Sequential Meta-Heuristic Approach for Solving Large-Scale Ready-Mixed Concrete–Dispatching Problems

Abstract

Finding a practical solution for the allocation of resources in ready-mixed concrete (RMC) is a challenging issue. In the literature, heuristic methods have been mostly used for solving the RMC problem. The introduced methods are intended to find a solution in one stage but the amount of infeasible allocations in their initial solutions is their main challenge, as these infeasible solutions need postprocessing efforts. This paper introduces a sequential heuristic method that can solve RMC problems in two separate stages without any need for postprocessing. It was found that the depot-allocation problem is more complicated than truck allocation and the combination of these two subproblems threatens the efficiency of the solution. Another contribution of this paper is proposing a new formulation for minimizing the number of trucks. A genetic algorithm (GA) has been selected for implementing the proposed idea and for evaluating the large-scale data-set model. The data set covers an active RMC for a period of 1 month. The comprehensive tests show that sequential GA is more robust than traditional GA when it converges 10 times faster with achieved solution at 30% less cost.

Simulation Model for Scenario Optimization of the Ready-Mix Concrete Delivery Problem

“…Conclusion of the aforementioned research is that by combination of the exact solution methods and Variable Neighborhood Search approach (VNS) is applicable for solving medium sized real-time instances of concrete delivery problems. Authors Maghrebi et al [16,17] suggested robust heuristic method and sequential heuristic method for solving large-scale RMC problems using the genetic algorithms (GA) for successful solving large scale ready-mix concrete delivery problems. Same authors in their article [18] tested the dispatchers' accuracy for solving addressed problems comparing them to mathematical optimization models which showed high accuracy of experts' decisions, but confirmed the need of optimization tool as well.…”

Column Generation‐Based Approach for Solving Large‐Scale Ready Mixed Concrete Delivery Dispatching Problems

“…(), Maghrebi et al. (), and Srichandum and Rujirayanyong (). Despite developments in this area, the solution structure among most introduced methods is quite similar, especially in the genetic algorithm (GA)‐based method where the chromosome structure consists of two merged parts: the first part defines the sources of deliveries; the second part expresses the priorities of customers.…”

“…It has been proved that an RMC optimization problem is an NP-hard problem (Yan et al, 2008;Asbach et al, 2009). Therefore, to deal with this problem, heuristic methods have been widely used in the literature such as Cao et al (2004), Feng and Wu (2006), Garcia et al (2002), Maghrebi et al (2013b), Maghrebi et al (2014d), and Srichandum and Rujirayanyong (2010). Despite developments in this area, the solution structure among most introduced methods is quite similar, especially in the genetic algorithm (GA)-based method where the chromosome structure consists of two merged parts: the first part defines the sources of deliveries; the second part expresses the priorities of customers.…”

Using Lagrangian Relaxation to Solve Ready Mixed Concrete Dispatching Problems

“…While this approach is shown to reduce the number of side constraints in the obtained MIP, the number of decision variables may increase significantly. This decomposition approach was subsequently used along with different solution methods for the multidepot RMCDP (13)(14)(15)(16)(17)(18). The present paper builds on this research and introduces a novel solution method based on Lagrangian relaxation.…”

“…To determine the values of the Lagrange multipliers, they are initialized with the marginal values (dual) of the LP relaxation of the initial MIP defined by Equations 4 to 15. A generic subgradient optimization algorithm is then used to iteratively update the values of the Lagrange multipliers (17). Convergence of the sub gradient algorithm is achieved when the maximal relative gap between two consecutive values of the Lagrange multipliers is lower than a predefined value.…”