Manufacturing methods for assessing the impact of wrinkles in wind turbine blades

Wrinkles are defects prone to occur during the manufacturing process of wind turbine blades. Wrinkles can emerge at various locations of the structure, exhibiting a range of diverse shapes. Different manufacturing steps can lead to different defect types. This work presents different manufacturing methods to embed artificial wrinkle defects in laminates representing a typical defect that can be found in a wind turbine blade. Several methods are tested at coupon scale to design a critical wrinkle defect that can lead to the blade structural knock-down on the mechanical performance during high cycle fatigue operation. Following the selection of the defect types for investigation at the test coupon scale, the corresponding defect types are embedded while manufacturing a 12.6 m wind turbine blade.


Introduction
In the production of large structures, such as wind turbine blades, manufacturing procedures can lead to different sources of defects.Wrinkles are one such defect that can initiate failure mechanisms, including fiber fracture and delaminations [1].Different steps in the manufacturing process can lead to different types of wrinkles.It can arise from the poor attachment of plies due to low interlaminar shear stress affecting compaction during layup, geometric changes on tooling surfaces, errors during operation, flaws in the layup process, distorted fibers, abrupt change in ply thickness, among other sources [2,3].Wrinkles can emerge at various locations assuming diverse configurations, of which out-of-plane waviness appears at a higher frequency [4].Several configurations are mentioned in the literature, divided into two major groups: uniform and nonuniform ply waviness [5].The embedded uniform pattern is a configuration replicated in labs making it feasible to examine laminates with such defects [6].Additionally, folded wrinkles are considered an extreme case in which the wrinkle is folded back to itself with a maximum deviation of the fiber misalignment [4,7].Concerning the presence of gaps in a laminate, it can create regions with a high tendency of resin accumulation after curing [8], referred to as resin pockets.Different methods are adopted when it comes to producing artificial defects in a lab [9,10,11,12].Wrinkles are introduced in a laminate during vacuum-assisted resin transfer molding (VARTM).A wrinkled shape is placed at the top of a vacuum bag, and pressure is manually applied to it after the infusion process but before the resin is fully cured [9].An alternative method is adopted to induce wrinkles in a sandwich laminate by introducing a plastic rod between a face sheet and a balsa core, followed by applying a vacuum to hold the defect shape.Subsequently, 1293 (2023) 012029 IOP Publishing doi:10.1088/1757-899X/1293/1/012029 2 the vacuum is released, the plastic rod is removed, and the vacuum is reapplied with the resin infusion carried controlled at an angled table [10].A different approach involves the utilization of 90-degree thin strips of pre-impregnated fibers (prepreg) positioned at specific locations of the laminate to induce the formation of wrinkles without adding foreign material [11].
The present work outlines various manufacturing methods leading to out-of-plane wrinkles of different severity types for a glass fiber reinforced (GFR) composite laminate.The severity types elected are submitted to a screening fatigue test at R=-1 to define a defect type that falls in the criteria for assessing its impact on realistic cycle conditions for fatigue failure of a wind turbine blade.Within the field of mechanics of materials, a screening test is a preliminary test performed to identify characteristics or properties of a certain structure.After selecting the severity types through the coupon test, the same types of wrinkles are embedded in a 12.6 m wind turbine blade.
This work proposes a protocol for selecting wrinkle types likely to occur in a wind turbine blade.The protocol integrates a method chain comprising defect design, manufacturing procedures, testing programs, and qualitative evaluation of fracture characteristics.The ultimate purpose is the selection of manufacturing procedures to create out-of-plane wrinkles similar to those found in a wind turbine blade that can potentially lead it to failure at high cycle fatigue.The protocol is implemented by upscaling the defect manufactured at the test coupon level to the structure level.The method chain is presented in section 2.

Methodology 2.1. Types of Wrinkles
Feasible wrinkle profiles for glass fiber wind turbine blades serve as the basis for the adopted defect configurations evaluated in this work.The five wrinkle types illustrated in Figure 1 were produced using three distinct manufacturing processes: parameter control, non-controlled parameter, and resin pocket.The methods parameter control and resin pocket use a cured UD insert for the central wrinkle, while the non-controlled parameter creates a fold based wrinkle.Wrinkle types one to four were created using a layup of [+45/ − 45/0/0/0/0/0]s, while a layup of [0/0/0/0/0]s was applied for defect type five.The root section, a structurally thicker region, exhibits a higher sensitivity to wrinkles formations when compared to other sections.To ensure accurate representation of the defects 3 being examined, it is crucial that the thickness of the test coupon specimens replicates the equivalent section thickness of the laminate, which imitates a region near the root section of the wind turbine blade.In addition, a critical factor in selecting the defect parameters was to ensure the defect size was sufficiently small to keep it in the blade during operation without repair.The specimens were produced and sliced in 25 mm wide samples resulting in uniform volumes measuring 400 × 25 × 11 mm 3 .The manufacturing procedures are detailed in section 2.2.

