Search

This paper develops a vendor-managed inventory (VMI) model for a multiple-vendor, single-manufacturer supply chain, in which the first stage members can be traders and/or producers and the second stage member is a manufacturer. The model utilizes a realistic transportation cost which is dependent on the sizes (small- or medium-sized) of trucks. It can determine suitable sizes and numbers of trucks that minimize the transportation cost. A genetic algorithm (GA) technique, implemented in MATLAB software, is used to determine the best solution to the problem. A case study in the instant noodle industry is conducted to demonstrate the usefulness of the proposed model. Based on the experimental results, the VMI model has reasonable behaviors using sensitivity analysis. To reduce the inventory level of raw materials, the penalty cost may be set at a relatively high level or the upper inventory limits may be set at relatively low levels, without significantly affecting the average total cost per period of the entire supply chain. When the vendors are allowed to make decision independently, the solution is still the same as the solution from the proposed VMI model, which means that the manufacture does not take advantage of the vendors.

This paper develops a two-phase LP-based heuristic for the Capacitated Vehicle Routing Problem (CVRP). It considers three objectives: (1) minimizing the total costs of fuel consumption and overtime, (2) maximizing the total personal relationships between customers and drivers, and (3) balancing the delivery weights of vehicles. The two-phase LP-based heuristic (cluster-first route-second) is proposed. First, in the clustering stage, three LP-based clustering models (denoted by C1, C2, and C3) are developed. Customers are grouped into clusters based on real distances between the customers for C1, personal relationships between the customers and drivers for C2, and the delivery weights of vehicles for C3. Second, in the routing stage, an LP-based traveling salesman problem model is used to form a route for each cluster, to minimize the total costs of fuel consumption and overtime labor. The experimental results from a case study of Thai SMEs show that when the C2 clustering model is applied, the performances are the best. Significant contributions of this paper include: (1) it is an original paper that proposes the C2 clustering model, and it has the best performances based on the experimental results, and (2) the proposed two-phase LP-based heuristic methods are suitable for practical use by SMEs since the required computational time is short, and it has multiple models with different objectives that can be selected to match a user's requirements.

This paper proposes an algorithm implemented in a web application to find feasible route for planning in truck transportation. The logistic route planning problem has attracted a lot of effort, however, most of the earlier researches are not be directly applicable to this problem because practical logistic constraints such as the width of the road or height of the tunnel are not considered. Furthermore, drivers’ basic needs in taking the time off for relaxation and lunch are involved into the delivery time computation as well. The web application has an ample amount of trials which were compared to a database of transportation in real-time. The results from a case study of Thai Adhesive and Sealant Company show a significant improvement, approximately about thirty-five percent, in forecast preciseness while the mistake rate is dropped. Moreover, this application can be applied to other industries that require feasible trucking logistics planning.

In new product development, identifying a product concept (criterion) and a market segment is a critically important issue. To do so, it is necessary to deal with uncertainties in an evaluation process. In this paper, we introduce (a) an alternative function principle based approach for coping with inter-relation uncertainties in human criteria assessment and heterogeneity of decision makers, in determining criteria weights for a multiple criteria group decision making problem (b) an alternative intersected area based approach in segmenting consumers for new product development. A case study is used to illustrate the discussed approaches. The proposed approaches are helpful to alleviate the tediousness of the mathematical operations and also make the fuzzy linguistic approach for a multiple criteria group decision making problem more convenient and practical to be used.

In practice, companies with dominant power influence partner operations for their benefits. This paper proposes the supply chain (SC) planning models that determine the optimal sale, pricing, production, and inventory plans of each SC member based on the dominant policy. The theoretical contributions of this paper include a development of linear mathematical functions and explicitly considering the inventory turnover as one of the business performance measures in addition to profit and revenue. The fuzzy multi-objective linear programming models (FMOLP) under dominant systems are developed. A case study in plastic fashion accessory industry in Thailand is performed. The performance measures are translated into satisfaction levels and plotted in a radar chart with an area of possibility, which clearly shows the sensitivity degree of uncertainty. Practical contribution is that this paper recommends suitable pricing and inventory plans based on the dominant policy. [Submitted: 15 August 2018; Accepted: 18 March 2019]

Practitioners have generally performed aggregate production planning and marketing promotion planning separately. However, there is a link between them; demand, which is used as input data in aggregate production planning, is always affected by marketing promotion. Therefore, it is beneficial to determine both plans simultaneously, which may result in higher profit for the organization. The objective of this paper is to provide a general optimization model that can help making decision in marketing promotion in accordance with production planning in order to maximize the profit for the company. The performance of the model is evaluated by a real application in a consumer product factory in Thailand.

