Research Article A Heuristic Approach Based on Clarke-Wright Algorithm for Open Vehicle Routing Problem

The Scientific World Journal Volume 0, Article ID 879, pages http://dx.doi.org/0.55/0/879 Research Article A Heuristic Approach Based on Clarke-Wright Algorithm for Open Vehicle Routing Problem Tantikorn Pichpibul and Ruengsak Kawtummachai Faculty of Business Administration, Panyapiwat Institute of Management, 85/ Moo, Chaengwattana Road, Bangtalad, Pakkred, Nonthaburi 0, Thailand Correspondence should be addressed to Ruengsak Kawtummachai; ruengsak@yahoo.com Received 5 July 0; Accepted 8 October 0 Academic Editors: A. Miné, D. Simson, and D. Talia Copyright 0 T. Pichpibul and R. Kawtummachai. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. We propose a heuristic approach based on the Clarke-Wright algorithm (CW) to solve the open version of the well-known capacitated vehicle routing problem in which vehicles are not required to return to the depot after completing service. The proposed CW has been presented in four procedures composed of Clarke-Wright formula modification, open-route construction, two-phase selection, and route postimprovement. Computational results show that the proposed CW is competitive and outperforms classical CW in all directions. Moreover, the best known solution is also obtained in 97% of tested instances (60 out of 6).. Introduction The open vehicle routing problem (OVRP) was firstly solved by Sariklis and Powell [] in their paper on distribution management problems. The characteristics of OVRP are similar to the capacitated vehicle routing problem (CVRP), which can be described as the problem of determining a set of vehicle routes to serve a set of customers with known geographical coordinates and known demands. A route represents a sequence of locations that a vehicle must visit. The distances between customer locations and between themandthedepotarecalculatedorknowninadvance.for each route, the vehicle departs from the depot and returns to the depot after completing the service. The CVRP involves a single depot, a homogeneous fleet of vehicles, and a set of customers who require delivery of goods from the depot. The objectiveofthecvrpistoconstructafeasiblesetofvehicle routes that minimizes the total traveling distance and/or the total number of vehicles used. Furthermore, the route must satisfy the constraints that each customer must be visited once, the demands of customers are totally satisfied, and the vehicle capacity is not exceeded for each route. In contrast, in the OVRP the companies either do not have their own vehicle available or the vehicles are inadequate to serve their customers. In this situation, the subcontracted vehicles will be hired from logistics outsourcing companies. Therefore, the transportation cost only depends on the traveling distance from depot to customers in which vehicles do not return to the depot and the maintenance cost does not occur. In another situation, the vehicles may return to the depot by following the same route in reverse order to collect items from the customers. The real-world case studies of OVRP are presented including a train plan model for British Rail freight servicesthroughthechanneltunnel[], the school bus routing problem in Hong Kong [], the distribution of fresh meat in Greece [], the distribution of a daily newspaper in the USA [5, 6], a lubricant distribution problem in Greece [7], and a mines material transport vehicle routing optimization in China [8]. The CVRP is one of the most important and widely studied problems in the area of combinatorial optimization. It comprises the traveling salesman problem (TSP) and the bin packing problem. The main distinction between OVRP and CVRP is that in CVRP each route is TSP which requires a Hamiltonian cycle [9], but in OVRP each route is a Hamiltonian path. Held and Karp [0] andmillerand Thatcher [] have shown that the TSP is classified as NPhard (Non-deterministic Polynomial-time hard) problem. In addition, the Hamiltonian path has also been shown to be NP-hard []. Besides, the CVRP and OVRP are NP-hard

