Вы находитесь на странице: 1из 10

Energy 36 (2011) 556e565

Contents lists available at ScienceDirect

Energy
journal homepage: www.elsevier.com/locate/energy

Multi-criteria analysis for the selection of space heating systems in an industrial building
Damiana Chinese*, Gioacchino Nardin, Onorio Saro
DiEM e Department of Energy Technologies, University of Udine, Italy

a r t i c l e i n f o
Article history: Received 22 February 2010 Received in revised form 28 September 2010 Accepted 2 October 2010 Available online 11 November 2010 Keywords: Industrial space heating systems Analytic Hierarchy Process Energy system choices Energy management

a b s t r a c t
Decisions on space heating of industrial buildings involve several conicting objectives and solutions prospected by economic optimization (e.g. life cycle cost minimization) models may be far from the real preferences of decision makers. To overcome this limit, multi-criteria decision analysis which has hardly been used at single building level, especially in industrial contexts, could be a helpful methodology. This paper presents the application of a well-known multi-criteria approach, the Analytic Hierarchy Process (AHP), to the selection of space heating systems for an industrial building. We discuss the technologies available for industrial heating, criteria elicited from the decision maker and the ranking of alternatives identied with our AHP model. As very little is reported in literature about industrial energy system choices, our study, although focused on an individual case, may shed some light on decision making in this sector. To this end, we also compare our results with evidence on residential heating systems choices derived from literature. Investment costs are the most important criterion for industry, whereas qualitative attributes and operational costs are most important for homes. Qualitative attributes also significantly affect industrial heating system choices and, as the AHP is particularly effective in handling these aspects, we suggest it could be used for tactical energy planning models. 2010 Elsevier Ltd. All rights reserved.

1. Introduction The heating requirements of manufacturing rms are considerable and, even if process heating is dominant in some sectors (e.g. pulp and paper, glass and ceramics), space heating of industrial buildings is generally a major component of energy requirements. Given recent oscillations and growth spurts in fossil fuel prices manufacturing companies are concerned with nding reliable low cost alternatives for space heating: relevant decisions have thus recently increased in importance. The potential for improvement and energy savings is generally remarkable because a number of technologies and sources can be exploited, including renewable ones, which, thanks to recent changes in market conditions, have turned out to be cost-effective even for small scale applications [1]. Decisions on how to heat industrial buildings have thus also increased in complexity, not only because they involve many

* Corresponding author. DiEM, Dipartimento di Energetica e Macchine, University of Udine, Via delle Scienze 208, 33100 UDINE, Italy. Tel.: 39 0432 558024; fax: 39 0432 558027. E-mail address: damiana.chinese@uniud.it (D. Chinese). 0360-5442/$ e see front matter 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.energy.2010.10.005

technological alternatives but also because they imply many e sometimes conicting e criteria. First of all, limited investment capitals and possible difculty accessing credit may prevent companies from opting for sustainable energy sources, which usually pay off in some years but require large initial expenditures. Thus, while life cycle cost (LCC) is an important indicator of economic performance in literature (see e.g. [2,3]), minimum life cycle cost facilities may seldom be preferred in reality. Secondly, solutions which are most advantageous both from a capital and a life cycle cost perspective may nevertheless be rejected on grounds other than economic, as observed for example in the case of bioenergy systems [4]. According to Bucholz et al. [4], this is because objective experts focusing on minimum cost and technically optimal solutions often fail to properly involve stakeholders, i.e. to allow them to formulate judgments based on their normative values. The involvement issue is especially critical in the case of group decision making, but is also relevant in consultancy relations to single clients [5]. Another crucial issue concerns the information type used for decision making: costs represent quantitative, hard information, which is mostly used in simulation and optimization models for the design of thermal systems [6] and in a number of decision support tools for the design of industrial heating systems available in literature (e.g. [7e9]). However,

D. Chinese et al. / Energy 36 (2011) 556e565

557

decision makers often attach fundamental importance to soft, qualitative information [10], which signicantly affects nal choices but may be difcult to incorporate into these quantitative tools for thermal system designs. In order to help people make decisions in line with their preferences in cases where there is more than one criterion, a number of formal methods have been devised. These methods are generally known under the common label multiple-criteria decision analysis (MCDA) or e interchangeably e multiple-criteria decision making (MCDM). A recent review of these methods and their application in the energy eld can be found in Lken [11]. This paper aims to show, through a case study, how MCDA could foster a conscious selection from among the manifold technical options for industrial heating and allow the identication of decision makers needs by dealing with the complex issues discussed above. Looking at our case study and literature on energy system choices, we will also point out some distinctive features of choice of heating systems in an industrial context as compared to a residential context. 2. Methodology: MCDA methods for the selection of heating systems The selection of heating systems for individual buildings is a relatively new eld of application for MCDA methods. In fact, although MCDA has been widely used in energy planning since the Eighties [12], in the past its application was almost exclusively at a national or regional level: Lken [11], in particular, points out that MCDA has never been applied to local energy systems involving several energy resources. Indeed, at the even smaller scale of building energy systems there are few applications of MCDA methods. Multi-objective optimization tools for the design of building envelopes [13] or heating systems [14] have been proposed for the technical optimization of system structure on a quantitative basis: multi-objective models are, in fact, suitable for solving continuous decision making problems [10], such as the optimal thickness of insulation for example. On the contrary, so called multi-attribute approaches are assumed to have a predetermined, limited number of alternatives: in fact, solving a multi-attribute decision making problem is a selection process, as opposed to a design process [10]. Up until now, just a couple of examples of treating heating system selection problems with multi-attribute approaches could be found in literature. A wellknown multi-attribute approach, the ELECTRE III method [15], was applied to the selection of the heating system for an historical building by Thiel and Mroz [16]: they compare three alternative heating systems for a historical building based on quantitative criteria such as primary energy consumption, total cost of usage and qualitative criteria, e.g. the difculty of implementation, expressed as abstract numbers between 0 and 1. Only traditional alternatives (fossil fuels and electric heating) are considered in this case and the main emphasis is on heating distribution systems and their interaction with the existing structural elements of the historical building. Kaklauskas et al. [17] introduce and widely apply their own multi-attribute method (COPRAS) to residential buildings, mainly dealing with building envelope design and refurbishment problems (e.g. [18]) in order to support contractor selection for public buildings. General policy making is the objective of Jaber et al. [19] and of Alanne et al. [20]: the former use the AHP method [21] and fuzzy sets [22] to evaluate conventional and renewable energy sources for household heating, the latter focus on micro-CHP and residential heating systems and handle uncertainties on the actual preferences of type decision makers for type heating systems by means of the multi-criteria method PAIRS [23]. Thus far, industrial buildings and heating systems seem not to have been considered, but they seem to be of interest for their importance in national energy balance, on one