This paper proposes a new inventory control system called the inventory/distribution plan (IDP) control system for a one-warehouse/multi-retailer supply chain. In the IDP control system, a proposed mixed-integer linear programming model is solved to determine an optimal IDP that controls the inventories of the supply chain. The efficiency of the IDP control system is compared to that of the echelon-stock R , s , S control policy, where R is a periodic review interval, s is a reorder point, and S is an order-up-to level, at various fill rates. The experimental results show that when the system faces non-stationary demands, the IDP control system significantly outperforms the echelon-stock R , s , S control system because it can give lower total costs for all ranges of fill rates.

CNC end milling is a widely used cutting operation to produce surfaces with various profiles. The manufactured parts’ quality not only depends on their geometries but also on their surface texture, such as roughness. To meet the roughness specification, the selection of values for cutting conditions, such as feed rate, spindle speed, and depth of cut, is traditionally conducted by trial and error, experience, and machining handbooks. Such empirical processing is time consuming and laborious. Therefore, a combined approach for determining optimal cutting conditions for the desired surface roughness in end milling is clearly needed. The proposed methodology consists of two parts: roughness modeling and optimal cutting parameters selection. First, a machine learning technique called support vector machines (SVMs) is proposed for the first time to capture characteristics of roughness and its factors. This is possible due to the superior properties of well generalization and global optimum of SVMs. Next, they are incorporated in an optimization problem so that a relatively new, effective, and efficient optimization algorithm, particle swarm optimization (PSO), can be applied to find optimum process parameters. The cooperation between both techniques can achieve the desired surface roughness and also maximize productivity simultaneously.

University timetabling is an NP-hard problem, which means that the amount of computation required to find solutions increases exponentially with problem size. Timetabling is subject to hard constraints that must be satisfied in order to produce feasible timetables and soft constraints, which are not absolutely essential. This paper describes the Stochastic Optimisation Timetabling Tool (SOTT) that has been developed for university course timetabling. Genetic Algorithms (GA), Simulated Annealing (SA) and random search are embedded in the SOTT. The algorithms include a repair process, which ensures that all infeasible timetables are rectified. This prevents clashes and ensures that the rooms are sufficiently large to accommodate the classes. The algorithms also evaluate timetables in terms of soft constraints: minimising student movement; avoiding fragmentation in the timetables for students and lecturers; and satisfying lecturers’ preferences for the timing of classes. The algorithms were tested using two sets of timetabling data from a collaborating university. Both GA and SA produced very good timetables, but the results obtained from SA were slightly better than those using GA. However, the GA was 54% faster than SA.

This paper deals with the problem of Supplier Selection and Order Allocation (SSOA) in a fuzzy sense. The demand and delivery lead time are treated as fuzzy numbers. The fuzzy number is first transformed into interval numbers. After doing some arithmetic operations, those fuzzy interval numbers are defuzzified to a crisp quantity. This crisp quantity is further used as an input parameter in the model. Essentially, the main approach in this paper is based on the function principle and the pascal triangular graded mean approach. The SSOA problem is constructed as a Multiple Criteria Decision Making (MCDM) problem aiming to optimize the order quantities placed to many suppliers. The problem is solved by a fuzzy linear programming technique. A numerical example is also given for the illustration of the discussed issues.

This paper aims to develop a finite capacity material requirement planning (FCMRP) system based on TOC philosophy (TOC-MRP) for multi-stage assembly factory that has some bottleneck stations. The proposed TOC-MRP system tries to load and schedule operations on bottleneck stations in a manner that they are free of idle time and overtime. The schedules on non-bottleneck stations will be arranged until they are not conflicting with those on the bottleneck stations. The non-bottleneck stations are allowed to have idle time and overtime if necessary. To analyse whether TOC-MRP is effective, it is compared with a FCMRP method that does not adopt TOC philosophy. The experimental results reveal that the TOC-MRP outperforms the FCMRP without TOC philosophy.

