Industrial Systems, Manufacturing and Management (ISMM) Final Projects 2015
Constance Deperrois (cyd20@cam.ac.uk)
Inference of independencies in a supply network through the analysis of delays and disruptions by an aircraft manufacturer.
Supervisors: Philip Woodall and Pascal Wichmann
This dissertation will look at the propagation of disruptions and delays within a supply network in the aircraft industry, in order to draw conclusions on the information that an aircraft manufacturer may infer from the suppliers’ failures to deliver on time.
Due to the complexity of the manufacturing operation, huge numbers of aircraft parts and subassemblies are outsourced to tier 1 suppliers. In addition, each of these suppliers has its own suppliers, and so on. This forms a multi-layered supply network, whose nodes are linked by supply relationships. Each node may also provide for several others in multiple layers, upstream and/or downstream, reciprocally or not. Although aircraft manufacturers are aware of this complexity, they typically lack detailed information about their supply network, having only access to data about their tier 1 suppliers. When an incident occurs at a node of the supply network, it propagates to each consecutive node with a certain probability, and its impact increases with another specific probability. At the end of the network, the aircraft manufacturer experiences the resulting delay of the supply. Because of the high costs associated with these risks, it is a current problem in the industry to understand better the supply network and to predict such interruptions. This research work will review relevant literature to improve the understanding of supply networks and define the scope of the project. The following step will be the design and coding of a computer model of supply networks to simulate the propagation of delays and disruptions. The model should account for the complexity of the actual supply network, as well as for the cost associated with incidents, and may encompass some of the following: inventory levels and turnover, value added by each node… Simulations of the model will reproduce the propagation of delays and incidents, and quantify their impacts on downstream nodes. One key objective is to understand how the aircraft manufacturer can infer hidden parts of its supply network from his observations with a certain level of confidence. Depending on the conclusions of the simulations, recommendations on how the aircraft manufacturer can predict some delays and mitigate the risks associated with supply chain disruption will conclude this work.
Martin Blechta (mb2016@cam.ac.uk)
Using external data for obtaining supply chain intelligence and predicting supply chain disruptions
Supervisors: Philip Woodall and Pascal Wichmann
The dissertation is expected to cover the area of using external data to obtain supply chain intelligence and identification of supply chain disruptions in particular. The dissertation will answer the following research questions:
- What causes high frequency, low impact supply chain disruptions and how do the various causes manifest themselves through external publicly available data?
- Can this data be used to predict the disruptions and what is the confidence of the predictions?
- How long before the disruption event can the data provide a prediction?
- How difficult is it to capture this data automatically?
- How difficult is it to use this data to make useful predictions for the organisation regularly?
Yan Ming Tiang (ymt21@cam.ac.uk)
Data quality problems arising from the use of publically accessible supplier-related data
Supervisors: Philip Woodall and Pascal Wichmann
Industrial Organisations today face many different data quality problems such as inconsistent formatting of data, identification of data entry mistakes and identification of duplicate entry. There are several software packages currently existing in the market to solve this problem such as Google Refine but they are too expensive or not very effective. Currently, Boeing is interested in gathering more information on its supply chain model through mining public data. However, there are several data quality problems involved in dealing with consistency and ambiguity of public data sources. This dissertation will address this issue by identifying specific problems faced by Boeing in their projects through interviews with key stakeholders, conducting a literature review on occurrences of these problems and existing ways of solving them, and proposing new algorithms to solve this problem. The new algorithms will then be applied to the problems identified and results found will be discussed and compared to existing software packages.
Luc Wijffels (lw500@cam.ac.uk)
Mining data of warehouse management systems to estimate order preparation time
Supervisors: Vaggelis Giannikas and Philip Woodall
Information regarding the time needed for the preparation of an order (i.e. time between the moment an order is received and the moment it is ready for dispatch) can be used to support several decisions for the management of logistics operations. Within the warehouse, this information can give estimates about the number of orders that cannot be picked during a normal shift or suggest ways different orders can be prepared (picked and packed) together to improve overall performance. Further, this information can be used to integrate warehousing and transportation operations by allowing the latter to be planned and scheduled based on the current and future status of the warehouse. This project aims at investigating ways data mining techniques can be used to make estimates regarding the preparation of orders in a warehouse as well as examine ways this information can be used to support planning decisions.