hand, and for specic aspects that differentiate them from residential and civic buildings on the other. Energy conversion and heating distribution technologies used in industrial buildings are diverse and, above all, values and criteria involved in dealing with energy problems [24] may be quite dissimilar from those adopted by private citizens for their homes. By investigating an industrial case study with a multi-criteria method, this paper will bring such particularities to light. In the following sections, we will briey present the features and energy requirements of the industrial building concerned and the scope of heating technologies which have been deemed feasible in this case. We will then present a multicriteria model constructed with the AHP approach [21], focusing on the criteria elicited from the building owner and the resulting priorities. 2.1. The AHP approach The AHP approach is a popular multi-attribute approach aimed at arriving at a rank order of alternatives characterized by single synthesizing values representing the decision makers preference. The steps required to analyze a decision problem with the AHP are [25]: 1. To break down the decision problem into a hierarchy consisting of the overall goal, one or more level of criteria and subobjectives and of the alternatives to be evaluated, which represent the lowest level of the hierarchy. The hierarchic structure is a base axiom [21] of AHP. 2. To assess the relative importance of the objectives and the contribution of alternatives to the objectives via pairwise comparison. Although an alternative absolute measurement system is also available, pairwise comparison is the typical approach of AHP, based on cognitive psychology considerations [21]. 3. To synthesize the pairwise comparison matrices to obtain weights/priorities, thereby calculating a consistency ratio to check the degree of inconsistency in the comparisons. The latter should be reconsidered if the consistency ratio exceeds 10% [21]. 4. To synthesize the individual comparison matrices by multiplying the evaluations obtained for single alternatives with their respective relative importance and by additively aggregating the obtained results to calculate overall preferences. Sensitivity and scenario analyses can then be performed to test how robust is the preference ranking of alternatives to changes in objective weights. Up until now the AHP method has been widely applied in the energy planning eld, especially for power generation and renewable energy planning (see e.g. Ref. [26] for a review and Ref. [27,28] for more recent examples); examples of its application to general policy making as to household heating [19] and lighting [29] were also examined. In spite of some limitations and critical points [21] the AHP method is according to many authors a leading MCDA methodology [26] mainly due to its ease of use [19]. This is especially true for industry decision makers, who found the pairwise comparison underlying the process intuitively appealing and a good basis for discussion [25]. These antecedents and the positive attitude from industrial decision makers that we experienced were our motivation for choosing the AHP method for our case study. 3. Case study The industrial building concerned is a recently refurbished construction which will accommodate the operations of a small

558

D. Chinese et al. / Energy 36 (2011) 556e565

mechanical company manufacturing components for the appliance industry. The building has a rectangular plan of about 1700 m2 and a saw-tooth roof with a minimum internal height of 7.9 m and a maximum height of 10.2 m. The resulting net volume is about 15000 m3. The rm is located near Treviso (North Eastern Italy): the number of degree days in the area is 2422, with a heating season of 183 days/year and an outdoor design temperature of 5  C. The company manager initially contacted a number of potential suppliers (heating system manufacturers or external engineering companies) and received a variety of offers, substantially differing in energy source and distribution system and sometimes extremely focused either solely on energy conversion or solely on energy distribution, respectively. As the proposed alternatives were so diverse, our support was consequently required to help select the most suitable option, therefore our aim was not to design the system or suggest further alternatives, but to frame and evaluate the available ones. In Fig. 1 we summarize the possible combinations of energy conversion and heat distribution systems conceived for the building examined. Basically, we can group heat distribution systems into three classes, depending on the temperature of the owing heat carrier. High temperature systems, including radiant tubes and radiant modules, are radiant heating systems which fundamentally consist of a burner unit coupled with a heat exchanger tube where combustion ue gases ow. Tubes reach temperatures of 400e500  C, providing a very high radiation efciency, also by means of the reecting cover placed over the tubes, which focuses the thermal ow downwards. Radiant tube systems are usually formed by a number of identical units that measure about 5e15 m each and are suspended from the factory ceiling. These are controlled by a centralized equipment control system which allows heating to be limited to operative areas of the building only. Radiant modules are similar to tubes but consist of suspended radiationheating belts forming single closed circuits whose shape and length are customized according to the layout of the building and operations. A whole building can be thus heated by a single belt, fed by a single burner located outside the building [30].

Medium temperature systems considered include hot water air heaters and hot water radiant panels or strips. Hot water air heaters are watereair heat exchange units with air blowers installed inside, which better distribute heat throughout rooms. Hot water is generated by a single separate boiler and piped to each heat exchange unit. Heat exchange is convection based. If coupled with external chillers, the same units can also be used in summer for air conditioning. Hot water radiant panels comprise hot water pipes fastened to steel plates which can be connected to form heating circuits. Water temperatures are such that a consistent proportion of heat exchange occurs by radiation and heating is localized to operation areas. Finally, low temperature systems include wall or oor heating systems, where warm water (at about 40  C) is circulated through extended pipes located beneath the oor or within wall coverings. The main advantage of these systems is the opportunity to exploit low temperature heat sources, e.g. warm water from condensing boilers, from solar panels or from ground source heat pumps. In our case, since the rm has its own well for industrial water, a closedloop groundwater heat pump was considered a low temperature heat generation system and compared with natural gas condensing boilers. Natural gas is the cheapest fossil fuel locally available and therefore has always been considered by prospective suppliers as the traditional fuel to feed burners or boilers. Based on local climate conditions, the building structure and an indoor design temperature of 16  C given the activity level of workers, heat load was rstly estimated through building simulation software [31]; however, given the specicity of industrial heat distribution systems, we preferred to size equipment based on constructor engineering manuals [32,33]. Calculated heating system capacities are reported in Table 1. Yearly energy requirements (YER) were estimated by taking into account the local degree days, the indoor design temperature and an operating time (H) of 14 h per day and 5 days a week according to Eq. (1):

YER

Dtmax

DDcorr H

(1)

Radiant tubes
Gas burners High temperature

Radiant models
High efficiency gas boiler

Where Q is the heating system capacity, Dtmax is the difference between the indoor design temperature and the outdoor design temperature, k is a correction factor to account for less than nominal efciency during the season, DDcorr is the number of degree days corrected for an indoor temperature of 16  C instead of 20  C and for operations 5 days a week. Assuming an average COP equal to 3.8 for geothermal heat pumps, an equivalent of 0.82103 TOE for Nm3 of natural gas and of 0.187 TOE for MWh of electric energy, the yearly primary energy requirements are those reported in Table 1.