The Scientific World Journal F-n5-k procedure which refers several neighborhood structures from the works of Subramanian et al. [5] andgroër et al. [6]. Moreover, the numerical experiment of CW for solving OVRP benchmark instances is also presented.. The Proposed Clarke-Wright Algorithm CVRP solution (7.570) OVRP solution (6.896) Figure : The difference of best known solutions between CVRP and OVRP. BecauseCWisaheuristicalgorithm,itcannotguaranteethe best solution. Therefore, we introduce the modified version of the Clarke-Wright algorithm in which the parallel version of CW is implemented since it usually generates better results than the corresponding sequential version [, 7]. The flowchart of the proposed CW is given in Figure.First, the Euclidean distance matrix (c i,j )iscalculatedwiththe following equation: [ 5]. In Figure, thebestknownsolutionfortheovrp is very different from the CVRP, and we also refer to Syslo et al. [] that deleting the largest edge from a minimum Hamiltonian cycle does not necessarily yield the minimum Hamiltonian path in the network. The CVRP has attracted many researchers since Dantzig and Ramser [6] proposed the problem in 959. Many efficient heuristics were presented to solve the CVRP. Conversely, from the early 980s to the late 990s, OVRP received very little attention in the operations research literature compared to CVRP. However, since 000, several researchers have used various heuristics to solve OVRP, such as a tabu search [, 7, 8], an ant colony system [9], a variable neighborhood search [0], a particle swarm optimization [], and a genetic algorithm [8]. The CW was proposed by Clarke and Wright [] who introduced the savings concept which is based on the computation of savings for combining two customers into the same route. The CW is a widely known heuristic for solving the vehicle routing problem (VRP), and the applications of CW have continued since it was proposed in 96. Improvements to the CW solution include proposed new parameters to the Clarke-Wright formulation composed of the nearest terminal k for solving multidepot VRP [], deleting c j, for solving OVRP [], an estimate of the maximum savings value s max, and a penalty multiplier α for solving VRP with backhauls [5], route shape λ for solving CVRP [6, 7], weight μ for asymmetric solving CVRP [8], the customer demand ] for solving CVRP [9], and the cosine value of polar coordinate angles of customers with the depot cos θ i,j for solving CVRP [0]. Second is improvements to the CW solution by proposed new probabilistic approaches to the CWprocedurecomposedoftheMonteCarlosimulation, cache, and splitting techniques for solving CVRP [, ], the two-phase selection and route postimprovement for solving CVRP [, ]. Based on our review, there are very few works availableintheliteraturestomodifythecwforsolving OVRP (only Bodin et al. []). Therefore, this is our major contribution to improve the CW solution by using our simple, efficient, and competitive approach. In the proposed CW, we have modified the parallel version of CW to deal with OVRPandhavecombinedthiswitharoutepostimprovement c i,j = (x i x j ) +(y i y j ), () where x i,y i and x j,y j are the geographical locations of customer i and j. Second, the savings value between customer i and j is calculated as s i,j =c,j c i,j, () where c,j is the traveling distance between depot and customer j and c i,j is the traveling distance between customer i and j. Equation() ismodifiedbybodinetal.[] from the Clarke-Wright formulation which is shown in (). After calculation, all savings values are collected in the savings list as follows: s i,j =c,i +c j, c i,j. () Third, the values in the savings list are sorted in decreasing order. Finally, the route merging procedure starts from the top of the savings list (the largest s i,j ). Both customers i and j will be combined into the same route if the total demand does not exceed vehicle capacity and no route constraints exist. Each condition for route constraints is described by three casesoffivecustomersasshowninfigure. InFigure (a), ( nor ) have already been assigned to a route (--). In Figure (b), exactly one of the two customers ( or ) has already been included in an existing route (--) and customer () is not interior to that route (a customer is interior to a route if it is not adjacent to the depot in the order of traversal of customers). In Figure (c), both customers ( and ) have already been included in two different existing routes (- and --5), and customer () is also interior to its route (--5). The route merging procedure is repeated until no feasible merging in the savings list is possible. Furthermore, in case of nonrouted customers, each is assigned by a route that starts at the depot, visits the unassigned customer, and returns to the depot. The proposed CW is an iterative improvement approach designed to find the global optimum solutions. It has been presented in four procedures consisting of Clarke-Wright formula modification, open-route construction, two-phase selection, and route postimprovement. The details of these procedures are shown below.

