OPERATOR MODELING

摘要

Operator modeling is not easily performed by the industry. Static methods (time studies) are used to calculate the number of operators. The methods used are based on detailed observations of all operator activities, their classification and the measurement or estimation of the time needed to carry them out. These methods are very slow and expensive. Moreover, they are discontinuous and do not allow for frequent tracking of human capacity. In addition, operator productivity factors can only be introduced in simulation models as constant values, handicapping the representation of their variability. We have developed a system that permits automatic quantification of the number of operators needed for a given volume of material, eliminating the need of direct observations. The magnitudes measured are .suited for an easy representation of operator behavior and variability in a simulation model. The advantage of our system is to reduce the cost of the studies, decreasing the time needed, to allow for continuous tracking of human capacity and to improve the representation of operator behavior and variability in a simulator. The only requirement is a computerized lot tracking system in the work center. This paper shows how the number of operators required moving a given volume of material is calculated, through measuring the current number of different operators making transactions every hour and its variability. We also show how these parameters can be introduced in a simulator to represent operators. To understand the impact of operators in the capacity, machine OEE and cycle time is critical to achieve high performance fabs. This analysis typically required direct observations on the shop floor (Raviv 1998, Pollit and Matthews 1998, Mostley et al 1998). These methods are expensive and time consuming, for that reason they are not frequently undertaken. Indicators have been developed (Bonal et al. 1999) to assess the impact of direct labor of a work center in cycle time and throughput. These indicators are automatically measured and identify the work centers with major impact in cycle time. However these indicators don't quantify the impact on it, they only point out the work center with significant impact of the staffing. Also these indicators don't allow forecasting the behavior of the cycle time and throughput if the volumes are ramping up with the same number of operators. This paper shows how is possible to built a static capacity model for operators using data obtain from the Manufacturing Execution System (MES). As a static method, it gives the number of operators required for a given volume. Also this paper show how it is possible to introduce the operator factor in any line simulation model using again data obtained from the MES. With these types of models it is possible to forecast the number of operators required for a volume of wafers and a cycle time.