Hot water radiant panels


Medium temperature
Table 1 Capacities and primary energy demand of the examined heating systems. Heat distribution system Hot water air heaters Hot water radiant panels Radiant modules or radiant tubes Radiant oor heating systems Heating system Yearly primary capacity [kW] energy demand [TOE/year] 300 250 191 278 22.0 18.3 14.0 20.4 (natural gas condensing boiler) 11.7 (geothermal heat pump) 18.9 (natural gas condensing boiler) 10.9 (geothermal heat pump)

Condensing gas boiler

Hot water air heaters

Geothermal heat pump

Radiant wall heating


Low temperature

Radiant floor heating


Entails air conditioning Compatible with air conditioning
Fig. 1. Examined combinations of heat generation and distribution systems for the industrial building of concern. Radiant wall heating systems 258

D. Chinese et al. / Energy 36 (2011) 556e565

559

4. AHP model for the system concerned The decision problem was broken down into the hierarchy tree presented in Fig. 2. The lowest level is represented by the alternatives, which result from the combinations of energy conversion systems and heat distribution systems presented in Fig. 1. The intermediate level is formed by the criteria used for the evaluation. The top level contains the goal, which is the choice of the best heating system according to the criteria and sub-criteria reported at intermediate levels. These criteria were elicited from the industrial decision makers through a number of interviews, rstly dening a wider set of criteria, then identifying the most important through a tentative evaluation of criteria weights by pairwise comparison and removing the more minor with the aim of obtaining no more than nine nal criteria, as suggested by AHP originator Saaty based on cognitive psychology and mathematical consistency arguments [34]. Minor criteria that were removed include impact on company image and primary energy use to give two examples. Although use of sustainable energy sources and technologies may contribute to a greener company image, and the rm examined is attentive to these factors and has obtained international certication of social responsibility, in the results the nal weight of these criteria was quite low. This is probably due to the fact that the contribution of relatively invisible facilities such as heating systems to image was perceived as small and because it was considered that primary energy savings should rather be mirrored by lower operation and life cycle costs. In the end, the following nine criteria were retained: 1. Reliability, understood as the probability of fault free performance of the equipment over the reference time frame of 15 years, which was considered a reasonable useful life for the heating systems; 2. Time to repair, understood as the average time required for a technician to arrive on site, detect a fault and x the faulty equipment; 3. Lead time, i.e. the period of time between placing the order and having the system installed and functioning on site;

4. Source exibility, i.e. to what extent is it possible to operate the system with different fuels, especially in case of unexpected rises in prices or difculties in fuel procurement or delivery; 5. Layout exibility, i.e. how well does the system adapt to changes in manufacturing plant layout; 6. Comfort, understood as thermal comfort for workers in the building: we distinguish between the heating season and the possible air conditioning season by introducing sub-criteria winter and summer, respectively; 7. Economic uncertainty, i.e. likelihood that real economic parameters will signicantly differ from estimated ones e.g. due to unexpected increases in equipment installation or fuel costs; 8. Operation cost, expressed as the estimated average annual fuel expenses, derived from yearly energy requirements estimated with equation (1) and from current fuel prices. Rather than hypothesizing a trend in fuel prices, uncertainty is accounted for with criterion 7. Maintenance costs are difcult to quantify, so that aspect is dealt with qualitatively through criteria 1e2, which basically assume a reactive maintenance strategy reecting both the approach of the company as to heating facilities and the fact that preventive maintenance expenses are similar (and low) across the set of heating systems considered. 9. Capital costs, derived from the best quotations obtained from prospective suppliers. According to the AHP methodology, the relative importance of these criteria was assessed by pairwise comparisons elicited from company management. The resulting set of local priorities is presented on the rst line of Table 3, where criteria are arranged in descending weight order. Basically, criteria 8 and 9 are expressed by numeric values, which are reported in Table 2: they can be directly entered in the comparison process provided that they meet the homogeneity assumption of the AHP method, i.e. that the ratio between the most divergent is smaller than nine [21]. On the other hand, instruments for quantitative assessment of criteria 1e7 are not available in some cases (e.g. for source

Fig. 2. Hierarchic model for selection of heating system.

560 Table 2 Capital and operation costs for the heating systems considered. Heating system Gas burners and radiant tubes Gas burners and radiant modules Gas boilers and hot water air heaters Gas boilers and hot water radiant panels Condensing gas boilers and radiant oor heating Condensing gas boilers and radiant wall heating GSHP and radiant oor heating GSHP and radiant wall heating

