Professional Documents
Culture Documents
Optimization of Regional Hazardous Waste
Optimization of Regional Hazardous Waste
www.elsevier.nl/locate/wasman
Abstract
The planning and design of regional hazardous waste management system (RHWMS) involves selection of treatment and dis-
posal facilities, allocation of hazardous wastes and waste residues from generator to the treatment and disposal sites and selection
of the transportation routes. An improved formulation based upon multi-objective integer programming approach is presented to
arrive at the optimal con®guration of RHWMS components. This formulation addresses important practical issues like unique
characteristics of the hazardous wastes re¯ecting on waste±waste and waste±technology compatibility. A utility function approach
is presented to integrate both cost and risk related objectives. An illustrative case example is presented to demonstrate the useful-
ness of the improved formulation as a tool which can be used by environmental planning agencies in regional planning for hazar-
dous waste management. # 1999 Elsevier Science Ltd. All rights reserved.
Keywords: Hazardous waste management; Regional planning; Treatment and disposal facility location; Waste allocation; Waste±technology com-
patibility; Waste±waste noncompatibility; Cost and risk minimization
In the example problem, the locations of treatment must address the following practical considerations:
facilities were pre-decided. Although the authors men- diverse characteristics of dierent wastes; compatibility
tioned that the model can take care of multiple types of between dierent waste types; waste residue generated
waste, no explicit consideration was made for waste± from waste treatment facilities; formulation of the siting
waste non-compatibility and waste±technology compat- problem using 0±1 decision variables; consideration of
ibility in the formulation of the mathematical model. multiple objectives (i.e., risk, cost, and/or joint function
Zografos and Samara, 1990 [7] provided a combined of risk and cost). It is evident from the review of existing
location-routing model for hazardous waste transpor- models, that no single model incorporates all the above
tation and disposal. They used a 0±1 integer linear pro- mentioned practical considerations. The issue of `waste±
gramming approach for the model formulation. This waste compatibility' and `jointly considering risk and
model considers three objectives: (i) minimization of cost' are not addressed by any of them. Clearly, recog-
transportation risk, (ii) minimization of travel time, and nizing and incorporating these issues would help in for-
(iii) minimization of disposal risk. They did not consider mulating a more eective management model and thus
multiple types of hazardous waste. ReVelle et al., 1991 help in arriving at a better implementable solution.
[8] formulated a management model for spent nuclear
fuel. The model simultaneously identi®es the storage
facilities, assigns the generators to the identi®ed facil- 3. Proposed model formulation
ities and chooses routes for shipments. Method of
shortest path, 0±1 integer mathematical programming The regional network for hazardous waste management
and the weighting method of multi-objective program- consists of a number of nodes. The nodes may be the
ming were synthesized for the solution of the problem. sources generating dierent types of hazardous wastes,
List and Mirchandani, 1991 [9] presented a model potential treatment facilities, or potential disposal facil-
useful in making routing decisions. The model can be ities. Nodes are connected to each other by means of
used for ®nding material or waste shipment paths as transportation routes. The hazardous wastes generated at
well as for siting waste treatment facilities. They used the sources must be carried to the appropriate treatment
integer linear programming with a utility function facility. The treated waste residues are then transferred to
approach to deal with multiple objectives. a disposal facility. The activities of transportation, treat-
Stowers and Palekar, 1993 [11] developed a model ment, and ®nal disposal involve costs and they also pose
that selects the location of storage/ treatment facility some risk to the environment. Thus, the problem is to
and transportation routes for a single type of hazardous select an optimal con®guration of facilities for transpor-
waste/material. They assumed that the facility can be tation, treatment, and disposal so that hazardous waste is
located anywhere in the region. However, in practice managed with minimum cost and minimum risk to the
only a few candidate sites will be available because of environment. The model formulation starts with identi®-
the landuse pattern of the region. cation of decision variables. The decision variables in this
Jacobs and Warmerdam, 1994 [10] formulated a lin- case are, (i) waste quantities traveling on the transporta-
ear programming based model for the simultaneous sit- tion links, (ii) decision variables for the location of treat-
ing and routing of hazardous waste transport, storage, ment and disposal facilities, and (iii) waste quantities
and disposal operations for a single type of waste. being processed at the treatment or disposal facilities. The
Peirce and Davidson, 1982 [5] Jennings and Scholar, ®nal step is the formulation of mathematical equations for
1984 [6] and Jacobs and Warmerdam, 1994 [10] modeled the objectives and constraints. The objective/(s) can be
cost with linear functions, that is, the cost of hazardous minimization of cost and/or minimization of risk. Satis-
waste treatment or disposal is dependent upon per unit faction of the constraints is essential for a solution to be
waste quantity processed at the facility. However, the implementable. An example of a constraint is `the waste
initial capital cost for installing any facility may sig- quantity to be processed at a treatment or disposal facility
ni®cantly in¯uence the siting decision. The models by shall not exceed the allowable capacities of the respective
Zografos and Samara, 1990 [7], ReVelle et al., 1991 [8], facility'. Next step is to develop a utility function in order
List and Mirchandani, 1991 [9], and Stowers and Pale- to ®nd a compromise solution for minimization of risk as
kar, 1993 [10], are for a single type of hazardous waste well as cost. The mathematical expressions for the various
or hazardous material. Therefore, waste±waste non- objectives and constraints and assumptions made are as
compatibility and waste±treatment technology compat- follows.
