The Impact of High-Quality Information on Performance in Manufacturing

Doctoral thesis, 1995

The dissertation deals with the impact of high-quality information on performance in manufacturing. In a system where two material flows with sequence dependency are matched, the use of real-time information to monitor and control the sequencing buffer improves the system's ability to schedule many different models, compared with a system with no information at all. In the latter case, the system is upset considerably when more than four different models enter the system. The use of automatic data capture systems enhances productivity in manufacturing companies. Six case studies were carried out, all of which showed that the productivity improvements obtained by these systems are mainly in the area of human resource utilisation which was found in all six case studies. Improvements in machine utilisation, transporter utilisation and inventory levels were also found. The productivity can furthermore be improved if the installed system acts as a proactive system, giving information on the operational level about deviations and changed customer demands. Out of 31 registered disturbances, 15 resulted in a productivity loss because scheduling information was not available. A system providing scheduling information makes it possible for the scheduler to change a dispatch list or final assembly schedule in such a way that a less important job can be run ahead of a delayed important job, only to release capacity when the delayed job arrives. These kinds of systems mainly enhance productivity in an environment with shared resources. The impact of providing the scheduler with real-time information of high quality is investigated. It is shown that this enables him to reschedule the work force when stochastic events such as disturbances of different kinds affect the scheduling environment. It is also shown that the performance of the scheduled system improves a/ in direct relation to the number of work stations being scheduled and b/ with the reduction in lapse between the detection of a deviation and corrective action. Measures taken to prevent the propagation of disturbances can be classified as formal measures and informal actions. Formal measures are such measures which are defined in a computer system, and informal actions are such that are executed manually. None of the measures identified in this study were efficient when a holistic evaluation was made. This evaluation describes the ability of the measure to absorb disturbance effects, its cost and the degree to which it upsets the system. Some measures were, however, better than others. Partial delivery is a suitable action when materials shortages occur, subcontracting when machines break down, and moving over personnel when absenteeism occurs. When an ADCS is used, two types of effect can be experienced; direct effects and indirect effects. The direct effects are the result of the data capture process as such being enhanced, i.e. less effort is needed to capture data and to trace and correct faulty data, and the indirect effects are the effects experienced when timely and accurate data is fed into the data base from which a number of functions retrieve the data. The direct effects are normally easy to define. They affect the number of staff involved in data capture and handling, and in some cases safety stock levels. Indirect effects are more difficult to isolate. A general model showing the manufacturing process and the different ADCSs which can be found within each activity was developed. Indirect effects experienced as a result of enhanced decision-making when an ADCS is installed were; increased sales, decreased costs and decreased inventory investments. Both direct and indirect effects enhance the effectiveness of a company. It is shown how the use of a serial information flow respectively a parallel information flow with different time delays affects the performance of a system when expediting is carried out. The results indicate that a parallel information flow, i.e. information to the succeeding activities being given in advance, performs best if the information quality is high, i.e. the time delay is close to zero. When the time delay increases there is a threshold when the serial information flow performs better than the parallel information flow with poor information quality. The performance of the parallel system decreases with poor information quality. The same result is obtained with respect to the output, i.e. the output decreases with poor information quality.