Risk Assessment

Natech risk assessment and mapping, which is the primary functionality of RAPID-N, is carried out in two steps. In the first step, the structural damage probability of plant units located at plants is calcu-lated by using the hazard parameters of a scenario natural hazard as input. For this purpose, the on-site hazard parameters are calculated either by using hazard-specific property estimators, or by using pre-calculated values available as hazard maps. Manual entry of on-site hazard parameters is also sup-ported. The damage probabilities are calculated for possible damage states by utilizing fragility curves. By using risk states relating damage states to major events, probable consequence scenarios are deter-mined. The U.S. EPA RMP Guidance for Offsite Consequence Analysis methodology (U.S. EPA, 1999) is used to estimate the distance to the endpoints for toxic and flammable substances and the results are converted into risk maps. The risk assessment records are used to select natural hazard scenarios, define settings for natech risk assessment, store output data, and visualize the results.

The data fields of the risk assessment records are listed in Table 1.

Table 1. Risk assessment data fields

For each risk assessment a descriptive name should be specified. The date of the risk assessment is automatically assigned by the system. The type of risk assessment can be indicated as public or private. Public risk assessments are available to everyone, whereas private risk assessments can only be ac-cessed by the owners. The triggering natural hazard, for which risk assessment will be performed, should be selected from the available hazards in the database. Both historical and scenario hazards can be used. If hazard maps are available for the selected natural hazard, they are listed automatically in the hazard maps drop-down list. Selecting a hazard map from the list enforces the system to use the hazard map instead of property estimators to calculate on-site hazard parameters at the target plants.

For the estimation of damage probabilities, the system is capable of determining the most suitable dam-age classification and fragility curve for each plant unit individually. But if needed, a specific damage classification can be defined. Selecting a damage classification activates the fragility curve drop down list. This list contains fragility curves defined for the selected damage classification. By default, a fragility curve from this list is automatically assigned to each plant unit considering plant unit properties and the flexible fragility curve selection status. If needed, the system can be forced to use a certain fra-gility curve for all plant units by selecting a custom fragility curve from the list. Plant units, which are not compatible with the selected damage classification or fragility curve, can be exempted from risk assessment by activating the "Evaluate compatible plant units only" checkbox. Because plant unit information is mostly not readily available, many plants in the database may not have plant unit data. Normally, risk assessment is not possible for such plants. However, to give at least an idea about possible damage probabilities, the system puts imaginary typical plant units at such plants during the risk assessment and reports the results accordingly. If this feature is not needed, plants without plant units can also be excluded from the analysis by activating "Exclude plants without plant units" option (Figure 1).

By default, plants that should be covered in the risk assessment are automatically determined by the system. To speed-up the analysis, a user-defined cutoff distance is used to eliminate plants far away from the origin of the hazard. If the analysis should be limited to a selected plant, it should be selected from the plants available in the database. For a user-defined plant, the system allows manual on-site hazard parameters entry in addition to calculating hazard parameters by using property estimators or the hazard map. On-site hazard parameters should be unique, but fuzzy numbers can be specified for numerical parameters.

For consequence analysis and natech risk assessment, RAPID-N uses damage parameters specified in the risk states. For a given natural hazard scenario and taking plant unit properties into consideration, the system determines suitable risk states and in the order of their precedence, uses their damage parame-ters as input parameters for the calculations. It is also possible to enter custom damage parameters spe-cifically for each risk assessment. If they are available, custom damage properties overwrite damage pa-rameters obtained from the risk states. Custom damage parameters should be indicated in the damage parameters section. Fuzzy numbers are supported for numerical parameters.

Once all natech risk assessment data are specified, the risk assessment is carried out in two main steps. In the first step, the natural hazard damage estimation is performed to determine the damage probabil-ity of plant units located at the plants due to the natural hazard. In the second step, likely major events are identified and the severity of the consequences are calculated by using the U.S. EPA RMP Guidance for Offsite Consequence Analysis methodology.