Marie-Helene Stoltz (mhays2@cam.ac.uk)
Exploring the applications and benefits of augmented reality in warehouse operations
Supervisors: Vaggelis Giannikas and Jumyung Um
This projects aims to investigate ways augmented reality technologies and applications could be used by and provide benefit to warehouse operations in the future. Among others, the project will look into wearable technologies, such as the Google Glass and the way the can be used in enterprises. The student will examine this issue both from a more theoretical point of view (potential business cases, differences with current systems) and a practical point of view (development of application and experimental testing).
Prateek Tuladhar (prt33@cam.ac.uk)
Inter and intra warehouse performance measurement
Supervisors: Philip Woodall and Vaggelis Giannikas
This project aims to develop a new method for comparing the performance of warehouses despite the various differences between different warehouses and their operations. The method should also be applicable for the case where a single organisation has multiple, distinct warehousing projects that need to be compared; so that the organisation can identify weakly performing projects and target improvement initiatives towards these.
Brian Robertson (br356@cam.ac.uk)
Towards the implementation of interventionist strategies in warehouse picking
Supervisors: Vaggelis Giannikas and Wenrong Lu
Conventional, static order picking requires batch formation to generate static pick-lists, which is time-consuming and insufficient to cope with the rising number of daily orders and the decreasing lead time. In order to shorten the response time, Dynamic Order-Picking (DOP) Systems that allow for changes of a pick-list during a pick-cycle have been introduced. Such systems can also be useful for managing disruptions occurring in a warehouse. Previously developed DOP systems constrain the picker to travel on a heuristic (non optimal) route and will only update the picker if the pick-location for the new order is on the picker’s assigned route. A recently developed algorithm, however can re-assign a new route optimally based on the newly received orders regardless of the picker’s location. This project aims to investigate appropriate ways for using this algorithm to develop interventionist picking strategies as well as evaluate the benefits of these strategies in an industrial context.
Moritz Schattka (mms61@cam.ac.uk)
Experimental analysis of disruption management strategies in production systems
Supervisor: Alena Puchkova
This dissertation will evaluate the resilience of different disruption management strategies through the simulation of a production system and will validate the results by laboratory experiments. Firstly, the dissertation will identify the most important strategies for disruption management in the literature. A production system will be modelled in a computer simulation and the disruption management strategies will be applied. The model will then be subject to a number of disruptions to determine the strategies respective resilience. In order to verify the computer simulations, experiments will be conducted in the automation laboratory. The setup of the experiment will mimic the simulation model. A limited number of runs of the experiment in the laboratory will then allow the validation of the accuracy of all simulation results to a sufficiently high statistical certainty.
Julien Le Romancer (jl840@cam.ac.uk)
Lean vs resilient production: Optimization of production line configuration
Supervisor: Alena Puchkova
Lean manufacturing is now widely adopted in production. The key driver is the reduction of cost and waste (time, labour, material) by limiting inventories. The companies also have to handle disruptions that can be quality problems, resource breakdowns and uneven demand.
The first part of the dissertation will focus on spotting the conflicts between lean and resilience, such as reducing inventory vs redistributing inventory, just in time schedules (pull configuration) vs schedules with some built in redundancy and others. Production line of the lab will be simulated to test “conflicting” strategies. The goal is to find the optimal configuration of production line (best number and location of buffers, location of inspection points, pull/push strategy) that allows to reach the compromise. The last part of the dissertation will aim at using the results to give industrial guidelines for a general production system configuration.
David Enrique Gutierrez Blanco (deg31@cam.ac.uk)
Development of a resilient framework for production systems
Supervisor: Raj Srinivasan
The complex dynamics associated with manufacturing enterprises requires resilient properties. Those manufacturing organisations are faced with disruptions ranging from natural events (such as floods, hurricanes, earthquakes), transportation disruptions (such as road closures) and internal disruptions (such as quality issues, resource breakdowns, material delivery issues). The ability to identify, respond and cope with disruptions is becoming essential for these firms to operate in a global environment and be competitive at the same time. The aim of this research is to develop a framework for incorporating resilience principles within manufacturing enterprises. Concepts from supply chain resilience, risk management, manufacturing control systems, and manufacturing planning and scheduling will be explored to develop the framework. Additionally, performance measurements related to flexibility and responsiveness will be explored to determine performance indicators and techniques for assessing resilience. Case studies from various industries will be carried to collect the required data.
