The decision tree is a hierarchical structure of risk items used to calculate the critical value of a risk focus. You can define any number of formulae to weight the risk items by importance, and you can define as many risk-item levels as necessary.
Establish the decision tree based on different levels of risk and then associate the tree with a piece of equipment or groups of equipment. The only real user input is at the ranking level. In the illustration below, the ranking levels are shaded with diagonal lines. The ranking level at each branch offers you a list of values derived from the branch rankings at the bottom level of each branch, which are shaded with a checkerboard pattern.
In the example illustration, the top item in the tree represents the risk focus. There are two general "branches" to the hierarchy that contribute to the risk focus critical value: Consequences and Probability. Consequences are the particular consequences of a failure of a specific piece of equipment. Probability is the likelihood that a failure will occur. The lower branches in the tree represent specific items that contribute to risk determination.
For junctions at non-ranking level branches in the tree, you can perform the following functions:
Weight any individual item in the hierarchy. Assign weight to the item based on relative importance of the item (using a multiplier) to permit only the min/max ranking value to roll up the hierarchy.
Specify, for any individual item, a formula that calculates a score value based on the score values flowing into the junction from below. That formula is typically an arithmetic expression involving summation and/or multiplication and/or division and/or subtraction and/or averaging. However, formulae are not restricted to only arithmetic expressions. For example, there is a min/max function that permits selecting the minimum/maximum score among two or more of the factors at the next lower level.
Values of the next lower level can be referenced using the assigned code for the child. For example, the formulae 2*C1 + 3*C2 will take 2 times the result of the C1 branch plus 3 times the result of the C2 branch. To get the minimum value of all the children, use LEAST (C1, C2…Cn), and to get the maximum value, use GREATEST (C1, C2…Cn).
Normalize any score flowing into a junction, if required, so that dissimilar values can be more reasonably compared and manipulated within the context of accurate and effective risk assessment.
|
Datastream strongly recommends that you carefully plan the decision tree on paper before defining the tree in Datastream 7i. |
Follow these steps to define the decision tree.
Open the Equipment risk assessment form (INRISK).
Query for the risk focus for which to define the decision tree.
Click the Decision tree tab. Datastream 7i displays the Decision tree page.
The left side of the form displays the structure of the decision tree; the right side of the form contains 3 blocks, which display detailed information about the tree.
Complete the following steps in the top section of the form. This section is the Parent section and applies to the current branch of the risk focus.
Normalize—Select to indicate that the value calculated by the associated formula should be replaced by a normalized value.
Ranking level—Select to indicate that the branch is a ranking level, meaning that information flows up from lower branches to form a list of values for this branch. Once you identify a branch as a ranking level, the tree hierarchy can only go one level deeper.
Out of service—Select to indicate that this branch should not be used for critical value calculations.
Formula—Enter the formula Datastream 7i uses to calculate the numeric value for the branch. Click Validate to verify the formula.
Complete the following steps in the middle section of the form. This section contains the normalization functionality to apply to the Formula in the Parent section block.
Minimum value and Maximum value—Enter the minimum and maximum values for the risk item (current branch) to establish a range of scores.
Normalized value—Enter the numeric value linked to the current range. For example, if the minimum value is 10 and the maximum value is 20, a normalized value is 1. Therefore, any value that is greater than 10 and less than or equal to 20 (with the Normalize check box selected) will obtain a normalized value of 1.
Complete the following steps in the lower section of the form. This section contains the children of the risk item in the top section.
Code—Enter a unique code identifying the child.
Description—Enter a description of the child.
Ranking value—Enter a numeric value to describe the rank of the item. You can only enter a ranking level value if you selected Ranking level for the parent risk item.
Choose File | Save from the menu bar. Datastream 7i saves the information to the database. To define other levels of the tree, click on the appropriate branch in the Hierarchy section on the left side of the form. Then repeat the steps listed above.