency between 1990 and 2005 IEA [25]. Furthermore, the compound efficiencies (ec) for energy chains in 2005 range from 2 to 25% suggesting any device operating above an efficiency of 20% is converting energy in an efficient manner. Most of the inefficiency can be traced to the poor conversion of energy in end-use conversion devices (ed), which average only 18%. Looking specifically at this column, it can be seen that engines, which deliver motion, typically operate with relatively high efficiencies (12e27%) due to intense development motivated by economic drivers to reduce the weight of both fuel and the engine in Table 5 Comparing the efficiency of conversion devices. Energy chain Conversion efficiencies ef ee ed ec % % % % Aircraft engine Diesel engine Other engine Electric motor Petrol engine Motion average 93 93 92 93 93 93 100 100 78 32 100 77 27 21 25 56 12 24 25 20 18 17 12 17 Coal burner Oil burner Gas burner Electric heater Biomass burner Heat exchanger Heat average 90 93 91 93 95 93 93 100 100 100 32 100 17 76 19 15 13 24 7 13 14 17 14 12 7 6 2 10 Light device Cooler Electronic Other average 93 93 93 93 34 33 32 33 12 7 6 8 4 2 2 2 Overall Average 93 70 18 11 Notes: e ¼ exergy efficiency, with subscripts, f ¼ fuel transformation; e ¼ electricity generation; d ¼ end-use device conversion; c ¼ compound efficiency. 2066 J.M. Cullen, J.M. Allwood / Energy 35 (2010) 2059e2069 transport vehicles. This is particularly the case for aircraft engines where weight constraints have resulted in highly efficient designs. Electric motors are even more efficient (56%), because the upstream conversion losses from combustion are included in the intermediate conversion step of electricity generation (ee). In contrast, devices which combust fuels to provide heat operate at lower device efficiencies (7e19%), with the variance depending primarily on the temperature at which heat is delivered. This explains why natural gas, a high quality fuel used in many low-grade applications such as space heating, is combusted at lower efficiencies than coal, which has many higher temperature industrial applications such as steel production. Cooling, lighting and electronic applications have low efficiencies (6e12%), and additional losses result from the conversion of fuel to electricity, at an efficiency of 32%. However, the efficiencies calculated in Table 5 are not in themselves sufficient for ranking conversion devices. To be consistent, the analysis needs to consider both the device efficiency limit and the scale of energy flow. For example, it would be illogical to focus efforts on improving the low efficiency of steam engines (included under other engines), when this technology is no longer in common use. The resulting efficiency gains would not translate into significant reductions in energy use or CO2 emissions because the application lacks scale. 4.2. Theoretical energy and CO2 savings Theoretical energy savings can now be calculated for each complete energy chain, from fuel to useful energy. Using Equation (1), the target efficiency is set to unity and the current efficiency equals the compound efficiency for each chain (ec), from Table 5. The corresponding savings in CO2 emissions are calculated by equating the fossil fuel energy inputs with their direct CO2 emissions, and are reported in Table 6. This allows alternative energy chains to be compared and ranked, based on the potential for reducing energy demand and CO2 emissions, and for responses to be directed towards the conversion devices with the greatest improvement potential. Table 6 shows that 85% of conversion losses can be attributed to the provision of heat and motion (10.4 and 9.6 EJ respectively, out of a total 23.8 EJ). The top half of the table is dominated by heaters, burners and engines, and efforts should be focused on improving Table 6 Theoretical energy and CO2 savings. ec Energy Energy CO2 emissions CO2 savings demand EJ savings EJ GtCO2 GtCO2 Energy Chain 1 % Electric heater Diesel engine Electric motor Biomass burner Gas burner Petrol engine Cooler Coal burner Oil burner Heat exchanger Light device Electronic Other engine Aircraft engine 93 80 83 94 88 88 98 83 86 98 96 98 82 75 58 58 55 49 47 41 33 31 28 20 18 16 10 11 54 47 46 45 41 36 33 26 24 20 17 15 8 8 3.4 4.1 3.2 0.0 2.6 2.9 1.9 2.7 1.9 1.2 1.0 0.9 0.7 0.7 3.1 3.3 2.6 0.0 2.3 2.5 1.9 2.2 1.7 1.2 1.0 0.9 0.6 0.5 Heat Motion Other 90 83 98 233 175 67 210 145 65 11.7 11.6 3.9 10.4 9.6 3.8 Total 89 475 420 27.2 23.8 Notes: ec ¼ compound exergy efficiency; potential for saving energy h conversion losses. the efficiency of these