Global and local emission impact assessment of distributed cogeneration systems with partial-load models

Applied Energy 86 (2009) 2096–2106

Contents lists available at ScienceDirect

Applied Energy journal homepage: www.elsevier.com/locate/apenergy

Global and local emission impact assessment of distributed cogeneration systems with partial-load models Pierluigi Mancarella a,b, Gianfranco Chicco b,* a b

Imperial College London, Department of Electrical and Electronic Engineering, Exhibition Road, SW7 2AZ London, UK Politecnico di Torino, Dipartimento di Ingegneria Elettrica, Corso Duca degli Abruzzi 24, 10129 Torino, Italy

a r t i c l e

i n f o

Article history: Received 29 February 2008 Received in revised form 17 December 2008 Accepted 20 December 2008 Available online 3 February 2009 Keywords: Cogeneration Distributed generation Emission factor Emission reduction Environmental impact Greenhouse gases Local emissions Microturbine Natural gas

a b s t r a c t Small-scale distributed cogeneration technologies represent a key resource to increase generation efficiency and reduce greenhouse gas emissions with respect to conventional separate production means. However, the diffusion of distributed cogeneration within urban areas, where air quality standards are quite stringent, brings about environmental concerns on a local level. In addition, partial-load emission worsening is often overlooked, which could lead to biased evaluations of the energy system environmental performance. In this paper, a comprehensive emission assessment framework suitable for addressing distributed cogeneration systems is formulated. Local and global emission impact models are presented to identify upper and lower boundary values of the environmental pressure from pollutants that would be emitted from reference technologies, to be compared to the actual emissions from distributed cogeneration. This provides synthetic information on the relative environmental impact from small-scale CHP sources, useful for general indicative and non-site-specific studies. The emission models are formulated according to an electrical output-based emission factor approach, through which off-design operation and relevant performance are easily accounted for. In particular, in order to address the issues that could arise under offdesign operation, an equivalent load model is incorporated within the proposed framework, by exploiting the duration curve of the cogenerator loading and the emissions associated to each loading level. In this way, it is possible to quantify the contribution to the emissions from cogeneration systems that might operate at partial loads for a significant portion of their operation time, as for instance in load-tracking applications. Suitability of the proposed methodology is discussed with respect to hazardous air pollutants such as NOx and CO, as well as to greenhouse gases such as CO2. Two case study applications based on the emission data of real microturbines are illustrated in order to highlight the effectiveness of the proposed assessment techniques. The numerical results exemplify the emission impact of distributed cogeneration systems operating under general and realistic loading conditions with respect to average and state-ofthe-art conventional technologies. Ó 2008 Elsevier Ltd. All rights reserved.

1. Introduction Assessing the environmental impact from conventional (centralized) and decentralized generation paradigms is particularly relevant in today’s changing energy scenario that is witnessing a radical shift from the status quo towards more distributed energy systems. The adoption of cogeneration or Combined Heat and Power (CHP) systems for small-scale applications (below 1 MWe) is one of the key drivers to the diffusion of thermal prime movers for Distributed Generation (DG) [1]. CHP systems are effective in reducing the primary energy consumption with respect to the * Corresponding author. Tel.: +39 011 090 7141; fax: +39 011 090 7199. E-mail addresses: [email protected] (P. Mancarella), gianfranco. [email protected] (G. Chicco). 0306-2619/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.apenergy.2008.12.026

conventional Separate Production (SP) of heat (produced in boilers) and electricity (produced in power plants) [2]. The most adopted DG CHP technologies are fuelled on Natural Gas (NG) and include Internal Combustion Engines (ICEs) and, more recently, Microturbines (MTs) [3,4]. The reduction in the fuel consumption from such CHP systems could bring a corresponding reduction of global emissions of CO2 seen as a Greenhouse Gas (GHG) [5–7]. The evolution of the energy generation scenario envisages a deeper penetration of CHP systems inside urban areas, where local emissions of hazardous air pollutants such as NOx, CO, SOx, Particulate Matter (PM), Unburned Hydrocarbons (UHC), and so on, may pose serious concerns [8–12]. Indeed, in urban contexts dispersion in the atmosphere of pollutants from small-scale generators sited among buildings may be more difficult than,

P. Mancarella, G. Chicco / Applied Energy 86 (2009) 2096–2106

2097

Nomenclature Acronyms BAT best available technologies CHP combined heat and power CO2ER CO2 emission reduction DG distributed generation FESR fuel energy savings ratio GHG greenhouse gases GSP global separate production ICE internal combustion engine LHV lower heating value LSP local separate production NG natural gas MT microturbine PM particulate matter SP separate production UHC unburned hydrocarbons Symbols F fuel energy content (LHV-based) [kWht] M number of hourly time intervals in the period of observation N number of loading levels Q heat [kWht] W electricity [kWhe] X generic energy output [kWh]

for instance, for big power plants with high stacks [13]. In addition, also due to the high population density, there is a number of relative weak receptors (elderly and sick people, children, etc.), with other potential impacts of pollutant emissions referred to ecosystems, monuments, and so forth [14]. A further critical point is represented by the already high background emission level mostly due to road traffic pollution. As a consequence, air quality standards and emission level limits can be quite stringent in urban areas, and environmental assessments tend to be conservative. Nevertheless, often little attention is paid at a regulation and planning stage to the emission worsening that could be brought about by consistent operation of DG systems at partial loads. This could lead to biased environmental assessment of thermal DG that were to be evaluated only on the basis of the full-load performance, whereas load-tracking operation can frequently occur for both thermal and electrical applications. In the latter case, in particular, future power system portions operated as microgrids [15,16] could more and more include the adoption of small-scale CHP or micro-CHP systems. The complexity of the issues involved in environmental assessments of distributed energy systems in urban fabrics calls for adequate approaches and methodologies. In this outlook, a systematic framework for evaluating the emission impact of smallscale CHP systems under general partial-load conditions is presented in this paper. The distributed nature of DG systems with respect to centralized power plants is addressed through a conceptual distinction between local and global emissions. Specific models based on an emission factor approach are formulated for assessing global emissions, and for approximately representing the contribution to the environmental impact due to local emissions from sources close to the receptors. The outcomes obtained from the two models can be seen as representative of boundary conditions, providing useful information to assist the operators to better understand the results under the large uncertainties characterizing the data used in the study. In the framework introduced, the relevant quantities characterising energy

d m n

g l

duration weight mass of pollutant [g] number of hourly time intervals associated to a loading level efficiency emission factor [g/kWh]