The Scientific World Journal General version Start Initialize basic input data such as vehicle capacity, customer demand, etc. Calculate savings list by Clarke-Wright formula modification procedure Sort savings list in the decreasing order Proposed version Calculate route merging Call open-route construction procedure Is better solution? No Yes Sort savings list by two-phase selection procedure Update new savings list No Is iteration over? Yes Call route postimprovement procedure End Figure : The flowchart of the proposed CW. 5 5 5 (a) (b) Figure : The route constraints of route merging procedure. (c)

The Scientific World Journal Solution = ---- Solution = ---- Value =+++=8 Value =+++=8 (a) Solution = --- Solution = --- Value =++=5 Value =++=7 (b) Figure : The two possible CVRP and OVRP solutions... The Clarke-Wright Formula Modification Procedure. Due to the latest improvement of CW for solving VRP that we mentioned in the literatures above, many authors proposed new parameters which we applied to this procedure. Gaskell [6] andyellow[7] presentedarouteshape(λ) parameter which controls the relative significance of direct arc between two customers. Their proposed savings formula is as follows: s i,j =c,i +c j, λc i,j. () According to (), we also modified () bydeletingc j, for solving OVRP with the following equation: s i,j =c,j λc i,j. (5) The parameter could be varied as studied by Altinel and Öncan [9]. They used a simple enumerative approach to produce 880 different solutions (λ [0., ], μ [0,], ] [0, ]). After that the best solution will be chosen. In order to avoid time-consuming iterative solutions, we have therefore applied only single parameter (λ) to this procedure... The Open-Route Construction Procedure. In order to solve CVRP by the CW after the route merging procedure, this procedure is needed to create the solution. Each route has toconstructforthecloseroute(hamiltoniancycle)byassigning the first customer who starts at the depot, and the last customer who returns to the depot. As shown in Figure (a), the Euclidean distance represented in this paper is symmetric. Therefore, the two possible CVRP solutions (---- and ----) are similar. In contrast, in OVRP each route has to construct for the open route (Hamiltonian path) by only assigning the first customer who starts at the depot. As shown in Figure (b), the two possible OVRP solutions (--- and ---) are very different. Consequently, in this procedure, the two possible OVRP solutions are constructed, and then thebestsolutionwillbeselectedastheovrpsolution... The Two-Phase Selection Procedure. After the CW solution is produced from a standard savings list generated by sorting the savings values in the decreasing order, this savings list will be regenerated as a new one by sorting the savings values randomly with probability. Pichpibul and Kawtummachai [] introduced the two-phase selection procedure for CVRP. In this paper, we adjust this procedure to deal with an OVRP by based on an operation of the genetic algorithm [8]. Figure 5(a)showsageneticrepresentation of chromosomes for ten savings values. The savings list is represented by one chromosome, and each gene represents the savings value between customers i and j. Inthefirst iteration, the chromosome is the savings list sorted by the decreasing order, but in next iteration the chromosome is the savings list derived from the best one. In Figure 5(b), we select one gene from the top four genes (T =) in the chromosome by fitness proportionate selection or roulette wheel selection. Here T is tournament size which is a random number between three and six. In order to create a roulette wheel, the selection probability (p n )andcumulativeprobability(q n )

The Scientific World Journal 5 Gene (n) 5 6 7 8 9 0 s i,j s,6 s,6 s, s,5 s, s,6 s 5,6 s,5 s, s,5 0.6 0.6.6.6 0.58 8.8 8.8 0.00 0.00.7 (a) n s i,j s,6 s,6 s, s,5 0.6 0.6.6.6 New gene (n) P n 0.8 0.8 0. 0. s i,j s,6 q n 0.8 0.56 0.78.00 0.6 Random spin (r) = 0.8 (b) T= Gene (n) 5 6 7 8 9 s i,j s,6 s, s,5 s, s,6 s 5,6 s,5 s, s,5 0.6.6.6 0.58 8.8 8.8 0.00 0.00.7 (c) n 5 6 s i,j s,6 s, s,5 s, s,6 s 5,6 0.6.6.6 0.58 8.8 8.8 New gene (n) P n 0. 0.7 0.7 0.6 0.5 0.5 s i,j s,6 s 5,6 q n 0. 0.8 0.55 0.7 0.86.0 0.6 8.8 (d) T=6 Random spin (r) = 0.90 New gene (n) 5 6 7 8 9 0 s i,j s,6 s 5,6 s,6 s, s,5 s, s,6 s,5 s, s,5 0.6 8.8 0.6.6.6 0.58 8.8 0.00 0.00.7 (e) Figure 5: The example of the two-phase selection procedure. with savings value (s n )foreachgene(n)arecalculatedusing the following equations: s n p n = i T s i q n = p i i n for n T, for n T. After that, we spin the wheel with a random number (r = 0.8) from the range between 0 and. The one savings value (s ) will be selected to be a gene of a new chromosome by considering r and q n.ifr q, then select the first savings value s ;otherwise,selectthenth savings value s n ( n T) (6) such that q n < r q n.theselectedgeneisremoved from the chromosome which leaves nine savings values as shown in Figure 5(c). Figure 5(d)showsthesameselection process in the next iteration with parameters T = 6 and r = 0.90. Therefore, this procedure will be executed until the last gene of the chromosome is selected to be a gene of the new chromosome which is shown in Figure 5(e). When the new chromosome which represents the new savings list is generated, it is calculated by the route merging procedureandtheopen-routeconstructionprocedureto produce a new OVRP solution. After we compare two solutions, the new chromosome will replace the previous chromosome only if the new solution is better than the