devices. Lighting devices, electronics and aircraft engines together account for less than 10% of the potential savings. Efforts aimed at promoting compact fluorescent light bulbs (CFL) and reducing electronic standby losses, present easy gains due to their relatively low efficiencies and help raise public awareness of efficiency concerns, but will not make a significant impact on demand for energy. The conversion efficiency of aircraft engines is already high (27%), suggesting that improvement in engine efficiency will be difficult to achieve, and the available energy savings at the global level are small. Thus few technical options remain to improve the energy efficiency of flying, so a reduction in CO2 emissions from this sector would be more easily obtained by a reduction in the number of flights. 4.3. Where are efficiency gains most likely? The analysis has shown that conversion devices on average operate at only 11% of their theoretical potential. Yet, given the sizeable effort already in progress to improve device efficiency, it is unlikely that this idealda factor 10 improvementdwill be approached in the near future. Where should action and responses be focused? Is it better to prioritise efforts on improving coal-fired power stations or diesel engines? This is difficult to answer because the theoretical saving in both energy and CO2 emissions depends not only on the efficiency of the individual device, but also on the upstream efficiencies of all devices in the energy chain. A solution to this question can be found by performing a sensitivity analysis to assess the energy savings that would be achieved from a small independent change in efficiency for each type of conversion device. Applying an absolute efficiency change (i.e. increasing each value of e by 1%) to each device might be misleading, as achieving an equivalent gain in an already efficient device is likely to be more difficult than for a less efficient device. Instead, the conversion loss (which equals the theoretical energy saving) for each device is reduced by 1%, and a modified device efficiency is calculated using: e0 ¼ e þ ð1 eÞ1% ¼ 0:99e þ 0:01 (6) The efficiency of each device in turn was changed to the modified value (e0 ) and the resulting total global energy input required to deliver the same useful energy was calculated. This leads to a sensitivity analysis of energy savings for the same relative level of improvement in each device, and provides a more equitable way to compare and rank individual conversion devices, irrespective of the location of the device in the energy network. This sensitivity analysis is performed for individual conversion devices, as opposed to energy chains, and the results are shown in Fig. 4. The chart shows the reduction in energy and CO2 emissions resulting from a 1% reduction in the loss from each conversion device. Efforts to improve the efficiency of coal-fired power stations will deliver the most savings in the upstream fuel conversion and electricity generation processes, because coal dominates electricity generation. However, greater energy savings are available from focusing individually on: biomass burners, coolers, gas burners and petrol engines. Collectively, prioritising efficiency measures for end-use conversion devices over fuel transformation and electricity generation delivers more than five times the potential gain (28 EJ versus 5 EJ). This is a surprising result, given the emphasis placed on improving the efficiency of electricity generation, for example in the IEA report, Energy technology perspectives 2008 [2]. Biomass burners emerge as the single most important conversion device, where the largest energy savings can be achieved from an incremental improvement in efficiency. These burners are predominantly open fires, which burn wood, dung, crop waste, coal 2067 J.M. Cullen, J.M. Allwood / Energy 35 (2010) 2059e2069 Table 7 Loss mechanisms which cause energy to be degraded. Loss mechanism EJ Fuel conversion Electricity generation Device conversiona Internal heat exchangeb Heat exchange Exhaust Electrical resistance Heat loss Oxidationb Fuel loss Friction Fission Mixingb 0 0 0 0 0 0 34 0 0 0 25 24 7 15 19 15 0 10 15 2 51 49 47 34 26 29 0 12 0 6 76 73 54 49 45 44 34 22 15 8 Heat transfer Combustion Otherc 0 0 34 50 42 40 122 86 46 172 128 120 Total 34 132 254 420 Total loss Notes: a In end-use devices. b In combustion processes. c Includes friction, electrical, fission, fuel losses. Fig. 4. The ranking of individual conversion devices by sensitivity to theoretical efficiency improvement. and charcoal, to meet the energy