Subscripts e electrical i dummy index p pollutant t thermal y cogeneration x% percentage loading level Superscripts F fuel GSP global separate production LSP local separate production Q heat SP separate production W electricity X generic energy vector y cogeneration

efficiency and local and global emissions (formulated in terms of equivalent reference emission factors) are referred to the electrical output of the CHP system. This allows the development of analyses depending on generic operational and loading conditions of the CHP system. In particular, an equivalent load approach is introduced to take into account the wide range of loading levels (with the corresponding off-design emissions) at which a CHP system might be operated. Thereby, an integrated emission assessment for DG CHP systems is addressed by incorporating the equivalent load model into the local/global emission assessment methodology. The above issues are illustrated in the rest of the paper as follows. Section 2 introduces the representation of the energy efficiency and emission characteristics of cogeneration systems, with special focus on partial-load modelling. Section 3 describes and discusses the global and local emission assessment models, and introduces the relevant equivalent reference emission factors. Section 4 illustrates general issues related to off-design emission characterization for small-scale CHP equipment and presents the equivalent load approach. Section 5 reports the results from the methodology introduced here for two specific case study applications with commercially available MTs. The last section contains the conclusive notes. 2. Cogeneration energy efficiency performance and emission characterization 2.1. Cogeneration energy efficiency performance The energy efficiency performance of a CHP system is characterized by representing the partial-load operation conditions in function of the electrical output Wy (the subscript y points out cogeneration entries). Considering the fuel thermal input Fy (based on the Lower Heating Value – LHV) and the thermal output Qy, it is possible to define the electrical, thermal, and overall cogeneration efficiencies respectively as

2098

P. Mancarella, G. Chicco / Applied Energy 86 (2009) 2096–2106

gW ðW y Þ ¼

Wy ; F y ðW y Þ

gQ ðW y Þ ¼

Q y ðW y Þ ; F y ðW y Þ

3. Cogeneration emission assessment models

gy ðW y Þ ¼ gW ðW y Þ þ gQ ðW y Þ

ð1Þ

The efficiencies (1) generally depend on several variables besides the loading level, such as the technology, the heat recovery and generation system (e.g., for hot water or steam generation), the outdoor conditions, and so forth [17,19]. In addition, a key indicator for the characterization of the cogeneration systems operation is the heat-to-electricity cogeneration ratio [2]

ky ðW y Þ ¼

Q y ðW y Þ gQ ðW y Þ ¼ Wy gW ðW y Þ

ð2Þ

2.2. Cogeneration emission factor models The emissions of a given pollutant p from a certain combustion device are represented by an energy output-related emission factor model [17,20–22]:

mXp ðXÞ ¼ lXp ðXÞ  X

ð3Þ

mXp

where is the mass of the pollutant p emitted while producing the generic energy vector X, and the entry lXp is the emission factor or specific emissions [g/kWh] of the pollutant p with respect to X. The emission factor is a function of the specific combustion equipment and of its operating conditions. In particular, it can vary under offdesign operations (and thus with the loading level, as highlighted in (3)), with aging, and with the maintenance state. As such, when possible it is advisable to run specific on-site measurements to characterize the emission performance of given equipment. For cogeneration applications, the expression (3) originates different emission factor definitions, since X could be for instance represented by the LHV-based fuel energy input Fy [kWht], electricity output Wy [kWhe], or useful heat output Qy [kWht]. Application of (3) to these energy vectors (pointing out the dependence on electrical loads) yields W;y Q;y myp ðW y Þ ¼ lF;y p ðW y Þ  F y ðW y Þ ¼ lp ðW y Þ  W y ¼ lp ðW y Þ  Q y ðW y Þ

ð4Þ where myp is the pollutant mass emitted in cogeneration, and the right-hand sides contain the emission factors relevant to the specific considered energy vector. Double superscripts are used in the emission factor definition, with the first superscript indicating the energy vector and the second one the production reference (cogeneration, in the case of (4)). Taking into account the definitions (1) it is easy to show that

l

F;y p ðW y Þ

W;y p ðW y Þ

¼l

 gW ðW y Þ ¼ l

Q ;y p ðW y Þ

 gQ ðW y Þ

ð5Þ

and thus, from (2), Q ;y lW;y p ðW y Þ ¼ ky ðW y Þ  lp ðW y Þ

ð6Þ

3.1. Global and local emissions Classical energy assessment through energy saving indicators is based on a direct comparison of the CHP system production with the SP from conventional systems [2]. Following the same concepts, the emissions due to decentralized CHP systems could be directly compared to the emissions from conventional SP means. These SP reference emissions are conventionally modelled as coming from an equivalent power plant (standing for ‘‘centralized” electric power plants) and from an equivalent boiler (standing for residential boilers spread over the territory). [1,8–10]. However, a sheer analysis of the balance between DG CHP and SP emissions, as it is done for energy, might not be sufficient for a correct interpretation of the environmental impact issue, since it does not take into account at all the distributed nature of small-scale CHP systems. Indeed, considering the spatial impact of given pollutant typologies over the potential receptors could play a key role in the environmental evaluation, above all for DG units installed in urban areas. In fact, while the GHG impact for instance from CO2 emissions is to every extent global (global pollutants), other pollutants like NOx, CO and UHC have a relatively limited radius of impact (from few to some hundreds of kilometres, also depending on the specific site conditions) [11,13,23], so that they can be considered more as local pollutants. This distinction is reflected in the construction of specific emission models for CHP assessment [17,24]. Building up on these premises, a synthetic and general-purpose framework for assessing the impact from distributed energy resources against conventional generation in function of the operation point (i.e., taking into account off-design performance of the CHP units) is presented in the sequel. 3.2. Models for global emission assessment The global emission assessment model does not consider the spatial position of the emission sources with respect to the receptors. Therefore, for a given time span (e.g., 1 hour) during which the CHP unit is operated at a constant loading level Wy, the mass of pollutant emitted from the DG CHP unit is compared to the mass of pollutant emitted from generating in SP the same amount of cogenerated electricity Wy and heat Qy(Wy) (Fig. 1):

8 y W;y > < mp ðW y Þ ¼ lp ðW y ÞW y GSP mp ðW y Þ ¼ mQp ;SP ðW y ÞþmW;SP ðW y Þ ¼ lQp ;SP ky ðW y ÞW y þ lW;SP W y p p > : Q;SP W;SP ¼ W y ðlp ky ðW y Þþ lp Þ ð7Þ In particular, in (7) the emissions from electricity and heat in SP [g/kWhe] are assessed through the reference emission factors lW;SP p and lQp ;SP [g/kWht], respectively. The superscript GSP (Global Separate Production) points out that the conventional emissions

local m

global

y p

m py Qy

CHP plant

(η W , η Q )

Wy

m Qp , SP Qy equivalent boiler

Qy CHP plant

(η W , η Q )

Qy

m Qp , SP equivalent boiler

mWp , SP Wy

Wy equivalent power plant

Fig. 1. Emission assessment models for distributed cogeneration systems with respect to conventional generation means.