ibility issues are not applicable. The objectives addressed are:
2.1. Remarks on the existing models . minimization of total cost, which includes treat-
ment and disposal costs (sum of the capital cost,
A mathematical model which may be helpful in plan- operation and maintenance cost of the treatment
ning for a regional hazardous waste management system and disposal facilities), and transportation cost;
444 A.K. Nema, S.K. Gupta / Waste Management 19 (1999) 441±451
. minimization of total risk, which includes waste of the accident can be estimated based on the physical
treatment and disposal risk as well as risk involved and chemical properties of the hazardous waste
in waste transportation; and released, its quantity and the sensitivity of the exposed
. minimization of the composite objective function environment to the accidental release. The hazard
consisting of cost as well as risk. potential for the wastes can be determined using a
ranking procedure such as the DARE technique (Klee,
The problem is subjected to the following constraints: 1976 [14]) or Analytic Hierarchical Process (Saaty, 1980
[15]) based upon the characteristics of the wastes.
. mass balance of wastes at each node (i.e. all gen-
Once the potential undesirable events are identi®ed,
erator nodes, all intermediate nodes, and all desti-
with their probabilities and consequences of hazardous
nation nodes);
waste release to the environment estimated, the risk to
. allowable capacities for treatment and disposal
the environment may be quanti®ed by the following
technologies at speci®c sites;
expression:
. waste±treatment technology compatibility con-
straints; and Risk Probability of occurrence of the
. waste±waste non-compatibility constraints.
hazardous release event 1
Estimated consequences of the event
3.1. Assumptions
For the present problem, total risk can be expressed
in terms of addition of risk due to transportation and
. The region for which the planning for hazardous
risk due to treatment and disposal [Eq. (2)].
waste management is to be done is divided into
source nodes, intermediate nodes, treatment TOTR RT RS 2
nodes, disposal nodes, and transportation links.
. The population impacted by a hazardous waste
Where, TOTR Total risk; RT Transportation risk
management related activity is considered as an
and RS Site risk.
attribute of the corresponding treatment node,
Transportation risk for a link is a function of waste
disposal node or transportation link.
quantity being transported through the link, hazard
. It is assumed that the objective functions are linear
potential of the waste, probability of accident on the
or segmented linear in nature.
link and receptor population impacted in case of acci-
. The transportation costs are directly proportional
dent [Eq. (3)]. The total transportation risk is the sum-
to the network distance used, with the constant of
mation of transportation risks on all the links.
proportionality being independent of the value of
the distance. X X
RT 12L w2W fWwrs Hw PRrs Prs g 3
. The risk functions for hazardous waste transpor-
tation are directly proportional to the quantity of
Where, RT Transportation risk; Wwrs=waste quan-
waste being transported and the network distance
tity (of the waste type w), traveling between the nodes r
traversed.
and s; Hw hazard potential of the waste type w,
including waste treatment residues; PRrs probability
3.2. Formulation of the objectives of accident on the link l, joining nodes r and s;
Prs population impacted if accident occurs on the link
3.2.1. Estimation of risk l, joining nodes r and s; and L set of all the transpor-
Risk estimation involves identi®cation of factors tation links in the system. Site risk is a function of the
leading to an undesirable event (i.e. an accident), its waste quantity being processed at the treatment or dis-
probability, and the assessment of the probable out- posal facility, hazard potential of the waste, probability
comes of the undesirable event. It also involves the of accident at the facility, and receptor population
quanti®cation of the consequences of the event. An impacted in case of accident [Eq. (4)].