D. Chinese et al. / Energy 36 (2011) 556e565

Capital costs [kV] 23.8 28.5 56.0 62.8 63.0 83.6 99.0 124

Fuel costs [kV/year] 7.0 7.0 11.0 9.2 10.2 9.4 8.2 7.6

exibility). In other cases, engineering tools have been developed in literature to model the qualitative aspects considered in our decision problem. For instance, a subjective concept like comfort has been developed into various mathematical models over the years (see e.g. [35]) leading to well dened design conditions and comfort criteria standardized by norms [36]. Also, the quantitative denition and assessment of reliability and maintainability has been developed into a structured engineering discipline over the 20th century [37] whose concepts have been applied to some extent to HVAC systems in recent years [38,39]. However, both instances show similar problems: scarce data are available for the problem at hand, excessive time and resources would be needed for a quantitative evaluation considering the decision time horizon and so many assumptions (about system design and conguration, operation patterns, occupational prole etc.) would have to be made that even rened, long term simulations could provide little more than educated guesses. We deemed it reasonable, then, to make educated guesses straightforwardly, using the pairwise comparison methodology foreseen by AHP and considering the personal communications of designers, installers and company engineers, as well as manufacturer claims and our knowledge of the systems. It should be observed that qualitative pairwise comparison of possibilities has been already used in reliability engineering when data are unavailable e however in human rather than in equipment reliability assessment (see e.g. [40]) e and that recently AHP has too been incorporated into that kind of study, demonstrating the capability of this methodology to effectively handle qualitative decision variables [41]. The normalized priorities and rankings obtained from our individual comparison matrices are reported in Table 3. Consistency ratios were lower than 10% for each criterion and equal to 0 for criteria 8 and 9, where numerical data were directly entered. 5. Results and discussion The proposed AHP model was implemented with the decisionmaking software Super Decisions [42]. By structuring the problem and obtaining nal priorities with AHP for our case study we can highlight some distinctive features of industrial heating systems and choices. 5.1. Heating technology performance and industrial particularities Industrial buildings are characterized by particular feasible heating technologies (including but not limited to the ones we considered), which only partially overlap with available residential heating technologies. In particular, high temperature radiant systems are typically used for industrial buildings because these usually have high ceilings, large volumes and surfaces which are often only partially used, specic comfort conditions depending on workers activities (e.g. acceptable indoor temperatures are lower than in residential properties) and operating times and

management practices different to those of residential applications. By quickly enabling partial, localized heating, high temperature radiant heating allows fuel savings and reductions in operational costs; at the same time, these solutions have minimum capital and installation costs, so in our case they clearly outperform other technologies from an economic viewpoint, as reported in Table 2. However, these technologies have low reliability and comfort scores. Table 3 shows that better reliability performance is associated with lower temperatures and simpler systems with fewer components: highest scores are thus attributed to low temperature oor and wall heating, especially if associated with ground source heat pumps, while the reliability performance of hot water air heaters is poor because of their many mechanical components. The lower the heat carrier temperature, the better the winter comfort too, with oor heating obtaining better results than wall heating in winter because of more uniform performance in large rooms, while wall radiant systems are more effective than oor systems for cooling in summer because of better temperature distribution patterns. As for comfort in summer, top scores are obtained by GSHP with radiant systems because it is the only one which inherently provides the building with some air conditioning; second best results are obtained by systems which could provide air conditioning if a cooling unit were added to the system. With hot radiant heating, a separate distribution system for air cooling should also be foreseen to provide air conditioning. Overall comfort performance results from the combination of summer and winter scores, with winter performance being deemed ve times more important than summer performance by decision makers. It should be observed that in existing company buildings e like in most industrial buildings in the area e air conditioning is limited to ofces, while different adaptive measures are undertaken in manufacturing areas (e.g. increased natural ventilation or different working schedules during hot periods). Finally, as to other scores, experts, installers and decision makers tend to associate traditional heating systems with shorter lead times and time to repair and with lower economic uncertainty, especially regarding installation costs, even though low temperature systems are also more exible to fuel/heat source changes and seem therefore more robust against sudden variations in traditional fuel prices or availability. As for changes in layout, the best results are obtained by those systems which are either easily restructured to match new requirements (e.g. hot water air heaters) or are so widely spread over the building and guarantee such uniform thermal performance that they can basically accommodate any layout type without changes (e.g. oor heating). Derived from local alternative weights, overall performance and alternatives ranking depend on the relative weights of criteria elicited from decision makers through pairwise comparison. Little is reported in literature about attributes affecting industrial energy choices, mostly about power supply [43] and cogeneration [44,45], and we could not nd specic information about the relative weight of these attributes or the criteria considered for heating systems. Thus our empirical ndings e despite being based on a single case study e shed some light on specic features of industrial decision making in this eld. 5.2. Economic criteria weights and industrial particularities Economic performance as a whole (including capital costs, operation costs and uncertainty) accounts in our case for almost 50% of overall scores. It should be stressed that for our rm capital costs weigh twice as much as operation costs, which according to AHP scales means that the former are moderately more important than the latter. That capital costs are more important than yearly energy costs for industrial decision makers could be a general result in our opinion.

D. Chinese et al. / Energy 36 (2011) 556e565


a Figures in bold represent criteria weights. Figures in round brackets represent the local ranking of alternatives according to each criterion. Figures in square brackets represent the contribution of winter and summer comfort to the determination of total comfort performance.

561

In fact, a discrete choice modelling study carried out in Canada [45], which, from a survey of approximately 50 companies, calculates an implicit discount rate of 34% for rms stating preferences on new steam generation systems, may corroborate our ndings. That the discount rate is so high e if compared, for instance, with the 9% discount rate for individual decisions on home heating systems [45] e implies that rms attribute much more importance to immediate sums, which investments in energy technologies usually are, than to future ones, such as yearly energy costs (or savings) during the expected service life of heating equipment, which may last several years (e.g. for the technologies examined in our case we assume a service life of 15 years). For Newnan et al. [46] it is generally true for all kinds of investment that minimum attractive rates of return are signicantly higher for rms than for households and this depends on higher competition for protable investments at the small scale affordable for families. Moreover, the same authors categorize rms into two groups: unstable rms, who accept investments having a payback period of one year or less before taxes or a rate of return of 25e30% after taxes, and stable rms, who use payback analysis for small projects only and accept an after tax rate of return of 12e15%. We can observe that, for the reference interest rate of 34% (before taxes) mentioned above [45], the present worth factor for a uniform payment series for an analysis period of 15 years is about 2.9, which means that, if we assume that fuel expenses are constant over the time period and calculate LCC accordingly, we obtain gures of merit where yearly operational costs are weighed at about three times as much as capital costs. When directly asked about the reference interest rate to be used for calculations, the management of our case study company indicated a rate of 18% (before taxes) which leads to a present worth factor around 5. On the other hand, the relative weights for capital and operational costs indirectly derived with AHP in our case yield a present worth factor of around 0.5. With our previous assumptions (analysis period of 15 years for LCC calculations), this present worth value would only be obtained by an interest rate of 200%. However, it should be noted that LCC calculation and the weighting method of AHP work quite differently and lead to very different gures of merit. The expression for LCC with constant operational costs for each alternative and the expression for global priorities with AHP (calculated according to [47]) when only capital costs and xed yearly operational costs are considered as criteria are shown, respectively, in equations (2) and (3)

