DYNAMIC MODELLING AND OPTIMIZING THE PRODUCTION LINE IN THE BEVERAGE INDUSTRY USING WITNESS FOR MIXED DISCRETE AND CONTINUOUS EVENT SIMULATION

Abstract

Nowadays, market is becoming very challenging to industries due to shorter product life cycle and high competition. Consequently, organizations are compelled to improve efficiency and productivity in order to stay competitive in the market and fulfil their customer demands. However, this might be costly and time consuming especially if improvements they made using a static model were later found to be less effective in real world scenario. Hence, developing a fully dynamic model which will be customized for the organization will be desirable. In this context, Industry 4.0 simulation tools and sequence dependent algorithms play an important role in increasing the overall efficiency, reducing process time and cost, and in evaluating the feasibility of solutions at an early stage. In this industrial project, author used “WITNESS” , a Mixed Discrete and Continuous events simulation system tool, to develop a dynamic model of the production line in the beverage industry which was then used to simulate the current baseline data collected in the shop-floor to assess its current performance in terms of its level of optimization in cleaning changeover process, setup time and the total process time. Then, a sequence dependent algorithm was developed and used many times on different factors to optimize the process further. This algorithm was plugged into the Simulation model using Microsoft Excel as middleware between user and Witness. Witness simulation results were then analyzed to find optimum solution based on total process time of each batch in each line, the percentage of cleaning time and setup time in each line. All Witness findings were linked and updated simultaneously in Excel where the user was able to track the model performance during the model run time. Further improvements were found by linking the simulation tool with the algorithm in order to optimize production scheduling and sequencing which in turn minimizes the total production makespan. In this project, three alternative solution were examined. First, using random scheduling method. Second, fixing product color and finding optimal product sequencing using K-T algorithm. Third, fixing product color and finding optimal bottle sizes sequencing using K-T algorithm. The results showed that the second plan, i.e. fixing product color and finding optimal product sequencing, is the best option making 4.4% saving in total process time, 4.92% saving in total setup time and 1.03% saving in total cleaning time.