undesirable event in this context may be release of X X X
hazardous wastes due to an accident during transporta- RS k2K t2T w2W fWTDwtk Hw PRtk Ptk g
tion or due to an accident that occurred during treat-
ment or disposal of the hazardous wastes. The 4
probability of occurrence of an accident within a speci®ed Where, RS site risk; WTDwtk waste quantity (of
period (usually 1 year) can be estimated based on either the waste type w), to be processed at site k using tech-
historical data, or employing fault and event tree techni- nology t; PRtk probability of accident at the site k due to
ques, (Environment Protection Authority of Victoria, technology t; Ptk population impacted if accident occurs
1985 [12]; Petts and Eduljee, 1994 [13]). The consequences at site k due to technology t. The above expressions of risk
A.K. Nema, S.K. Gupta / Waste Management 19 (1999) 441±451 445
[Eqs. (3) and (4)] are limited to the receptor population Where, TOTC total cost.
in consideration, which is human population in the
present case. If information is available, then the equa-
tions can be extended to consider all types of receptors. 3.3. Composite cost-risk utility function
A possible extension is to use a representative popula-
tion, which may be the sum of the population of each To consider both risk and cost together in a single
receptor multiplied by their relative importance in the objective function, a normalized composite utility func-
ecosystem. tion is proposed. As measures of risk and cost have dif-
ferent units of expression, it is not desirable to combine
3.2.2. Estimation of cost them directly in a quasi-additive form (i.e. by giving them
The total cost of waste treatment or disposal (Fig. 1) weightings). The utility function, u is formulated as the
is assumed to consist of a ®xed cost (depending upon composite of weighted risk and cost utilities [Eq. (10)].
the capacity) for installing the facility, and a variable u Weighting to Risk
operational cost depending upon the quantity of the
Risk=Minimum achievable Risk
waste being treated [Eq. (5)]. The cost of insurance 10
against any accident may be included with the unit costs Weighting to Cost
of treatment, disposal and transportation. Cost=Minimum achievable Cost
that the addition will be equal to 1. The utility function WTDwtk ÿ Ctk zwtk 40 for all k andt 14
proposed should be minimized, for the given constraints.
X
An examination of Eq. 10 shows that the minimum w2W fWTDwtk g4Ctk for all k and t 15
achievable value of the utility function is 1, and the value
would be higher if the objectives of minimization of risks X nX X
and costs are not commensurate to each other. w2W i2I fWrki
ÿ Wrik g j2J Wrkj ÿ Wrjk
o
3.4. Formulation of the constraints ÿREStw WTDwtk 0 for all k andt
16
3.4.1. Mass balance at the nodes
Mass balances of wastes, at all the generation nodes, Where, Wr Quantity of waste residue; Ctk Capacity
intermediate nodes (the junction nodes of the transpor- of technology t at node k; REStw Residue generation
tation links), treatment nodes and disposal nodes are factor for waste type w at technology t; and zwtk 0ÿ1
formulated as follows. variable which represents the presence or absence of the
At the waste generation nodes the mass balance waste type w being processed at the treatment or disposal
between waste generation and waste handling is for- technology t at the site k.
mulated in Eq. (11). The equation states that all the
waste being generated at the node must be transported 3.4.2. Logical constraints for waste±waste and waste±
for treatment. technology compatibility conditions
X X The waste±waste and waste±treatment technology
j2J Wwij ÿ Wwji k2K fWwik ÿ Wwki g compatibility constraints are necessary to be included in
order to avoid non-feasible solutions. An example is
Qwi for all i and w 11 presented to illustrate the formulation of such logical
constraints.