Overall normalized priorities and rankings of alternative Economic uncertainty 0.026 0.048 Source exibility Lead time 0.049 Layout exibility 0.071 Time to repair 0.087 Summer [0.167] Winter [0.833]
a a

Comfort

0.110

Reliability

0.132

Operation costs

0.027 0.042 0.020 0.065 0.169 0.120 0.341 0.215

(7) (6) (8) (5) (3) (4) (1) (2)

0.021 0.039 0.018 0.072 0.275 0.131 0.293 0.152

(7) (6) (8) (5) (3) (4) (1) (2)

0.020 0.020 0.053 0.031 0.081 0.131 0.256 0.407

(7) (7) (5) (6) (4) (3) (2) (1)

0.112 0.156 0.347 0.262 0.028 0.048 0.019 0.027

(4) (3) (1) (2) (6) (5) (8) (7)

0.110 0.043 0.310 0.084 0.231 0.031 0.168 0.024

(4) (6) (1) (5) (2) (7) (3) (8)

0.018 0.021 0.062 0.041 0.171 0.171 0.258 0.258

(8) (7) (5) (6) (3) (3) (1) (1)

0.186 0.151 0.295 0.134 0.029 0.060 0.018 0.030

(3) (3) (1) (4) (7) (5) (8) (6)

0.151 0.151 0.304 0.177 0.079 0.079 0.030 0.030

(3) (3) (1) (2) (5) (5) (7) (7)

0.146 0.138 0.134 0.108 0.128 0.092 0.145 0.107

(1) (3) (4) (7) (5) (8) (2) (6)

Table 3 Local weights for criteria and alternatives in selection of heating systems.

0.159

0.151 0.151 0.096 0.116 0.104 0.112 0.130 0.140

(1) (2) (7) (5) (8) (6) (4) (3)

LCCi CCi

1 r n 1 $OCi r 1 r n
1 CCi

(2)
1 OCi

Capital costs

GPi wCC $LPCCi wOC $LPOCi wCC $P

1 jA CCj

wOC $P

1 jA OCj

(1) (2) (3) (4) (5) (6) (7) (8)

(3)
Where: CCi is the capital costs for the i-th alternative; OCi the operation cost for the i-th alternative; r is the interest rate; GPi is the global priority for alternative i; LPOCi is the local priority for alternative i with respect to capital costs; - LPOCi is the local priority for alternative i with respect to operation costs; - A is the set of all alternatives; - wCC and wOC are the relative, normalized weights of capital and operation costs taken as the only criteria for evaluation (in our case wCC 0.6667 and wOC 0.3333, so that the 2:1 ratio between capital and operation cost weights is respected). -

0.318

Local weightsa and rankingsa for each criterion and alternative

Gas & Radiant tubes Gas & Radiant modules Gas boilers & HW air heaters Gas boilers & HW radiant panels Cond.gas boilers & radiant oor heating Cond. gas boilers & radiant wall heating GSHP & radiant oor heating GSHP & radiant wall heating

0.268 0.223 0.114 0.102 0.101 0.076 0.061 0.051

562

D. Chinese et al. / Energy 36 (2011) 556e565

If, to enable comparison, we normalize LCCs using the ideal synthesis of the AHP approach, i.e. we take LCC as single criterion, apply the calculation procedure expressed by equation (3) for determining local LCC based priorities LPi and divide by the maximum calculated LPmax, we obtain the ideal priorities plotted in Fig. 3, where decisions based on LCCs, on ideal global priorities (GPi/GPmax) based on capital and operational costs from Eq. (3) and on ideal global priorities from the complete multi-criteria model are compared. It can be observed that LCC based rankings and cost based multi-criteria rankings are quite similar: in particular, cost based multi-criteria priorities calculated with our decision makers weights are for most alternatives very close to LCC based priorities calculated with the companys reference interest rate of 18%, and generally between 18% LCC based priorities and 34% LCC based priorities. As expected, with a 34% interest rate capital intensive solutions such as low temperature systems are more heavily penalized. Thus, the qualitative evaluation elicited with the pairwise comparison approach, which expresses subjective feelings and judgments by company management about desired economic performance by stating that capital costs are twice as important as operational costs, is in good accordance with the results obtained with objective measurements of expected economic performance expressed by the company through the adopted interest rate. On the other hand, Fig. 3 also shows that the introduction of noneconomic criteria signicantly changes priorities and overall ranking, boosting scores and position of low temperature solutions.

5.3. Non-economic criteria weights and industrial particularities As for qualitative criteria adopted in industrial energy choices, there is even less information in literature than about economic criteria. However, there is some empirical evidence on criteria adopted by residential users which can be used for comparison. In particular, Sadler [48] and Mahapatra and Gustavsson [49] present results from surveys carried out in Canada and Sweden, respectively, about the attributes taken into account by residential customers when expressing preferences regarding the adoption of new heating systems.

According to Sadlers empirical survey about the selection of heating systems involving almost 700 households in Canada, for residential decision makers the most important attribute is system reliability, followed by operating costs and energy prices. Comfort is the fourth most important attribute, followed by purchase prices. As Sadler [48] observes, this preference pattern is consistent with the relatively low implicit discount rate e about 8.4e8.8%, in line with the 9% value presented by Rivers and Jaccard [45] e which she also calculates from her empirical survey. A similar pattern is also found by Mahapatra and Gustavsson [49], who surveyed two different samples of about 700 Swedish homeowners each on their preferences regarding heating system selection. With slight differences between the baseline and the resurvey samples, those authors found that the annual cost of heating was the rst most important attribute considered by homeowners, followed by functional reliability. Indoor air quality e which can be compared with Sadler and our comfort attribute e comes third in the re-survey sample and fourth in the baseline survey, while investment cost is the fourth most important attribute in the resurvey and the third most important in the baseline, in both cases, however, with very slight score differences. The relative ranking of qualitative attributes reliability and comfort found in residential surveys is thus generally consistent with the results from our industrial case study, with reliability being more important than comfort e which makes sense, as an unreliable system cannot guarantee any comfort. However, economic aspects are more important for industrial decision makers than for homeowners and the relative importance of operating and capital costs is reversed. It is hardly possible to compare other attributes, as there are differences not only between residential and industrial users but also between the two surveys examined on the residential sector. Other attributes included by Sadler [48] are, in descending order of importance, concern for the environment, energy efciency labelling and system responsiveness, i.e. time required for heated rooms to reach the desired temperature. Instead, the residential survey by Mahapatra and Gustavsson [49] considers system automation, security in fuel supply, environmental benignity, low GHG emissions, increased market value

Fig. 3. Ideal priorities based on LCCs, on capital and operational costs weighted according to AHP and on the complete multi-criteria model for the case study.

