This paper presents three economic queueing models of push and pull types, discusses the limits of pull on Japanese production systems, and considers the problem of setting the economic leadtime by 2-stage design metbod. A single-stage lot/cell production system under make-to-order/stock is here treated, and is assumed to be Poisson arrival and exponential service in batch. First, we give the expected operatian cost of three queueing models. This cost consists of inventory/order, backlog, busy, and idle costs, and the bebaviours of these costs are showed as a function of mean processing time or fixed setup time. The optimal(or economic)traffic exists, and the minimum operating cost is found grapbically. Second, the push vs.pull system is discussed from a view of inventory or operating cost. One limitation of the pull system is that a reduction in setup-time can cause the inventory cost to inflate, and a minimum operating cost exists. Another limitation is that the pull system is specially better for the case that the inventory cost is important to managers, but that a backlog is allowed. Third, a scientific basis of economic leadtime setting is given by 2-stage de-sign method. The economic processing time is obtained at the first stage, and the economic leadtime can be set at the second stage under the economic processing time. The economic leadtime of a pull system is not necessarily better than that of a push system.

View full abstract

A simulation-based scheduling method for minimizing the due-date-deviation is proposed on the basis of a combination of the BFHS(Backward/Forward Hybrid Simulation)method and the parameter-space-search-improvement method. In order to minimize the due-date-deviation in a simulation-based scheduling, two functions are indispensable : the first function is to control operation-onset timing of jobs at all work centers to avoid inadequate early completion of jobs, and at the same time to determine job priorities so as to suppress the late completion of the job, and the second is to adjust the balance of the two reductions properly so that the final schedule achieves the minimization of the due-date-deviation. In this paper, a new schedule generation method named the BFHS/type D is first proposed, in which the information generated during the backward simulation is utilized to control operation-onset timings and job priorities in the forward simulation. Then, after the investigation of the backward simulation characteristics, two parameters are introduced in relation to due-date-deviation to manipulate directly the backward simulation process and consequently to manipulate indirectly the forward simulation process in the BFHS/type D method. Finally, a scheduling procedure is proposed to search out the best schedule with respect to the due-date-deviation on the space spanned by the two parameters. The effectiveness and efficiency of the proposed method is demonstrated by using a set of planning conditions obtained not only from a simple job-shop model, but also from a practical large-scale system.

View full abstract

This paper is concerned with Vehicle Routing Problem in a Local Area Transportation System. The system consists of depot, vehicles, customers and pieces of luggage. In the system, vehicles are loaded with luggage and go to customers from the depot according to the Schedule. At the customer, the vehicle is unloaded with a piece of luggage and goes to the next customer. When all luggage has been delivered, the vehicle goes back to the depot. Such a system is often seen in big cities. The efficiency of the system is generally achieved if much luggage is carried by a low number of vehicles over a short distance. Recently, an additional service has been introduced. In the service, arrival time of vehicle at each customer is specified. This constraint is called Time Window, where Time Window is defined with the earliest time and the latest time to arrive at. For the service, we can obtain safe vehicle's schedule if we assume that travel velocity of vehicle is low. However, the efficiency is lost. On the other hand, if we assume that the velocity is high, some Time Windows may not be completed. Possibility of this incompleteness is low, but reliability of the system is lost if it has occurred. In this paper, we propose Time Window Margins to measure the possibility of the incompleteness. We made an investigation on an actually operated system with this measurement and propose three scheduling methods based on Evolution Programs. Numerical investigation of proposed methods for both the efficiency and safety of delivery determined on the Time Window Margins is shown.

View full abstract

This study proposes a method for analysis of cognitive processes during viewing of television commercials(CM's). In this analysis method, we first identify semantic roles of elements appearing in each CM scene. Corresponding to those roles, we describe each CM scene in a semantic network form, and calculate semantic distances between any two scene elements. In this study, semantic distance refers to the scale of semantic similarity. Based on semantic distances and viewers' eye movement data, we define two measurements(rate of semantic deviation and semantic transition)relating to how viewers direct their attention to each CM. Viewers' cognitive processes are evaluated quantitatively based on the two measurements. A series of experiments was carried out to obtain eye movement data as well as individual attributes data(such as ages, preferences towards advertised products, and so forth)during viewing of brand-new CM's. The experiments were held within a week of those CM's broadcast debut. Applying the analysis method using the two measurements based on semantic structure of CM scenes to data obtained in the experiments, we could confirm that the proposed analysis method is able to detect individual differences in cognitive processes. In addition, there are implications that the CM design parameters(such as number of scene elements used)and the individual attributes(such as product preferences)may affect those differences. Considering these results, our analysis model seems to be appropriate as a useful method for the pre-evaluation of alternatives for a new CM by advertising companies.

View full abstract

The diffusion of automation and labor saving devices has given rise to the necessity of condition monitoring of equipment and persons from a remote place through a monitor. When an observer watches a work area through video cameras, the color of equipment differs from its actual color under the influence of sunlight from windows, shadow and lighting fixtures. Therefore, this research aims to analyze visibility in monitoring under several lighting conditions. First, an experiment is carried out to evaluate how the lighting sources affect the ability of the observer to distinguish colors of equipment through the monitor. Next, a statistical method is presented in order to estimate the limit of the color that could be recognized. Then, a method is formulated, which can convert the indistinguishable color on the monitor so that it becomes capable of being recognized.

View full abstract

Decoupling stock can be defined as an additional stock for making separate decisions in a multiple installation inventory system. Edward A.Silver proposed this concept in the early 1980's, and presented some qualitative analysis. Our purpose in this research is to analyze the decoupling stock quantitatively under(R, Q)policy. We derive three average inventory levels required for three different decision-making frameworks, i.e., centralized decision-making, decentralized decision-making with information sharing, and decentralized decisionmaking without information sharing. Based on these required inventory levels, we derive decoupling stock function in which average inventory level under centralized decision-making is subtracted from that under decentralized decision-making. The function of decoupling stock is justified by simulation, and the relationships between the lead-time and the decoupling stock, and between lot size and the decoupling stock are analyzed. The results show that the decoupling stock is reduced greatly by sharing information, such as inventory levels, lead times, and end demand distribution. Decoupling stock can also be reduced by shortening lead-time, and cutting lot sizes, and the key point is to reduce the lot size so that the average replenishment cycle lengths are close to the lead time of a neighboring stage in the upper stream.

View full abstract