Choosing a sustainable demolition waste management strategy using multicriteria decision analysis

Waste Management 29 (2009) 12–20

Contents lists available at ScienceDirect

Waste Management journal homepage: www.elsevier.com/locate/wasman

Choosing a sustainable demolition waste management strategy using multicriteria decision analysis Nicolas Roussat *, Christiane Dujet, Jacques Méhu LGCIE, INSA-Lyon, 20 Avenue Albert Einstein, 69621 Villeurbanne Cedex, France

a r t i c l e

i n f o

Article history: Accepted 21 April 2008 Available online 24 June 2008

a b s t r a c t This paper presents an application of the ELECTRE III decision-aid method in the context of choosing a sustainable demolition waste management strategy for a case study in the city of Lyon, France. This choice of waste management strategy takes into consideration the sustainable development objectives, i.e. economic aspects, environmental consequences, and social issues. Nine alternatives for demolition waste management were compared with the aid of eight criteria, taking into account energy consumption, depletion of abiotic resources, global warming, dispersion of dangerous substances in the environment, economic activity, employment, and quality of life of the local population. The case study concerned the demolition of 25 buildings of an old military camp. Each alternative was illustrated with different waste treatments, such as material recovery, recycling, landfilling, and energy recovery. The recommended solution for sustainable demolition waste management for the case study is a selective deconstruction of each building with local material recovery in road engineering of inert wastes, local energy recovery of wood wastes, and specific treatments for hazardous wastes. Ó 2008 Elsevier Ltd. All rights reserved.

1. Introduction The aim of this study is to compare different strategies of demolition waste management with the aid of a multicriteria analysis method. The comparison of these strategies is based on a study of the management of demolition wastes that were produced in a demolition site located in Lyon. The client for the demolition was the urban community of Lyon (The Grand Lyon, France). Different strategies were evaluated in relation to sustainable development objectives, i.e., by taking into consideration the waste management aspects in relation to environmental, economic, and social issues. Multicriteria methods are often used for problems of sustainable development (Klang et al., 2003; Munda, 2005; Maystre et al., 1994; Yelda and Shrestha, 2003; Pohekar and Ramachandran, 2004) and waste management (Brent et al., 2007; Dais et al., 1999; Hokkanen and Salminen, 1997a, b; Maystre et al., 1994). The decision criteria concern the local and global environmental aspects, the economic aspects of the waste management strategy, the impacts of the strategy on the quality of life of the local population, and the consequences on employment in the local area. Several criteria are used to assess the global environmental consequences of waste management strategies. These criteria take into account abiotic depletion, energy consumption, and the green-

* Corresponding author. Tel.: +33 4 72 43 63 61; fax: +33 4 72 43 87 17. E-mail address: [email protected] (N. Roussat). 0956-053X/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.wasman.2008.04.010

house effect. The evaluation of these criteria was done with a mass and energy balance using the concept of embodied energy and raw material content. Sanitary and environmental impacts due to the dispersion of dangerous substances (initially contained in the wastes) into the environment were quantified with the results of lysimeter leaching tests realised with reconstituted samples of demolition waste from a previous study. Economic issues concern the cost of the buildings’ demolition and the cost of waste disposal. The costs are paid by the urban community, and so are indirectly paid by the population via local taxes. The second economic aspect concerns the influence of the use of secondary raw materials on economic activity in the urban territory of Lyon. For the social issues, two aspects were considered. The first concerns employment in relation to the waste management strategy. This criterion takes into consideration the number of new jobs arising due to the strategy and the risks associated with these jobs. The second aspect concerns the quality of life of the local population. The impacts on quality of life for each strategy were valued through the nuisances due to the demolition site and truck traffic, and also the destruction of natural areas for depositing the waste. Finally, the multicriteria analysis, taking into consideration all the aspects described above, must rank the different strategies in relation to the sustainable development objectives. The ELECTRE III method was chosen because this method has been specifically adapted for environmental problems by various authors (Morissey and Browne, 2004; Salminen et al., 1998) and has been used in