D. Chinese et al. / Energy 36 (2011) 556e565

563

of building and time for collection of information as further attributes. We can observe that for residential customers environmental aspects also come after the main common attributes (reliability, comfort, costs and investment) and also after others, e.g. security in fuel supply and system automation. Security in fuel supply is similar to the attribute labelled as source exibility in our industrial case, while other attributes we elicited, such as layout exibility, time to repair and lead time, seem typical for industry as they reect the attitude to time management and the continuous change and improvement required for companies to be competitive.

5.4. How criteria weights affect alternative rankings While we observed that a direct comparison of industrial and residential priorities is difcult and an immediate derivation of attribute scores according to AHP scales for residential systems is not possible using the available literature, we tried nevertheless to understand how the different importance of attributes affects heating system choice. To do this, we developed two series of priorities for the industrial set of attributes elicited in this study, in such a way that the priority of industry specic attributes remained mostly unchanged and just the order of importance for common attributes was modied to qualitatively mimic the classications reported for the residential sector by Sadler [48] and Mahapatra and Gustavsson [49], respectively. In Fig. 4 we present the sets of criteria priorities thus obtained in tabular form, where Residential 1 refers to the attribute classication based on Ref. [48] and Residential 2 to the one based on Ref. [49]. The radar plot in the gure illustrates the ideal priorities obtained for various heating systems according to industrial and residential-like criteria; the score plot of the alternatives based on LCC at 18% interest rate is also drawn for the sake of comparison. We notice that, while the two residentiallike outcomes are quite similar, they substantially differ from the industrial classication of alternatives obtained in this case study. If we were to select the industrial heating system reasoning as residential customers would, GSHPs would be by far the rst choice and low temperature heating systems, even coupled with natural gas boilers, would be in theory second best alternatives mainly due
Fig. 5. Sensitivity analysis for heating system priorities with respect to capital costs of GSHPs.

to their high comfort and reliability performance. On the other hand, radiant tubes are the rst choice for industry especially because capital costs are so important; however, GSHP with radiant oor heating is a close second and just a small reduction (5e10%) in capital costs e easily achievable through common supplier discounts, rather than with special subsidies e would theoretically be enough to make it rst choice for the industrial decision makers too. This is shown in Fig. 5, which illustrates how normalized priorities of various heating systems change if a reduction coefcient (e.g. a subsidy) is applied to capital costs of GSHP solutions alone. That GSHP systems apparently have concrete application opportunities for the industrial case study concerned mainly depends on the importance of the main qualitative criteria (reliability, comfort), as we can deduce from the comparison between the industrial priority plot and the LCC based priority plot in Fig. 4. However, these considerations risk being predominantly theoretical. That the best solution for the system concerned is so sensitive to small changes in capital costs suggests that GSHP may not be stable as an optimal solution for our industrial case study. Actually this is conrmed by the sensitivity analysis presented in Fig. 6, illustrating the variation of normalized global priorities of heating alternatives for the system concerned to increases in the

GSHP & rad iant f loor heat ing


1

G as & Rad iant t ubes

0. 8 0. 6 0. 4 0. 2

G SHP & rad iant wall heat ing

LCC 18% Mult ricrit . I nd ust rial Mult icrit . Resid .1 Mult icrit . Resid . 2

Gas & Rad iant mod ules

Cond .gas boilers and rad iant f loor heat ing

Gas boilers & HW rad iant panels Cond .gas boilers and rad iant wall heat ing
Criteria weights Industrial (C.R. 0.089). Residential 1 (C.R. 0.083) Residential 2 (C.R. 0.008) Capital costs 0.318 0.099 0.110 Operation costs 0.159 0.181 0.266 Reliability 0.132 0.324 0.244 Comfort 0.110 0.144 0.128

Gas boilers & HW air heat ers

Time to repair 0.087 0.085 0.085

Layout flexibility 0.071 0.058 0.058

Source flexibility 0.049 0.045 0.045

Lead time 0.048 0.045 0.045

Economic uncertainty 0.026 0.019 0.020

Fig. 4. Comparison of ideal priorities for various heating systems according to industrial and residential criteria and life cycle costs.

564

D. Chinese et al. / Energy 36 (2011) 556e565

Fig. 6. Sensitivity analysis for the priority of heating systems with respect to the priority of criterion capital cost.

normalized priority of capital cost, where the priority of all other criteria is proportionally reduced. It can be observed that radiant tubes are the rst choice for all values of normalized priorities of capital costs exceeding 0.315; in our case study, the calculated value was 0.318 so we are just above the intersection in a quite unstable position. In the end, the company examined decided on radiant tubes. If we try to generalize, the real priority of capital cost for industrial customers is likely to be even greater than the value we calculated for our reference company, since the satisfactory rate of return used by the rm we investigated is quite lower than that reported in the literature on industrial case studies [45]. Thus, general market opportunities for more capital intensive solutions for industrial heating systems, such as GSHP with low temperature radiant heating in our case, are likely to be less promising than those we calculated here not due to limitations in the technologies but because of the importance that decision makers attach to different attributes, and especially because industrial heating systems are selected on a different basis to residential ones. 6. Conclusions While just a few applications of multi-attribute decision making methods to single buildings or small scale energy systems are reported in literature, in our experience this modelling approach was appreciated by the technical management of the company examined because of its ability to pinpoint the most important aspects of the decision problem, to make judgments on qualitative aspects more transparent and rational and to structure consultancy on a participatory and fair basis. A strong point of the adopted approach, i.e. the AHP, is the elicitation and pairwise comparison approach, which is characteristic of the method and a fundamental difference to other multi-attribute decision making methods adopted for this kind of problems, which usually require decision makers to assign weights e e.g. numbers between 0 and 1 e from scratch to various criteria (e.g. [16]). In our case, the pairwise comparison approach e although at times cumbersome because of the number of considered criteria e allowed us to obtain realistic classications of system performance as to qualitative criteria and to assign criteria weights that reect well the companys priorities and expectations, as we have seen especially for cost based criteria. Based on this case study and particularly because of the dominance relationships within the examined set of solutions, it is hard to understand whether AHP is capable of predicting real energy system choices by industrial companies better than approaches based on LCCs or even on investment costs alone. However, a rst important lesson we draw from our case study is that qualitative variables do matter in the selection of industrial

