Hybrid Flowshop Scheduling Problem at DrPaste

Abstract

DrPaste is a manufacturer company of pastry materials and tools. In this study, we consider sugar paste production of this manufacturer. The company wants to minimize the makespan to deliver the sugar paste to its customers faster. There are 5 stages of sugar paste production: i) refining the sugar, ii) preparing the raw materials, iii) mixing by boilers, iv) packaging the sugar paste, v) controlling with a metal detector. In stage 3, there are 12 identical machines (boilers) for mixing. The boilers are not dedicated to a specific color. Each boiler has a capacity of 65 kg. In stage 4, there are 4 employees responsible for taking the sugar paste from the boilers and packing manually into different package sizes. We focus on the 3rd and 4th stages of sugar paste production, which can be handled as a two-stage hybrid flowshop environment. There are 17 colors of sugar paste. In stage 3, there is a clean-up process before mixing a different color. Cleaning the dark colors from a boiler such as brown is more difficult than light colors like pink. We have 4 types of color groups; i) white, ii) light colors, iii) midtone colors, iv) black. There are sequence-dependent setup times before switching to a different color group. In stage 4, employees need to separate the sugar paste into 0.5 kg, 1 kg, 2.5 kg, and 5 kg packages according to order specifications. Under these circumstances, the job (color) sequence on each machine/employee plays an important role in stages 3 and 4. Therefore, we want to find the best job sequence in mixing and packing stages. Since the problem is NP-hard, we employ Genetic Algorithm to solve the problem. We also develop a decision support system (user interface) to implement the proposed solution approach easily in the company.