Since the design of supply chain network affects supply chain performance for prolong period, there are number of various models has been formulated to suit the needs of such business contexts. The purpose of this paper is to develop an alternative approach to formulate manufacturing supply chain network design problem involving multi-echelons, multi-commodities, product's architectures, and manufacturing processes. Two new conceptual structures, which are product- and network-related, will be introduced as a foundation for developing mathematical models. In addition to the methodology, a case study in a consumer product firm is used to demonstrate applicability of the suggested method. Moreover, some managerial insights are also elaborated form the experiment results of a case study.

A new model called MPS-T is presented for the scheduling of temporary workers in mixed model flow lines with both skilled and unskilled workstations. We classify workers in two dimensions: (i) permanent or temporary, and (ii) skilled or unskilled. While all permanent workers are skilled, temporary workers can be either skilled or unskilled. Two workstation restrictions are modeled in the MPS-T: (i) some workstations may be limited to only skilled workers if there is a significant operational complexity, (ii) each workstation is limited to a minimum ratio of permanent to temporary workers. Such workforce models are commonly seen in labor intensive export oriented factories located in the Far East. These factories experience an order behavior that is characterized by high variance and short lead times. Distinguishing between worker types, permits management to limit their fixed workforce costs and also rapidly access a cheaper labor pool to meet demand surges. MPS-T is modeled as a mixed integer program and the key decisions are: (i) What type of temporary workers to hire or layoff in each period, (ii) the number of workers to hire or layoff in each period, (iii) the assignment of the work time of workers to the work stations, and (iv) the usage of overtime hours. A case study example to study model behavior under different parametric conditions is reported.

The drum–buffer–rope (DBR) is a scheduling mechanism under the Theory of Constraints (TOC) philosophy. In DBR, ‘drum’ is a production schedule on the capacity-constrained resources (CCRs), which controls the speed of production for the whole system; ‘rope’ is a mechanism to release the required material to the CCRs; and ‘buffer’ is used to protect the CCRs from starvation due to statistical fluctuations. For a non-identical parallel machine flow-shop environment, estimating an efficient rope and time buffer for DBR implementation is not an easy task because of the complexity of non-identical parallel machine loading. This paper proposes a new scheduling method, which is called the modified DBR (MOD-DBR). It applies a backward finite capacity scheduling technique, including machine loadings and detail scheduling, instead of the rope mechanism in DBR. The scheduling performances of MOD-DBR are evaluated under variable processing time situations. The experimental results indicate that the MOD-DBR without a t...

A conventional material requirement planning (MRP) logic considers infinite capacity of machines. A planner must manually solve capacity problems (if any) after the material requirement plan has been generated. This paper tries to alleviate the capacity problems by developing a finite capacity material requirement planning (FCMRP) system for assembly operations. The FCMRP system is capable of automatically allocating some jobs from one machine to another and adjusting timing of the jobs considering a finite available time of all machines. The FCMRP system has been tested using real data from a steering-wheel factory. Effects of options – namely, scheduling, rearranging and adjusting options in the FCMRP system on the performance measures – are statistically analysed. The FCMRP system can reduce required overtime if the jobs are allowed to be completed earlier than the due dates. The required overtime can be further reduced if the jobs are allowed to be completed later than the due dates.

This paper presents the aggregate production planning for multiple product types where the worker resource can be transferred among the production lines. A mathematical model was formulated in spreadsheet format. Then the spreadsheet-solver technique was used as a tool to solve the model. A real situation of a manufacturing company was selected as a case study. The actual data was used to test and validate the proposed model. The optimal aggregate production plan provides the information on managing the available production capacity together with the useful workforce transferring plan. The obtained solutions were compared to those of another approach where the workers cannot be transferred among the production lines. The total cost is significantly reduced when the workers are allowed to transfer among the production lines.