2099

P. Mancarella, G. Chicco / Applied Energy 86 (2009) 2096–2106

are assessed by taking into account both reference equivalent boiler and equivalent power plant. From (7), it is straightforward to in funcdefine an equivalent global reference emission factor lW;GSP p tion of the cogenerated electricity Wy as

l

W;GSP ðW y Þ p

mGSP p ðW y Þ ¼ ¼ lQp ;SP  ky ðW y Þ þ lW;SP p Wy

ð8Þ

Given the cogenerated electricity Wy, the equivalent emission factor (8) allows the calculation of the relevant reference emissions from SP according to the global assessment model. 3.2.1. The CO2 emission reduction (CO2ER) indicator A relevant case of global emission assessment refers to CO2 emissions, given their global warming impact as GHG. At first approximation, the emission factor lFCO2 referred to the cogeneration thermal input Fy [kWht] can be estimated as a function of the fuel carbon content and of its LHV, and thus as a function of the fuel only [17,20], independently of the loading level. The relative emission reduction from cogeneration with respect to the separate production can then be evaluated through the CO2 Emission Reduction (CO2ER) indicator as a sub-case of the models introduced in [6,7,25] for multi-generation systems:

CO2ER ¼

y mGSP CO2 ðW y Þ  mCO2 ðW y Þ

¼1

mGSP CO2 ðW y Þ

¼1

lF;y CO2  F y ðW y Þ W;SP lCO2  W y þ lQCO;SP2  Q y ðW y Þ

lF;y CO2 Q;SP lW;SP CO2  gW ðW y Þ þ lCO2  gQ ðW y Þ ð9Þ

which points out the dependence on the loading level only through the cogeneration efficiencies. In particular, if the same fuel is used for both the CHP system and the conventional generation, and taking into account (5), it is possible to rewrite (9) as

CO2ER ¼ 1  g

1 W ðW y Þ gSP e

þ

gQ ðW y Þ gSP t

ð10Þ

SP where gSP e and gt represent the conventional energy efficiencies for the reference power plant and boiler, respectively. In this case, the environmental benefits can be evaluated as a function of only the energy efficiencies involved in the analysis, and the expression of the CO2ER indicator becomes indeed equivalent to the expression of the classical Fuel Energy Savings Ratio (FESR) indicator [2], as detailed in [6,7,25].

3.3. Models for local emission assessment Large power plants, considered as references for conventional electricity production, are often far from the urban areas where most receptors are located. Focusing on NG-fuelled technologies, the most significant local emissions refer to NOx, CO and UHC [17–19,26]. The radius of impact of these pollutants is relatively limited. Therefore, the local emission assessment model takes into account only residential boilers spread over the urban fabric as the comparison reference for environmental impact analysis. Thus, considering a given time span (e.g., 1 hour) during which the CHP unit is operated at a constant loading level Wy, the mass of the generic pollutant p emitted from cogeneration is compared to the mass emitted by a reference boiler producing the same amount of heat Qy(Wy) (Fig. 1):

(

myp ðW y Þ ¼ lW;y p ðW y Þ  W y Q;SP mLSP ðW y Þ ¼ lQp ;SP  Q y ðW y Þ ¼ lQ;SP  ky ðW y Þ  W y p ðW y Þ ¼ mp p

ð11Þ In analogy to (7), the superscript LSP (Local Separate Production) points out that for local comparisons only distributed boilers are used. Hence, it is possible to define an equivalent local reference in function of the cogenerated electricity emission factor lW;LSP p Wy as

lW;LSP ðW y Þ ¼ p

mLSP p ðW y Þ ¼ lQp ;SP  ky ðW y Þ Wy

ð12Þ

For a better understanding of the models introduced, it is crucial to highlight that the expressions (11) and (12) are a function of the cogenerated electricity Wy even though the emissions accounted for as a conventional reference are related to only heat production from boilers (Fig. 1). Indeed, the reference amount of heat is linked to the cogenerated electricity Wy through the cogeneration ratio ky ðW y Þ, as it can be appreciated from (11) and (12). 3.4. Discussion on the emission assessment models The identification of the emitted amount of pollutant is only one aspect of the environmental impact evaluation, to be followed by an analysis of the pollutant dispersion in the atmosphere [13] and of the corresponding harmful effects over given receptors [11–14]. In this outlook, the local emission assessment model represents an attempt of taking into account the radius of impact of certain pollutants (relevant to the position of the receptors), besides their emitted amount, thus going beyond the application of classical balances between CHP emissions and (global) SP emissions. However, such an approach is only valid at first approximation. In fact, referring for instance to the NOx or UHC emission impact as local, the formation of secondary pollutants such as aerosol [27] or ozone [28] is not addressed. This may occur within radii of even some thousands of kilometres, so that a global emission model (entailing power plant emissions) should rather be used. However, how to take into account the dynamics of such phenomena still represents an open debate, and could be assessed only through advanced chemical models of the pollutant dispersion in the atmosphere [13], heavily depending on the specific site and on the relevant atmospheric conditions. In addition, even for site-specific assessments, most commercial software adopted to check the compliance of the resulting air quality with given standards in urban areas implement relatively simple Gaussian models [13] that do not contain chemical details. As a typical outcome, the radius of impact of the primary pollutants included in the analyses run with such models is normally limited to few kilometres [29], and as such a local comparison with only boiler emissions seems to be adequate. Hence, for general-purpose analyses aimed at a preliminary assessment of the impact on air quality in urban fabrics due to CHP systems, the ‘‘local emission approximation” represents an excellent standpoint. More generally, being the modelling uncertainties expectedly high according to the above considerations, the results of the local environmental analyses can be conveniently expressed in terms of the boundaries represented by the local and global emissions. In other words, given that neither model is sufficient in evaluating the impact from local pollutants (the local one being more conservative, while the global one being too optimistic in terms of assessing the local environmental pressure from DG), the comparison of the results from the two approaches intrinsically yields relevant information on the extent of the approximation introduced.

2100

P. Mancarella, G. Chicco / Applied Energy 86 (2009) 2096–2106

Table 1 Reference emission factors and energy efficiencies from conventional generation systems (electricity-only and heat-only generation). Reference technology

Average BAT

4. The equivalent load approach 4.1. Generalities on off-design emissions

Heat generation [g/kWht]

Electricity generation [g/kWhe]

lW;SP NOx

lW;SP CO

lW;SP CO2

gSP e (%)

lQNO;SPx

lQCO;SP

lQCO;SP2

gSP t (%)

0.5 0.1

0.3 0.2

700 363

40 55

0.2 0.05

0.04 0.02

270 210

85 95

When the radius of impact of the pollutants and the position of sources and receptors do not affect the results (as in the case of GHG emissions, for instance), the global emission model can be used outright. The emission assessment results are extremely powerful for obtaining indications independent of the specific site. These results could be fruitfully used for energy planning (e.g., assessment of the air quality change due to widespread diffusion of CHP systems, to be compared to other energy scenarios) or regulatory purposes (e.g., to set up suitable emission tests or establish adequate emission limits). In this respect, the results may strongly depend on the choice of the reference models for characterizing the conventional SP of heat and electricity, as discussed below. 3.5. Reference models for the conventional separate production of heat and electricity The numerical values of the SP reference emissions from the equivalent power plant and boiler need to be set up within a common framework, especially when this methodology is applied for regulatory purposes (see for instance [30–33] for energy assessments), since they could change significantly the outcomes of the analysis. For both energy and environmental assessment, in general two main lines can be undertaken to establish the separate production references:

