JEC COMPOSITES MAGAZINE - Issue #96 - April 2015 - 54

health monitoring

Fig. 2: Experimental setup for the first prototype (all units in mm)

the resistance of the sensors to the manufacturing process.
The experiments were performed for the healthy structure and,
subsequently, for different known damage conditions artificially
promoted in the blade. Eight different typical wind turbine damages like debondings, delaminations, impacts and cracks were artificially induced into the blade. The experimental setup is shown
in Figure 2. The annotations enclosed in circles (S1 to S6) denote
the position of a group of sensors, each group consisting of four
sensors (two at the leading edge and two at the trailing edge).
The measurements were repeated five times for each damaged
condition and 10 times for the healthy structure (5 times for the
baseline model and 5 times for verification data).
The second prototype was used, as mentioned before, to validate
the blade design and the manufacturing technology used. This
goal was achieved by means of a certification test campaign performed in LEA installations near Pamplona, Spain.
The blade was attached to a metal ring used to install the assembly on the test bench. All the tests were performed with the blade
in cantilever. The load was applied at different points depending
on the specific test. The main goal was to try to emulate the real
load distribution. An automatic load application system based on
water deposits was used. The experimental setup can be seen in
Figure 3.

Tests and models
The test campaign consisted of five different tests. The first four

tests were intended to certify the
load envelope of the blade. In
these tests, the blade was loaded
up to 100% of the designed
ultimate load in four different
positions: first, edgewise trailing
edge to leading edge (TTL);
second, edgewise leading edge
to trailing edge (LTT); third,
flapwise suction side to pressure side (STP) and finally, flapwise
pressure side to suction side (PTS). The load was applied in steps
in order to allow the OBR measurements.
The fifth test originally consisted in loading the blade to failure in
the PTS configuration. However, it was decided that if the blade
did not fail at 130% of the ultimate load, another test would be
performed. This time, artificial damage was induced in order to
"compare" the damaged condition with the healthy one (first
trial). Because this condition happened, it was possible to perform a study to compare the healthy structure against a damaged
condition.
The idea that the blade could reach 130% of the ultimate load was
based on all the known uncertainties associated with the design
and manufacturing of these prototypes, which involved new design methodologies, materials and manufacturing technologies.
Due to some issues during the test for the known artificial damage (first trial), the test had to be aborted. The blade was loaded
for a long time and some additional natural damages occurred.
The unloading process was treated like another damage condition
(second trial). Before performing the last trial, the first artificial
damage was increased in order to have three damaged conditions.
For both experiments, two different models were built with the
healthy data. The first model was based on PCA and the second
one on h-NLPCA. For further details about the methodology
and the mathematical background, the reader is directed to [1].
Later, data for damaged conditions were projected into the PCA
and h-NLPCA models and damage indices and their associated
damage thresholds were calculated. The idea was to achieve the
first level of SHM: damage detection. The most representative
damage index in the experiments performed was the Q index,
which indicates how well each sample fits the PCA or h-NLPCA
model.
Some representative results are presented in Figures 4 and 5.
Figure 4 shows the results of the experiment performed with the
first prototype, in particular the Q index for the PCA model (left)
and the Q index for the h-NLPCA model (right). Both results
correspond to the measurements from FBGs.

Fig. 3: Experimental setup, TTL position

54 jec composites magazine / No96 April 2015

JEC COMPOSITES MAGAZINE - Issue #96 - April 2015

Table of Contents for the Digital Edition of JEC COMPOSITES MAGAZINE - Issue #96 - April 2015