REPAIR TECHNIQUES FOR COMPOSITE MATERIALS APPLICABLE TO WIND TURBINE BLADES D. J. Lekou1, I. Velasco Mateos2, K. Rossis1, A. M. van Wingerde3, T. K. Jacobsen4, P. Vionis1 1
CRES, Wind Energy Dept., 19th km Marathonos Ave., 19009 Pikermi, Greece, Tel. +30 2106603300, Fax. +30 2106603301, email:
[email protected]
2
FIBERBLADE EOLICA, S.A., Engineering Dept., Edificio 1-G, Plaza del Europa s/n, Ciudad del Transporte 31119, Imarcoain (Navarra), Spain, Tel. +34 948 314489, Fax. +34 948 314070, email:
[email protected] 3
Knowledge Centre WMC, P. O. Box 523, 2600 AM Delft, The Netherlands, Tel. +31 15 2783729, Fax: +31 15 2782308, email:
[email protected]
4
LM Glasfiber A/S, Rolles Moellevej 1, Lunderkov, DK-6640, Denmark, Tel. +45 79840473, Fax: +45 75586202, email:
[email protected]
SUMMARY Within the framework of an EU-funded research project, entitled OPTIMAT BLADES, repair techniques are evaluated in order to assess their suitability in applications on composite material parts of wind turbine blades. The objective of these investigations is to develop repair methodologies applicable to the load carrying laminates of wind turbine blades, so as to avoid possible rejection of products both during production and service life. This becomes more crucial as blades become larger and even a localized deficiency will lead to the destruction of the whole blade, if repairs are not accepted as a standard procedure. The typical repair procedure, namely the damage identification of the part, the decision for repair, the removal of the damaged area, the application of the repair patch and the quality inspection of the repair, is step by step surveyed and adapted for the special needs for application on wind turbine blades. Practical stress calculation procedures are selected from a literature review and proposed for the evaluation of the restored strength during the repair design. Applications are experimentally evaluated through tensile testing both under static and fatigue loading of repaired coupons. Test results are also used to compare the repair methods applied. Among the evaluated techniques, the most promising method was found the scarf repair with a slope of 1:50. Test data reveal that a strength restoration of over 80% can be achieved. INTRODUCTION Within the framework of a European funded research project, entitled OPTIMAT BLADES (ENK6-CT2001-00552), repair techniques are reviewed and adapted in order to enable their suitability in applications for wind turbine blades. The objective of these investigations is to develop repair methodologies applicable to the load carrying laminates of wind turbine blade structures, so as to avoid possible rejection of products both during production and service life. Currently there are no recommendations available for repairing structural parts of blades or for assessing the loadcarrying capability of damaged areas. Thus, blades those are damaged or have productions deficiencies in their laminated structural parts are being destroyed even if the damage or deficiencies are local. As the blades become larger more material is wasted due to such localized deficiency. To this end, the typical repair procedure, namely the damage identification on the structural part, the decision for repair, the removal of the damaged area, the application of repair and the repair quality inspection [1], is step by step surveyed and adapted for the special needs of wind turbine blades application. The location, type and importance of damaged zone are identified and semi-empirical stress calculation procedures are proposed, as selected from literature (e.g. [1], [2]), for a preliminary repair design. A minimum target value for the repair efficiency is stated with respect to both strength and stiffness. Repair techniques are surveyed and evaluated on aspects like complexity, suitability to highly stressed, load carrying laminated parts and application on site if possible. The most promising techniques are selected and applied to the repair of flat specimens. Within an extensive testing campaign these repaired specimens are tested in uniaxial tension and results with respect to both strength and exhibited stiffness are compared with that of flawless specimens, which are tested to form the necessary baseline. Applied non-destructive inspection techniques for the quality assurance of the repaired area are also shortly assessed. Static test data show that the most promising from the selected repair techniques is the scarf repair with a slope of at least 1:50, which leads to a strength restoration of over 80%. Verification of behavior under fatigue loading is carried out for the selected repair method with similar results. REVIEW OF REPAIR METHODS Within the framework of OPTIMAT BLADES it was selected to concentrate on flaws that are found on parts of the primary structure of the wind turbine blades, e.g. the girder part of the blade and not the sandwich parts that include structural foam. Among the various defects or damages that can be found on that location the restoration was chosen to be limited on holes and delaminations. Reviewing the available repair methods for application to composite
materials in general, two were found suitable for the repair of the load carrying laminates on wind turbine blades, namely the scarf repair and the plug/patch. For both patch repair techniques of interest the damaged laminate is cut out down to the depth of the deepest flaw and either a scarf or plug/patch repair scheme is applied. Fig. 1 displays the two repair systems selected.