In general, it is not possible to formulate comprehensive models for emission characterization of pollutants other than CO2 (for which the model outlined in Section 3.2.1 applies). In fact, the actual emissions strongly depend on the combustion dynamics of each specific device, whose modelling can be extremely complex and is out of the scope of this work. Nevertheless, some theoretical considerations on off-design emissions from combustion equipment can be carried out (see for instance [18,21,24,26]). However, it is crucial to highlight that such considerations are meant to be only indicative of possible equipment behaviours, and may be unsuitable for practical applications. In general, at partial-load emissions of CO and UHC, intrinsically related to combustion efficiency, can increase substantially, and the same could occur for NOx due to changes in the combustion characteristics. In addition, non-linearities in the emission profiles could typically take place. As confirmed by several studies [4,34–37], this holds true particularly for MTs, that in general are optimised for operation at high loading level, in analogy to the bigger gas turbines. Conversely, NG-fuelled ICEs typically exhibit a different behaviour: although full-load NOx and CO emissions are even ten times higher than MTs [17–19,38], ICEs are generally characterized by a relatively flat emission profile at variable loads, being historically adopted for a large range of load-tracking as well as back-up applications. As far as other emerging CHP technologies such as Stirling engines or fuel cells are concerned, it is expected that more relevant information on emission profiles will be available after a wider product commercialization. As general mathematical models cannot be drawn, the off-design emission characterization of specific equipment is usually based upon discrete emission profiles sampled at different operation points, given by manufacturers or gathered through field measurements. The emission data used in the case studies in Section 5 are indeed elaborated from field trial reports. 4.2. Formulation of the equivalent load approach

1. Comparison of cogeneration to average conventional technologies, thus implicitly assuming that cogeneration is displacing a wide range of separate production technologies. 2. Comparison of cogeneration to the Best Available Technologies (BAT), thus assuming that the production displaced by cogeneration refers to updated conventional means (such as low-emission high-efficiency boilers and combined cycles). Following these two streamlines, Table 1 indicates the reference emission factors and energy efficiencies from conventional SP technologies used in the case study examples of Section 5. More specifically, data for average thermal electricity generation from the power system in Italy have been estimated on the basis of information available from different sources, including estimated transmission and distribution losses. Average emissions for the thermal production are estimated by assuming an equivalent fuel mix as input to residential boilers in urban areas. Similarly, BAT data refer to NG-fuelled state-of-the-art boilers and combined cycles (including again an efficiency penalty for transmission and distribution losses). The fuel-based emission factor for natural gas is assumed as lFCO2 ¼ 200 g=kWht (referred to the LHV). The emission analyses in the sequel are focused on NOx and CO, as the most relevant pollutants from NG-fuelled cogeneration systems. However, the analyses shown can be easily extended to other pollutant typologies (such as UHC, for instance), as well as to PM, SOx or other pollutants, for studies related to fuels such as diesel or oil.

Provided an off-design characterization is available, time-domain simulations could be run for specific CHP systems (for instance in electricity tracking or heat tracking operation), obtaining case study-dependent emission results. A more general assessment, although still conventional, could be carried out by determining an equivalent load to be supplied by the CHP system, which takes into account pre-defined degrees of partialload operation. More specifically, this can be done by setting proper time duration weights for certain discrete loading levels, and by weighting the emissions relevant to each load accordingly.1 In practice, let us consider the duration curve representing the ordered sequence of the M hourly energy values W 1 ; . . . ; W M [kWhe] defined in the period of observation, sorted in descending order. The duration curve contains N discrete loading levels. Each loading level is characterised by a hourly energy W i and a corresponding number of hours ni, for i ¼ 1; . . . ; N, from which it is possible to define the duration weight di = ni/M. On these bases, the equivalent electrical load is calculated as the weighted average

W¼

N X

di W i

ð13Þ

i¼1

1 A similar approach is adopted to test non-road engine applications according to the standard ISO 8178 [39].

2101

P. Mancarella, G. Chicco / Applied Energy 86 (2009) 2096–2106

Therefore, the relevant emissions according to the different assessment models are calculated as

8 N P > > myp ðWÞ ¼ di  lW;y > p ðW i Þ  W i > > > i¼1 > > < N P mLSP di  lQp ;SP  ky ðW i Þ  W i p ðWÞ ¼ > i¼1 > > > > N > > > mGSP ðWÞ ¼ P di  ½lQ;SP  ky ðW i Þ þ lW;SP   W i : p

i¼1

p

p

ðcogenerationÞ ðreference local emissionsÞ

ð14Þ

ðreference global emissionsÞ

and the corresponding electrical output-related emission factors become

8 y > > lyp ðWÞ ¼ mpWðWÞ > > > > > > ðcogenerationÞ > > > > LSP < LSP lp ðWÞ ¼ mp WðWÞ > > ðreference local emissionsÞ > > > > GSP > GSP > > lp ðWÞ ¼ mp WðWÞ > > > : ðreference global emissionsÞ

30-kWe MT unit. Its characteristic energy and emission data represented at the three different sampled operating points (i.e., three

ð15Þ

In addition, it is also possible to calculate average electrical and thermal efficiencies gW ðWÞ and gQ ðWÞ in correspondence of the equivalent loads by interpolation of the data or by using specific models available for the CHP efficiencies (1). 4.3. Utilization of the equivalent load approach The duration and loading level of the various steps in (13) could be derived from an actual or forecast duration curve of the CHP system. In alternative, the double degree of freedom obtained by modifying the weights and the loading levels in (13) could be exploited to run generic scenario analyses. In particular, an integrated assessment of the operation for different loading configurations could be run by opportunely and conventionally weighting the different operation points, giving birth to different operational scenarios. The approach illustrated can be used under different time horizons and for different purposes. For instance, for energy planning an elementary time interval of one hour and a broad period of observation such as 1 year could be used, constructing the equivalent load synthesizing the annual CHP operation under a wide range of seasonal load variations. Conversely, by adopting a relatively short elementary time interval (e.g., 1 min) within a daily or weekly operation time span, it would be possible to focus more specifically on some dynamics of the CHP system operation, for instance representing in more detail the local emissions resulting from the application of automatic load-tracking strategies. In general, it could be possible to spot whether and to what extent specific operational conditions could generate particularly severe environmental pressure. The proposed approach could be also adopted for regulatory purposes. For instance, the duration weights and the loading levels could be established a priori within a specific regulation framework, limiting the emissions corresponding to the equivalent load so identified to a certain threshold. 5. Case study applications 5.1. Case 1: 30-kWe MT