heating systems: this is conrmed by the way multi-criteria rankings change if we incorporate a qualitative variable. A second observation which is conrmed by our case study is that the basis of industrial decisions on heating systems is different from the residential sector: capital costs are more important than operational costs and considered qualitative attributes only partially overlap with residential ones. In our view, these observations have implications especially for future research on tactical energy policy making and modelling, i.e. for designing means (e.g. subsidies, advertisement etc.) to put strategic energy plans into practice. If, as we found, the industrial sector attaches more value to capital costs than to operational costs e and further survey based empirical investigations would be useful to conrm this e then different subsidy structures should be conceived to foster sustainable energy technologies in the residential and industrial sector respectively (e.g. by privileging equipment grants for industry and fuel or power cost reductions for homes). Secondly, as qualitative aspects do matter and the AHP proved especially efcient in dealing with these facets and, according to literature, especially in forecasting consumers behaviour based on qualitative attributes of goods when decision problems are more complex [50,51], further research could be carried out to use the AHP for incorporating so far omitted qualitative aspects into tactical energy policy models, e.g. in bottom up or hybrid energy policy modelling [45]. Up until now, the AHP has been widely used in literature for strategic policy modelling, i.e. to identify the best energy plans for countries based on a plurality of sometimes conicting macro-objectives, e.g. social costs, employment, environmental impact etc. (e.g. [19]): its application to tactical modelling could be a promising eld for future research. Future research could also be directed to the application of other multi-criteria methods to similar industrial and residential problems from the adopters perspective, with the purpose of highlighting similarities and differences depending on methodology selection, rather than on internal boundary conditions such as technologies considered, eld of application and the views and values of decision makers. Acknowledgements Kind hints from Rozann W. Saaty on assumptions and usage of Super Decisions are gratefully acknowledged. References
[1] Schnitzer H, Brunner Ch, Gwehenberger G. Minimizing greenhouse gas emissions through the application of solar thermal energy in industrial processes. Journal of Cleaner Production 2007;15(13e14):1271e86. [2] Gustafsson SI. Optimisation of insulation measures on existing buildings. Energy and Buildings 2000;33(1):49e55. [3] Hasan A, Vuolle M, Sirn K. Minimisation of life cycle cost of a detached house using combined simulation and optimization. Building and Environment 2008;43(12):2022e34. [4] Buchholz T, Rametsteiner E, Volk TA, Luzadis VA. Multi criteria analysis for bioenergy systems assessments. Energy Policy 2009;37(2):484e95. [5] Mulligan J, Barber P. The client consultant relationship. In: Sadler Ph, editor. Management consultancy: a handbook for best practice. London: Kogan Page; 2001. p. 83e102. [6] Bejan A, Tsatsaronis G, Moran M. Thermal design and optimization. New York: Wiley; 1996. [7] Bojic M, Lukic N, Trnobransky K. Linear programming applied to an industrial building with several available hot refuse ows. Energy 1995;20 (10):1067e74. [8] Heidari Tari M, Sderstrm M. Modelling of thermal energy storage in industrial energy systems the method development of MIND. Applied Thermal Engineering 2002;22(11):1195e205. [9] Chinese D, Meneghetti A. Optimisation models for decision support in the development of biomass-based industrial district-heating networks in Italy. Applied Energy 2005;82(3):228e54.