needs of people living in the developing world. The reason biomass burners top the sensitivity list is due to the inefficiency of the burners, averaging only 7%, and the scale of usedmore than a third of the world's population burn biomass for cooking and heating, according to Warwick and Doig (p. 1 [39]), and the IEA [[26], p. 115] report that solid biomass accounts for 10% of global energy supply. In this analysis, biomass burners do not contribute to CO2 emissions, because it is assumed that the CO2 released during combustion is equivalent to the CO2 absorbed when growing the biomass. However, if the biomass is not replaced, for example in areas where deforestation is a problem, then net CO2 emissions to the atmosphere result. Improving the efficiency of biomass burning stoves is technically very easy, and has the added benefit of reducing respiratory illness from the inhalation of smoke, which is ‘the single biggest killer of children under five years of age’ [[40], p. 24]. Options to improve stove technology or use fuels with lower CO2 emissions are held back by the lack of international political backing, limited funding and the logistical problem of disseminating new technology, due to the enormous number of open fires in use. 4.4. Understanding how energy quality is degraded The global map of energy conversion (Fig. 3) shows that only a small fraction of the available energy supply is converted to useful energy in conversion devices. This fraction represents the theoretical minimum amount of energy that is required to provide the same amount of final service (assuming the downstream passive system does not change). During the conversion process the remaining energy is degraded to low temperature heat and finally ‘lost’ to the surrounding environment. In Table 7 this loss of energy quality (or exergy) is divided into ten engineering loss mechanisms. Understanding the causes of energy degradation in conversion processes helps to direct research priorities and future technical innovation. Heat transfer processes are identified as the most significant source of loss (at 172 EJ, more than 40%). This stems from the irreversible nature of heat transfer across a finite temperature difference, and reflects the ill-considered use of high quality energy sources (fossil fuels and electricity) for low temperature applications. Combustion processes are a significant source of losses (128 EJ, 30%), especially from internal heat exchange when cold reactants mix with hot combusted products. The majority of combustion losses cannot be avoided without separating the reactant and product streams, suggesting that long-term technical opportunities lie in devices which convert chemical energy directly to electricity. Surprisingly, friction does not feature prominently in the analysis indicating that the research activity in the fields of lubrication and tribology, though important for preventing material wear and hence reducing equipment costs, have limited scope for reducing energy use. 5. Conclusion Developing more efficient energy conversion devices is essential if efforts to reduce CO2 emissions are to be successful. The global map of energy efficiency presented in this paper allows conversion devices to be ranked according to their theoretical improvement potential. The analysis makes three novel contributions to our understanding of energy efficiency by:  determining the average global conversion efficiency of devices along each individual energy chain, and presenting this analysis in a visually accessible format;  combining the scale of energy flow and the theoretical limits to efficiency to identify key areas where technical innovation is likely to deliver gains;  allocating, for the first time, the global loss of energy quality to technical loss mechanisms, to direct future priorities. 2068 J.M. Cullen, J.M. Allwood / Energy 35 (2010) 2059e2069 Simple options for improving device efficiency include moving the average global efficiency towards best practice and reducing excess capacity from over-design. For example, the global average efficiency of energy use in light devices is calculated to be 4%, still far below advanced technologies such as CFL and light emitting diodes (LED) with efficiencies above 20%. Similarly, electricity generation in advanced gas-turbine plants is approaching efficiencies of 60%, yet the global average is nearer to half this value. Many conversion devices are also over-designed for excess capacity so operate well away from their optimal efficiency point. This is the case with vehicle engines, which at normal cruising conditions