Combining DC-programming and steepest-descent to solve the single-vehicle inventory routing problem

The single-vehicle cyclic inventory routing problem (SV-CIRP) is concerned with a repeated distribution of a product from a single depot to a selected subset of retailers having stable demands. If a retailer is selected for replenishment, the supplier collects a retailer-related fixed reward. The objective is to determine the subset of retailers to cyclically replenish, the quantities to be delivered to each, and to design the vehicle delivery routes so that the expected total distribution and inventory cost is minimized while the total collected rewards from the selected retailers is maximized. The resulting distribution plan must prevent stockouts from occurring at each retailer. In this paper, the underlying optimization problem for the SV-CIRP is formulated as a mixed-integer program with linear constraints and a nonlinear objective function. An optimization approach combining DC-programming and Branch-and-Bound within a steepest descent hybrid algorithm, denoted by DCA-SDHA, is developed for its solution. The approach is tested on some randomly generated problems and the obtained results are compared with results from the standard steepest descent hybrid algorithm (SDHA). These encouraging results show that the proposed approach is indeed computationally more effective and is worth being further investigated for the solution of medium to large instances.     

Integrating order acceptance decisions with flexible due dates in a production planning model with load-dependent lead times

We consider a tactical planning problem, which integrates production planning decisions together with order acceptance decisions, while taking into account the dependency between workload and lead times. The proposed model determines which orders to accept and in which period they should be produced, so that they can be delivered to the customer within the acceptable flexible due dates. When the number of accepted orders increases, the workload and production lead time also increase, and this may result in the possibility of missing customer due dates. This problem is formulated as a mixed integer linear programme for which two relax-and-fix heuristic solution methods are proposed. The first one decomposes the problem based on time periods, while the second decomposes it based on orders. The performances of these heuristics are compared with that of a state-of-the-art commercial solver. Our results show that the time-based relax-and-fix heuristic outperforms the order-based relax-and-fix heuristic, and the solver solution as it yields better integrality gaps for much less CPU effort.     

Measuring complexity in mixed-model assembly workstations

In an effort to maintain or increase their market share and at the same time prevent costs from escalating, manufacturing organisations are increasingly using their current manufacturing system to produce custom output. As a consequence, the large number of product variants increases significantly the complexity of manufacturing systems, both for the operators as for the support services. This is especially true in automotive industry, where customisation is increasing at a rapid pace. To counter the ensuing loss of productivity, a more fundamental approach to dealing with this complexity in manufacturing processes is required. In order to investigate the impact of complexity on production performance, one must first delineate the concept and then identify as unambiguously as possible highly complex workstations. This article defines complexity at the workstation level and proposes a complexity measure for mixed-model assembly workstations. Based on data from several leading automotive companies from Belgium and Sweden, some statistical models are proposed to characterise workstations complexity. The models are described and their validity and accuracy are discussed.     

Models for robust tactical planning in multi-stage production systems with uncertain demands

We consider the problem of designing robust tactical production plans, in a multi-stage production system, when the periodic demands of the finished products are uncertain. First, we discuss the concept of robustness in tactical production planning and how we intend to approach it. We then present and discuss three models to generate robust tactical plans when the finished-product demands are stochastic with known distributions. In particular, we discuss plans produced, respectively, by a two-stage stochastic planning model, by a robust stochastic optimization planning model, and by an equivalent deterministic planning model which integrates the variability of the finished-product demands. The third model uses finished-product average demands as minimal requirements to satisfy, and seeks to offset the effect of demand variability through the use of planned capacity cushion levels at each stage of the production system. An experimental study is carried out to compare the performances of the plans produced by the three models to determine how each one achieves robustness. The main result is that the proposed robust deterministic model produces plans that achieve better trade-offs between minimum average cost and minimum cost variability. Moreover, the required computational time and space are by far less important in the proposed robust deterministic model compared to the two others.     

Integrated production quality and condition-based maintenance optimisation for a stochastically deteriorating manufacturing system

This paper investigates the problem of optimally integrating production quality and condition-based maintenance in a stochastically deteriorating single- product, single-machine production system. Inspections are periodically performed on the system to assess its actual degradation status. The system is considered to be in ‘fail mode’ whenever its degradation level exceeds a predetermined threshold. The proportion of non-conforming items, those that are produced during the time interval where the degradation is beyond the specification threshold, are replaced either via overtime production or spot market purchases. To optimise preventive maintenance costs and at the same time reduce production of non-conforming items, the degradation of the system must be optimally monitored so that preventive maintenance is carried out at appropriate time intervals. In this paper, an integrated optimisation model is developed to determine the optimal inspection cycle and the degradation threshold level, beyond which preventive maintenance should be carried out, while minimising the sum of inspection and maintenance costs, in addition to the production of non-conforming items and inventory costs. An expression for the total expected cost rate over an infinite time horizon is developed and solution method for the resulting model is discussed. Numerical experiments are provided to illustrate the proposed approach.     

