QRA - Personnel Risk Metrics
In this article:
Introduction
When preparing a QRA for assessing the fatality risk, risk metrics need to be selected for calculation. In some instances, this is guided by local regulations specifying mandated metrics and associated Risk Acceptance Criteria (RAC). In other, regulations might provide overarching directives, requiring the quantification of risks to individuals and groups without specifying the exact metrics. In mature industries such as Oil and Gas, the operator often defines the risk metrics and RAC.
This insight aims to simplify the understanding of risk metrics to be applied for a QRA in process industries, and how to choose and use the right risk metrics in different scenarios.
More specifically, the insight will define and discuss important aspects of personnel risk metrics that are typically applied in QRA for process industries. The discussion will include applications, similarities, advantages, and disadvantages. The risk metrics discussed are:
Fatality Accident Rates (FAR) – Estimate the frequency of accidents with fatal consequences;
Individual Risk (IR) – Assessing individual risk from different angles;
Potential Loss of Lives (PLL) - A metric showing potential lives lost in case of an accident;
In addition, the insight will discuss and recommend how to apply different plant operating modes and time-limited operations.
Personnel risk metrics
Several different risk metrics have been developed for QRAs. The optimal risk metrics to apply for a specific QRA will depend on the project and purpose of the QRA. Even though it is not significant extra work to add additional risk metrics in a QRA, it is advisable typically to limit it to 2-3 risk metrics that are complementary (not overlapping), which captures the purpose of the QRA. Some risk metrics express the same risk in slightly different ways and are not recommended to be used together in the same QRA. This insight will define and discuss important aspects of personnel risk metrics that are typically applied in QRA for process industries.
Fatality Accident Rate (FAR)
The FAR value is the number of fatalities in a group of personnel per 100 million exposed hours. The number of 108 exposed hours is roughly equivalent to the number of hours at work in 1000 working lifetimes.
FAR is therefore normally used to express risk to 1st and 2nd party personnel that are part of the operation of a specific plant rather than risk to 3rd party (risk to society).
It is important to note that FAR is based on the number of hours of exposure e.g. hours that a group of individuals spends on a plant or are performing specific operations. Therefore FAR is a good risk metric for summarizing historic accident statistics of different types of plants, operations, trades, etc. that can be applied for direct risk level comparison.
It also gives some distinct advantages when comparing risk levels of permanently manned plants with intermittently manned plants compared to some of the other risk metrics. A topic that will be discussed later in the article.
The FAR can be calculated for different types of personnel groups operating a specific plant and will in this case be an expression of the individual risk of this personnel group. The average FAR of all personnel groups can also be calculated for a plant. Typically, a RAC for both the average FAR as well as for worst risk personnel group i.e. a stricter requirement is often used for the average than the personnel group with the highest risk.
It is also possible to calculate the FAR of a specific area of a plant to see what areas have the highest risk.
In the case of FAR for onshore plants, the exposure hours are based on the hours at work whereas for offshore plants normally the hours spent offshore are applied even though approximately half of the time will be off-duty spent inside an accommodation.
Individual Risk (IR)
IR is a metric that quantifies the risk to an individual within a specified area over a given period. It's often expressed as the probability of a person being subject to a specified level of harm (e.g., fatality) per unit of time. Individual risk in this insight will include discussions for the following aspects:
Location Specific Individual Risk (LSIR);
Individual Risk Per Annum (IRPA)
ISO-risk curves.
Location-Specific Individual Risk (LSIR)
LSIR is used to indicate the risk at a particular location e.g. area or module of a plant. It is the risk for a hypothetical individual who is positioned for 24 hours per day, 365 days per year. For onshore QRA, the geographical variation of LSIR is normally represented as iso-risk contour plots which is discussed in its section below.
Since in reality, no people will remain continuously at one location, this is not a realistic risk measure of what an individual is exposed to. However, it provides a good overview of which areas of a plant are hazardous and where to avoid having personnel located if not necessary. Often RAC is not applied for LSIR even though in principle it could make sense as a criterion when designing a plant.
LSIR is normally applied as an intermediate result in QRA’s for calculating the individual risk of different personnel groups.
Individual Risk Per Annum (IRPA)
IRPA is applied to establish the risk of an individual being part of a certain personnel group operating a plant. Examples of personnel groups could be administrative personnel, process operators, maintenance technicians, etc. The IRPA for different personnel groups is then calculated for different personnel groups taking the actual time spent on average per year in different areas of the plant. Time not spent in the plant is normally per definition assumed to have zero risk as it has nothing to do with the plant. The exception is the transport risk to and from an offshore installation.
It is important to identify all different personnel groups that have a significant difference in risk levels when calculating the IRPA or at least make sure that the worst-risk case personnel group has been identified. Normally an individual risk exceeding 10-3 per year is considered intolerable for a process plant.
There are some complications with IRPA if individuals do not work full time on the same plant but may spend time on different plants. This will be discussed separately later in this insight.