This paper is to analyse impacts of pricing policies, inventory policies, and benefit sharing mechanism on supply chain (SC) performances. The novelties and theoretical contributions of this paper include a development of linear mathematical functions to represent pricing policies, especially 'every day low price' (EDLP), and explicitly considering cash as one of the performance measures in addition to profit and revenue. A mix integer linear programming (MILP) model with pricing mechanism and policy constraints is developed. A case study in plastic fashion accessory industry in Thailand is performed. Experiments are conducted to demonstrate how the proposed model is applied. The performance measures are translated into satisfaction levels and plotted in a radar chart to facilitate performance comparison among policies. Practical contribution is that this paper recommends suitable combinations of policies to match with the priority of profit, revenue, and cash performances. [Received 10 January 2018; Revised 12 June 2018; Accepted 19 June 2018]

This paper discusses a methodology to determine an appropriate sequence for implementing engineering and management (E&M) techniques for enhancing the effectiveness of the production and inventory control (P&IC) system. Firstly, direct relationships among E&M techniques are analysed using the fuzzy interpretive structural modelling (FISM) approach. A procedure based on a max-min fuzzy composition of the direct relationships is also applied to determine indirect relationships among E&M techniques. Then, an appropriate implementation sequence of these E&M techniques is determined based on the driver power and dependence scores that are derived from the overall direct and indirect relationships among these techniques.

In a discrete parts manufacturing environment, material requirement planning (MRP) is carried out without considering the manufacturing resource capacity. As a result, during implementation, adjustments in planned orders may become necessary. This paper presents a finite capacity material requirements planning algorithm (FCMRP) to obtain capacity-based production plans. These plans need no costly adjustments to satisfy the capacity constraints. Performance of the FCMRP system, when studied through a set of test examples, has been found to be superior to the existing MRP system.

An analytical method for predicting the optimum number, location, and signal sound level of auditory warning devices is proposed. Factors which influence the perception of alarm signals, namely, ambient sound level, machining workstations (locations and their generated sound levels), workers' locations, and recommended signal sound level reaching workers are considered in the development of the objective function and constraints. Solving the multiple alarm location problem with a nonlinear programming technique yields the minimum number of auditory warning devices, their locations, and the minimum required sound level of warning signals. Two examples are given to demonstrate the applications of the predictive model.Relevance to industryThe method presented in this paper enables engineers to determine the optimum number and location of auditory warning devices for manufacturing facilities to ensure adequate perception of warning signals. Since the proposed method is analytical in nature, it helps reduce trial-and-error effort normally spent in locating the alarm devices, thus saving both time and money.

Cost estimation is an upstream activity in manufacturing and service business. Accurate cost estimation promotes profitable decision making. In this study, a cost model was developed to estimate paint cost in a spraying process of wooden toys. The required amount of paint was formerly believed as a function of only product surface area. However, data collected earlier indicated that volume and projected area also had significant contribution. Altogether, these parameters triggered uncertain paint loss which complicated a cost model. In order to solve this problem, we propose a cost estimation method based on a fuzzy theory. While the traditional parametric cost function used a constant paint consumption rate, the proposed fuzzy model allowed the loss to be dependent on product parameters. Predictive results from both models were compared with actual data and their performances were measured. Overall performance of the fuzzy model is more superior especially for larger products in an upright placement on the spraying rack. With an accurate cost estimation system, enterprises can plan for more efficient material requirements, and pricing strategies.

This paper developed a Finite Capacity Scheduling (FCS) system for make-pack production based on a real case of an adhesive factory. The FCS determines production quantity of each machine to conform with resource capacities and due date of customer orders while minimizes related total cost. The total cost includes total production, inventory, and cleaning cost. A Mixed Integer Linear Programming (MILP) model is formulated and solved by LINGO software. The computational time is very long since the model has a lot of integer variables. Thus, the model is solved for a reasonable time and the best but not optimal solution is reported with the lower bound. This paper tries fixed horizon and rolling horizon scheduling methods. The fixed horizon plans for an entire horizon of 30 days. The rolling horizon plans for a sub-horizon of 10, 15, and 17 days. An overlapping of sub-horizons is applied to reduce end-of-horizon effect. Three scenarios (high, normal, and low) of demands are considered. The fixed horizon method is applied first to all scenarios of demand. If the best solution is far away from the lower bound, the rolling horizon method is applied. The results indicated that the rolling horizon method may significantly reduce the total cost with the same computational time. Moreover, the rolling horizon method is more applicable for a dynamic situation where customers frequently change orders. The proposed MILP model can generate reasonable solutions and they are useful for scheduling decision of make- pack production.