operate well below their optimum efficiency, because of the requirement to have reserve power for acceleration. Designs which avoid or smooth out these peaks in power demand, such as hybrid power systems in vehicles, deliver much higher conversion efficiencies. How much of the theoretical efficiency improvement could be realised in practice? Beyond the simple gains described above, it is necessary to consider the technical and socio-economic barriers preventing advances in energy efficiency and look for alternative technology chains to deliver useful energy. There are many technical factors that prevent designers from approaching theoretical efficiency limits. For example, combustion processes, because they convert fuel into heat, are constrained by Carnot's Law and the adiabatic flame temperature of the fuel. This means that the efficiency of power generation is unlikely to rise much above 65% and current efforts to improve efficiencydfor example, increasing the heat addition temperature by using novel materials, preheating combustion reactants, extracting mechanical work from turbines prior to steam productiondwill give only incremental gains. To approach the thermodynamic limit would require avoiding combustion altogether, by converting the chemical energy in fuels directly into electricity (and then motion) in devices such as fuel cells. Significant socio-economic barriers also limit the uptake of new efficient designs. These include market imperfections (such as a lack of adequate information and financing, higher perceived costs, and differential benefits to the owner and user) and behavioural barriers (for example, consumer trends and habits, and the rebound effect). It is important that theoretical measures of efficiency gain, such as the work presented in this article, are in turn evaluated against such socio-economic considerations. Nevertheless, the overriding lesson from this analysis is to begin focusing research initiatives and directing efficiency policy towards the technical devices in which the greatest gains can be found. Only 11% of primary energy is converted into useful energy, thus the theoretical gains available are substantial. Further work is required to find the practical improvement potential for conversion devices, subject to engineering and technical constraints. In addition, estimating the potential gains from improving passive systems, for which there are no theoretical limits, would allow the overall gains from fuel to final services to be found. Nevertheless, a consistent framework for ranking energy efficiency opportunities is now in place, and can be used for directing future research and policy decisions in the field of energy efficiency. Acknowledgements The work of the first author is supported by the Overseas Research Scheme and the Cambridge Commonwealth Trust. References [1] Kaya Y. Impact of carbon dioxide emission control on GNP growth: interpretation of proposed scenarios. Paris: IPCC Energy and Industry Subgroup. Response Strategies Working Group; 1990. [2] IEA. Energy technology perspectives 2008. Paris: International Energy Agency; 2008a. [3] Efficient use of energy. In: Ford KW, Rochlin GI, Socolow RH, Hartley D, Hardesty DR, Lapp M, Dooher J, Dryer F, Berman SM, Silverstein SD, editors. Conference on the technical aspects of the more efficient use of energy, Princeton, NJ, July 8eAugust 2, 1974. New York, United States: American Institute of Physics; 1975. [4] Cullen JM, Allwood JM. The efficient use of energy: tracing the global flow of energy from fuel to service. Energy Policy 2010;38(1):75e81. [5] Dyer CH, Hammond GP, Jones CI, McKenna RC. Enabling technologies for industrial energy demand management. Energy Policy 2008;36(12):4434e43. [6] IEA. World energy outlook 2006. Paris: International Energy Agency; 2006. [7] Blok K. Improving energy efficiency by five percent and more per year? Journal of Industrial Ecology 2004;8(4):87e99. [8] Raupach MR, Marland G, Ciais P, Le Quere C, Canadell JG, Klepper G, et al. Global and regional drivers of accelerating CO2 emissions. Proceedings of the National Academy of Sciences of the United States of America 2007;104 (24):10288e93. [9] Smil V. Long-range energy forecasts are no more than fairy tales. Nature 2008;453(7192):154. [10] Bressand F, Farrell D, Hass P, Morin F, Nyquist S, Remes J, et al. Curbing global energy demand growth: the energy productivity opportunity, Tech. Rep. McKinsey Global Institute; 2007. [11] Vattenfall. Global climate abatement map, Vattenfall, www.vattenfall.com; 2007 [accessed 20.02.07]. [12] Sims R, Schock R, Adegbululgbe