6 The Scientific World Journal previous one. This acceptance criterion is referred to a basic variable neighborhood search which is to accept only improvements [9]. Our approach is continued until the stopping criterion, which is the number of global iterations for two-phase selection, is satisfied... The Route Postimprovement Procedure. In order to find any further improvements for the best solution found when the stopping criterion of the two-phase selection procedure is satisfied, we have developed the route postimprovement procedure to generate different routes in our best solution. In order to explore the whole neighborhood of our best solution, we focus on the order of customers in single route called intraroute and multiple routes called interroute. The neighborhood structures that we used are several well-known move operators found in the works of Subramanian et al. [5] and Groër et al. [6] including shift moves (-0, -0, -0), swap moves (-, -, -), and λ-opt moves (λ {,}). The shift moves remove customers and insert them in another place. The swap moves select customers and exchange them. The λ-opt moves remove edges between customers and replace them with new edges. Our scheme is adapted from the local search strategy found in Çatay [0] by first applying a local neighborhood search to improve our best solution in each route. Then, a larger neighborhood search is applied across each pair between routes, respectively, by using all eight move operators with equal probability. This procedure is repeated until the stopping criterion, which is the number of consecutive iterations without any improvements in the best found solution, is satisfied.. Computational Results The proposed CW was coded in Visual Basic 6.0 on an Intel Corei7CPU860clockedat.80GHzwith.99GBofRAM under Windows XP platform. The numerical experiment used five well-known data sets of Euclidean benchmarks (composed of 6 instances) of the OVRP consisting of Augerat et al. [] in data sets A, B, and P, Christofides and Eilon []indatasete,andfisher[]indatasetf.theinput data is available online at http://www.branchandcut.org/ (last access /00). The best known solutions which are available online at http://www.hha.dk/ lys/ (last access 7/0) are obtained by a branch-and-cut algorithm from the work of Letchford et al. []. In our approach, some parameters have to be preset before the execution as shown in Table. Table describes the development of the proposed CW in detail. The benchmark problem sizes that we highlighted in this paper are classified as small-scale (less than 50 customers) andmedium-scale(between5to00customers)withdifferent features, for example, uniformly and not uniformly dispersed customers, clustered and not clustered, with a centered or not centered depot. All problems also include capacity constraints and minimum number of vehicles used restrictions. The first benchmark in data sets A, B, and P was proposed by Augerat et al. []. For the instances in data set A, both customer locations and demands are randomly generated. The customer locations in data set B are clustered Table : The parameters used in the proposed CW. Parameter The Clarke-Wright formula modification procedure Route shape (λ) The two-phase selection procedure Number of tournament sizes Number of global iterations for two-phase selection Values 0..0 (increment of 0.) 0 (random number) 5,000 The route postimprovement procedure Number of consecutive iterations without any improvements in the best found solution 500 Probability to select each move operator 0.5 Abbreviation CW- CW- CW- Table : The different features of the proposed CW. Details Improve CW solution with Clarke-Wright formula modification procedure Improve CW- solution with two-phase selection procedure Improve CW- solution with route post-improvement procedure instances. The modified version of other instances is data set P. In these data sets, the problem ranges in size from 6 to 69 customers including the depot. The second benchmark in data set E was proposed by Christofides and Eilon []. In this data set, the customers are randomly distributed in the plane and the depot is either in the center or near to it. The problem ranges in size from to 0 customers including the depot. The third benchmark