Where, i waste generation node; j intermediate Example: let us assume site k as the potential site for a
node; k waste treatment or disposal node; w waste technology t. There are three types of wastes (W1 , W2 ,
type; Wwrs quantity of waste type w traveling between and W3 ) being generated in the region. All the three waste
the nodes r and s (where, r and s represent any pair of types are compatible with the technology t. The waste
nodes and can be replaced with i, j, or k), and types W1 and W3 are considered as compatible with each
Qwi quantity of waste type w generated at node i. other (i.e. waste types W1 and W3 can be treated by the
At the intermediate nodes, the waste quantities arriv- same treatment technology). However, combination of
ing must be balanced with the waste quantities being waste types W1 ÿ W2 and W2 ÿ W3 are considered as
transported away from the intermediate node [Eq. (12)]. non-compatible (i.e. these combinations should not be
X X allowed at the same treatment technology).
i2I Wwji ÿ Wwij k2K Wwjk ÿ Wwkj If we use a 0±1 variable ytk to represent the presence
or absence of the technology t at the site k, and a 0±1
0 for all j and w 12 variable zwtk to represent the state of waste type w being
treated at technology t at site k, then the waste±waste
At a treatment or disposal node all the quantity of and waste±technology compatibility constraints may be
waste or waste residue coming in must be treated [Eqs formulated as follows.
(13) and (14)]. At a treatment or disposal node the
compatibility of the waste type with the technology is Logical conditions Logical constraint
checked using 0±1 variables (discussed later). The waste
quantities being processed should not exceed the capa- if either z111 or z211 or z311 1 z111 z211 z311 ÿ y11 51
city of the treatment or disposal facility [Eq. (15)]. The
waste residue being generated at a treatment node must 17
be sent to the disposal nodes [Eq. (16)].
then y11 1 z111 ÿ y11 40 18
X nX X
w2W i2I fWwki ÿ Wwik g j2J Wwkj ÿ Wwjk
z211 ÿ y11 40 19
o
WTDwtk 0 for all k and t
z311 ÿ y11 40 20
13
eitherz111 1; or z211 1 z111 Z211 41 21
Where, WTDwtk waste quantity (of the waste type w),
treated or disposed at technology t, at node k. either z211 1; or z311 1; z211 z311 41 22
A.K. Nema, S.K. Gupta / Waste Management 19 (1999) 441±451 447
4.2. Results and discussion ``e''. Nodes 1 and 13 have been chosen for land®ll in the
cases of ``a'' and ``b'', whereas, only one node has been
The example problem has been solved for the follow- selected as the weighting to cost is 0.5 and more. Fig. 3
ing ®ve sets of joint functions of cost and risk: (a) (a±e) shows that the transportation routes vary sig-
minimization of risk, which has been achieved by ni®cantly for dierent cases. Fig. 4 and Table 5 shows
assigning zero weighting to cost; (b) weighting to risk : that risk and cost has an inverse relationship. A com-
0.7 and weighting to cost : 0.3; (c) weighting to risk : 0.5 promise solution can be obtained by judiciously choos-
and weighting to cost : 0.5; (d) weighting to risk : 0.3 ing the weightings to cost and risk.
and weighting to cost : 0.7; (e) minimization of cost, The observations can be summarized as follows:
which has been achieved by assigning zero weighting to
risk. Fig. 3 (a±e) show the results obtained for various . total cost and total risk of the system have inverse
combinations of weighting given to risk and cost. The relationships, i.e. minimization of cost and mini-
®gures show the selected treatment technologies, dis- mization of risk are con¯icting objectives. How-
posal sites and waste quantities being transported on ever, the extent of the inverse relationship would
various routes. Relationship between cost and risk is depend upon the con®guration of the network,
shown in Fig. 4 and Table 5. Nodes 2 and 15 have been other attributes of the network like population
chosen for incineration and solidi®cation/chemical ®xa- aected, probability of accident on the link etc.,
tion technologies, respectively, for the cases ``a''±``d'', and the cost-risk data.
whereas node 2 is chosen for solidi®cation/chemical . Solution for minimum cost and minimum risk may
®xation and node 15 is chosen for incineration in case of dier in technologies associated with the sites,
allocation of the wastes to the technologies, rout-
ing of the hazardous wastes and residues, and the
Table 2
choice of the land®ll sites.
Waste±waste non-compatibility information
. Ideally waste treatment facilities and disposal facil-
Waste types Remark ity should be at the same site unless demanded by
Metal plating waste and petrochemical waste Non-compatible site speci®c constraints. Availability of an attached
Metal plating waste and pesticides waste Compatible
Petrochemical waste and pesticides waste Non-compatible
(or close by) land®ll site in¯uences the locational
decision of the treatment/disposal technologies.