N. Roussat et al. / Waste Management 29 (2009) 12–20

many cases for waste management decisions (Dais et al., 1999; Hokkanen and Salminen, 1997a,b; Maystre et al., 1994). 2. Case study and alternatives for demolition waste management The case study is the demolition of an old military camp with an area of 34 ha. In this site, 25 buildings were demolished. Wastes produced by this demolition are summarized in Table 1. Wastes produced by demolition include metals (steel, aluminium, copper, and zinc), inert wastes (tiles, concrete, bricks, ceramic materials, stones, clinker), wood, plaster, plastic (PVC), flat glass, glass wool, and hazardous waste (bituminous roof and fluorescent lamps containing mercury). The alternatives represent the different possible strategies for demolition waste management. Each alternative has different environmental, economic, and social consequences. Nine alternatives (called A1–A9) are presented below. All alternatives are realistic and technically possible. Alternative 1: In this alternative, buildings are demolished without sorting of the different materials. The wastes of this demolition are dumped illegally. Alternative 2: Buildings are also demolished without sorting of different constituents, but after demolition, the wastes go to a sorting platform for construction and demolition waste. After sorting of the different fractions of wastes, metals are recycled while all other categories of waste go to a landfill. When buildings are demolished without preliminary deconstruction, the sorting of different wastes is difficult; hazardous wastes are not sorted and are mixed with inert wastes. Alternative 3: Buildings are demolished without sorting and wastes go to the sorting platform. After sorting, metals are recycled and inert wastes are recovered in road engineering. Non-hazardous solid wastes (woods, plastics) are burned in an incinerator. Energy produced by incineration is recovered in electricity and in thermal power. In this case, like alternative 2, some hazardous wastes are mixed with inert wastes. Alternative 4: The first step of this alternative is the selective deconstruction of each building, i.e., all non-hazardous and hazardous components are removed before demolition of the building structures. In this case, all hazardous wastes are sorted. Each waste type, except metals and flat glass, which are recycled, go to sanitary landfills. Alternative 5: As in alternative 4, the first step is the selective deconstruction of buildings before demolition of their structures. Inert wastes are recovered in road engineering. Metals, flat glass,

Table 1 Quantity of wastes produced by the demolition Wastes

Tonnes

Steel Aluminium Tiles Concrete Wood Bricks Ceramic materials Copper Stones Glass wool Flat glass Gypsum plaster PVC Clinker Zinc Bituminous roof Fluorescent lamps

118 2.2 229 18,575 378 595 255 0.3 318 233 21.9 114 18 1260 2.3 210 0.3

13

and glass wool are recycled. Wood wastes are used as fuel for district heating, and other non-dangerous wastes go to a landfill. All hazardous wastes are sorted and go to a sanitary landfill for hazardous wastes. Alternative 6: The difference between this alternative and alternative 5 is that inert wastes are used to produce new concrete blocks. The making of concrete blocks with recycled aggregates requires the addition of natural aggregates and increasing of the cement content in relation to standard concrete block composition, in order to have the same technical characteristics of usual concrete blocks. Alternative 7: The first step of this alternative is the selective deconstruction of each building before demolition of their structures. Inert wastes are recovered for use in aggregates for road engineering. Metals, flat glass, glass wool, and PVC are recycled. Wood wastes are used to make particle board. Hazardous wastes are treated and recovered. Bituminous roofs are recovered in energy for cement plants, and mercury from fluorescent lamps is salvaged and reused. Alternative 8: The only difference between this alternative and alternative 7 is that inert wastes are used to produce new concrete blocks, as in the alternative 6. Alternative 9: This alternative is the same as alternative 7, except that wood wastes are used as fuel for district heating. Table 2 summarizes the different alternatives with the means of sorting and waste treatments used. 3. Methodology 3.1. Criteria The criteria taken into consideration for this study are summarized in Fig. 1 and are described in more detail below. Fig. 1 represents the three spheres of sustainable development, which include its social, environmental, and economic aspects. Each decision criteria used in this study takes place in these spheres and interspheres of sustainable development. These criteria are called C1– C8. Sub-criterion j of a criterion i is called Cij. 3.1.1. C1: ‘‘lost energy” This criterion measures the lost energy between the initial construction product and the use of demolition waste. In the first step, the embodied energy of all construction products that comprise the different buildings is considered. This embodied energy represents the initial energy investment necessary for construction products. In the second step, the embodied energy is compared with the energy used and produced by the waste management practice. For example, if waste goes to the landfill, all the embodied energy of this waste is lost. If waste is used as fuel, the lost energy is the difference between the produced energy and the initial embodied energy. Produced energy is the energy usable and produced by the energy recovery. This energy is not equivalent to the net calorific value, and takes into consideration the efficiency of the recovery process. For example, if wood is used to produce electricity, the energy recovered is the product of the net calorific value of wood and the efficiency of the power station. For the recovery of waste in a new product, the lost energy is the sum of the embodied energy of the waste with the energy consumed in the recovery of this waste, reduced by the energy which would be required to make the same new product with natural raw materials (without the use of wastes). For recycling, the embodied energy of product is saved, and only the energy used in the recycling process is lost. 3.1.2. C2:‘‘depletion of abiotic resources” This criterion is divided in two sub-criteria. The first sub-criterion (C21) concerns a balance of the flows of abiotic materials. This

14

N. Roussat et al. / Waste Management 29 (2009) 12–20

Table 2 Waste treatments used for each alternative

Demolition Deconstruction sorting Platform Illegal dump Inert waste in Landfill Inert wastes in road engineering Inert wastes in concrete blocks Non-hazardous wastes in landfill Non-hazardous wastes in incinerator Recycling of metals Recycling of flat glass and glass wool Recycling of PVC Wood wastes in particles boards Wood wastes used to district heating Hazardous wastes in sanitary landfill Specific treatments for hazardous wastes