in data set F is the real-life problem given by Fisher []. Instances F-n5-k and F-n5-k7 represent a day of grocery deliveries from the Peterboro and Bramalea, Ontario terminals, respectively, of National Grocers Limited. Instance F-n7-k represents the delivery of tires, batteries, and accessories to gasoline service stations. The depot is not centered in both instances. The problemrangesinsizefrom5to5customersincluding the depot. We discuss each benchmark problem in which the percentage improvement between CW solution (cws) and obtainedsolution(obs)iscalculatedasfollows: cws obs Percentage improvement =( ) 00. (7) cws Moreover, the percentage deviation between obtained solution (obs) and the best known solution (bks) is also calculated as follows: Percentage deviation =( obs bks ) 00. (8) bks The computational results for OVRP benchmark instances of Augerat et al.[], Christofides and Eilon [], and Fisher [] are reported in Tables 5. We do not only consider the

The Scientific World Journal 7 Table:ComputationalresultsbetweentheproposedCWandCWfordatasetsA,B,P,E,C,andF. Number Instance Solution Percentage improvement CW CW- λ CW- CW- CW- CW- CW- CPU time (seconds) A-n-k5 6.99 55.9.7 9.8 87..57 9.5 0.50.89 A-n-k5 96.6 58.5.5 5.5.5 7.669..5 5.9 A-n-k6 566.6 56.8.7 6.0 6..58 8.06 8. 5.89 A-n-k5 57.7 50.09. 59.5 508.7.966 5.07 7. 5.675 5 A-n6-k5 689.8 57.70.8 55. 59.6 7.08.7.660 6. 6 A-n7-k5 579.7 5.6.9 5. 86. 6.57.86 6.8 6.689 7 A-n7-k6 7.86 6.76.9 66.6 58.07.0.55 8.87 6.85 8 A-n8-k5 55.8 50.78.6 5.9 98.00 5.796.859 9.98 7.09 9 A-n9-k5 70.55 590.7. 55.80 59.68 6.58.5.980 7.568 0 A-n9-k6 70.55 558.86.9 57.57 5.07.9.007 6.09 7.9 A-n-k6 757.90 7.. 6.77 67.9.0 5. 8.50 0. A-n5-k6 7.97 6.56.0 76.5 68.67.7.5.7 0. A-n5-k7 89. 759.0.6 699.86 685.6 8. 5.6 7.95 0.75 A-n6-k7 78.00 65.. 59.57 58.5.569 9.570 0.90.59 5 A-n8-k7 789.9 756.58. 76.5 669.8.56 7.96 5.6.5 6 A-n5-k7 86.8 70.85.5 665.9 655.8.907 8.95 9.76 5.6 7 A-n5-k7 95.5 808..5 7.60 709.7 5.08.95 5.58 6. 8 A-n55-k9 80.06 76.9.9 696.5 669.06 8..58 6.58 7. 9 A-n6-k8 979.5 85.5. 85. 78.8.05 6.77 0.0.959 0 A-n65-k9 8.5 800.. 78. 78.59 5.05 7.7.70.5 A-n69-k9 9.87 798.6.0 77.7 757.76 5.96 7.998 9.6 8.600 Average percentage improvement of data set A 0.6.877 7.9.60 B-n-k5 8.68 67.00 0.9 6.80 6.7.7.9 5.6.65 B-n-k5 5. 506.6 0.9 59.59 58.76 6.80 5.0 5.5 5.556 B-n5-k5 599.6 595.5. 567. 557. 0.669 5. 6.98 5.98 B-n8-k6 500.6 8.. 50.7 5.6.95 9.97 0.989 7.09 5 B-n9-k5 8. 5.0..70.5 7.50.09 5.590 7.60 6 B-n-k6 59.5 507.5.8 9. 8.07 5.97 8.96 0.0 8.59 7 B-n-k6 8. 8.57.6.0 8.7 0.0 0.56.9 9. 8 B-n-k7 575.5 560.5.9 5.6 50..60 0.97.895 0. 9 B-n5-k5 60.7 5.7.9 509.56 88.07.788 5.5 8.887 0.57 0 B-n5-k6 59.90 0.8..5 0.8 6. 6.68.97 0.57 B-n50-k7 57. 9.9.9 6.07 7.5 8. 6.969 8.69.65 B-n5-k7 70.8 68..0 656.0 65..9 6.79.78.509 B-n5-k7 8.90 65.7.8 50.07.9.67 6.798 8.67.586 B-n56-k7 97.5 7..5 6.06 0.8.69 6.856 5.0 7.09 5 B-n6-k0 950.8 950.8.0 857.90 87.07 0.000 9.70.9.7 6 B-n6-k9 57. 5.70. 58.7 50.7 5.6.67 9.07.76 7 B-n68-k9 80.8 777.68. 758.69 70.7 6.57 8.6 5.50 7.87 Average percentage improvement of data set B.9 9.00.78.58 P-n6-k8 5.89 5.89 0. 5.06 5.06 0.000 0.5 0.5.7 P-n9-k 98.5 7.9.9 68.57 68.57.77.97.97.565 P-n0-k 0.0 8.50.9 70.8 70.8.7 8.98 8.98.7 P-n-k 09.9 80.7.8 68.5 6.88. 9.897.9.79

8 The Scientific World Journal Table : Continued. Number Instance Solution Percentage improvement CW CW- λ CW- CW- CW- CW- CW- CPU time (seconds) 5 P-n-k 06.00 8.58.8 7.6 67.9 0.88 6.765 8.80.888 6 P-n-k8 70.6 5.5.6 5. 5.87 6.679.89 6.588.0 7 P-n-k8 09.7 07.8.7 0.8 0.87 0.79.586.9.608 8 P-n0-k5 0.65 95.7.5 70.6 9.55 5.9.889 6.90 7.79 9 P-n5-k5 59.9..5 96.6 9.8.7.6.69 0.59 0 P-n50-k7 68. 7.69.5 0.56 97.8.6 5.98 5.67.7 P-n55-k7 5.5 6.90. 5.69.58 9.55.8 9.80 6.67 P-n55-k8 505.7 76..8..55 5.79.558 8. 6.9 P-n55-k0 555.5 50.56.6 88.65. 9.90.995 9.98 7.87 P-n60-k0 58.0 59..8 50.8 8.09 7.69.86 7.507 0.99 5 P-n65-k0 69. 59.5.6 5.57 5.50 8.77 8. 9.59 9.66 Average percentage improvement of data set P 7.5.86 5.058 9.6 E-n-k 86.9 60.6. 5.6 5.6 9.69.955.955.75 E-n-k 97.8 56.86..9.98 8. 0.68 0.90.56 E-n-k 6.0 576.9. 58.0 5.6 8.965 8.65 9.6. E-n5-k5 9.0 77.78.0 5.67 6.06.09 8.85 5.60 6.5 5 E-n76-k0 697.0 6.8.9 587.5 567. 7.97 5.76 8.65 68.97 6 E-n0-k8 807. 7.8.0 69.88 6.6 0.6.99 0. 8.87 Average percentage improvement of data set E 7.98.0 6.8.598 F-n5-k 65. 55.85.0 78.0 6.90.886.7.58 0.65 F-n7-k 08.9 9.8.0 87.67 77.00 8. 9.899 5.0.55 F-n5-k7 0. 856..9 86.0 775.80 7. 0.996.96.58 Average percentage improvement of data set F.70 7.707.507 55.059 Italic number indicates the infeasible solution (the number of vehicles used is inadequate). Table : Computational results between the proposed CW and MA for data sets A, P, and F. Number Instance Solution Percentage deviation Best Known MA Our CW MA Our CW A-n-k5 87. 87. 87. 0.000 0.000 A-n-k5.5.5.5 0.000 0.000 A-n-k6 6. 6. 6. 0.000 0.000 A-n-k5 508.7 508.5 508.7 0.068 0.000 5 A-n6-k5 59.6 59.6 59.6 0.000 0.000 6 A-n7-k5 86. 86. 86. 0.000 0.000 7 P-n9-k 68.57 68.57 68.57 0.000 0.000 8 P-n0-k 70.8 70.8 70.8 0.000 0.000 9 P-n-k 6.88 6.88 6.88 0.000 0.000 0 P-n-k 67.9 67.9 67.9 0.000 0.000 P-n0-k5 9.55 9.55 9.55 0.000 0.000 P-n5-k5 9.8 9.8 9.8 0.000 0.000 P-n50-k7 97.8 07.7 97.8.605 0.000 F-n5-k 6.90 6.90 6.90 0.000 0.000 5 F-n7-k 77.00 77.5 77.00 0.57 0.000 Average percentage deviation of data sets A, P, and F 0.95 0.000