The damage estimation procedure follows the steps listed below:

1. In the first step of the damage assessment, the plants for which the risk assessment will be done are determined. If a custom plant is selected, it is used for the analysis. Otherwise, for each plant available in the database, the distance to the origin of the hazard is calculated and compared to the cutoff distance. Plants that are within the cutoff distance are selected for risk assessment. If the "Exclude plants without plant units" option is activated, plants without plant units are ex-cluded from the list of selected plants.
2. For each selected plant, on-site hazard parameters are calculated. If a hazard map is specified, first the hazard parameters found in the hazard map are interpolated to find the parameter values at the plant location. Then property estimators are used to estimate additional parameters by using the interpolated hazard parameter data as input. Plants located outside the hazard map boundary are excluded from the risk assessment. If no hazard map is specified, source hazard parameter data from the system’s database is used as the input data and on-site hazard parameters are calculated by using property estimators. If on-site hazard data is available either as a on-site hazard data record for the plant or as manual input to the risk assessment (for the selected plant), they are directly included in the list of on-site hazard parameters.
3. For each plant, plant units that should be included in the risk assessment are determined. For this purpose, plant unit properties are determined for each plant unit by using available proc-ess unit property estimators and taking user-defined plant unit properties as input. If a certain fragility curve is specified for risk assessment and the "Evaluate compatible plant units only" op-tion is checked, plant units that are not compatible with the fragility curve are excluded from the analysis. Otherwise all plant units are used.
4. For each plant unit, a suitable fragility curve is determined. If a fragility curve is specified explicitly for the risk assessment, that fragility curve is used directly. Otherwise, the validity conditions of the available fragility curves are tested against the plant unit properties and fragility curves appropri-ate for the plant unit are added to the list of candidate fragility curves. If a damage classification is specified explicitly for risk assessment, only the fragility curves defined for the damage classification are used for the evaluation of candidate fragility curves. Among the candidate fragility curves, the curve with the highest number of validity conditions is selected as the fragility curve for the damage estimation. If no candidate curves are found, the system checks the "Flexible fragility curve selec-tion" option. If this option is active, the curve fulfilling the maximum number of validity conditions is selected for the damage estimation. Otherwise, the plant unit is excluded from the risk assess-ment procedure.
5. By using on-site hazard parameters calculated in step 2 and the fragility curve selected in step 4, damage probabilities are calculated for each damage state of the fragility curve.

The risk assessment procedure follows the steps listed below:

1. If the damage probability calculated for a plant unit is less than 1%, the unit is excluded from the risk assessment.
2. If information on substances found in the plant unit is not available, the damage probability of the plant unit is reported and the unit is excluded from the risk assessment.
3. For each damage state having an occurrence probability greater than 1%, the risk state correspond-ing to the damage state is found. For this purpose, risk states available in the database for the dam-age state are examined and their validity conditions evaluated for the plant unit. Similar to the suitable fragility curve selection procedure, the risk state for which the validity conditions are ful-filled and which has the highest number of conditions is selected for the risk assessment. Both proc-ess unit and substance properties are used to evaluate validity conditions.
4. According to the damage parameters of the selected risk state and custom damage parameters specified in the risk assessment record, the distance to the endpoint calculations are conducted by using the U.S. EPA RMP Guidance for Offsite Consequence Analysis methodology (U.S. EPA, 1999). RAPID-N includes a self-implementation of the U.S. EPA RMP methodology by using the property es-timation framework. All equations and rules required for the analysis are entered into RAPID-N as property estimators. Hence, suitable combinations of equations are automatically selected by the system. Use of the property estimation framework allows parameters used by the equations to be modified and alternative equations to be substituted easily. For example, endpoint distance criteria, which are set as second-degree burns, can be changed for first or third-degree burns, or the atmos-pheric dispersion model, which uses lookup tables, can be substituted with a more advanced model. By modifying the damage parameters different scenarios can be evaluated without any modifica-tions to the system code and without any disturbance to the existing system functionality.

The results of the risk assessment are presented as summary reports and interactive risk maps showing natech event probabilities and the areas possibly affected by the events.

A sample risk assessment report and corresponding risk map are given in Figure 2.