5.1.1. Description of the equipment As a first illustrative example of partial-load emission assessment of a small-scale CHP system, let us consider a NG-fuelled

loading levels referred to the electrical capacity) are reported in Table 2. The emission characteristics are elaborated from Ref. [34]. Although a complete emission mapping is not available, the partial-load CO and NOx emission profiles prove to be strongly nonlinear. In particular, the NOx emissions increase dramatically at half load and drop to a minimum at 75% loading. On the contrary, CO emissions exhibit the highest value at 75% loading, while at half load the figures are slightly lower than the full-load ones and equal to about one tenth of the 75%-load emissions. Regarding the energy performance, at lower load the electrical efficiency decreases, while the thermal efficiency increases, as a consequence of the higher amount of thermal power discharged and thus available for heat recovery. The resulting effect is that the overall cogeneration efficiency remains roughly constant. 5.1.2. Partial-load characteristics and CO2 emission assessment Fig. 2 shows the MT energy performance characteristics for the three given loading levels (the points are interpolated with a second-order model), as well as the CO2 emission reduction characteristics, evaluated through the CO2ER indicator defined in Section 3.2.1. More specifically, the CO2 emission reduction is evaluated for average technologies through the general model (9), whereas for BAT SP technologies, both NG-fuelled like the MT analysed, the model (10) is applied. The SP figures are the ones given in Table 1. For every loading level, the CO2 emission reduction is excellent if compared to average Italian technologies, ranging from about 40% at full load to about 35% at half load. Given the symmetry between thermal and electrical assessment in the indicator (9), and SP being gSP e much lower than gt , the CO2ER is more sensitive to partial-load variations in the prime mover electrical efficiency than in its thermal one. Hence, the emission reduction decreases at partial load in consequence of the CHP electrical efficiency decrease, in spite of an increase in the CHP thermal efficiency. The results become consistently worse if comparing the MT to the BAT reference, with marginal emission reduction at high loads (about 5%), which becomes even negative at half load. The latter poor results reflect the fact that the maximum electrical performance from MTs is currently limited, also due to technological scale reasons. 5.1.3. Partial-load NOx and CO emission assessment Figs. 3 and 4 show, respectively, the NOx and the CO emission characteristics (in terms of electricity-related emission factors W;y lW;y NOx and lCO ) at the three sampled load levels. In addition, the

Table 2 Partial-load characteristics for the 30-kWe MT used in the case study (Case 1). Loading level (%)

100 75 50

Efficiencies (%)

Emission factors [mg/kWhe]

gW

gQ

gy

lW;y NOx

lW;y CO

27 26 23

49 51 54

76 77 77

100 10 2400

550 5000 350

2102

P. Mancarella, G. Chicco / Applied Energy 86 (2009) 2096–2106

efficiencies and CO 2 reduction

0.8

overall efficiency

0.7 0.6 0.5 0.4

thermal efficiency CO2ER (average technologies)

0.3

electrical efficiency

0.2 0.1

CO2ER (BAT)

0.0 -0.1

50

60

70

80

90

100

loading level [%]

Fig. 2. 30-kWe MT partial-load characteristics: efficiencies and CO2 emission reduction.

NOx emissions [mg/kWhe]

2400

MT global (average technologies) local (average technologies) global (BAT) local (BAT)

2000 1600 50% load 1200

100% load

800

75% load

400 0 0.22

0.23

0.24

0.25

0.26

0.27

0.28

electrical efficiency Fig. 3. 30-kWe MT partial-load emissions and comparison with local and global emissions from separate production: NOx case.

75% load

1000

From (13), the equivalent load can then be expressed as 100% load

W ¼ d100%  W 100% þ 0:5  ð1  d100% Þ  W 75% þ 0:5  ð1  d100% Þ  W 50%

50% load

ð16Þ

100

10

1 0.22

5.1.4. Determination of the equivalent load Following the model introduced in Section 4.2 and the concepts discussed in Section 4.3, it is possible to run scenario analyses through equivalent loads based on pre-set loading levels and corresponding duration weights. For instance, let us consider the following hypotheses: 1. The MT runs at full load for given time spans assumed as parameters, ranging from 0% to 100% at discrete 10% steps. 2. The complementary operation time span is equally shared between operation at 75% of full load and at half load.

10000

CO emissions [mg/kWhe]

tures are plotted by taking the CHP electrical efficiency gW (at the three considered part-load values) as the independent variable, to point out the simultaneous efficiency and emission change at partial load. In practice, the difference between the MT actual emissions and the local emissions yields the absolute additional (positive or negative) pressure due to CHP according to the local assessment model. Likewise, the difference between the MT actual emissions and the global emissions yields the absolute additional (positive or negative) pressure due to CHP according to the global assessment model. From inspection of Fig. 3, when the MT operates at high loading (75% and 100% of the capacity) and at high electrical efficiency, its NOx emissions are below the average local and global values. In addition, the MT emissions are even lower than the global ones corresponding to the BAT, and slightly lower (at 75% loading) or of the same order of magnitude (at full load) of the local emissions for BAT. Instead, when the loading level drops at 50% of the full load, the NOx emissions increase dramatically and become far higher than both global and local emissions, even in comparison with average SP technologies. Similar considerations can be drawn for CO emission assessment (Fig. 4). However, the outcomes are qualitatively different, because of the ‘‘inverse” relation in the partial-load emission profile for CO and NOx [21,24]. Moreover, in the CO case the partial-load emission increase is such that a logarithmic scale for the vertical axis is more suitable to represent the values. In particular, while at full load and half load the specific emissions are of the same order of magnitude of the average global emissions, at 75% loading they are one order of magnitude higher.

MT global (average technologies) local (average technologies) global (BAT) local (BAT) 0.23

0.24

0.25

0.26

0.27

0.28

electrical efficiency Fig. 4. 30-kWe MT partial-load emissions and comparison with local and global emissions from separate production: CO case.

same figures also contain the equivalent local reference emission and lW;LSP (according to the model (12)) and the factors lW;LSP NOx CO and lW;GSP equivalent global reference emission factors lW;GSP NOx CO (according to the model (8)) for both average technologies and BAT (with the reference values in Table 1). For the sake of clarity, rather than in direct function of the three loading levels, the pic-

with d100% ¼ 0; 0:1; 0:2; . . . 1. According to the objective of the specific study, as indicated in Section 4.3 the relevant emission models can be computed for a specified operation time (e.g., 1 year). For instance, applying a duration weight equal to 0.2 to the full load means that for 20% of its operation time the MT operates at full load, while for 40% of the time it operates at 75% load and for 40% of the time it operates at half load. 5.1.5. Results of the integrated emission assessment for different partial-load operational scenarios The CO2 emission assessment results for the equivalent loads calculated on the basis of (16) are shown in Fig. 5. The range of variability of the equivalent load, in percentage of the rated one, is comprised between W = 100% (corresponding to d100% ¼ 1 in (16)) and W = 62.5% (for d100% ¼ 0). The CO2ER values shown have been computed as described in Section 5.1.2 and by using average CHP electrical and thermal efficiency inputs in correspondence of W (Section 4.2). These average efficiencies have been calculated

2103

P. Mancarella, G. Chicco / Applied Energy 86 (2009) 2096–2106

10000

MT

average thermal efficiency

global (average technologies)

0.5 CO2ER (average technologies)

0.4 0.3

average electrical efficiency

0.2 0.1

CO2ER (BAT)