D. Chinese et al. / Energy 36 (2011) 556e565 [10] Malckzewski J. GIS and Multicriteria Decision Analysis. New York: Wiley; 1999. [11] Lken E. Use of multicriteria decision analysis methods for energy planning problems. Renewable and Sustainable Energy Reviews 2007;11(7):1584e95. [12] Hobbs BF, Meier P. Energy decisions and the environment: a guide to the use of multicriteria methods. Norwell: Kluwer; 2000. [13] Diakaki Ch, Grigoroudis E, Kolokotsa D. Towards a multi-objective optimization approach for improving energy efciency in buildings. Energy and Buildings 2008;40(9):1747e54. [14] Wright JA, Loosemore HA, Farmani R. Optimization of building thermal design and control by multi-criterion genetic algorithm. Energy and Buildings 2002;34(9):959e72. [15] Roy B. Multicriteria methodology for decision aiding. Norwell: Kluwer; 1996. [16] Thiel T, Mroz T. Application of Multi-Criterion Decision Aid method in designing heating systems for museum buildings. Informatica 2001;12 (1):133e46. [17] Kaklauskas A, Zavadskas EK, Raslanas S. Multivariant design and multiple criteria analysis of building refurbishments. Energy and Buildings 2005;37 (4):361e72. [18] Kaklauskas A, Zavadskas EK, Raslanas S, Ginevicius R, Komka A, Malinauskas P. Selection of low-e windows in retrot of public buildings by applying multiple criteria method COPRAS: a Lithuanian case. Energy and Buildings 2006;38(5):454e62. [19] Jaber JO, Jaber QM, Sawalha SA, Mohsen MS. Evaluation of conventional and renewable energy sources for space heating in the household sector. Renewable and Sustainable Energy Reviews 2008;12(1):278e89. [20] Alanne K, Salo A, Saari A, Gustafsson SI. Multicriteria evaluation of residential energy supply systems. Energy and Buildings 2007;39(12):1218e26. [21] Saaty TL. Fundamentals of decision making and priority theory with the Analytic Hierarchy Process. Pittsburgh: RWS Publications; 2001. [22] Zadeh LA. Fuzzy sets. Information and Control 1965;8:338e53. [23] Salo AA, Hmlinen RP. Preference assessment by imprecise ratio statements. Operations Research 1992;40(6):1053e61. [24] Nagesha N, Balachandra P. Barriers to energy efciency in small industry clusters: multi-criteria-based prioritization using the analytic hierarchy process. Energy 2006;31(12):1969e83. [25] Hbner R, Gnther H. Using AHP for strategic production site assessment: a case study from specialty chemicals industry. In: Gnther H, Mattfeld DC, Suhl L, editors. Management Logistischer Netzwerke. Heidelberg: PhysicaVerlag; 2007. p. 3e22. [26] Pohekar SD, Ramachandran M. Application of multi-criteria decision making to sustainable energy planning e a review. Renewable and Sustainable Energy Reviews 2004;8(4):365e81. [27] Kaya T, Kahraman C. Multicriteria renewable energy planning using an integrated fuzzy VIKOR & AHP methodology: the case of Istanbul. Energy 2010;35 (6):2517e27. 7th International Conference on Sustainable Energy Technologies, 7th International Conference on Sustainable Energy Technologies, June 2010. [28] Jiang-Jiang W, You-Yin J, Chun-Fa Z, Xu-Tao Z, Guo-Hua S. Integrated evaluation of distributed triple-generation systems using improved grey incidence approach. Energy 2008;33(9):1427e37. [29] Ramanathan R, Ganesh LS. Energy alternatives for lighting in households: an evaluation using an integrated goal programming-AHP model. Energy 1995;20(1):63e72. [30] Fraccaro Srl. Product information, radiant modules, http://www.fraccaro.it/ Depliant/GIRAD/girad_en.pdf [accessed 19.02.10]. [31] University of Illinois and the Ernest Orlando Lawrence Berkeley National Laboratory, Energy Simulation Software, http://apps1.eere.energy. gov/ buildings/energyplus/pdfs/engineeringreference.pdf [accessed 19.02.10].

565

[32] CarliEuklima. Product technical documentation for radiant heating systems, http://www.carlieuklima.com/Calor/Eurad/Documentation/documentation. php [accessed 19.02.10]. [33] Doninelli M. Radiant oor heating systems (in Italian). available at: http:// www.calef.us/it_IT/calef/Details/Handbooks/pdf/quaderno_4_it.pdf [accessed 19.02.10]. [34] Saaty TL, Ozdemir MS. Why the magic number seven plus or minus two. Mathematical and Computer Modelling 2003;38(3e4):233e44. [35] Hens HSLC. Thermal comfort in ofce buildings: two case studies commented. Building and Environment 2009;44(7):1399e408. [36] Pfafferott J, Herkel S, Kalz DE, Zeuschner A. Comparison of low-energy ofce buildings in summer using different thermal comfort criteria. Energy and Buildings 2007;39(7):750e7. [37] Saleh JH, Marais K. Highlights from the early (and pre-) history of reliability engineering. Reliability Engineering & System Safety 2006;91(2):249e56. [38] Kwak R, Takakusagi A, Sohn J, Fujii S, Park B. Development of an optimal preventive maintenance model based on the reliability assessment for airconditioning facilities in ofce buildings. Building and Environment 2004;39 (10):1141e56. [39] Myrefelt S. The reliability and availability of heating, ventilation and air conditioning systems. Energy and Buildings 2004;36(10):1035e48. [40] Kirwan B. A guide to practical human reliability assessment. London: Taylor & Francis; 1994. [41] Park KS, Lee J. A new method for estimating human error probabilities: AHPSLIM. Reliability Engineering & System Safety 2008;93(4):578e87. [42] Creative Decisions Foundation. Super decisions software. available at: http:// www.superdecisions.com/indextables.php3 [accessed 19.02.10]. [43] Sderberg M. A choice modelling analysis on the similarity between distribution utilities and industrial customers price and quality preferences. Energy Economics 2008;30(3):1246e62. [44] Bonilla D, Akisawa A, Kashiwagi T. Modelling the adoption of industrial cogeneration in Japan using manufacturing plant survey data. Energy Policy 2003;31(9):895e910. [45] Rivers N, Jaccard M. Useful models for simulating policies to induce technological change. Energy Policy 2006;34(15):2038e47. [46] Newnan DG, Lavelle JP, Eschenbach TG. Engineering economic analysis. New York: Oxford University; 2002. [47] Saaty RW. Decision making in complex environments, the Analytic Hierarchy Process (AHP) for decision making and the Analytic Network Process (ANP) for decision making with dependence and feedback, super decisions tutorial. available at: www.superdecisions.com/Tutorial2003.doc; 2003 [accessed 19.02.10]. [48] Sadler M. Home energy preferences & policy: applying stated choice modelling to a hybrid energy economy model. Research project no. 342, School of Resource and Environmental Management, Simon Fraser University, Burnaby. available at: http://ir.lib.sfu.ca/bitstream/1892/9850/1/b31855751.pdf; 2003 [accessed 19.02.10]. [49] Mahapatra K, Gustavsson L. An adopter-centric approach to analyze the diffusion patterns of innovative residential heating systems in Sweden. Energy Policy 2008;36(2):577e90. [50] Meiner M, Decker R. An empirical comparison of CBC and AHP for measuring consumer preferences, on line proceedings of the ISAHP 2009 symposium, July 29eAugust 1, 2009, Pittsburgh, Pennsylvania, USA. available at: http:// www.isahp.org/2009Proceedings/FinalPapers/49 MeissnerDecker Compa risonCBCandAHP ConsumerPreferences REV FIN.pdf [accessed 19.02.10]. [51] Helm R, Steiner M, Scholl A, Manthey L. A comparative empirical study on common methods for measuring preferences, Jenaer Schriften zur Wirtschaftwissenschaft 20/2004, Friedrich-Schiller-Universitt Jena, http://www. wiwi.uni-jena.de/Papers/wp-sw2004.pdf [accessed 2/19/2010].

Вам также может понравиться