Validated dynamic flow model: Aeolus deliverable no.: 2.5

The purpose with this deliverable 2.5 is to use fresh experimental
data for validation and selection of a flow model to be used for
control design in WP3-4. Initially the idea was to investigate the
models developed in WP2. However, in the project it was agreed to
include and focus on a additive model structure suggested by
University of Lund the WP4 leader. This particular model structure
has the advantages that it fits better into the control design frame
work used by WP3-4 compared to the model structures previously
developed in WP2. The different model structures are first
summarised. Then issues dealing with optimal experimental design is
considered. Finally the parameters are estimated in the chosen
static and dynamic models and a validation is performed. Two of the
static models, one of them the additive model, explains the data
well. In case of dynamic models the suggested additive model turns
out not to be useful for prediction of the flow. Moreover, standard
Box Jenkins model structures and multiple output auto regressive
models proves to be superior as they can give useful predictions of
the flow.