CO emissions [mg/kWhe]

efficiencies and CO 2 reduction

0.6

1000

local (average technologies)

100 global (BAT) 10 local (BAT)

0.0 60

65

-0.1

70

75

80

85

90

95

100

1

equivalent load (% of rated load)

60

65

70

75

80

85

90

95

100

equivalent load (% of rated load) Fig. 5. 30-kWe MT energy performance and CO2 emission reduction for different loading scenarios.

by interpolating the three sampled points available, and are shown in Fig. 5 as well. The results confirm the partial-load modifications already discussed in Section 5.1.2. In particular, it is worth pointing out that, in spite of an overall efficiency increase and due to the average electrical efficiency drop, also the CO2ER decreases for lower equivalent loads, so that the environmental performance in terms of GHG emission reduction is worse than expected by a full load-only analysis (up to becoming marginally positive for the lower equivalent load considered, in the comparison with BAT references). The emission assessment results for the different equivalent load configurations are shown in Figs. 6 and 7 for the NOx and the CO cases, respectively. In the NOx case (Fig. 6), the relevant local and global emissions from SP do not change significantly for different equivalent loads, and their spread is clearly defined and almost constant. On the contrary, the actual MT emissions increase of one order of magnitude passing from full load to lower loading levels, according to the weights described above. More specifically, the full-load emissions are comparable to the BAT local emissions and are lower than all the other comparative references. Hence, at full load the MT boasts excellent environmental performance even with respect to the most conservative SP reference (local BAT emissions). Then, MT emissions increase significantly for lower loads, and cross the average local emissions for an equivalent load of about 90%. This means that the actual additional NOx emission pressure due to the considered 30-kWe MT operating at high equivalent loads would be negligible with respect to the one of traditional boilers adopted for thermal-only production. However, for lower equivalent loads

NOx emissions [mg/kWhe]

1400

MT

1200

global (average technologies)

1000 800 local (average technologies)

600

global (BAT)

400

the average emissions become higher than the average ones from local boilers, so that the additional emission pressure from DG starts getting significant, and become even higher than the average global ones (representing for DG the optimistic threshold with reference to current technologies before yielding a critical additional environmental impact) for an equivalent load below 70%. Therefore, in the range between 70% and 90% of the equivalent load it can be concluded that the MT emissions lie within the boundary levels of the local and global emission reference thresholds calculated with respect to current (average) technologies. On the other hand, MT deployment might lead to significant additional environmental pressure with respect to BAT technologies already for equivalent loads below 96% (at which point the MT emissions cross the BAT global reference). Regarding the CO emissions (Fig. 7), again a logarithmic scale is preferred for the vertical axis in order to better represent the values. An increasing environmental pressure can be appreciated for decreasing equivalent loads. However, in this case the MT emissions are always higher or far higher than any comparative reference. These results lead to the conclusion that the actual additional CO emission pressure due to the presence of such MTs could be quite relevant with respect to the traditional boilers adopted for thermal-only production (local balance) as well as to both boilers and power plants (global balance), for all the operating conditions calculated with the above assumptions. 5.2. Case 2: 60-kWe MT 5.2.1. Equipment characteristics and CO2 emission assessment As a second numerical application of the models and analyses proposed, let us consider a NG-fuelled 60-kWe MT unit. In analogy to the system studied in Section 5.1, the characteristic energy and emission data are represented at three different sampled operating points (Table 3), elaborated from [37]. In this case, the partial-load CO and NOx emissions increase monotonically at partial load. In particular, whereas the NOx emission increase is relatively limited, the CO emissions increase dramatically passing from full load to half load. Table 3 Partial-load characteristics for the 60-kWe MT used in the case study (Case 2).

local (BAT) 200 0

Fig. 7. 30-kWe MT local and global emission assessment for different loading scenarios: CO case.

Loading level (%)

60

65

70

75

80

85

90

95

100

equivalent load (% of rated load) Fig. 6. 30-kWe MT local and global emission assessment for different loading scenarios: NOx case.

100 75 50

Efficiencies (%)

Emission factors [mg/kWhe]

gW

gQ

gy

lW;y NOx

lW;y CO

26 24 20

52 56 57

78 80 77

70 80 120

45 2200 10,250

2104

P. Mancarella, G. Chicco / Applied Energy 86 (2009) 2096–2106

overall efficiency

0.7 0.6

thermal efficiency

0.5

CO2 reduction (average technologies)

0.4 0.3

electrical efficiency

0.2 0.1

CO2 reduction (BAT)

0.0 -0.1

50

60

70

80

90

10000

100

100% load

100 75% load 10

loading level [%]

Regarding the energy performance, the electrical efficiency decreases at partial load, while the thermal one increases, for the same reasons discussed in Section 5.1.1. The overall efficiency exhibits a maximum at 75% as a synthesis of the two trends. The three efficiencies are shown in Fig. 8, together with the CO2ER indicator, evaluated as described in Section 5.1.2 and related to the SP figures in Table 1. When assessed with respect to average Italian emissions, the CO2 emission reduction lies within the range 41–32%, decreasing with the loading level particularly because of the CHP electrical efficiency reduction, as already pointed out for the 30-kWe MT. Again, if the comparison is run relatively to the BAT reference, the emission reduction varies from about 6% at full load to becoming negative below 65% of the nominal capacity. 5.2.2. Partial-load NOx and CO emission assessment Figs. 9 and 10 respectively show the electricity-related emission W;y factors lW;y NOx and lCO , plotted against the electrical efficiency gW in correspondence of the three considered loading levels (as in Figs. 3 and 4). The same graphs contain the equivalent local reference and lW;LSP , as well as the equivalent global emission factors lW;LSP NOx CO and lW;GSP , for both average techreference emission factors lW;GSP NOx CO nologies and BAT. From Fig. 9 it can be appreciated how, notwithstanding an increase at partial load, the NOx emissions remain below the SP emissions for all SP models and references and for all the loading levels.

1 0.18

100% load 600 400

75% load

200

0.20

0.22

0.24

0.26

0.24

0.28

electrical efficiency Fig. 9. 60-kWe MT partial-load emissions and comparison with local and global emissions from separate production: NOx case.

0.26

0.28

On the contrary, in the CO case (Fig. 10) at full load the MT emissions are of the same order of the local BAT emissions, but increase substantially at 75% and even more at 50% of full load (a logarithmic scale is again used), thus reaching values well above all the SP references. 5.2.3. Results of the integrated emission assessment through the equivalent load approach In order to carry out an integrated emission assessment suitable for more realistic evaluations, the same equivalent load as in (16) is built, and the same relevant emissions are calculated. The corresponding CO2 emission results are plotted in Fig. 11, together with the equivalent efficiencies, all calculated as in Section 5.1.4. The numerical outcomes again confirm that the environmental performance under various and realistic loading conditions becomes much worse than expected by a full load-based analysis. The emission evaluation for the equivalent loading levels is reported in Figs. 12 and 13 for NOx and CO, respectively. Concerning NOx (Fig. 12), at every equivalent load the MT emissions are below the SP references, as expected from the sample point characteristics in Fig. 9. Instead, for CO (Fig. 13) the actual MT emissions increase of two orders of magnitude passing from full load to half load. In particular, whereas the full-load emissions are relatively very low and comparable to BAT local emissions, they explode already at 96% equivalent load, skyrocketing above all the SP references. Hence, although at full capacity the considered MT

0.6

50% load

800

0.22

Fig. 10. 60-kWe MT partial-load emissions and comparison with local and global emissions from separate production: CO case.

efficiencies and CO 2 reduction

1000

0.20

electrical efficiency

MT global (average technologies) local (average technologies) global (BAT) local (BAT)

1200

NOx emissions [mg/kWhe ]

50% load

1000

Fig. 8. 60-kWe MT partial-load characteristics: efficiencies and CO2 emission reduction.