The Scientific World Journal 9 Table 5: Computational results between the proposed CW and the other algorithms for data sets E and F. Number Instance Solution Best Known B PR LGW FOH Our CW E-n5-k5 6.06 6. 6.06 6.06 6.06 6.06 E-n76-k0 567. 57.5 567. 567. 567. 567. E-n0-k8 69.7 6.6 6.76 69.7 69.7 6.6 F-n7-k 77.00 77. 77.00 77.00 78.09 77.00 5 F-n5-k7 769.66 78. 770.7 769.66 769.66 775.80 Number Instance CPU time (seconds) B PR LGW FOH Our CW E-n5-k5 88.8 0 6. 7.6 6.5 E-n76-k0 67.5 50. 9.0 68.97 E-n0-k8 5. 80 9.5 9.6 8.87 F-n7-k 98. 00 9.5 0..55 5 F-n5-k7 000. 590 58. 7.5.58 Bold number indicates the best known solution was obtained. improvements of our solutions over CW solutions as shown in Table, but also show the performance of our solutions by comparing the proposed CW with the algorithms for OVRPasshowninTables and 5 by using the following abbreviations:maformirhassaniandabolghasemi[], B for Brandão [], PR for Pisinger and Ropke [5], LGW for Li et al. [6] and FOH for Fleszar et al. [0]. Results from Tables 5 indicate that the proposed CW can find high quality solutions within reasonable time, especially for small and medium scale problems. Out of 6 problems, we find the optimal solutions for 60 problems with up to customers. For two problems with 00 customers, the percentage deviations between our solutions and the optimal solutions are very low (E-n0-k8 and F-n5-k7). Nevertheless, in thoseoptimalandnearoptimalsolutions,therearefour problems (A-n-k5, P-n50-k7, F-n7-k, and E-n76-k0) for which our results in Tables and 5 are better than the others. Moreover, our results in Table show that the proposed CW always performed better than CW. These indicate that the proposed CW is effective and efficient in producing high quality solutions for well-known benchmark problems. The important details of our improvement are discussed below. The average percentage improvements between CW solutions and our solutions for benchmark of data sets A, B, P, E, and F are presented in Table.WehavefoundthatCWsolutions were improved by the average of 5.7%. This finding shows that data set F has the highest average improvement and data set B has the lowest average deviation. The greatest improvements of CW solutions are, respectively, presented in top three instances including A-n9-k6 (6.09%), A-n5-k7 (5.58%), and F-n5-k7 (.96%). We can conclude that theproblemswhichhavethefeatureslikeclusteredcustomers canbesolvedbycwbetterthantheproblemswhichhave the features like dispersed customers. In addition, the results show that the proposed CW can solve both above-mentioned problems to obtain optimal or near optimal solutions. The performance of route postimprovement procedure can be described as the reduction of the percentage deviation between CW- and CW- solutions. The average reduction of thepercentagedeviationbetweencw-solutionsandcw- solutions for benchmark of data sets A, B, P, E, and F are.65,.857,.98,.699,and.806.wehavefoundthatcw- solutions were reduced by the average of.88%. The greatest reductions of CW- solutions are, respectively, presented in top three instances including B-n6-k9 (.769%), P-n50-k7 (0.866%), and A-n5-k6 (0.60%). According to Table, some CW- solutions can be reduced to obtain the optimal solutions. Another finding from our work is the infeasible solutions produced by CW in which the number of vehicles used is inadequate. This finding is referred to Vigo [7]that CW does not allow for the control of the number of routes of the final solution. The solution found for a given instance can, in fact, requiremorethank routes to serve all the customers, hence being infeasible.. Conclusions In this paper, we have presented a new heuristic approach basedonclarke-wrightalgorithmtosolvetheopenvehicle routing problem (OVRP). We have modified the Clarke- Wright algorithm with three procedures composed of Clarke- Wright formula modification, open-route construction, and two-phaseselectionandhavecombinedthemwitharoute postimprovement procedure in which the neighborhood structures composed of shift, swap, and λ-opt move operators are used to improve our best solution. We also have done experiments using six well-known data sets of OVRP (composed of 6 instances) obtained from the literatures. The numerical results show that our approach is competitive and our solutions outperform Clarke and Wright [] in all directions. Moreover, it also generates the best known solutions in 97% of all instances (60 out of 6). During the development of our approach, we have mentioned the ideas related to the CW that deserve more attention in further studies. Consequently, it may be interesting to develop a more powerful postimprovement procedure. An additional study is to extend the proposed CW to deal with

