PSA tools produce many kinds of results. Some are common to most of the tools, e.g., probability/frequency of some group of consequences, importance factors, sensitivity analyses. They fall into different categories. The following three categories are so frequent that is it worth to normalize the way they are stored into XML files.
A first (and good) way to encode minimal cut sets consists in using the representation of formulae defined by the Model Exchange Format. However, it is often convenient to attach some information to each product, which is not possible with the formulae of the Model Exchange Format. An alternative XML representation for sums of products (sets of minimal cut sets are a specific type of sums of products) is given in Listing 9.1. More attributes can be added to tags “sum-of-products” and “product” to carry the relevant information.
sum-of-products =
element sum-of-products {
name?,
attribute description { text }?,
attribute basic-events { xsd:nonNegativeInteger }?,
attribute products { xsd:nonNegativeInteger }?,
product*
}
product =
element product {
attribute order { xsd:positiveInteger }?,
literal*
}
literal =
element basic-event { name }
| element not { element basic-event { name } }
Listing 9.1 The RNC schema for the XML representation of sums-of-product
Statistical measures are typically produced by sensitivity analyses. They are the result, in general, of Monte-Carlo simulations on the values of some parameter. Such a measure can come with moments (mean, standard deviation), confidence ranges, error factors, quantiles, etc. The XML representation for statistical measure is given in Listing 9.2.
statistical-measure =
element measure {
name?,
attribute description { text }?,
element mean {
attribute value { xsd:double }
}?,
element standard-deviation {
attribute value { xsd:double }
}?,
element confidence-range {
attribute percentage {
xsd:double { minExclusive = "0" maxExclusive = "100" }
},
attribute lower-bound { xsd:double },
attribute upper-bound { xsd:double }
}?,
element error-factor {
attribute percentage {
xsd:double { minExclusive = "0" maxExclusive = "100" }
},
attribute value {
xsd:double { minExclusive = "0" }
}
}?,
quantiles?,
}
quantiles =
element quantiles {
attribute number { xsd:positiveInteger },
quantile+
}
quantile = element quantile { bin-data }
bin-data =
attribute number { xsd:positiveInteger },
attribute mean { xsd:double }?,
attribute lower-bound { xsd:double }?,
attribute upper-bound { xsd:double }?
Listing 9.2 The RNC schema for the XML representation of statistical measures¶
Two or three dimensional curves are often produced in PSA studies. A typical example is indeed to study the evolution of the system unavailability through the time. The XML representation of curves suggested by the Model Exchange Format is given in Listing 9.3.
curve =
element curve {
attribute name { xsd:NCName }?,
attribute description { text }?,
(attribute X-title { xsd:string },
attribute Y-title { xsd:string },
attribute Z-title { xsd:string }?)?,
(attribute X-unit { unit },
attribute Y-unit { unit },
attribute Z-unit { unit }?)?,
element point {
attribute X { xsd:double },
attribute Y { xsd:double },
attribute Z { xsd:double }?
}*
}
unit = "seconds" | "hours"
Listing 9.3 The RNC schema the XML representation of curves