0 0.18

MT global (average technologies) local (average technologies) global (BAT) local (BAT)

100000

CO emissions [mg/kWhe ]

efficiencies and CO 2 reduction

0.8

average thermal efficiency

0.5 CO2ER (average technologies)

0.4 0.3 0.2

average electrical efficiency

0.1

CO2ER (BAT)

0.0 60 -0.1

65

70

75

80

85

90

95

100

equivalent load [%]

Fig. 11. 60-kWe MT energy performance and CO2 emission reduction for different loading scenarios.

P. Mancarella, G. Chicco / Applied Energy 86 (2009) 2096–2106

NOx emissions [mg/kWhe ]

1200

global (average technologies)

1000 800 local (average technologies) 600

global (BAT)

400

local (BAT) MT

200 0 60

65

70

75

80

85

90

95

100

equivalent load [%] Fig. 12. 60-kWe MT local and global emission assessment for different loading scenarios: NOx case.

MT

CO emissions [mg/kWhe ]

10000

mulation of an equivalent load model with relevant weights brings about a smoothing effect on the specific emission trends. This makes the results less critically dependent on the emission data available (that can exhibit large variations for different loading levels, as in the cases analysed). Different partial-load emission profiles can thus be easily handled within the equivalent load approach. As such, this approach is suitable to address general analyses, for instance aimed at assisting policy development, by definition independent of the specific equipment. From a numerical standpoint relevant to the application cases considered, a particularly critical situation can be pointed out for both MT types with regard to CO emissions. In this case, in order to preserve the air quality standards, a suitable solution could be the installation of CO abatement systems, which of course would increase the energy system cost and might substantially reduce the plant economic profitability, above all for small-scale units. The same need for installing abatement systems could be in general envisaged in the presence of binding constraints over NOx emissions (or other pollutants), for instance for the MT in Case 1. 6. Conclusion

global (average technologies) local (average technologies)

1000

100 global (BAT) 10

1

2105

local (BAT)

60

65

70

75

80

85

90

95

100

equivalent load [%] Fig. 13. 60-kWe MT local and global emission assessment for different loading scenarios: CO case.

exhibits excellent environmental performance, its deployment might lead to significant additional environmental pressure in terms of CO even at very high equivalent loading levels, as for the 30-kWe MT analysed above. 5.3. Comments on the numerical results The numerical analyses run for the two considered case studies exemplify how it is generally possible to obtain different partialload emission trends in dependence of the specific equipment. This is in line with the considerations drawn in Section 4.1, according to which no mathematical model can represent pollutant emissions other than CO2 in a general way. However, regardless of the partial-load emission profile being highly non-linear or simply monotonic, with the adoption of the proposed methodology the emissions in the considered cases increase at lower equivalent loads, determining a worse environmental performance for both local and GHG emissions with respect to the rated conditions. In this respect, neglecting partial-load operation might lead to completely wrong environmental assessment. The loading levels and duration weights used in the formulation of the equivalent load result in relatively smooth trends of variation of the specific emissions when the equivalent load varies. In the cases analysed, the specific equivalent emissions increase in a monotonic way with decreasing loading levels. However, in principle the specific emissions could exhibit non-monotonic trends, depending on the partial-load emission profiles that can be encountered in practical applications. The salient aspect is that for-

This paper has presented a general methodology to assess the emission impact due to adoption of DG CHP systems in urban areas under general operating conditions (including, in particular, partial-load operation). Simplified models based upon a distinction between local and global emissions have been presented and their suitability has been discussed for different air pollutants. For various pollutants locally produced by the CHP system, the boundary values from the two emission models presented are key outcomes of the proposed approach. More specifically, the results from the local emission model represent a lower boundary of the actual emissions, since large power plants are assumed to be far enough not to be considered within the impact radius, and the effects of chemical phenomena on a larger scale are neglected in the model; on the other hand, the results from the global emission model provide the upper boundary. Availability of the boundary values enables to make evaluations independent of the specific characteristics of the sites, and to focus on the details of the emissions from cogeneration technologies operating in various loading conditions. Different levels and durations of partial-load operation are represented in a synthetic way through the equivalent load model introduced in this paper. This enables to get a fuller picture of the actual emission pressure from CHP system under real operating conditions. The characteristics of generality and synthesis of the proposed approach are of clear interest for energy planners or policy makers. In particular, this approach can be useful to obtain indicative results for broad emission impact assessments, extended in time (such as scenario studies) or in width (region- or nation-wide), when complex models for evaluating the diffusion of the pollutants cannot be used (for unavailability of the large amount of detailed data needed) or when their results could be affected by large and even difficult to quantify uncertainties (due to uncertainty on the input data and on the model itself). The results of the case studies run by considering two typical MT types available on the market have highlighted how the fullload emission characterization can be unsuitable to estimate the environmental impact due to actual operating conditions. Critical situations have emerged in the specific case for what concerns high NOx emissions at low equivalent load (for the 30-kWe MT) and high CO emissions at basically any loading condition (for both MT types). The equivalent load model, encompassing partial-load behaviour, can be a significant value-added option with respect to more classical analyses run considering only global emissions