A1

A2

A3

x

x

x

x

x

A4

A5

A6

A7

A8

A9

x

x

x

x

x

x

x x

x x

x

x x

x x

x x

x

x

x

x

x x

x x

x x

x

x x

x x

x x

x x x x

x x x x

x x x x

x

x

x

Environment C1 «lost energy» C2 «depletion of abiotic resources» C3 «global warning» C4 «dispersion of dangerous substances in environment»

C7 «quality of life»

C6 «economy activity»

C8 «employment»

Social

C5 «financial cost of demolition»

Economy

Fig. 1. Criteria taken into account for a sustainable demolition waste management.

sub-criterion has the same philosophy as criterion C1 ‘‘lost energy”, but flows of energy are replaced with flows of abiotic resources in tonnes. The second sub-criterion (C22) is a recovery ratio that considers the non-renewable character of materials contained in wastes. In fact, this rate is a recovery ratio weighted with the factor of abiotic depletion of every material contained in the waste. To weigh the recovery ratio, the abiotic depletion potential (ADP) is used, which was developed by The Institute of Environmental Sciences (CML) of Lendein University (Van Oers et al., 2002) and used in life cycle assessment. These two sub-criteria were considered to have the same importance, i.e., the weight of criterion C2 has been divided equally between each sub-criterion C21 and C22 for the use of the ELECTRE III method. 3.1.3. C3:‘‘Global warming” This criterion is a balance of greenhouse gases, and has the same philosophy of criterion C1 ‘‘lost energy”. A total balance of greenhouse gases in kilograms of CO2-equivalent was made for the life cycle of each building constituent.

3.1.4. C4: ‘‘Dispersion of dangerous substances into the environment” This criterion takes into consideration the dispersion of pollutants into the environment by leachate when demolition wastes are used in road engineering or disposal in landfills. Pollutant concentrations in leachate, according to the sorting level of the demolition waste, were studied in a previous survey on the leaching behaviour of hazardous demolition wastes mixed with inert wastes (Roussat et al., 2008). Each pollutant concentration (As, Ba, Cd, Cr, Cu, Hg, Mo, Ni, Pb, Sb, Se, fluorides, chlorides, sulphates, DOC (dissolved organic carbon), and phenol index) for the first eluate (liquid-to-solid ratio, L/S = 0.1) was compared with the limit value of the European Council Decision of 19 December 2002 establishing criteria and procedure for the acceptance of waste at landfill (2003/33/EC) published in the official journal of European communities of 2003.01.16 for inert waste. This comparison allows the estimation of the performance of each alternative for the dispersion of pollutants as a function of the pollutant concentrations in the leachate. If the pollutant concentration for the first eluate is lower than the limit value, the performance of the alternative for this pollutant is contained between 0 and 1, where 1 represents the best perfor-

15

N. Roussat et al. / Waste Management 29 (2009) 12–20

to 0.1 for the first category, 0.25 for the second, 0.8 for the third, and 1.25 for the last. Ci is the concentration in percent of the limit value where the performance begins to decrease. This concentration is equal to 110% for the three first categories, and 120% for the last category. Fig. 2 shows the performance of the alternatives for this leachate concentration in relation to each category of pollutants. Finally, 16 performances are obtained (one performance for each pollutant). To determine global performance, two sub-criteria (C41 and C42) are considered. The first (C41) is the sum of positive performances of each pollutant (when pollutant concentration is lower than the limit value) and the second (C42) is the sum of negative performances obtained for the other pollutants. These two sub-criteria do not have the same importance. The sum of negative performances is an important factor because this sub-criterion represents the sanitary and environmental risks due to waste management. The sum of positive performances is useful in comparing two alternatives which have leachate concentrations lower than the limit values. So, the sum of negative performances was considered to be three times more important than the sum of positive performances, i.e., the weight of sub-criteria C42 is three times more important than the weight of sub-criteria C41.

mance, i.e., the consequences of pollutant dispersion are considered to be negligible. It was arbitrarily decided that this is true if the concentration is lower than 60% of the limit value. The performance is null when the concentration is equal to the limit value. Between these two extreme cases, the performance P of the alternative for each pollutant is calculated in relation to the concentration in percent of the limit value C% with the equation:

PðC % Þ ¼ 1 

C% X ð2  ðC %  0:6ÞÞ2 : 0:6

In fact, the nearer a concentration is to the limit value, the more quickly the performance decreases. If the pollutant concentration is higher than the limit value, the performance of the alternative for this pollutant is negative. Around the limit value, a low exceedance is not significant, but a large exceedance (2 or 3 times the limit value) is unacceptable. Each pollutant does not have the same environmental impact. For example, an exceedance of the limit value in arsenic is more dangerous than an exceedance of the limit value in sulphates. Performances must be different according to each considered pollutant. Pollutants were classified into four categories in relation to their dangerous characteristics, which are given in Table 3. These categories were defined in relation to the European limit values for drinking water (European Directive 98/83/CE). The first category is composed of the most dangerous pollutants. For each category, the performance P for each pollutant is calculated with the concentration in percent of the limit value C% with equation:

PðC % Þ ¼ 

3.1.5. C5: ‘‘Financial cost of demolition” This criterion represents the cost of the demolition paid by the urban community. This cost (in k€) takes into consideration the costs of deconstruction, demolition, waste transport, and waste treatment.

