SUMMARY

Railway infrastructure consists of fixed facilities that support the movement of rolling stock from one point to another. A typical railway infrastructure system comprises of the following subsystems:

(1) Track; (2) Bridges; (3) Electrical;

(4) train authorisation; and (5) telecommunications.

The life cycle of railway infrastructure components consists of the following phases: (1) planning and specification;

(2) design; (3) construction; (4) operation; (5) research; and

(6) maintenance and retirement phases.

Railway infrastructure that is reliable and safe for the movement of trains can be achieved by the execution of effective maintenance strategies. The maintenance of railway infrastructure can be effectively performed by executing the following steps of the maintenance cycle:

1. Identification of the need for maintenance; 2. Maintenance cost justification;

3. Resource allocation planning; 4. Scheduling;

5. Assignment of tasks;

6. Execution of maintenance activities; and 7. Feedback

OP ERATI ONAL RIS K ANA LYS IS FOR THE M ANAGE MENT OF R AI LWAY INFR AS TRUC TURE MAINTE NAN CE

P HUMZILE DHLAM INI MAY 2010 70

Operational risks are identified using various information sources and techniques for identifying operational risk from these information sources such as checklists, organisational charts and organisational flow charts.

The analysis of operational risk can either be performed using methods that are qualitative, quantitative or a combination of both methods. Qualitative operational risk analysis often involves the expression of operational risk in terms of risk map rating scales. The most common qualitative operational risk analysis methods are risk self assessments, risk process flow analysis and scenario analysis. Quantitative operational risk analysis involves the numerical estimation of operational risk. The actuarial approach and stress testing are examples of quantitative operational risk analysis methods. A combination of qualitative and quantitative operational risk analysis methods can be used by developing causal models. Causal modelling involves the development of graphical representations of events, their causes and a simulation that derives their cumulative probability distributions. Methods such as neural networks and Bayesian networks can be used for causal modelling.

The proposed operational risk analysis methodology for the management of infrastructure maintenance is done by developing Bayesian network causal models. Two causal models are developed for each identified operational risk for forecasting the operational risk frequency and severity. The identification of operational risk is performed during a discussion with experts by primarily using historical data. Operational risk analysis is composed of the following stages:

• causal model building;

• causal model data collection; and • causal model data processing.

Causal model building involves the identification of operational risk causes and causal model formation. Causal model data collection entails the collection of objective historical data from the organisation’s operational risk database and subjective data from face-to-face interviews with experts. During causal model data processing, the probability distribution of the operational risk frequency, severity and their causes are obtained using the collected data and a Bayesian causal network computer program. The research methodology was selected to be a case study of ARL, a South African railway organisation. The management of ARL’s infrastructure maintenance activities occur in seventeen (17) depots that are situated nationwide; each depot is managed by a depot engineer. The limitation of the current ARL operational risk methodology is that it does not assist the depot engineer to forecast the effect of railway infrastructure maintenance activities on operational risks that are caused by railway infrastructure failure.

The proposed operational risk analysis methodology allows depot engineers to forecast the following:

• the frequency of operational loss events that are caused by the failure of one or more components of their railway infrastructure region;

• the cost of rehabilitating the railway infrastructure after operational loss events such as theft, train accidents, natural disasters and sabotage; and • the impact that preventative maintenance activities can make on the

probability of the frequency and severity of operational loss events.

The case study was limited to analysing the operational risk of the train derailments that occur in the Johannesburg region caused by railway infrastructure component failure. The operational risk event of train derailments and their causes were identified using the ARL operational risk register for three consecutive financial years. Thereafter, the train derailment frequency and cost causal models were constructed. Objective data was obtained from the ARL risk register and the Johannesburg finance department database. Subjective data was obtained from face-to-face interviews with an engineering technician, chief engineering technician and a senior engineer. The probability of one contributing factor to one cause of train derailments was decreased sequentially; the resulting probability distribution of the train derailment frequency and cost of rehabilitating decreased as well.

The forecasted frequency and cost of rehabilitating railway infrastructure after train derailments were compared with the actual figures of the 2008/2009 financial year. The forecasted train derailment frequency exceeded the actual amount by 15 train derailments. Therefore, the difference between the forecasted and actual probability of a train derailment that is caused by infrastructure component failure is 0.77%. The actual rehabilitation costs exceeded the forecasted by 0.33%.

5.2 CONCLUSION

An engineer who manages the maintenance of railway infrastructure can greatly contribute to society by decreasing the amount of operational risk events that are caused by railway infrastructure component failure as a result of a decrease in the following:

• the amount injuries and fatalities of members of a railway company and the public;

OP ERATI ONAL RIS K ANA LYS IS FOR THE M ANAGE MENT OF R AI LWAY INFR AS TRUC TURE MAINTE NAN CE

P HUMZILE DHLAM INI MAY 2010 72

• the amount of legal restrictions that are imposed by a country’s government or railway safety regulator; and

• the amount of money that the company loses due to train delays, claims and the rehabilitation of railway infrastructure.

This dissertation proposes an operational risk analysis methodology that transfers the approach of operational risk analysis from a macro level to a micro level. The objective of proposing this approach is to provide engineers with the tools to manage railway infrastructure maintenance more effectively and efficiently. The proposed operational risk analysis methodology assists these engineers in forecasting the impact that maintenance activities have on operational risks that are caused by railway infrastructure component failure. This allows engineers, to forecast the probability that their targets for reducing operational risks that are caused by railway infrastructure failure will be met. Additionally, the proposed methodology enables the forecasting of the cost of rehabilitating railway infrastructure after the occurrence of an operational risk event. The proposed operational risk analysis methodology can be used during various phases of the maintenance cycle. The forecasted cost of rehabilitating the track after the occurrence of an operational loss event can be used to justify the funds that the organisation should use on the maintenance activities that can prevent these events.

The proposed operational risk analysis methodology was made for the engineers that maintain railway infrastructure. However, other technical employees of railway companies can use it. Additionally, engineering consulting companies can use this methodology to assist companies in decreasing the amount of operational risks that are caused by railway infrastructure.

A more detailed causal model is likely to produce more accurate forecasts; the author suggests that the following contributing factors of operational risk causes may be added for increasing the accuracy of the forecasts:

• the volumes of the trains that are passed, • the climate of the region,

• the resources that is available for preventative maintenance etc.

An increase in the train volume results in an increase in the probability of wear occurring on the rail. Defects such as broken rail can occur as a result of wear. Extreme temperatures can potentially result in defects such broken rail and slack. The lack of resources for preventative maintenance increases the likelihood of operational risk events occurring.

5.3 RECOMMENDATIONS FOR FURTHER