The usual p chart for the monitoring and control of product attributes is inadequate for processes with very high yields. Recently, instead of the p chart, some alternative charts for very-high-yield processes have been presented based on cumulative count conforming items. However, these charts are sensitive for detecting upward shifts in p, but not sensitive for detecting downward shifts in p. In the real field, the fundamental aim of monitoring and improving actions of process quality is always to realize zero-defect processes. From this viewpoint, this paper presents simplified charts for detecting the downward shifts in p more sensitively which indicates further quality improvements in very-high-yield processes. The Average Number of Observations to Signal (ANOSs) of the proposed charts are compared with those of conventional charts through computer simulations.

View full abstract

The evaluation of work capabilities of aged persons has important meaning in the establishment of some indices of self-support in social activities and daily life. In addition, those indices can be used to make good use of aged persons' abilities in jobs and social activities. In this research, equipment is developed for measuring human physical work capabilities in jobs and daily life, such as the strength in a person's back, quickness, force of fingers and hands, walking ability and posture. This equipment is used to measure the work capabilities of a wide range of aged persons over 55 years old. The results shows the relation between age and work capabilities, relative personal work capabilities and the mutual relation among work capabilities.

View full abstract

This paper proposes an optimization approach for the production planning and scheduling of a job shop. Production planning must satisfy the forecasted demand with minimum production cost, while scheduling is responsible for generating the detailed allocation of operations among machines based on the given production plan. In general, the planning phase of the job shop cannot utilize the maximum production time because the scheduling phase often requires some unavoidable idle time in order to satisfy the precedence constraint of operations. In addition, the scheduling problem is determined after the production plan is frozen. As it is difficult to formulate and solve an entire planning and scheduling model, our approach solves the planning and scheduling problems iteratively. A Lagrangean relaxation method is adopted to solve the production planning problem formulated as a mixed-integer program. In the scheduling phase, the record of schedules obtained before is first utilized in order to reduce the computational workload, and if no record is available or no feasible solution is obtained, the scheduling problem is solved by a branch and bound method to minimize the makespan. If the makespan of the optimal schedule is greater than the available production time, a new capacity constraint to the planning problem is generated and the enlarged production planning problem is solved again. The generated constraint requests that the sum of processing time of operations belonging to the critical path must be equal to or less than the available production time. When the scheduling phase finds a feasible schedule, a feasible production plan and its detailed production schedule are obtained. Computational experiments have shown that the proposed approach can find feasible solutions with a relatively small number of added constraints, and the degree of minimization of makespan may affect the quality of the proposed approach in terms of both finding feasible solutions and obtaining lower cost solutions.

View full abstract

This paper reports on the planning of potted flower production conducive to optimum utilization of a greenhouse. Four kinds of flowers, Cyclamen, Begonia, Zygocactus and Clivia, were selected, taking into account their growth properties, profitability and efficient utilization of bench space. Production schedules were formulated based on existing production data. These schedules were represented with mathematical formulas by applying linear programming, and the optimum solution for the scheduling of the flowers was determined. In addition, regarding the optimum shipping of cyclamen in particular, by increasing the capacity of the pots the flowers are to shipped in, an attempt was made to derive a greenhouse utilization plan that would be more profitable than other production plans in common use. These results might be regarded as useful when considering efficient greenhouses utilization.

View full abstract

A layout problem in a cellular manufacturing system can be recognized as an unequal-area layout problem because each cell contains a different number of machines. In this paper, a two-objective (minimization of total material flow and total layout area) cell layout problem is formulated under the condition that each cell has a rectangular shape and an aspect ratio range. The aspect ratio is defined as the ratio of the width and height of a rectangular cell. At first, we define two dissimilarity coefficients based inter-cell part movements and cell shape. The dissimilarity coefficient based on inter-cell part movement is defined for the minimization of total material flow and the other coefficient, based on the cell shape, is for the minimization of total layout area. Using a hierarchical cluster analysis that combines the above two coefficients, a new layout method is proposed for the two-objective cell layout problem. The basic idea to solve the cell layout problem is to repeat the layout procedure of two adjacent cells, which are selected by the clustering method. By applying the proposed method to several examples, non-dominated layout solutions that have small total material flow and total layout area can be easily found with short computation time. Using this method, a large layout problem containing many cells can also be solved in a short amount of time.