C% X ðC %  FdÞ2 :

3.1.6. C6: ‘‘Economic activity generated by recycled materials” The aim of this criterion is to measure economic activity created by the use of wastes such as raw material or fuel. Indeed, the use of waste as a secondary raw material avoids the importation of natural resources from other territories and generates economic activity in the territory. The performance of this criterion is measured by value added (in k€) produced by companies in the urban community, i.e., the difference of economic value between the wastes and the produced secondary raw material (or energy).

Ci

Fd is the factor of danger (without scientific signification). The more hazardous a pollutant is, the lower this factor is. This factor is equal

Table 3 Categories of pollutants according to their toxicities for the valuation of criterion C4 ‘‘dispersion of dangerous substances in the environment” 1 (pollutants the most toxic) 2 3 4

As, Sb, Cd, PB, Hg, Se Ni, Cr, Mo, phenol Fluorides, Cu, Zn, Ba Sulphates, Chlorides, DOC

3.1.7. C7: ‘‘Quality of life” This criterion is subdivided into three sub-criteria. The first sub-criterion (C71) is about the areas destroyed, or preserved, by waste management in the urban community. This sub-

0 0%

50%

100%

150%

200%

250%

300%

350%

-10

performances

Group Group Group Group

-20

-30

-40

-50

-60

Concentration in % of limit value group 1

group 2

group 3

group 4

Fig. 2. Performances by group of pollutant for the criterion C4.

400%

16

N. Roussat et al. / Waste Management 29 (2009) 12–20

criterion takes into account the natural area necessary for landfills and the sorting platform. The use of inert wastes to produce aggregates allows the preservation of natural areas used for quarries. This sub-criterion is measured in hectares. The second sub-criterion (C72) takes into consideration the duration of demolition. Indeed, the longer the duration of the demolition, the more important the nuisances are for the local residents. The duration is noted in months. The last sub-criterion (C73) concerns the impact due to the traffic of the trucks carrying the wastes. This criterion is valued with the formula as follows:

C73 ¼ number of trucks  ðinterurban distance þ 23  urban distanceÞ  1:104 : In this formula, the number 23 represents the ratio of population density between the urban area and the interurban area in the urban community of Lyon. These three sub-criteria do not have the same importance. The sub-criteria C71 and C73 are more important than sub-criterion C72 because they have long-term impacts. Nuisances due to the duration of demolition are limited in the time. The sub-criteria C71 and C73 are twice as important as sub-criterion C72. 3.1.8. C8: ‘‘Employment” This criterion takes into account the number of jobs created due to waste management. These jobs are related to deconstruction and demolition of buildings and waste treatment. This number of jobs is diminished by a factor Fa in relation to occupational accidents. The demolition sector is an accident-prone field, and this factor gives more importance to non-dangerous jobs. This factor Fa was determined from French statistics of occupational accidents, and is defined as follows:

Fa ¼ 1 

If Tg  : 1000 Tgn

where If is the frequency index of accident per thousand employees for the considered job, Tg is the rate of serious accidents for the considered job, and Tgn is the national rate of serious accidents.

cj(a,b) 1

0

 Concordance. Alternative a outranks alternative b if a sufficient majority of criteria are in favour of alternative a.  Non-Discordance: When the concordance condition holds, none of the criteria in the minority should be opposed too strongly to the outranking of b by a. The main difference between ELECTRE III and the other ELECTRE methods is that outranking relations can be interpreted as fuzzy relations. In ELECTRE III, the assertion that a outranks b is characterized by a credibility index which permits knowing the true degree of this assertion.

pj

gj(a)-gj(b)

Fig. 3. Pseudo-criterion used in ELECTRE III to determine concordance index.

