Financial investments in large scale engineering sectors are subject to a variety of risks which affect the productivity and the profitability of the industry in the long run. Therefore, it is essential to quantify these risks and build mathematical models for analysis, so that better and more informed decisions can be made.
This PhD research work focuses on the financial risk modelling of raw materials production engineering facilities and applies the developed concepts and tools to mining and mineral processing. These sectors are specifically chosen as there are a lot of operational, environmental and other site specific hazards involved and at present there is no streamlined methodology for financial risk assessment. The approach developed in this research uses heuristic methods and expert inputs and is given a quantitative foundation using a robust modelling technique with an engineering basis, as opposed to the qualitative and subjective methods currently employed.
Current methods of estimating risk or forecasting future contingencies rely only on past historical data of failure and insurance claims. These often miss out on the finer details of the specific process and grossly over or under estimate the risks. Many other modelling techniques based on conservation laws employ mass balance and material flow through the whole process chain to obtain a more accurate model. However, the process systems data for mass and material flows through each component of the system is difficult to obtain and in many cases is a trade secret. Thus, it is difficult for an independent assessor or an external insurer to have a clear impression of the risk involved in the operation. The modelling techniques developed in this research aim to identify an optimal balance between these two extremes and handle missing or inaccessible information in the process flow chain. As such, and beyond its academic value, the work is of significant practical value and especially useful for insurance underwriters, financial risk managers, business consultants and the raw materials production systems operators.
The thesis describes the development of a novel risk modelling tool based on a systems engineering approach. The risk model uses finite state machines with a set of logic for transitioning between the different states of the system. The flexibility of the systems risk modelling tool allows for features like incorporation of maintenance schedules, modelling of storage elements, redundancy schemes etc. which are implemented and validated with short case studies. These bespoke model refinements are useful for the process systems studied in this thesis.
The work presented next then describes how elements from queuing theory have been integrated with the proposed risk modelling methodology. This is especially useful for process systems which can be naturally modelled as a combination of servers and queues (e.g. transportation networks, automated manufacturing operations, computer networks etc.). A validation study is carried out to show the effectiveness of the proposed methodology.
The systems based risk modelling methodology, is augmented with fuzzy logic for incorporating expert knowledge and handling the uncertainties in the model parameters. This is of practical use as it allows for risk quantification in cases where only qualitative inputs are available from an external risk assessor.
Next, a Bayesian network based formalism is coupled with the systems level risk model for incorporating the effects of catastrophic risks. This allows for situations where information for the various parameters are partially available. Validation studies are conducted on both the fuzzy and Bayesian augmented risk frameworks to show the effective working of both methodologies.
Finally, the developed risk modelling tools are used to characterise the risk profiles of a mineral processing operation and various mining operations. The research would help investors and insurers in the mining and mineral processing sector to make appropriate investment decisions, by quantitative assessment of the risk factors, and hence minimise financial losses. It would also be useful for the management of the process operation to identify the bottlenecks in the system and propose appropriate risk mitigation strategies for the same. To this end, a multi-objective optimisation is also done to obtain various Pareto optimal risk vs. reliability tradeoffs for the mineral processing operation. Since the modelling paradigm used is a generic one, it is expected that it can easily be adapted to other scenarios and would be of benefit for modelling financial risks in other industrial domains.