View full abstract

This paper presents an efficient approach for a decision maker to obtain the best compromise solutions for multiobjective discrete optimization problems. We translate an original problem into a surrogate problem using a surrogate multiplier. The surrogate problem with only one objective is solved by an algorithm based on the modular approach. The use of a modular approach makes it possible to solve a large-scale multiobjective discrete optimization problem. The pareto optimal solution of the surrogate problem is equivalent to the pareto optimal solution of the original problem. In other words, the solution of the original problem can be obtained by solving the surrogate problem. The decision maker can obtain pareto optimal solutions suited for sense of values by interactively operating surrogate multipliers. A simple example shows the effectiveness of our interactive decision-making algorithm.

View full abstract

In the Kanban system, the order release for each process is determined on the basis of actual demand, that is, without demand forecasts, and it leads to avoiding the influence of forecasting error. For the Kanban system, we have proposed a reactive Kanban system which can react to unstable changes in product demand. However, in the reactive Kanban system, only the mean changes in product demand are considered, and constant and known variances of product demand are assumed. Supposing not only the mean but also the variance of product demand fluctuates dynamically, the variance changes as well as the mean changes must be detected, and a response to the changes must be determined. Additionally, monitoring mean changes is affected by variance changes. Therefore, in this paper, we propose a reactive Kanban system for a multi-stage production system with unstable changes, not only the mean but also the variance of product demand. The proposed system includes a reactive controller that monitors jointly the mean and variance changes in product demand and responds to the changes. For detecting unstable mean and variance changes in product demand, time series data for product demand are grouped into batches, and the mean and variance are utilized as new time series data. For the new time series data, the exponentially weighted moving average charts for mean and variance are applied to detecting unstable changes. If an unstable change is detacted, the number of Kanbans is adjusted as a reaction to the unstable changes detected, a mathematical model of the reactive system is formulated, and the performance of the propose system is analyzed and compared with previous systems by simulation experiments. The results show the effectiveness of the proposed system.

View full abstract

[in Japanese]

View full abstract

Regression models or mixed-effect models are often supposed for the data taken from the production process to comprehend the dependence of effect variables upon cause variables. Then, the results of the analyses based on such models for this kind of data sometimes largely depend on a few observation values and a quite different result may be obtained if the observation values are removed in the analysis. So, the influence of observation value on the result of the analysis should be evaluated, which is expected to make the result of the analysis more robust. So far, various statistics of diagnoses have been developed for the regression analysis and their properties have been investigated in detail. Leverage is an important statistic of diagnosis to evaluate the stability of the result of the regression analysis by detecting influential observations based on sensitivity analysis. For the mixed-effect model, Christensen, Pearson and Johnson [1] studied the Case-Deletion diagnosis. In this paper, a generalized leverage for the mixed-effect model is proposed and the characteristics of the proposed statistics are investigated. Further, the proposed generalized leverage under the mixed-effect model is shown to be useful for the data taken from the production process.

View full abstract

In this paper, we discuss a single-stage, single-product, multi-period production planning problem under the condition of uncertain demand. Our primal assumption is that the forecasted product demand is given via a set of discrete scenarios, i.e., demand quantity and its realization probability. We also assume that the planner must determine the production plan before he or she knows any realization of the product demand. In the traditional approach, the problem is modeled as a stochastic linear program based on the realization order of product demand, and thus the model often produces a wide difference between the minimum and maximum production quantities for two or later periods. The preceding activities such as ordering materials may have difficulty in planning operations because many possible production quantities with wide variations are provided from the production plan. In this paper, we consider the situation such that a limited quantity of additional production can be conducted if necessary after receiving the realized demand quantity provided that the required production capacity is allocated in advance. In this mechanism, the planner decides the basic production quantity and the slack capacity for additional production. Using numerical experiments, we have found that by keeping a slack capacity, the level of inventory at the end of the first period is lower than that of the traditional approach under some conditions, and the proposed mechanism can control the difference between the maximum and minimum production quantities.