0 The Scientific World Journal othervariantsofthestudiedproblemssuchassimultaneous pickup and delivery (VRPSPD) or time windows (VRPTW). References [] D. Sariklis and S. Powell, A heuristic method for the open vehicle routing problem, The Journal of the Operational Research Society,vol.5,no.5,pp.56 57,000. [] Z. Fu and M. Wright, Train plan model for British rail freight services through the channel tunnel, TheJournalofthe Operational Research Society,vol.5,no.,pp.8 9,99. [] L.Y.O.LiandZ.Fu, Theschoolbusroutingproblem:acase study, The Journal of the Operational Research Society, vol. 5, no. 5, pp. 55 558, 00. [] C. D. Tarantilis and C. T. Kiranoudis, Distribution of fresh meat, Journal of Food Engineering,vol.5, no.,pp.85 9, 00. [5]R.Russell,W.-C.Chiang,andD.Zepeda, Integratingmultiproduct production and distribution in newspaper logistics, Computers & Operations Research,vol.5,no.5,pp.576 588, 008. [6]W.-C.Chiang,R.Russell,X.Xu,andD.Zepeda, Asimulation/metaheuristic approach to newspaper production and distribution supply chain problems, International Journal of Production Economics,vol.,no.,pp.75 767,009. [7] P. P. Repoussis, D. C. Paraskevopoulos, G. Zobolas, C. D. Tarantilis, and G. Ioannou, A web-based decision support system for waste lube oils collection and recycling, European Journal of Operational Research, vol.95,no.,pp.676 700, 009. [8] S. Yu, C. Ding, and K. Zhu, A hybrid GA-TS algorithm for open vehicle routing optimization of coal mines material, Expert Systems with Applications,vol.8,no.8,pp.0568 057,0. [9] N. Christofides, Worst-case analysis of a new heuristic for the travelling salesman problem, Tech. Rep. 88, Graduate School of Industrial Administration, Canegie Mellon University, 976. [0] M. Held and R. M. Karp, The traveling salesman problem and minimum spanning trees, Operations Research,vol.8,pp. 8 6, 970. [] R. E. Miller and J. W. Thatcher, Complexity of Computer Computations,PlenumPress,NewYork,NY,USA,97. [] M. Syslo, N. Deo, and J. Kowaklik, Discrete Optimization Algorithms with Pascal Programs, Prentice Hall, Englewood Cliffs, NJ, USA, 98. [] P. Toth and D. Vigo, The Vehicle Routing Problem, SIAM Monographs on Discrete Mathematics and Applications, SIAM Publishing, Philadelphia, Pa, USA, 00. [] J. Brandão, A tabu search algorithm for the open vehicle routing problem, EuropeanJournalofOperationalResearch,vol.57,no., pp. 55 56, 00. [5] M. Salari, P. Toth, and A. Tramontani, An ILP improvement procedure for the Open Vehicle Routing Problem, Computers & Operations Research,vol.7,no.,pp.06 0,00. [6] G. B. Dantzig and J. H. Ramser, The truck dispatching problem, Management Science,vol.6,pp.80 9, 959. [7] C.D.Tarantilis,D.Diakoulaki,andC.T.Kiranoudis, Combination of geographical information system and efficient routing algorithms for real life distribution operations, European Journal of Operational Research,vol.5,no.,pp.7 5,00. [8] Z.Fu,R.Eglese,andL.Li, Anewtabusearchheuristicforthe open vehicle routing problem, The Journal of the Operational Research Society,vol.56,no.,pp.67 7,005. [9] X. Li and P. Tian, An ant colony system for the open vehicle routing problem, in Ant Colony Optimization and Swarm Intelligence, M. Dorigo, L. M. Gambardella, M. Birattari, A. Martinoli, R. Poli, and T. Stützle,Eds.,vol.50ofLecture Notes in Computer Science, pp. 56 6, Springer, Berlin, Germany, 006. [0] K. Fleszar, I. H. Osman, and K. S. Hindi, A variable neighbourhood search algorithm for the open vehicle routing problem, EuropeanJournalofOperationalResearch,vol.95,no.,pp. 80 809, 009. [] S. A. Mirhassani and N. Abolghasemi, A particle swarm optimization algorithm for open vehicle routing problem, Expert Systems with Applications,vol.8,no.9,pp.57 55, 0. [] G. Clarke and J. W. Wright, Scheduling of vehicles from a central depot to a number of delivery points, Operations Research,vol.,pp.568 58,96. [] F. A. Tillman, The multiple terminal delivery problem with probabilistic demands, Transportation Science, vol., pp. 9 0, 969. [] L. Bodin, B. Golden, A. Assad, and M. Ball, Routing and scheduling of vehicles and crews. The state of the art, Computers & Operations Research,vol.0,no.,pp.6,98. [5] I. Deif and L. D. Bodin, Extension of the Clarke-Wright