Fuzzy relations are introduced in ELECTRE III with the use of pseudo-criteria. These pseudo-criteria allow building a concordance and discordance index for each criterion, and so to construct outranking relations. 3.2.1. Concordance index and discordance index To compare a pair of alternatives (a, b), the assertion ‘‘a outranks b” is evaluated, for each criterion, with the help of pseudocriteria. Fig. 3 gives the structure of a pseudo-criterion; cj(a, b) represents the concordance index for the criterion j, and is valued with the difference gj(a)  gj(b). The pseudo-criterion is built with two thresholds, denoted by qj and pj: qj represents the indifference threshold and pj the preference threshold. If the difference gj(a)  gj(b) is lower than qj, there is no difference between a and b (so cj(a, b) = 0), and if gj(a)  gj(b) is higher than pj, a is strictly preferred to b (so cj(a, b) = 0). For a criterion j and a pair of alternatives (a, b), the concordance index can be defined as follows:

8 > < gjðaÞ  gjðbÞ 6 qj () cjða; bÞ ¼ 0; ; qj < gjðaÞ  gjðbÞ < pj () cjða; bÞ ¼ gjðaÞgjðbÞqj pjqj > : gjðaÞ  gjðbÞ > pj () cjða; bÞ ¼ 1: A global concordance index Cab, for each pair of alternatives (a, b), is computed with the concordance index cj(a, b) of each criterion j: n P

C ab ¼

wj  cjða; bÞ

j¼1 n P

: wj

j¼1

3.2. Description of ELECTRE III method The problem of multicriteria analysis is formulated by using a set of alternatives (A1, A2, A3. . .) and a set of criteria (C1, C2, C3. . .), which can be divided into sub-criteria. Criteria represent the performance of the alternative in relation to the decision objectives, with each criterion being a decision aspect. The evaluation of criterion j for alternative a is called gj(a). Depending on if the goal is to maximize or minimize the criterion, the higher or lower it is, the better the alternative is in relation to this criterion. ELECTRE III is a method used to rank problems using binary outranking relations. The construction of these outranking relations is based on two major concepts as in other ELECRTE methods (Figueira et al., 2005):

qj

where wj is the weight of criterion j. A discordance index dj(a, b) is also taken into consideration for each pair of alternatives and each criterion. This discordance index is evaluated with the help of pseudo-criteria with a veto threshold vj which represents the maximum difference gj(a)  gj(b) acceptable to not reject the assertion ‘‘a outranks b”. If the difference gj(a)  gj(b) is lower than pj, there is no discordance and so dj(a, b)=0 and if gj(a)  gj(b) is higher than vj, dj(a, b) = 1. This index can be represented as follows:

8 > < gjðbÞ  gjðaÞ 6 pj () djða; bÞ ¼ 0; ; pj < gjðbÞ  gjðaÞ < vj () djða; bÞ ¼ gjðbÞgjðaÞpj vjpj > : gjðbÞ  gjðaÞ > vj () djða; bÞ ¼ 1: Finally, the index of credibility dab of the assertion ‘‘a outranks b” is defined as follows:

dab ¼ C ab

Y 1  djða; bÞ 1  C ab j2F

with F ¼ fj 2 F; djða; bÞ > C ab g:

When a veto threshold is exceeded for at least one criterion, the index of credibility is null, i.e., the assertion ‘‘a outranks b” is rejected. 3.2.2. Ranking procedure Two complete preorders are constructed through a descending distillation procedure and an ascending distillation procedure.

17

N. Roussat et al. / Waste Management 29 (2009) 12–20

These procedures are not presented in this paper; for details, see Maystre et al. (1994) and Roy and Bouyssou (1993). Briefly, we can say that a descending distillation procedure is the ranking from the best alternatives to the worst, and an ascending distillation procedure is the ranking from the worst alternatives to the best. The two rankings are commonly not the same, and it can be difficult to obtain a final ranking. If an alternative is incomparable with the other alternatives, it does not have the same ranking in descending distillation and ascending distillation. The software ELECTRE III, developed by the LAMSADE (Laboratory of Analysis and Modelling of Decision Aid Systems) of the University Paris-Dauphine is used to realise the ranking procedure. 4. Results

w1 þ w2 þ w3 ¼ 1 with w1 ¼ w2 ¼ w3 ; And for i–1; 2; 3; wi ¼ 1: 4.2.3. Sensitivity analysis The ranking of alternatives can depend on the values of the different thresholds. A sensitivity analysis is necessary to justify the final recommendation. In the present study, thirteen variations of thresholds were considered, as follows: (1) (2) (3) (4) (5)

4.1. Assessment matrix of criteria Table 4 gives the evaluation of criteria for each alternative. The last column gives the preference direction for each criterion. If the preference direction is positive, the higher the evaluation of the criterion is, the better the alternative is, and vice versa. 4.2. Parameters needed by ELECRTE III 4.2.1. The thresholds The values of preference, indifference, and veto thresholds for each criterion are shown in Table 5. For criterion C1, the indifference threshold is q1 = 10 toe (tonnes oil equivalent, 420 GJ), i.e., if the difference between two alternatives for this criterion is lower than 10 toe, these alternatives are indifferent for this criterion. The preference threshold is p1 = 50 toe (2100 GJ). These two thresholds, even though they have physical meaning, were chosen arbitrarily but their influences on the result were studied in the sensitivity analysis. For criterion C3 (global warming), the thresholds are the same as those for criterion C1, but tonne oil equivalent is transformed into tonne carbon dioxide equivalent. (1 toe  4 t CO2). For the other criteria, the thresholds were determined according to the value amplitudes in the matrix of criteria (Table 3). Only criterion C41, concerning the dispersion of pollutants into the environment, has a veto threshold. The value of this threshold agrees with a concentration of arsenic twice as high as the limit value in concentration (cf.3.1). 4.2.2. The weights This study is within the framework of sustainable development; environmental aspects, social issues, and economic criteria must be considered, and must have the same importance. Considering Fig. 1, each sphere and inter-sphere of sustainable development must have the same importance, and so must have the same weights. The weights have been defined as follows:

