Risk analysis

What is a risk analysis

A risk analysis is the study of random events that may have a negative effect on the human health, the environment, property, market reputation and production regularity. Failure rate and the severity of an unwanted consequence is used to define risk through a risk matrix. Risk acceptance criteria are used to determine if the risk is acceptable or not. Risk analyses are proactive, meaning that they identify and evaluate the risk before any losses occur. New and modified technology, procedures and operations where potential losses may occur should be subjected to a risk analysis. The main steps in a risk analysis are boundary definition, identifying problems, causes, consequences, criticality ranking and finally suggestion and evaluation of risk reducing actions. Most people and engineers carry out risk analyses in everyday life without even thinking about it. The reason for this is that these analyses are done informally. The problem with these analyses is the lack of formal methodology and documentation. This prevents other people from participation in the analysis work, but also in reviewing the risk analysis and in transfer of new knowledge between employees. Several aspects are important to achieve a good risk analysis including knowledge about the object to be analysed, participants ownership to the method, practical tools to support decisions and finally multidisciplinary involvement from the end user.

Qualitative risk analyses

Qualitative risk analyses methods are used to describe single events and thereby not the total system risk. However, a well-performed qualitative risk analyses is much more valuable than quantitative analysis based on limited and uncertain input data. The strength of the qualitative risk analysis is to assess every aspect that may affect the risk by systematically reasoning, relying on expert competence and previous experiences. Especially in the case where limited input data is available the qualitative analysis preferable. The following links give introductions to commonly used qualitative risk analyses. Most descriptions also include a guideline on how to actually perform the analysis.

• Operational problem analysis (OPERA)
• Failure Mode Effect and Criticality Analysis (FMECA)
• Action Error Analysis (AEA)
• Specific Operational Risk Analysis (SORA)
• Reliability centered maintenance (RCM) analysis
• HAZOP
• CRIOP 
• Safe job analysis (SJA)
• Software analysis

Quantitative risk analyses

The quantitative risk analyses use experience data, such as failure rates of failure modes, as input and mathematical models to analyse the risk. Use of software tools for calculations are normal. As opposed to the qualitative risk analyses, the quantitative risk analysis is very well suited to assess the total system risk. Quantitative analyses should include a sensitivity evaluation to see how much the results are affected by variations in the input data. The following links give introductions to commonly used quantitative analyses (no guidelines).

• Fault tree analysis (FTA)
• Event tree analysis (ETA)
• PDS method
• Markov analysis
• Quantitative risk analysis (QRA)
• Safety and emergency preparedness analysis (SEPA)

Further discussion

Typical sources for information when discussing causes and consequences are:

• Accident reports
• Incident reports
• Accident investigations
• Questionnaires
• Interviews
• Technical reports
• Expert judgements
• Basic cause databases
• Calculations
• Models
• Tests

Have control of the following before starting a quantitative analysis:

• Use of internal statistics may be risky due to small samples (number of events)
• External statistics may be risky due to different item design and operational conditions
• Underlying statistical distributions doesn't always hold (will have item burn-in and aging effects)
• Statistics may be risky since it may not include any relevant event observations

Input statistics are often generic data for different technology and different physical and environmental conditions. Data adjustments may be needed but are often not easily made. The quantitative analysis are most robust when used for relative comparisons.