2106

P. Mancarella, G. Chicco / Applied Energy 86 (2009) 2096–2106

balances and full-load performance. This is particularly true for CO, whose emissions may increase tremendously at partial load, as in the cases shown in this paper with reference to MTs with different CO emission characteristics. Although the specific results are not meant to be generalized, the synthetic and simple methodology proposed is highly suitable for running effective analyses with indicative or conventionally established emission data, regardless of the specific application. This aspect is particularly useful to formulate sound emission policies for distributed energy systems. Works in progress are aimed at adopting the methodology introduced here for a comprehensive comparative assessment of different distributed cogeneration equipment (MTs, ICEs, fuel cells, and so on) from a technical, environmental and economic standpoint. Future extensions of the methodology envisage to entail the presence of different input fuels to the DG system, such as bio-masses [40] or hydrogen [41], which call for adopting specific approaches, for instance based upon life-cycle assessments [40– 43]. Acknowledgments This work has been in part supported by the Regione Piemonte, Torino, Italy, under the research grant C65/2004 ‘‘Territorial sustainability of distributed energy generation and interactions with electro-energetic systems”. References [1] Pepermans G, Driesen J, Haeseldonckx D, Belmans R, D’haeseleer W. Distributed generation: definition, benefits and issues. Energy Policy 2005;33:787–98. [2] Horlock JH. Cogeneration-combined heat and power. Malabar, FL, USA: Krieger; 1997. [3] Kaikko J, Backman J. Technical and economic performance analysis for a microturbine in combined heat and power generation. Energy 2007;4:378–87. [4] Colombo LPM, Armanasco F, Perego O. Experimentation on a cogenerative system based on a microturbine. Appl Therm Eng 2007;27:705–11. [5] Meunier F. Co- and tri-generation contribution to climate change control. Appl Therm Eng 2002;22:703–18. [6] Chicco G, Mancarella P. Assessment of the greenhouse gas emission from cogeneration and trigeneration systems. Part I: models and indicators. Energy 2008;33:410–7. [7] Mancarella P, Chicco G. Assessment of the greenhouse gas emission from cogeneration and trigeneration systems. Part II: analysis techniques and applications cases. Energy 2008;33:418–30. [8] Strachan N, Farrell A. Emission from distributed vs. centralized generation: the importance of system performance. Energy Policy 2006;34:2677–89. [9] Hadley SW, Van Dyke JW. Emissions benefits of distributed generation in the Texas market. Report prepared for the Gas Technology Institute. Oak Ridge National Laboratory, Oak Ridge, Tennessee, USA; 2003. [December 2008]. [10] Pehnt M. Environmental impact of distributed energy systems – the case of micro cogeneration. Environ Sci Policy 2008;2:25–37. [11] Allison J, Lents J. Encouraging distributed generation of power that improves air quality: can we have our cake and eat it too? Energy Policy 2002;30:737–52. [12] Gulli F. Social choice, uncertainty about external costs and trade-off between intergenerational environmental impacts: the emblematic case of gas-based energy supply decentralization. Ecol Econ 2006;57:282–305. [13] Arya SP. Air pollution meteorology and dispersion. NY: Oxford University Press; 1999. [14] Dincer I. Environmental impact of energy. Energy Policy 1999;27:845–54. [15] Lasseter RH, Akhil A, Marnay C, Stephens J, Dagle J, Guttromson R, et al. The CERTS microgrid concept. White paper on integration of distributed energy resources, prepared for transmission reliability program, Office of Power Technologies, US Department of Energy, April 2002. [December 2008]. [16] Hatziargyriou N, Asano H, Iravani R, Marnay C. Microgrids. An overview of ongoing research, development, and demonstration projects. IEEE Power Energy 2007;5:78–94.

[17] EDUCOGEN. The European educational tool on cogeneration, December 2001. [December 2008]. [18] Distributed Energy Resources Emissions Survey and Technology Characterization. [December 2008]. [19] US Environmental Protection Agency. Catalog of CHP technologies. [December 2008]. [20] Cârdu M, Baica M. Regarding the greenhouse gas emissions of thermopower plants. Energy Convers Manage 2002;43:2135–44. [21] Cârdu M, Baica M. Regarding the relation between the NOx content and CO content in thermo power plants flue gases. Energy Convers Manage 2005;46:47–59. [22] Rosen MA. Allocating carbon dioxide emissions from cogeneration systems: descriptions of selected output-based methods. J Clean Prod 2008;16:171–7. [23] Heat GA, Granvold PV, Hoats AS, Nazaroff W. Intake fraction assessment of the air pollutant exposure implications of a shift toward distributed electricity generation. Atmos Environ 2006;40:7164–77. [24] Canova A, Chicco G, Genon G, Mancarella P. Emission characterization and evaluation of natural gas-fueled cogeneration microturbines and internal combustion engines. Energy Convers Manage 2008;49:2900–9. [25] Chicco G, Mancarella P. A unified model for energy and environmental performance assessment of natural gas-fueled poly-generation systems. Energy Convers Manage 2008;49:2069–77. [26] Lozza G. Gas turbines and combined cycles (in Italian). Esculapio, Progetto Leonardo, Italy: Bologna; 2000. [27] Redington AL, Derwnt RG. Calculation of sulphate and nitrate aerosol concentrations over Europe using a Lagrangian dispersion model. Atmos Environ 2002;36:4425–39. [28] Jenkin ME, Clemitshaw KC. Ozone and other secondary photochemical pollutants: chemical processes governing their formation in the planetary boundary layer. Atmos Environ 2000;34:2499–527. [29] Torchio MF, Genon G, Poggio A, Poggio M. Merging of energy and environmental analyses for district heating systems. Energy, in press, doi:10.1016/j.energy.2008.01.012. [30] Cardona E, Piacentino A. Cogeneration: a regulatory framework toward growth. Energy Policy 2005;33:2100–11. [31] Directive 2004/8/EC of the European Parliament and of the Council of 11 February 2004 on the promotion of cogeneration based on a useful heat demand in the internal energy market and amending Directive 92/42/EEC. Official Journal of the European Union, 21.2.2004; L 52: 50–60. [December 2008]. [32] COGEN EUROPE. Position statement on the definition of harmonised efficiency reference values. February 2005. [December 2008]. [33] Deliberation no. 42/02 of the Italian AEEG (Electricity and Gas Authority, Autorità per l’Energia Elettrica e il Gas) 2002. [December 2008]. [34] Petrov AY, Zaltash A, Rizy TD, Labinov SD. Environmental aspects of operation of a gas-fired microturbine-based CHP system. Oak Ridge National Laboratory report 2001. [December 2008]. [35] Hanssen JH, et al. Operating experiences from 18 microturbine applications for CHP and industrial purposes. Danish Gas Technology Centre, report C0404. [December 2008]. [36] Greenhouse Gas Technology Center. Greenhouse cas (GHG) verification guideline series: natural gas-fired microturbine electrical generators. March 2002. [December 2008]. [37] Greenhouse Gas Technology Center. Environmental technology verification report: Capstone 60 kW microturbine CHP system. September 2003. [December 2008]. [38] Onovwiona HI, Ugursal VI. Residential cogeneration systems: review of the current technology. Renew Sust Energy Rev 2006;10:389–431. [39] Emission Test Cycles – ISO 8178. [December 2008]. [40] Chevalier C, Meunier F. Environmental assessment of biogas co- or trigeneration units by life cycle analysis methodology. Appl Therm Eng 2005;25:3025–41. [41] Raugei M, Bargigli S, Ulgiati S. A multi-criteria life cycle assessment of molten carbonate fuel cells (MCFC) – a comparison to natural gas turbines. Int J Hydrogen Energy 2005;30:123–30. [42] Voorspools KR, Brouwers EZ, D’haeseleer WD. Energy content and indirect greenhouse gas emissions embedded in ‘emission-free’ power plants: results for the low countries. Appl Energy 2000;67:307–30. [43] Koroneos C, Dompros A, Roumbas G, Moussiopoulos N. Life cycle assessment of hydrogen fuel production processes. Int J Hydrogen Energy 2004;29:1443–50.