A, Fenhann J, Konstantinaviciute I, Moomaw W, et al. Energy supply. In: Metz B, Davidson OR, Bosch PR, Dave R, Meyer LA, editors. Climate change 2007: mitigation. Contribution of Working Group III to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge, United Kingdom and New York, NY, USA: Cambridge University Press; 2007. [13] Rosen M, Dincer I, Kanoglu M. Role of exergy in increasing efficiency and sustainability and reducing environmental impact. Energy Policy 2008;36 (1):128e37. [14] Reistad GM. Available energy-conversion and utilization in United-States. Journal of Engineering for Power-Transactions of the ASME 1975;97(3): 429e34. [15] Ertesvag I. Society exergy analysis: a comparison of different societies. Energy 2001;26(3):253e70. [16] Nakicenovic N, Gilli PV, Kurz R. Regional and global exergy and energy efficiencies. Energy 1996;21(3):223e37. [17] Ertesvag I, Mielnik M. Exergy analysis of the Norwegian society. Energy 2000;25(10):957e73. [18] Çengel Y, Boles M. Thermodynamics: an engineering approach. 5th ed. McGrawl Hill; 2006. [19] Ahern JE. The exergy method of energy systems analysis. New York: John Wiley & Sons; 1980. [20] Hammond GP. Industrial energy analysis, thermodynamics and sustainability. Applied Energy 2007;84(7e8):675e700. [21] Kotas T. The exergy method of thermal plant analysis. London: Butterworths; 1985. [22] Szargut J, Morris DR, Steward FR. Exergy analysis of thermal, chemical, and metallurgical processes. New York: Hemisphere; 1998. [23] Szargut J. Exergy method: technical and ecological applications. Great Britain: WIT Press; 2005. [24] IEA. 2005 balance table for the world. Paris: International Energy Agency, www.iea.org; 2008b [accessed July 2008]. [25] IEA. Worldwide trends in energy efficiency. Paris: International Energy Agency; 2008c. [26] IEA. Energy statistics manual. Paris: International Energy Agency; 2005. [27] USDOE. Energy use, loss and opportunities analysis. U.S. Manufacturing & Mining, Industrial Technologies Program, U.S. Department of Energy; 2004. [28] Ayres RU, Ayres LW, Pokrovsky V. On the efficiency of US electricity usage since 1900. Energy 2005;30(7):1092e145. [29] IEA. Energy use in the new millenium: trends in IEA countries. Paris: International Energy Agency; 2007a. [30] Dunbar W, Lior N. Sources of combustion irreversibility. Combustion Science and Technology 1994;103(1e6):41e61. [31] Prins M, Ptasinski K. Energy and exergy analyses of the oxidation and gasification of carbon. Energy 2005;30(7):982e1002. [32] Dunbar W, Moody S, Lior N. Exergy analysis of an operating boiling-waterreactor nuclear-power station. Energy Conversion and Management 1995;36 (3):149e59. [33] Durmayaz A, Yavuz H. Exergy analysis of a pressurized-water reactor nuclearpower plant. Applied Energy 2001;69(1):39e57. [34] Rakopoulos CD, Giakoumis EG. Comparative first- and second-law parametric study of transient diesel engine operation. Energy 2006;31(12):1927e42. [35] Turgut ET, Karakoc TH, Hepbasli A. Exergetic analysis of an aircraft turbofan engine. International Journal of Energy Research 2007;31(14):1383e97. [36] Mecrow B, Jack A. Efficiency trends in electric machines and drives. United Kingdom: Foresight Programme, Office of Science and Innovation; 2006. [37] IEA. Key world energy statistics 2007. Paris: International Energy Agency; 2007b. [38] Cullen JM, Allwood JM. The role of washing machines in life cycle assessment studies. Journal of Industrial Ecology 2009;13(1):27e37. J.M. Cullen, J.M. Allwood / Energy 35 (2010) 2059e2069 [39] Warwick H, Doig A. Smoke e the killer in the kitchen: indoor pollution in developing countries. London: ITDG Publishing; 2004. [40] Gordon B, MacKay R, Rehfuess E. Inheriting the world: the atlas of children's health and the environment. Geneva: World Health Organization; 2004. [41] Summers CM. The conversion of energy. Scientific American 1971;225(3):148e60. [42] O'Callaghan P. Design and management for energy conservation. Oxford: Pergamon Press; 1981. [43] Culp A. Principles of energy conversion. 2nd ed. McGrawl Hill; 1991. 2069 [44] Gilli PV, Nakicenovic N, Kurz R. First- and second-law efficieincies of the global and regional energy system. Austria: International Institute for Applied Systems Analysis (IIASA); 1996. [45] Hammond G, Stapleton A. Exergy analysis of the United Kingdom energy system. Proceedings of the Institution of Mechanical Engineers Part A-Power and Energy 2001;215(A2):141e62. [46] Warr B, Schandl H, Ayres R. Long term trends in resource exergy consumption and useful work supplies in the UK, 1900e2000. Canberra: CSIRO Sustainable Ecosystems; 2007.