-------------------------------------------------------------------------------------
A LOW-FIDELITY MODEL FOR THE PRELIMINARY DESIGN PHASE OF FULL-LATTICE WIND SUPPORT
STRUCTURES: DATASET USED FOR THE DATA-DRIVEN MODEL.
-------------------------------------------------------------------------------------

GENERAL DATASET DESCRIPTION:
This data set includes two pieces of information: (1) the frequency of the first five
modes of a large variety of straight lattice support structures and (2) the first 
five frequencies and the tuning coefficients relative to the low-fidelity model
created for each of the lattice structures. This information was used to develop a
surface response model in the work of [1].

[1] Vergassola M, Cabboi A, van der Male P, Colomés O. A low-fidelity model for the
	preliminary design phase of full-lattice wind support structures. 2022. Preprint
	available at DOI: 10.13140/RG.2.2.35071.89760.

For questions you can contact Marco Vergassola at M.Vergassola@tudelft.nl

NOTE: The dataset is available under the licence CC BY-SA 4.0
-------------------------------------------------------------------------------------

PRACTICAL INFORMATION:
The data are stored in .mat files and divided per mode frequency, from 1 to 5. The
acronym HFM and LFM stands for high- and low-fidelity model, respectively. The parameters
considered to generate the lattice structures (HFM) are: braces diameter, legs diameter,
width of the bay, height of the bay and number of bays. These are also linked to the
dimensions of the matlab arrays. The structural parameter were varyed as follows:

diameter_braces = 0.5:0.25:1.5;
diameter_legs = 1:0.5:3.5;
width_bay = 4:2:22;
height_bay = 4:2:22;
num_bays = 4:1:12;

The dimensions of these five arrays were used to create the results arrays. The entries
of the array are regulated by the parameter arrays. For example, to retrieve the 2nd
natural frequency of a lattice structure having diameter_braces = 0.5, diameter legs = 1.0,
width_bay = 4, height_bay = 4 and num_bays = 4, one has to go to LFM_freqs_mode2(1,1,1,1,1).
If one wants to retrieve the same frequency but for a structure having a larger value of
diameter_legs, let's say diameter_legs = 2, then the correct entry would be
LFM_freqs_mode2(1,3,1,1,1). Not all the geometric combination were meaningful, therefore
some constraints were imposed on the dataset:

a) 0.8 <= height_bay/width_bay <= 2.5
b) height_bay*num_bays > width_bay

All the matrix entries corresponding to cases that violate at least one of these constraints
are set as 'NaN'. For the LFM, a extra .mat file is present, namely LFM_tuning_coeffs. These
matrix contains the three tuning coefficients obtained from the minimization problem explained
in the referenced research and a check value. These are stored in the same fashion as the
frequencies but the matrix has an extra dimension to allocate the four aforementioned value,
i.e. tuning coeff 1, 2 and 3 plus the check value. The check value is an indication of the
success or not of the minimization problem. If this value  equal 0 then the minimization problem
was solved without any problem and down to the required tolerance. If this value equal 1, the
minimization problem was terminated by an error or by the limit of iterations. In either cases,
the result should be discarded.