Civil Engineering Reference
In-Depth Information
For better understanding the concepts that are discussed in this section it is
necessary to gain an adequate knowledge of the uncertainty and sensitivity con-
cepts. Therefore I will briefl y introduce the main concepts related to uncertainty and
sensitivity analysis, and then describe how these tools can improve our energy con-
scious building design. For detail discussion on this subject see (Saltelli et al.
2008
)
and (Aven
2003
).
13.3
Uncertainty and Sensitivity Analysis Defi nitions
By defi nition an uncertainty analysis (UA) is a tool that helps the modeler to study
the changes in the output of his model when he changes the inputs to his model. The
modeler gains multiple benefi ts from the uncertainty analysis. The most important
advantage of performing an uncertainty analysis is that it can help the modeler to
make better decisions by better understanding the system which is under investiga-
tion. This better decision making can be achieved from the fact that uncertainty
analysis provides proper ground for more informed discussions and communica-
tions among the professionals whom are working on designing or improving a
system by making quantifi ed output values accessible for them.
Sensitivity analysis (SA) on the other hand is a useful method that helps the
modeler to calculate the uncertain parameters which have the most infl uences on the
outputs of the model. There are many apparent benefi ts to this type of analysis
because once the most infl uential parameters in a model (system) are identifi ed
more attention can be placed on improving the real-life targets by improving the
most effective elements of the system.
13.4
Uncertainty Quantifi cation Techniques
In general two main approaches for quantifi cation of uncertainty are known to be
external and internal approaches; both of these rely on statistical techniques.
13.5
External Methods
In an external method approach the mathematical structure of the simulation soft-
ware are assumed to remain unaffected. It means the methodologies that are used to
calculate different outputs will not change. Instead here the external factors such as
how does the simulator select the inputs and how does he describe the model are
assumed to be changed. Therefore this method basically evaluates the effects of
uncertainty from the outside of the simulation program. It means that in an external
method in a computer simulation model, the simulator changes the input parameters