Table 3
Details of treatment facilities
Table 4
Details of disposal facilities
Fig. 3. Eect of variation in weightings to cost and risk on the solution of the example problem.
Table 5 References
Results of example problem
[1] Batstone R, Smith Jr JE, Wilson D. The safe disposal of hazar-
Sr.No. Weighting Weighting Total cost Total risk dous wastes, the special needs and problems of developing coun-
to cost to risk ($/year) (x 10ÿ3) tries, vols. I±III. World Bank Technical Paper Number 93, 1989.
[2] Ministry of Environment and Forests, Government of India.
1 1.0 0.0 11,06,600 4,060.63
Guidelines for management and handling of hazardous waste, 1991.
(Min. cost
[3] Sloan WM. Site selection for new hazardous waste management
solution)
facilities. WHO Regional Publications European Series: No. 46,
2 0.7 0.3 11,26,560 3,106.11
1993.
3 0.5 0.5 11,63,760 2,949.31
[4] Lagrega MD, Buckingham PL, Evans JC. The Environmental
4 0.3 0.7 11,76,560 2,934.91
Resources Management Group. Hazardous waste management.
5 0.0 1.0 12,06,560 2,914.11
New York: McGraw-Hill, 1994.
(Min. risk
[5] Peirce JJ, Davidson GM. Linear programming in hazardous
solution)
waste management. Journal of the Env Engrg (ASCE)
1982;108(EE5):1014±26.
[6] Jennings AA, Scholar RL. Hazardous waste disposal network
compatibility and consideration of multiple objectives,
analysis. Journal of Environmental Engineering (ASCE)
such as cost and risk. The example problem discussed in this 1984;110(2):325±42.
paper demonstrates the usefulness of the improved for- [7] Zografos KG, Samara SSA. Combined location-routing model
mulation. Further the proposed model can help decision- for hazardous waste transportation and disposal. Transportation
makers to plan for the following management strategies: Research Record 1990;1245:52±9.
[8] ReVelle C, Cohon J, Shobrys D. Simultaneous siting and routing
. How to minimize risks for a given budget? in the disposal of hazardous wastes. Transportation Science
. How to re¯ect issues related to total risks, site 1991;25(2):138±45.
[9] List G, Mirchandani P. An integrated network/planar multi-
risks and transportation risks etc.?
objective model for routing and siting for hazardous materials
. How to decide on capacities of treatment/disposal and wastes. Transportation Science (ASCE) 1991;25(2):146±56.
facilities? [10] Jacobs, TL, Warmerdam, J.M. Simultaneous routing and siting
. How to use waste±waste compatibility constraints? for hazardous waste operations. Journal of Urban Planning and
. How to set up a management model, given limited Development (ASCE), 1994:120(3):115±31
[11] Stowers CL, Palekar US. Location models with Routing con-
information on risks and cost?
siderations for a single obnoxious facility. Transportation Science
In order to minimize risk for a given budget the model (ASCE) 1993;27(4):350±62.
should be subjected to a constraint such that the total [12] Environment Protection Authority of Victoria. Hazard assess-
ment of industrial waste disposal facilities, Pub. 214, 1985.
cost [Eq. (9)] can be equal to or less than the given bud- [13] Petts J, Eduljee G. Environmental impact assessment for waste
get. Site risks and transportation risks can be minimized treatment and disposal facilities. New York: John Wiley and
individually by using Eq. (3) and (4), respectively, in the Sons, 1994.
A.K. Nema, S.K. Gupta / Waste Management 19 (1999) 441±451 451
[14] Klee AJ. Models for evaluation of hazardous wastes. Journal of [17] Freeman HM. Standard handbook of hazardous waste treatment
Environmental Engineering (ASCE) 1976;102(1):111±25. and disposal. New York: McGraw-Hill, 1988.
[15] Saaty TL. The analytic hierarchy process, planning, priority set- [18] UNEP IE/PAC. Land®ll of hazardous industrial wastes, a train-
ting, resource allocation. New York: McGraw-Hill, 1980. ing manual. Technical Paper No. 17, 1994.
[16] Donnachie R. The eects of land ban restrictions: on site±o site [19] Wentz CA. Hazardous waste management. New York: McGraw
economics. Pollution Engineering 1991;23(10):58±65. Hill, 1989.