(6) (7) (8) (9) (10) (11) (12) (13)

All indifference thresholds are null. 8i; q0i ¼ 0

8i; q0i ¼ 0:8  pi All preference thresholds are divided by 2. 8i; p0i ¼ pi =2 v0i ¼ 5  vi All thresholds were based on value amplitudes and were defined as follows: 8i ; q0i ¼ 10%DC i max; p0i ¼ 50%DC i max i; v0i ¼ 80%DC i max For i = 1,2,3, q0i ¼ 0 and p0i ¼ qi For i = 1,2,3,q0i ¼ pi and p0i ¼ 3pi For i = 5,6, q0i ¼ 0 and p0i ¼ qi For i = 5,6, q0i ¼ pi and p0i ¼ 3pi For i = 7,8, q0i ¼ 0 and p0i ¼ qi For i = 7,8, q0i ¼ pi and p0i ¼ 3pi For i = 4, q0i ¼ pi and p0i ¼ 3pi and m0i ¼ 5mi The veto threshold for criterion C42 was deleted.

8i; p0i ¼ 2  pi ;

4.3. Results of analysis Basic solution and the solutions of each sensitivity analysis are given in Fig. 4. The results are represented on a graph where the abscissa is the position of the alternative in the descending distillation and where the ordinate is the position of the alternative in the ascending distillation.

Table 5 Thresholds for each decision criterion Thresholds

q (indifference)

p (preference)

C1 C21 C22 C3 C41 C42 C5 C6 C71 C72 C73 C8

420 1000 10% 40 1 7 50 20 1 1 4.6 1

2100 5000 50% 200 5 20 100 80 4 2 23 3

V (veto)

46

Table 4 Matrix of the evaluations of criteria

C1 (GJ) C21 (t) C22 (%) C3 (tco2) C41 C42 C5 (k€) C6 (k€) C71 (ha) C72 (months) C73 C8

A1

A2

A3

A4

A5

A6

A7

A8

A9

Preference direction

38,723 33,207 0 3375 11.36 320.9 203.7 0 0 1 21.5 0

34,913 32,085 0.2 3127 12.78 148.4 463.0 11.7 4.9 1 47.9 3.7

25,596 2123 5 3547 12.78 148.4 356.2 44.8 10.7 1 27.7 4.5

34,842 32,095 0.2 3115 12.86 9.9 552.5 11.7 5.4 3.5 39.7 10.3

22,570 1445 0.2 3090 12.86 9.9 295 95.9 11.2 4 1.5 11.5

39,773 17,485 0.2 4135 17 0 383 95.9 11.2 4 22.7 11.5

19,500 868 99 3160 12.86 9.9 264 116.8 11.2 4 2.7 11.3

34,525 16,958 99 4295 17 0 352 116.8 11.2 4 23.9 11.3

16,486 958 99 3653 12.86 9.9 264 164.9 11.2 4 1 11.4

  +  + +  + +   +

18

N. Roussat et al. / Waste Management 29 (2009) 12–20

basic solution

A7

Sensitivity analysis n˚1

A9

1 A5

3

A1

4 A6

5 A3

6

A4

7 A2

8

A8

2

A8

ascending distillation

ascending distillation

2

3 A3

4

A6

5

A4

A2

6

A1

7

9 9

8

7

6 5 4 3 descending distillation

2

1

9

8

7

5

4

3

2

1

Sensitivity analysis n˚3 A7 A5

1

1

A6 A8 A9

ascending distillation

2 A3

3

6

descending distillation

Sensitivity analysis n˚2

ascending distillation

A5

8

9

A1

4 A4 5 A2 6 7 8

A9

A8 A6

2

A1

3

A7 A5

4 A3

5

A4

6

A2

7 8 9

9

9

8

7

6

5

4

3

2

9

1

8

descending distillation

1

7

6

5

4

3

1

A8

A7

A9

A5 2

ascending distillation

A6 3 A3 A1

5

1

Sensitivity analysis n˚5

A9 A7

A5

2

4

2

descending distillation

Sensitivity analysis n˚4

ascending distillation

A9 A7

1

A4

6 A2 7

A8 A6 3 A1 4 A4 5

A3

A2 6 7

8 9

8

9

8

7

6

5

4

3

2

1

descending distillation

8

7

6

5 4 3 descending distillation

2

1

Fig. 4. Basic solution and solutions of sensitivity analysis with ELECTRE III.