Manufacturing Methods
The present investigation has explored three distinct manufacturing methods, which are depicted and identified in Figure 2, Figure 3, and Figure 4, respectively.One of the limiting factors in the manufacturing of test coupons with artificial defects is the need to ensure consistent replication of the geometrical parameters within a given batch of samples.
The parameter control method (Type 1, 2 and 3) ensures that predefined geometrical parameters, such as amplitude and wavelength, are maintained during production.The process is divided into two distinct phases.Phase 1 entails manufacturing an insert that conforms to the pre-defined geometrical parameters through the utilization of a cast mold.The insert comprises two unidirectional (UD) layers arranged on the aluminum cast mold in a stacked configuration.The defect amplitude and aspect ratio are defined according to the specifications outlined in Figure 1.The wrinkle defect is introduced transversely to the orientation of the UD fibers.The vacuum-Assisted Resin Infusion (VARI) process is used to impregnate the UD plies with resin.The Hexion epoxy resin system uses a mixing ratio of 100/28 (by weight).The curing process involves treating the resin-impregnated insert to a temperature of 40 • C at 12 h, followed by subsequent heating at 80 • C at 10 h.The objective of Phase 2 is to integrate the UD insert into the symmetry axis of the layup sequence [+45/ − 45/0/0/0/0/0]s along with the remaining plies.The second Biax (biaxial) ply is folded to align with the center of the insert containing the wrinkle defect.The folding process emulates the undulation that may occur during the layup process for the blade and, therefore, can be the reason for causing wrinkles.Additionally, the folded wrinkle facilitates increased compaction of the material during the infusion process, reducing the resin content between the undulated plies.The same resin mixing ratio and curing cycles implemented during Phase 1 are applied to Phase 2. The manufacturing procedure is repeated for the classification of wrinkles type 1, type 2, and type 3, as depicted in Figure 1.
The non-controlled parameter method (Type 4) comprises a single phase as indicated in Figure 3.The only wrinkle present is the result of the fold at the second biax layer in the composite layup.Subsequently, the layers are assembled with the same layup and undergo the same resin infusion procedure while adhering to the equivalent curing cycle employed in the parameter control Methodology.The laminate contains the wrinkle classification type 4, resulting in a less pronounced wrinkle and a reduced potential for retention of resin pockets.
The resin pocket method (Type 5) follows two phases, as pointed out in Figure 4.The initial phase is focused to produce a two-layer UD insert by applying the identical manufacturing principle adopted in the initial phase of the parameter control method.Subsequently, in phase 2, the aforementioned insert is embedded at the center of a ply stacking sequence comprising [0/0/0/0/0]s.The stacked configuration is confined between two aluminum plates placed at an angled position for a controlled resin infusion process.This approach effectively contains the formation of wrinkles in the pre-cured UD insert, thereby mitigating its propagation to adjacent plies in the stacking sequence.By not implementing the biax fold underneath the insert, the space between the flat ply and the wrinkle insert is filled by resin during the infusion process.The current approach adopts identical curing cycle and temperature conditions as those described in the previous methods.

Testing Procedures
The purpose of the experimental tests is to select a manufacturing process and wrinkles size to be embedded into a DTU designed 12.6 m blade [13,14] to investigate delamination growth initiated from wrinkle defects during a fatigue test campaign.The DTU blade is described in references 13, 14.Screening fatigue experimental tests assess the structural performance of the produced laminates with diverse wrinkle classifications.The tests aim to identify a wrinkle classification that could persist within the blade structure undetected, ultimately resulting in high cycle fatigue damage.A wrinkle classification is defined by the manufacturing method and the geometrical features.
The experimental protocol is performed for fatigue with R = −1, normally defined as load or stress ratio with R = ε min /ε max .The fatigue test is carried out on a single representative sample for each wrinkle classification.Whereby considered representative of the full batch of samples within the same defect classification.Figure 5 show the test setup incorporating consumer-grade Nikon D7500 cameras to capture the cross-sectional and outer surfaces of the specimens.The cameras track the crack initiation and propagation during the test.Additionally, a thermal camera is installed to track the temperature gradient associated with the fracture progression.Extensometers are positioned at a distance of 40 mm from the center of the wrinkle to enable strain measurement.The order of tests follows the indication in Table 1 and Figure 6.The experimental program starts with the sample manufactured with the resin pocket method for wrinkle classification type 5.The linear strain amplitude is established to its maximum limit avoiding off-axis buckling.It is kept constant throughout the fatigue test.As the strain amplitude applied drove the representative sample to low cycle failure, the subsequent test was performed using a reduced strain amplitude compatible with what is intended to be used for the mid-span area of the 12.6 m blade.The representative samples produced with the parameter control method and classification type 1 and type 2 were tested with high cycle failure observed.As the latter two classification types fulfilled the requirements for the defect choice, defect classification type 3 was not tested further.Considered less critical by the smooth wrinkle configuration, the characteristic sample for non-controlled parameter classification type 4 was tested in a staircase approach.The test strain amplitude is initially incremented from classification type 2 and gradually increased if crack propagation remains undetected.The latter test resulted in a runout without failure detected after 4 million cycles.

