2023
Simulation for Decision Support in Disaster Management, Healthcare and Manufacturing Inventory Management
19 July 2023, 14.15 - 16.30 - University of Exeter and Zoom
More details
Stephan Onggo, University of Southampton.
Title: Disaster management needs complex systems simulation
Abstract: A disaster disrupts a complex system of systems where socio-economic system, physical system, ecological system, and other systems interact. Therefore, disaster management is dealing with a very complex system. Researchers have proposed various modelling and simulation methods to help decision makers make better disaster management decisions. One of the methods is agent-based simulation (ABS). Researchers often use of ABS because of its ability to model complex systems. Furthermore, ABS enables us to combine both quantitative and qualitative evidence into one formal model. The complexity of disaster management highlights the need to integrate quantitative and qualitative evidence when making decisions. Therefore, ABS has the potential to make an impact on disaster management. This talk will present some examples of ABS used in disaster management and highlight the need for considering the complex behaviours commonly observed during disasters drawn from the literature and a workshop that we organised in Beirut in March 2023.
Alison Harper and Tom Monks, University of Exeter
Title: A framework to share healthcare simulations on the web using free and open source tools and Python
Abstract: We present a simple framework to support sharing of executable healthcare discrete-event simulation models in python over the web. Our sharing framework is based on combining remote version control repositories with free and open source software - Jupyter Notebooks, Jupyter Books, and streamlit - along with free digital infrastructure provided by Binder, streamlit.io, GitHub pages, and open science repositories such as Zenodo. The framework enables executable models to be shared with users of different technical abilities: from coders to software literate users. We provide an applied example including a full web application of an executable model. Our framework aims to support NHS organisations to preview, validate, and use models. Academic research teams can also benefit from enhanced scrutiny of their work and long term archiving of models.
Yogendra Singh, University of Exeter
Title: Dynamics of transition from make-to-stock to make-to-order manufacturing system
Abstract: Purpose: To create a rich, realistic, and relevant model of a manufacturing supply chain echelon and to understand how three different lead times influence its dynamics. The three lead times are the lead time experienced by the customer, the lead time for the shop floor to produce finished goods inventory (FGI), and the lead time for the supplier to deliver raw material inventory (RMI).
Method: We use value stream mapping (VSM) to summarise recent empirical casework to develop a generic representation of a manufacturer. The manufacturer receives and satisfies orders from customers, sets production targets, and issues supplier replenishment orders. The VSM is then converted into a difference equation model that we simulate in Excel, and a z-transform based block diagram and transfer function model, from which we obtain analytical results. We assume demand is a first-order auto-regressive process forecasted with conditional expectation.
Findings: We derive the inventory optimal replenishment rule for setting the in-house production targets. We also derive the optimal policy for generating replenishment orders to issue to an external raw material supplier. We obtain expressions for the bullwhip effect in both the in-house production and the supplier orders. We reveal how the demand and lead times affect the variance of the FGI and RMI levels as well as the variance of the production and supplier orders. We compare our new results with the established results in the literature. In many cases there is a direct mapping from existing models to our model. However, there is a structural difference in the RMI variance ratio when the customer lead time is less than the production lead time. Finally, we conclude with an economic analysis of the FGI and RMI. Interestingly we find under negatively correlated demand, the RMI costs are not always increasing in the customer lead time. There is an odd-even lead time effect which means this is not always economically advantageous to increase the customers lead time in MTS settings.
Originality: When the customer lead time is greater than or equal to the production lead time (and one period to account for the sequence of events), the supply chain echelon operates in a make-to-order (MTO) mode; when the customer lead time is shorter, it operates in a make-to-stock (MTS) mode. We believe this is the first paper that is capable of understanding the dynamics when MTS transition into a MTO system.
Limitations: We take a linear approach to our modelling work. For example, we have not considered the impact of capacity constraints, non-negative production orders, non-negative FGI and RMI, and random yields from the production system and/or supplier. It would also be interesting to identify optimal production planning and supplier scheduling algorithms that account for the impact of the bullwhip effect on the capacity usage of the production system and the supplier.