5. Discussion For the fourteen results, the rankings of alternatives were very stable, except for A1. The instability of A1 shows that this alternative is not comparable with the rest of the alternatives and must be excluded from the final recommendation. The instability of alternative A1 showed that the ELECTRE III method allows the detection of alternatives that are non-coherent with the objectives of the initial problem and the exclusion of these alternatives. Alternative A9 has a stable ranking and arrives in the first position, for ascending and descending distillation, for all cases. A7 also has a stable ranking and is the same as A9 for four cases. A5 always follows these two actions. These three actions were always at the top of the rankings. The common point of these three actions is the in situ material recovery of inert wastes in road engineering.

The difference between alternative A5 and alternatives A7 and A9 concerns the treatment of hazardous wastes. In A5, hazardous wastes go to a landfill, whereas in A7 and A9 these wastes are treated and recovered. The treatment of hazardous wastes allows the improvement of demolition waste management. Between A7 and A9, the difference concerns the management of wood wastes. In the first case (A7), wood wastes are used to made particle board, and in the second case (A9) wood wastes are used as fuel for district heating. It appears, but is not certain, that local energy recovered from wood wastes is better than material recovered for the sustainability of land. A8 and A6 often have the same position, even if A8 appears to be better than A6. The common point of these scenarios is the recovery material of inert wastes in concrete blocks. The recovery of inert wastes in concrete blocks requires more energy and abiotic

19

N. Roussat et al. / Waste Management 29 (2009) 12–20

Sensitivity analysis n˚6

Sensitivity analysis n˚7

A9

A7

A7 1

1 A5

2 A8

3

ascending distillation

ascending distillation

2 A1

4 A6 5 A3 6

A4

7 A2

A1 3 A8 A6

4 A4 5 A3

6 A2 7

8

8

9 9

8

7

6

5

4

3

2

9

1

8

7

6

descending distillation

4

3

2

1

A7

A9

Sensitivity analysis n˚9 A7

1

A9

1 A5

A5

2 ascending distillation

2 A1 3 A8 A6

4

5

descending distillation

Sensitivity analysis n˚8

ascending distillation

A9

A5

A3 5 A4 6 A2

A8

3 A1 4

A6 5 A3

6 A4

7

7

8

8

A2

8

7

6

5

4

3

2

8

1

7

6

5

4

3

2

Sensitivity analysis n˚10

Sensitivity analysis n˚11 A5 A7

A7 A9

A1

A8

2 A5

A8

3

ascending distillation

2 ascending distillation

A9

1

1

A3 4 A6 5 A4 6 A2

A6 3 A3

4 A4 5 6 A2 7

7

8

8 9

8

7

6

5

4

3

2

8

1

7

6

descending distillation

5

4

3

2

1

descending distillation

Sensitivity analysis n˚12

A7

1

Sensitivity analysis n˚13

A9

A7

1 A5

A9

A5

2

2 ascending distillation

ascending distillation

1

descending distillation

descending distillation

A8 3 A1 4 A6 5 A3

6

A4 7

A8

3 4

A6

A1 5 A4 6 A3 7 A2

A2 8

8 8

7

6

5 4 3 descending distillation

2

1

9

8

7

6

5

4

descending distillation Fig. 4 (continued)

3

2

1

20

N. Roussat et al. / Waste Management 29 (2009) 12–20

Synthesis rankings A7

1

A9

ascending distillation

A5

2

A8

3 A6

A1

4

A3

5 A4

6 A2

7 8 8

7

6 5 4 3 descending distillation

2

1

Fig. 5. Synthesis rankings of each alternative.

resources that the primary production of concrete blocks, and A8 and A6 are not better for the socio-economic aspects than the other alternatives. A2 is the worst solution (A1 is excluded from ranking). A3 and A4 are slightly ahead of A2 in the rankings. A4 has a poor ranking because inert wastes go to a landfill, and A3 has poor performance because of dispersion of dangerous substances into the environment. The stability of the results allows the construction of a synthesis graph, as in Fig. 5. This graph takes into consideration the mean ranking of each alternative. In this graph, alternatives A9, A7, and A5 appear to be the best strategies for demolition waste management. It can be concluded that the essential point for a sustainable demolition waste management practice is selective deconstruction with local material recovery for inert wastes. 6. Conclusion The first objective of this article was to present criteria that can be taken into consideration to evaluate the sustainability of demolition waste management. These criteria take into account the waste management aspects relative to economic, environmental, and social issues. These criteria concern the global environmental aspects (energy consumption, abiotic depletion, global warming), dispersion of pollutants into the environment, the economic consequences of the waste management strategy (cost of demolition and local economic activities due to wastes recovery), the impacts of the strategy on the quality of life of the local population (nuisances due to traffic and duration of demolition, destruction of natural areas), and the consequences on local employment. The second objective of this study was to show the usefulness of multicriteria methods for choosing a method of sustainable waste management. Indeed, in sustainability perspectives, it can be difficult to make a choice because many aspects, often in conflict, must be taken into consideration. The use of multicriteria methods can clarify the decision. The ELECTRE III method is particularly adapted for sustainable development problems because this method searches for the best compromise between all decision criteria and not the solution where only some criteria are optimised. Indeed, a sustainable solution is a compromise between all aspects of sustainable development, and not a solution which optimises only some aspects. In addition, the fuzzy outranking relations of ELECTRE III give a judgement more refined than other ELECTRE methods using non-fuzzy outranking relations.