View full abstract

The purpose of this study is to propose that modal shift is effective as an environmental countermeasure. Therefore, we present the optimal modal shift pattern considering environmental problems. In such research, it is necessary to consider the emission of pollutants though overall energy flow, from the input process of primary energy to energy consumption. In other words, it is essential to focus on import, conversion, transportation, and consumption processes of energy in addition to dealing with the physical distribution process. An "energy input-output model" that clarifies the overall energy flow is presented in our study. This model consists of energy input, conversion, transportation and consumption sectors. Bi-objective linear programming is used to formulate a model which minimizes the emission of carbon dioxide and nitrogen oxide. The optimal modal share of freight transport, which minimizes NO_x and CO_2 emissions considering the overall processes of energy flow, is identified in this analysis.

View full abstract

This paper considers the job-shop scheduling problem to minimize total weighted tardiness since the due date of each job has been promised to its customer and it is most important to meet due dates. A heuristic algorithm based on the tree search procedure is developed for solving the problem. A certain job-shop scheduling problem to minimize the maximum tardiness subject to fixed sub-schedules is solved at each node of the search tree, and the successor nodes are generated where the sub-schedules of plural operations are fixed. Thus, a schedule is obtained at each node and a sub-optimum solution is determined among the obtained schedules. Computational results show that the proposed algorithm can find better solutions with little computation time and the number of generated nodes hardly increases even if the number of machines increases.

View full abstract

Cellular manufacturing is a plan for reducing lead times in batch production systems. Usual methods of cell formation are based on machine-component incidence, production sequence, product volumes and so on. This study proposes a new cell formation with a reduction in setup times between the machines in the same cell. Similarity measure, clustering method and performance measure are defined for evaluating solutions of the proposed cell formation. Using numerical examples, differences in cell formation and performance evaluation between the usual methods and the proposed method are also investigated.

View full abstract

In the just-in-time (JIT) production system, single-unit production and conveyance are applied to a production line without conveyors which manufactures different kinds of relatively small parts. To achieve this at a low production cost, a U-shaped layout and multi-function workers are originated. The multi-function worker takes charge of multiple machines, and visits each of them once in one cycle. When he arrives at one of these machines, he waits for the end of processing of the preceding item if it is not completed, performs required operations, and then walks to the next machine. In this paper, we deal with a U-shaped production line operated by a single multi-function worker. We assume that processing times of items, operation times of the worker and his walking time between machines are sequences of mutually independent random variables, and the processing times have Erlang distributions. We formulate this production line into a Markov chain embedded at the departures and arrivals of the worker at each machine and propose a method for evaluating the performance measure such as the expected cycle time of the production line. By applying this method, we examine the effects of phases of Erlang distributions and the variances of the distributions of the operation and walking times on the expected cycle time. We also observe the effects of the partitioning or unifying of stations when imbalances exist among the stations in the production line.

View full abstract

This paper proposes a branch and bound algorithm for solving job-shop scheduling to minimize the makespan. The proposed algorithm is based on single-machine scheduling problems with ready and due times constraints that relax the original problem. New powerful elimination rules associated with the single-machine problem are used at any node to select some disjunctions directly or at least to determine whether or not a potentially optimum solution can be obtained. Some problems with ten jobs and ten machines are solved by the proposed algorithm and the results are reported.

View full abstract

In this paper, we study a decision-making model which enables indication of how the most valuable design should be selected from the viewpoint of value engineering. To be concrete, by the decision-making model described here, "the permissible product cost by function" and "the performance degree of the function (basic specification) as a valuation basis" can be established and the "expected level (minimum line of satisfaction against a valuation basis)" of each valuation basis can be also identified. After that, we construct a decision-making model which enables calculation of "a characteristic value for selection" of each design based on "expected level" and to consider" a characteristic value in the situation" for the whole design. Further, it should be a decision-making model based on "reformed analytic hierarchy process (AHP), " the improved "pairwise comparison method of AHP, " that is to be adapted in the thinking process of value analysis. By the way, the pairwise comparison method of AHP means relative measurement approach, the most normal AHP developed by Dr. Saaty. It is expected that this decision-making model eliminates regrets and anxieties which decision-makers often have after making their decision, being compared with the previous one (relative measurement approach of AHP). As a result, our organized model will be able to contribute to the speeding up of product development activities by decreasing the "trial and error" process for decision-making during product design activities.

View full abstract