ISO-risk contour plots
In onshore QRAs, it is normally always required to calculate Individual Risk contours or ISO-risk contour plots. It is curved with a certain LSIR value typically 10-5 per year, 10-6 per year, 10-7 per year, etc. surrounding the plant are calculated. These are used to assess the risk to 3rd parties or the public for land zoning purposes. For instance, residential areas should be located outside a certain ISO-risk contour e.g. hospitals outside an even lower ISO-risk contour.
As discussed under LSIR the ISO-risk contours will be based on that people are always present 24 hours 365 days a year. It is even also often assumed that people will not attempt to escape or move away from exposure to a hazard.
ISO risk contours are normally not a good risk metric for an offshore plant or inside the fence of an onshore process plant as the spatial risk variation inside the plant will often be discontinuous and very regional. Hence ISO contours are most useful for assessing risk to 3rd party outside the plant boundary.
An example of ISO-risk contours has been provided below:
Potential Loss of lives (PLL)
PLL is probably the simplest risk metric to understand. Simply put it is the estimated or projected number of fatalities of operating a plant for a certain period of time. PLL is typically expressed as fatalities per year. The PLL can be broken down in PLL for different personnel groups and also in PLL for 1st / 2nd parties and 3rd parties respectively or it can simply be the total combined PLL.
The PLL to 1st and 2nd parties is very dependent on the number of personnel taking part in operating a plant and due to this dependence normally a RAC is not applied for PLL.
PLL is a useful risk measure for As Low As Reasonably Practicable (ALARP) evaluations considering different plant concepts and cost-benefit analyses (CBA). Here the delta PLL of different concepts for the lifetime of the plant is converted into a monetary scale (roughly speaking by putting a value on human life) and compared to the cost of implementing the different plant concepts.
Group risk - F-N curves
Group (or societal) risk is the risk experienced in a given period by the whole group of personnel exposed. It reflects the severity of the hazard and the number of people in harm's way. The group risk is often expressed as F-N curves showing the relationship between the cumulative frequency (F) and the number of fatalities (N).
F-N curves are frequency-fatality plots, showing the cumulative frequencies (F) of events involving N and more fatalities. They are derived by sorting frequency-fatality pairs of each outcome of each accidental event and summing them to form cumulative frequency-fatality coordinates for the plot.
Hence F-N curves are graphical measures of group risk that show the relationship between frequency and size of the accident. F-N curves are typically more difficult to calculate than other risk metrics and can be confusing to interpret. Especially for onshore QRAs as it may involve people in surrounding areas of the plant at significant distances from the plant.
The special interest in group risk is related to that it is generally more accepted by society that a single person is killed in more frequent events than a group of people (say 10) is killed in a single event that is less frequent even though the PLL of both accidents may be the same.
RAC for F-N curves are curves themselves that the calculated F-N-curve needs to be below in all points (N-values) to meet the RAC.
An example of F-N curve and RAC has been provided below:
Different plant operation modes
Some plants may operate in different operation modes during the year. E.g. an offshore oil & gas production platform may have the following operation modes during a year:
Normal operation;
A combined operation with production and a drilling rig drilling a new well or performing a well intervention on existing wells;
Construction and maintenance campaign together with production.
Here two approaches can be taken for quantifying the risk. Either calculate the average risk metrics of the different operation modes weighted to the duration of the individual operation modes during the year or calculate the risk for each operation mode separately.
It is highly recommended to deal with this situation by calculating the risk for each operation mode separately for the following reasons:
If the risk of different operation modes is averaged you may have an operation with a risk exceeding RAC that only becomes below the acceptable risk levels because it is preceded or followed by a period of low-risk operation. It is considered good practice that risk levels are below the RACs in all operation modes;
Duration of different operation modes may vary significantly from year to year and you end up trying to estimate a worst case. This problem is avoided by considering operation modes separately.
When assessing risk separately per operation mode there is still the option to calculate the average risk of the year in addition in case this risk is of interest.
Time-limited operations or intermittent operations
Some operations may be of limited duration or a plant may only be manned intermittently e.g. an offshore normally unmanned installation (NUI). This can represent a challenge when assessing if the risk is acceptable compared to RAC. For instance, an individual may spend 10% of his work time on a specific NUI per year and work 90% on one or more other installations. Here it is important to reflect on what risk is of interest in the specific case. Is it the risk the individual experiences in reality or is it the risk level of the NUI specifically that is of interest? If you are designing the NUI, it is the risk level of the NUI that is of interest, and risk averaging with other installations should be avoided. In this case, it is recommended to assume the individual spends all his work time during the year on the NUI as a hypothetical case when calculating IRPA. Alternatively, the IRPA RAC should be reduced to 10% of the yearly accepted RAC. Otherwise, you can design a very hazardous installation that is only acceptable because personnel only spend part of their time on the installation. This complication for IRPA is avoided by applying FAR that does not suffer from this problem.
A similar situation arises for time-limited operations. If QRA methodology is applied for assessing time-limited operations by comparison to annual RAC there is a big risk of accepting a very hazardous operation just because it is short-term and is averaged out by other low-risk operations. Either the time-limited operation should be evaluated for the operation mode separately as if it goes on continuously or QRA methodology and normal RAC should not be applied.
