ABSTRACT
Since the power generation capacity of a wind turbine has a direct correlation with increased blade length there is a general trend in the wind industry to move towards larger blades. Critical design issues associated with larger blades are: weight, strength and blade stiffness, reliability, manufacturing, installation and service costs and testing. Definitely, full scale testing becomes quite expensive. Therefore, testing components reduced in size and containing critical parts like adhesive bond lines seems to be an interesting alternative. As the adhesives are one of the main load carrying materials in many modern wind turbine blades, the bond line strength investigation is of vital importance. Recently, linking the gap between coupon size material characterization and the material performance on a blade structure, a lot of effort has been invested in the development of sub-component tests for structural evaluation prior to the construction of a prototype. In the present study, two GFRP I-beams are loaded incrementally to failure. The particular subcomponents simulate the complex stress state developed in the adhesive bonding between the spar cap and the shear web connection of a blade. The bond line has a thickness of 5 mm. Typical Acoustic Emission (AE) load-hold proof tests are performed at intermediate loading stages, in order to locate and characterize damage processes at relatively low loads.
INTRODUCTION
During the last decade the size of the Wind Turbine Rotor Blade has increased to over 60m and is expected to reach 90m within the next years. The load carrying components are made from materials that have a high weight and cost-specific strength as well as stiffness. For the same reason Wind Turbine Rotor Blades are adhesively bonded structure. Structural evaluation under realistic stress fields will contribute to optimize the blade design and reduce costs. As the adhesives are one of the main load carrying materials in many modern Wind Rotor Turbine Blades, the bond lines are often of great importance. Lately, a lot of effort has been invested in the development of sub-component tests for material evaluation and possible certification prior to the construction of a prototype blade. The investigation of the sub-components mechanical behavior can be beneficial for the structural characterization under realistic stress field in terms of damage initiation and propagation mechanisms [1-2].
Non Destructive Testing (NDT) techniques are widely used to assess damage accumulation in tested materials, components or systems. Particularly, Acoustic Emission NDT technique has been already implemented to monitor Wind Turbine Rotor Blades during static and fatigue tests [3-8].
Acoustic Emission (AE) is defined in E 1316, Terminology for Nondestructive Examinations, as the class of phenomena whereby transient elastic waves are generated by the rapid release of energy from localized sources within a material. An AE sensor (usually piezo-electric based) mounted onto the component, detects the high frequency mechanical shock wave and converts its into an electronic signal that is amplified by a preamplifier and processed by the AE instrument. AE is generally transient, occurring in discrete bursts. AE systems process these bursts as AE "hits" by analyzing various aspects of the waveforms associated with each hit.
The AE system is very sensitive and can detect much weaker signals than those normally audible. The signals can be characterized in terms of their features such as Amplitude, Energy, rise time etc. Their on-line or post analysis can reveal failure location and damage processes taking place in the structure. AE monitoring can also be used to determine damage severity [9-10].
AE events can be located in different ways. The simplest one is zonal location, in which the event is picked up by one sensor. Linear Location means the location is calculated between two sensors using the difference in time of arrival of the AE waves. Planar location means the location is calculated using the difference in time of arrival of the AE wave on three different sensors providing the x-y coordinates.
In the present study, two I-beams, representing the adhesive loading of the spar cap to web bonding are loaded incrementally to failure. The I-Beam assembly comprises two flanges of a multidirectional layup made of GFRP (upper and lower), a shear-web sandwich-like structure with a polymer foam core and GFRP skins and a 5 mm thick epoxy bonding paste joining flanges with shear web. The first I-Beam is incrementally loaded up to failure in order to measure the Ultimate Strength (US). During the test the structure is monitored with AE. The second I-Beam is subjected to intermediate loading stages while typical AE load-hold proof tests are performed in order to locate and characterize damage processes at relatively low loads.
I-BEAM DESIGN
Two different Beam geometries are tested. One with a symmetric and one with an asymmetric web design. The height of the beams is 110mm and the bond line height 5mm. The bond line width is 13mm for the symmetric and 19mm for the asymmetric beam. The length of both beams is 1 meter. The I-beam consists of three different materials. The flanges and the skin of the sandwich shear-web are made of glass/epoxy. The lay-up of the flanges is [±45,04]2. The thickness of the 45o lamina is 0.325mm and the thickness of the 0o lamina is 0.7426mm. That leads to a nominal thickness of 5.25mm for each flange. The lay-up of the shear web’s skins is [±45]2, with average thickness 2.6mm. The core of the shear web is a closed cell, cross-linked polymer foam that combines high specific stiffness and strength. The thickness of the web core is 10mm. A thick bonding paste is used to glue together the three parts, the two flanges and the shear-web. The adhesive is a solvent free epoxy based bonding paste. The system is hot thixotropic and particularly suitable for bonding laminates.
TEST PROCEDURE AND INSTRUMENTATION
Two cantilever I-Beam tests are performed. Four aluminum pieces are glued, using the commercial adhesive Araldite 420A/B on the two sides of the upper flange and the bottom side of the lower flange. The aluminum pieces are used in order to adjust the I-Beams to the loading frame. A piece of the shear web is removed in order to increase the section modulus (the further the material of a structure is from its neutral axis the larger the section modulus becomes and the larger the section modulus the larger the bending moment resistance). Finally, two wooden blocks are placed between the load introduction point and the I-Beam in order to protect the flange and avoid local damage. Figure 1 illustrates how the I-beams are adjusted to the loading frame.