algorithm for solving the vehicle routing problem with backhauling, in ProceedingsoftheBabsonConferenceonSoftware Uses in Transportation and Logistics Management, A. E. Kidder, Ed., pp. 75 96, Babson Park, Mass, USA, 98. [6] T. J. Gaskell, Bases for vehicle fleet scheduling, Operational Research Quarterly,vol.8,pp.8 95,967. [7] P. Yellow, A computational modification to the savings method of vehicle scheduling, Operational Research Quarterly, vol., pp. 8 8, 970. [8] H. Paessens, The savings algorithm for the vehicle routing problem, European Journal of Operational Research,vol., no., pp. 6, 988. [9] İ. K. Altinel and T. Öncan, A new enhancement of the Clarke and Wright savings heuristic for the capacitated vehicle routing problem, The Journal of the Operational Research Society, vol. 56,no.8,pp.95 96,005. [0] T. Doyuran and B. Çatay, A robust enhancement to the Clarke- Wright savings algorithm, The Journal of the Operational Research Society, vol. 6, no., pp., 0. [] A. A. Juan, J. Faulin, R. Ruiz, B. Barrios, and S. Caballé, The SR-GCWS hybrid algorithm for solving the capacitated vehicle routing problem, Applied Soft Computing Journal, vol. 0, no., pp.5,00. [] A. A. Juan, J. Faulin, J. Jorba, D. Riera, D. Masip, and B. Barrios, OntheuseofMonteCarlosimulation,cacheand splitting techniques to improve the Clarke and Wright savings heuristics, The Journal of the Operational Research Society, vol. 6,no.6,pp.085 097,0. [] T. Pichpibul and R. Kawtummachai, New enhancement for Clarke-Wright savings algorithm to optimize the capacitated vehicle routing problem, European Journal of Scientific Research,vol.78,pp.9,0. [] T.PichpibulandR.Kawtummachai, AnimprovedClarkeand Wright savings algorithm for the capacitated vehicle routing problem, ScienceAsia, vol. 8, pp. 07 8, 0. [5] A. Subramanian, L. M. A. Drummond, C. Bentes, L. S. Ochi, and R. Farias, A parallel heuristic for the vehicle routing

The Scientific World Journal problem with simultaneous pickup and delivery, Computers & Operations Research, vol. 7, no., pp. 899 9, 00. [6] C. Groër, B. Golden, and E. Wasil, A library of local search heuristics for the vehicle routing problem, Mathematical Programming Computation,vol.,no.,pp.79 0,00. [7] G. Laporte, M. Gendreau, J. Potvin, and F. Semet, Classical and modern heuristics for the vehicle routing problem, International Transactions in Operational Research,vol.7,no.-5,pp. 85 00, 000. [8] M. Gen and R. Cheng, Genetic Algorithms and Engineering Design, John Wiley & Sons, New York, NY, USA, 997. [9] V.C.Hemmelmayr,K.F.Doerner,andR.F.Hartl, Avariable neighborhood search heuristic for periodic routing problems, European Journal of Operational Research, vol.95,no.,pp. 79 80, 009. [0] B. Çatay, A new saving-based ant algorithm for the vehicle routing problem with simultaneous pickup and delivery, Expert Systems with Applications,vol.7,no.0,pp.6809 687,00. [] P. Augerat, J. Belenguer, E. Benavent, A. Corbern, D. Naddef, and G. Rinaldi, Computational results with a branch and cut code for the capacitated vehicle routing problem, Research Report 99-M, Universite Joseph Fourier, Grenoble, France, 995. [] N. Christofides and S. Eilon, An algorithm for the vehicle dispatching problem, Operational Research Quarterly, vol. 0, pp.09 8,969. [] M. L. Fisher, Optimal solution of vehicle routing problems using minimum K-trees, Operations Research,vol.,no.,pp. 66 6, 99. [] A. N. Letchford, J. Lysgaard, and R. W. Eglese, A branch-andcut algorithm for the capacitated open vehicle routing problem, The Journal of the Operational Research Society, vol.58,no., pp. 6 65, 007. [5] D. Pisinger and S. Ropke, A general heuristic for vehicle routing problems, Computers & Operations Research, vol., no. 8, pp. 0 5, 007. [6] F. Li, B. Golden, and E. Wasil, The open vehicle routing problem: algorithms, large-scale test problems, and computational results, Computers & Operations Research, vol., no. 0, pp. 98 90, 007. [7] D. Vigo, Heuristic algorithm for the asymmetric capacitated vehicle routing problem, European Journal of Operational Research,vol.89,no.,pp.08 6,996.

Journal of Industrial Engineering Multimedia The Scientific World Journal Applied Computational Intelligence and Soft Computing International Journal of Distributed Sensor Networks Fuzzy Systems Modelling & Simulation in Engineering Submit your manuscripts at Journal of Computer Networks and Communications Artificial Intelligence International Journal of Biomedical Imaging Artificial Neural Systems International Journal of Computer Engineering Computer Games Technology Software Engineering International Journal of Reconfigurable Computing Robotics Computational Intelligence and Neuroscience Human-Computer Interaction Journal of Journal of Electrical and Computer Engineering