Integrated production planning and preventive maintenance in deteriorating production systems

This paper discusses the issue of integrating production planning and preventive maintenance in manufacturing production systems. In particular, it tackles the problem of integrating production and preventive maintenance in a system composed of parallel failure-prone production lines. It is assumed that when a production line fails, a minimal repair is carried out to restore it to an ‘as-bad-as-old’ status. Preventive maintenance is carried out, periodically at the discretion of the decision maker, to restore the production line to an ‘as-good-as-new’ status. It is also assumed that any maintenance action, performed on a production line in a given period, reduces the available production capacity on the line during that period. The resulting integrated production and maintenance planning problem is modeled as a nonlinear mixed-integer program when each production line implements a cyclic preventive maintenance policy. When noncyclical preventive maintenance policies are allowed, the problem is modeled as a linear mixed-integer program. A Lagrangian-based heuristic procedure for the solution of the first planning model is proposed and discussed. Computational experiments are carried out to analyze the performance of the method for different failure rate distributions, and the obtained results are discussed in detail.     

Analysing the effectiveness of vendor-managed inventory in a single-warehouse,multiple-retailer system

This paper considers a two-stage supply chain, consisting of a single warehouse and multiple retailers facing deterministic demands, under a vendor-managed inventory (VMI) policy. It presents a two-phase optimisation approach for coordinating the shipments in this VMI system. The first phase uses direct shipping from the supplier to all retailers to minimise the overall inventory costs. Then, in the second phase, the retailers are clustered using a construction heuristic in order to optimise the transportation costs while satisfying some additional restrictions. The improvement of the system's performance through coordinated VMI replenishments against the system with direct shipping only is shown and discussed in the comparative analysis section.     

Analysis of the single-vehicle cyclic inventory routing problem

The single-vehicle cyclic inventory routing problem (SV-CIRP) consists of a repetitive distribution of a product from a single depot to a selected subset of customers. For each customer, selected for replenishments, the supplier collects a corresponding fixed reward. The objective is to determine the subset of customers to replenish, the quantity of the product to be delivered to each and to design the vehicle route so that the resulting profit (difference between the total reward and the total logistical cost) is maximised while preventing stockouts at each of the selected customers. This problem appears often as a sub-problem in many logistical problems. In this article, the SV-CIRP is formulated as a mixed-integer program with a nonlinear objective function. After a thorough analysis of the structure of the problem and its features, an exact algorithm for its solution is proposed. This exact algorithm requires only solutions of linear mixed-integer programs. Values of a savings-based heuristic for this problem are compared to the optimal values obtained for a set of some test problems. In general, the gap may get as large as 25%, which justifies the effort to continue exploring and developing exact and approximation algorithms for the SV-CIRP.     

Optimization of the integrated fleet-level imperfect selective maintenance and repairpersons assignment problem

Journal of Intelligent Manufacturing - Industrial environments such as manufacturing and transportation industries usually involve fleets of identical systems that must carry out several missions...     

The two-stage recombination operator and its application to the multiobjective 0/1 knapsack problem: A comparative study

In this paper, we first propose a new recombination operator called the two-stage recombination and then we test its performance in the context of the multiobjective 0/1 knapsack problem (MOKP). The proposed recombination operator generates only one offspring solution from a selected pair of parents according to the following two stages. In the first stage, called genetic shared-information stage or similarity-preserving stage, the generated offspring inherits all parent similar genes (i.e., genes or decision variables having the same positions and the same values in both parents). In the second stage, called problem fitness-information stage, the parent non-similar genes (i.e., genes or decision variables having the same positions but different values regarding the two parents) are selected from one of the two parents using some fitness information. Initially, we propose two different approaches for the second stage: the general version and the restricted version. However, the application of the restricted version to the MOKP leads to an improved version which is more specific to this problem. The general and the MOKP-specific versions of the two-stage recombination are compared against three traditional crossovers using two well-known multiobjective evolutionary algorithms. Promising results are obtained. We also provide a comparison between the general version and the MOKP-specific version.