PhD Student Researcher
“Minimising the makespan in batch production
facilities through scheduling optimisation models”
The current research project aims to apply and develop a scheduling optimisation model in order to minimise the makespan in batch manufacturing facilities. The industrial partner of the research project is GlaxoSmithKline (GSK). GSK operates a high-volume batch production facility to produce skincare products (creams, lotions, gels and liquids) that utilises a number of large mixing vessels that require using various shared utilities and resources.
Scheduling is a decision-making process for short-term activities. By solving a scheduling problem, a limited set of resources can be allocated to produce certain products with specified recipes without violating or exceeding the limits of such resources. These resources can include equipment, utilities, manpower, raw materials, storage tanks…etc.
The objective of the scheduling process may vary between minimisation of energy consumption, decreasing the timespan of production, increasing production throughputs to meet market requirements, and decreasing the environmental loads resulted from a production facility. The solution of a scheduling problem defines how many batches need to be executed (batching), where these tasks will be executed (assignment) and when and in what sequence the tasks will be executed (timing and sequencing). Such decisions are limited by the constraints in the manufacturing facility.
The manufacturing cycle in the studied batch manufacturing facility consists of four phases; dispensing, manufacturing, storage and filling, which are processed in a sequential manner as shown in Figure 1.
Figure 1 Overall manufacturing cycle
In the dispensing phase, the raw materials of each batch are prepared, weighed, and sent to the manufacturing room. Each product is manufactured by applying different heating, cooling, and mixing steps in a predetermined sequence, which is called the product recipe. The duration's of heating, cooling, and mixing steps depend on the product. The manufacturing phase is performed in several large mixing vessels, which require the usage of various utilities including hot water, cooling water and electricity to produce the required batches.
The assignment of the manufacturing vessel depends on the batch size and its product family as the manufacturing vessels are not identical and differ in size and mechanical design. After a vessel is used to produce a product batch, it is cleaned before the next use. Some batches of the same product can be manufactured multiple times in the same manufacturing vessel without being cleaned up to a certain number of batches. This manufacturing model is defined as campaign manufacturing. If a product campaign equals three, for example, this means that the same vessel can be used to produce three batches in series of this product. The vessel will then be cleaned at the end of the third batch. Once the manufacturing phase is completed, the product is sent from the manufacturing vessel to a storage vessel(s) depending on the size of the product batch and its product family. The filling lines are fed from the storage vessels until the filling phase is completed. Each product is filled in a specific filling line for a known filling duration. Multiple resources are required to produce every batch. These resources include workers, manufacturing vessels, storage vessels, filling lines and utilities.
This project started with collecting two different data groups from the manufacturing facility. The first data group included the product code, batch code, batch size, start and end time for each recipe step and the assigned manufacturing vessel for the manufactured batches between April 2017 and April 2018. These data were collected from the manual batch manufacturing records and then digitised. The second data group included the standard dispensing time, standard manufacturing time, assigned manufacturing vessel, number of allowable batches per campaign, standard cleaning time, number of required storage vessels, shelf-life in storage vessels, filling line and filling duration for each product. After the data collection and digitising were performed, the scheduling model was developed using the Optimisation Programming Language (OPL) in the IBM ILOG CPLEX environment and solved using its constraint programming (CP) optimiser. Constraint Programming (CP) was originally developed to solve feasibility problems. However, constraint programming has been successful in solving optimisation problems of combinatorial natures especially in sequencing and scheduling applications.
Figure 2 shows the elements of the scheduling optimisation model. The results of the scheduling optimisation model included when to start/end each phase in the manufacturing cycle, how the resources were assigned to each batch and in what sequence the batches were manufactured. The results from the model were validated qualitatively and the results showed that a decrease in the manufacturing cycle makespan could be achieved by using the scheduling optimisation model.
Figure 2 - Scheduling optimisation model
The scheduling optimisation model is being further developed as a multi-objective scheduling optimisation problem which will include the energy consumption minimisation as well as makespan minimisation.
Mohamed Awad holds a bachelor and master’s degree in chemical engineering from Cairo University, Egypt. Since his graduation, he was working as an assistant lecturer in the chemical engineering department of Cairo University in addition to working as process and chemical engineer in multiple companies. Before starting his PhD project, Mohamed was a lead process engineer in ThyssenKrupp Industrial Solutions. He was able to achieve large savings in different projects by applying process optimisation methodologies. Mohamed has a passion in optimisation which was also clear not only from his professional career but also from his master’s thesis which was entitled, “Optimization of small-scale liquefaction processes using genetic algorithm”.
Mohamed published the following two papers from his PhD work in addition to co-authoring a third:
Awad M. et al. “The Identification of Utility Constraints in a Batch Manufacturing Facility”. In: The 17th International Conference on Manufacturing Research (ICMR). Belfast UK; 2019. http://ebooks.iospress.nl/volumearticle/52654
Awad M. et al. “Energy Consumption Profiling - A Case Study in a Batch Manufacturing Facility”. In: International Conference on Innovative Applied Energy (IAPE’19). Oxford, United Kingdom; 2019. https://wearecatalyst.org/wp-content/uploads/2019/10/Energy-Consumption-Profiling-gsk-International-Conference-on-Innovative-Applied-Energy.pdf
Mulrennan, K., Awad, M., Donovan, J. et al. “Modelling the electrical energy profile of a batch manufacturing pharmaceutical facility”. Int J Data Sci Anal (2020) https://link.springer.com/article/10.1007%2Fs41060-020-00217-1
Mohamed also presented his work as poster presentations at three conferences. He was also the winner of the best poster competition in 2019 Irish MedTech conference. Olympic handball and digital drawing are his main hobbies.