The results of the multicriteria analysis showed that selective deconstruction is an important step for sustainable demolition waste management. A waste management strategy is not effective without a good sorting of different wastes. For inert wastes, recovery in aggregates for road engineering is a better solution than the use of these aggregates to produce concrete blocks. The poor technical characteristics of recycled aggregates that are involved in the making of concrete blocks are not environmentally and economically efficient in comparison to concrete blocks produced with natural aggregates. In this case study, the recovery of hazardous wastes improves demolition waste management because bituminous roofs represent an important calorific value that is not exploited if these wastes go to landfill, and fluorescent lamps contain a significant quantity of mercury, which can be recycled. Local authorities will be able to use the results of this study to provide guidelines for future demolition of buildings. In addition, the urban community could use the decision criteria and the method that were used in this study to choose other sustainable strategies within the community. Acknowledgements This work was supported by the region Rhône-Alpes (France) and The Grand Lyon (Urban Community of Lyon, France).

References Brent, A.C. et al., 2007. Application of the analytical hierarchy process to establish health care waste management systems that minimize infection risks in developing countries. European Journal of Operational Research 181, 403–424. Dais, D., Loizidou, M., Papadopoulos, A., Fatta, D., 1999. Multicriteria analysis model for the management of used tyres: Case study of Attika-Greece. In: Proceeding Sardina 9, Seventh International Waste Management and Landfill Symposium. Figueira, J., Mousseau, V., Roy, B., 2005. ELECTRE METHODS 134–164 Chapter of book. In: Figueira, J., Greco, S., Ehrgott, M. (Eds.), Multiple Criteria Analysis – State of the Art Surveys, International Series in Operations Research and Management Sciences, vol. 78. Springer, New York, USA. Hokkanen, J., Salminen, P., 1997a. Locating a waste treatment facility by multicriteria analysis. Journal of Multi-Criteria Decision Analysis 6, 175–184. Hokkanen, J., Salminen, P., 1997b. Choosing a solid waste management system using multicriteria decision analysis. European Journal of Operational Research 98, 19–36. Klang, A., Vikman, P.-A., Btattebø, H., 2003. Sustainable management of demolition waste – an integrated model for the evaluation of environmental, economic and social aspects. Resources, Conservation and Recycling 38, 317–334. Maystre, L.Y., Pictet, J., Simos, J., 1994. Méthodes multicritères ELECTRE. Description, conseils pratiques et cas d’application à la gestion environnementale. Presses polytechniques et universitaires romandes, Lausanne, Switzerland. Morissey, A.J., Browne, J., 2004. Waste management models and their application to sustainable waste management. Waste Management 24, 297–308. Munda, G., 2005. Multiple criteria decision analysis and sustainable development, 954–986 Chapter of book. In: Figueira, J., Greco, S., Ehrgott, M., (Eds.), Multiple Criteria Analysis – State of the Art Surveys, vol. 78, International Series in Operations Research and Management Sciences. Springer, New York, USA. Pohekar, S.D., Ramachandran, M., 2004. Application of multi-criteria decision making to sustainable energy planning – A review. Renewable & Sustainable Energy Reviews 8, 365–381. Roussat, N., Méhu, J., Abdelghafour, M., Brula, P., 2008. Leaching behaviour of hazardous demolition waste. Waste Management 28, 2032–2040. Roy, B., Bouyssou, D., 1993. Aide Multicritère à la Décision: Méthodes et cas. Edition Economica, Paris, France. Salminen, P., Hokkanen, J., Lahdelma, R., 1998. Comparing multicriteria methods in the context of environmental problems. European Journal of Operational Research 104, 485–496. Van Oers, L., de Koning, A., Guinnee, J.B., Huppes, G., 2002. Abiotic resource depletion in LCA – Improving characterisation factors for abiotic resource depletion as recommended in the new Dutch LCA Handbook. Leinden: The Road and Hydraulic Engineering Institute of the Dutch Ministry of Transport, Public Works and Water Management. Yelda, S., Shrestha, R.M., 2003. Multi-criteria approach for the selection of alternative options for environmentally sustainable transport system in Delhi. Transportation Research A37, 717–729.