FIG 1. The adjustment of the I-Beam to the loading frame.
The first I-Beam is incrementally loaded up to failure while the second I-beam is loaded in steps. The load level of each step is defined according to the US of the I-Beam measured in the first experiment.
The two I-Beams are instrumented with one LVDT, positioned in the center, measuring the deflection during the uniaxial bending test. A PAC SPARTAN system, PCI-DSP4 (4 channels) is used while surface mounted AE sensors, R15 with 40dB internal preamplifier, monitor the activity. The sensors are located in such a way to perform Linear Location. At the start of each test, lead break tests are performed to check the mounting of the sensors, as specified in standard ASTM E976-94 [11]. The I-Beams are loaded in the flapwise direction in displacement control using rate 2.5mm/min. In addition to the aforementioned instrumentation, Digital Image Correlation (DIC) is used in order to measure the resulting strain field in the bonding paste. Results of the DIC will not be discussed in this paper.
Figure 2a,b presents the set up of the two experiments. In the first experiment, the transducers are positioned on the shear web while in the second one on the upper flange, for safety reasons. Figure 3 depicts the load hold test profile which consists of four blocks. The first block is a triple loop of a loading up to 20% of the US, holding for 60 sec and unloading. The second block is a double loop of a loading up to 40% of the US, holding for 30 sec and unloading. The third block is two single loops of a loading up to 60% of the US, holding for 20 sec, unloading and reloading up to 80% of the US, holding for 10 sec and unloading. Finally the fourth block is a single loop up to 90% of the US holding for 10 sec unloading and reloading up to failure. The first and the second block are named ‘nominal operating load ‘, the third block ‘examination load’ and the forth block ‘maximum test load’.
FIG 2. The position of the transducers during the first (a) and the second test (b).
FIG 3. The Load Hold Unload Reload test Profile.
RESULTS
The first I-Beam fails at 17.1KN. Figure 4 presents a 3D bar distribution graph of the hits versus time per channel. Figure 5 presents a 3D bar distribution graph of the planar event location (counts to peak are selected to be plotted for each event). Figure 6 depicts the position of the four transducers along the shear web. Useful information can be extracted by examining figures 4,5. The first and the second transducer record events which might originate from the friction between the two wooden blocks and the loading piston. A hits peak (transducer number 3) at 500 sec indicates the existence of a defect approximately located at x=0.5m and y=0.075m. However this defect does not seem critical for the structural integrity of the I-Beam.
FIG 4. A 3D bar distribution graph of the hits versus time for each channel.
FIG 5. 3D distribution bar graph of the planar event location (counts to peak are selected to be plotted for each event)
FIG 6. The position of the four transducers in the shear web.
Post failure analysis of the I-Beam may confirm the previous observation. Figure 7a,b illustrates the damaged I-Beam. Although the failure is sudden, it can be observed that the damage is initiated at the bonding paste of the I-beam’s lower part, close to the clamped region. A crack is propagating across the adhesive, then into the adherent, where it propagated in an interlaminar mode, producing joint separation and causing severe damage to the flange.
FIG 7. The damaged I-beam
The second I-Beam is subjected to a load-hold proof test. This test is appropriate so as to investigate in which load level the damage might be critical for the I-Beam’s structural integrity. Figure 8 presents the recorded hits during the whole period of the proof holding test while a more detailed view of the events follows in figure 8. Almost zero activity is recorded during the first block – upper left graph of figure 9. During the first loop of the second block the activity starts at 3.26 KN and keeps on during the holding period. That might be an indication of damage initiation. On the other hand, during the second loop of this block there is no significant activity in the holding and unloading phase, correlating the activity of the first loop with cracks coming from the Araldite 420 A/B paste. Moreover, during the third block the felicity ratio is 0.97 which proves that the structure withstands the ‘examination load’.
FIG 8. The recorded hits during the proof holding test.
FIG 9. A detailed view of the four blocks.
Finally, the events, occurring during the unloading period of the forth block and the felicity ratio of 0.897 which is calculated in the reloading phase of the forth block, prove the existence of severe damage. The second I-Beam fails at 16.1 KN. Although the failure seems similar to the first I-Beam’s failure, it is hard to conclude if it is caused due to shear web buckling or due to failure at the bonding paste, see figure 10.
FIG 10. The damaged I-Beam
CONCLUSIONS
Two I-Beams, simulating the adhesive complex stress field of the spar cap to web bonding were loaded incrementally up to failure. The first one was incrementally loaded measuring the US while the second was loaded in steps performing load-hold proof tests. Acoustic Emission technique was used trying to locate and evaluate the damage process. Examining the results of the first test, one can notice that the AE technique succeeded to locate a defect early in the loading procedure. However, the defect did not seem critical for the I-Beams structural integrity. This observation might be confirmed by the results of the second test. These results showed that severe damage to the I-Beam appeared at almost 90% of the US.
However, in order to comprehend better the results and confirm firmly their validity, more tests and an advanced analysis of the Acoustic Emission’s recordings are needed.