Fracture Characteristics
After evaluating the test images, it was determined that the dominant fatigue failure mechanism is represented by a critical crack at the maximum amplitude wrinkle, resulting in multiple delaminations along the interfaces and a kink band as the final failure.The failure sequence for each of the samples tested for the various manufacturing processes are depicted in Figure 7.
The characteristic samples classification type 1, type 2, and type 5 experience a critical crack length around the wrinkle insert during the fatigue test.After testing the sample resultant from the non-controlled parameter method, it was noticed that a crack initiates around the fold ply.However, propagation is not observed along the interface.The specimens manufactured through the parameter control method passed the initial assessment adhering to the failure criteria in high cycle fatigue.Additionally, the wrinkle defects embedded through the latter method are considered sufficiently fine to remain within a thick blade section despite surface repair.Regarding manufacturing, the latter method is equally feasible for embedding in the blade.Defect classification type 5 is ruled out due to difficulties controlling the throughthickness wrinkle shape propagation during blade manufacturing.Therefore, the parameter control method is selected to assess wrinkles impact on the blade undergoing fatigue loads.Wrinkle classification type 1 and type 2 are chosen to compare the impact of two severity levels in the blade.These selected wrinkle defects are selected for embedding in the DTU 12.6 m blade to evaluate crack growth in wrinkle defects under high cycle fatigue conditions.
Figure 7: Crack initiation and propagation detected during fatigue tests for various defect configurations.

Defect in the Blade
The defect classifications selected at the coupon scale are embedded in a 12.6 m blade.The defect is embedded in the spar cap on the pressure side mid-span within a region where the strain amplitude can drive crack growth for wrinkle type 1 and type 2, as shown in Figure 8.The parameter control manufacturing method is adjusted to accommodate the structural up-scaling.Wrinkle type 1 and type 2 are manufactured with the same method, applying cast molds designed to meet the pre-defined geometric parameters unique for each wrinkle type in phase 1.The method is changed at phase 1 by producing an insert with one UD layer submitted to evenly spaced cut-outs a the fiber direction, performed before the infusion.Subsequently, the cut-out spaces are filled in with dry UD fibers.This procedure is necessary for the viability of the resin infusion process in the blade.Manufacturing trials conducted for a thick laminate using the insert produced in phase 1 in its pristine version proved that it restrains the resin infusion, resulting in a laminate with the accumulation of dry spots.After modification of the method, the defects are effectively embedded in the blade without dry spots.

Conclusions
This work presents a protocol to select a manufacturing method for evaluating wrinkles impact on the fatigue life of a wind turbine blade.Different out-of-plane wrinkle defects are derived from the manufacturing methods presented.Of the three methods presented, one method fulfilled the criteria concerning the feasibility of manufacturing up-scaling and subsequently expected fatigue delamination crack growth during high-cycle fatigue loading.After evaluating the fatigue damage mechanisms, delamination is found to be the driving mechanism of fatigue failure.The proposed methods were implemented and tested, demonstrating the concept for future studies to investigate how different wrinkles severity classifications can impact the structural performance of wind turbine blades.

Figure 1 :
Figure 1: Cross-section of different types of wrinkles defects obtained from various manufacturing methods with an assigned layup and geometrical parameters.

Figure 2 :
Figure 2: Schematics of the manufacturing method parameter control represented in two phases.

Figure 3 :
Figure 3: Schematics of the manufacturing method non-controlled parameter represented in a single phase.

Figure 4 :
Figure 4: Schematics of the manufacturing method resin pocket represented in two phases.

Figure 5 :
Figure 5: Experimental setup for testing coupon specimens embedded with various classifications of wrinkle defects.Yellow is the side view, blue is the surface view and red is the thermal camera.

Figure 6 :
Figure 6: Testing program depicting the number of cycles versus percentage strain with regions outlined for crack initiation and final failure.

Figure 8 :
Figure 8: Defects classifications type 1 and type 2 embedded in a 12.6 m blade.

Table 1 :
Testing order and corresponding strain amplitude for each defect classification.