(A)
External patch
Filler Parent structure
(B) Scarf patch
Smooth or small steps Fig. 1 Plug/patch (a) and scarf (b) repair systmems The basis of the repair design follows logical repair criteria. Some of the parameters of repair criteria, which were addressed during the analysis, at least from a theoretical point of view are listed in Table 1. Table 1 Repair criteria # 1 2 3
Criterion Static strength and stability Repair durability Stiffness requirements
Assessment Full (80%) strength restoration; Deformation resistance Fatigue load spectrum Deflection limitations; Load path variations
Due to geometrical constraints for the testing of coupons the damage was modeled as a through the width of the specimen delamination by a channel width of 10mm. The proportion of the laminate that is considered to be "damaged" is taken as either the 1/3 or the 2/3 of the total thickness. For the scarf repair, in the current approach, the "damaged" region is removed, leaving a straight channel. The straight channel is then tapered to a predetermined slope and over this region a patch with tapered (or stepped ends) is bonded to the parent laminate. This repair configuration has the benefit of a nearly uniform shear stress distribution in the adhesive layer. In addition, due to the lack of eccentricity in the load, the patch peel stresses are low. Therefore, scarf repairs are considered highly efficient and are particularly suited to external repairs of thick laminates (similar to those in wind turbine blade load carrying parts) because of the unlimited thickness of material that can be joined and the smooth surface contour that can be produced. Single-sided scarf patches can also be employed to repair partthrough or full-penetration damage. Scarf repairs are usually based on patches with a ply configuration similar to the parent material and patches are generally co cured to avoid the severe fit-up problems encountered with procured patches. To cure the patch and adhesive, pressure may be applied by a vacuum bag - heater blanket procedure. The process used during the manufacturing of the coupons was selected by each manufacturing participating. For the scarf repair case, with simple analysis, when the patch and parent laminate stiffnesses are balanced and the thermal coefficients of expansion matched, the shear and normal stresses are given by the following equations respectively [1]: P sin (2θ ) τ= Eq. 1 2t P sin 2 (θ ) σ= Eq. 2 t The allowable load in the joint is given by the next relationship [1]: 2t Pall = Eε all t = +D Eq. 3 tan θ where t is the laminate thickness and D is the hole diameter (equal to the delamination pattern). In case of a plug/patch repair, none of the damaged plies need to be removed. The repair scheme is to adhesively bond a doubler patch over the damaged region and stiffen it. However, in the current approach for application of the repair to the flat coupons, the "damaged" region should be removed, leaving a straight channel. The cavity was then
filled Over this region a patch with tapered (or stepped ends) is bonded to the parent laminate. This repair configuration is similar to that of a tapered single-overlap joint; the taper is most important to reduce peel and shear stresses that would otherwise cause failure of the patch. Nevertheless, a taper is not required for patches of only a few ply thickness, which was also the case considered. External patches can be employed reasonably successfully, depending on the stressing requirements of the area, to repair skins of thickness up to about 16 plies. This type of repair will be the most widely employed, since external patches are relatively easily applied under field conditions. Strength recoveries of 50-100% of ultimate allowables of the parent material can be achieved, depending on the laminate thickness. The main problem with external patches is that, there is an eccentric load path that results in quite severe bending in the patch and peeling stresses in the adhesive and composite. Out-of-plane bending under compressive axial loads can also significantly reduce the buckling stability. However, these effects are greatly reduced if the patched region is supported by a substructure, such as foam core, that reacts out the bending. Several options exist for the patch; it may be made with similar ply configuration of the parent laminate; or it could be made with a quasi-isotropic lay-up or a standard lay-up (to reduce the danger of lay-up and application errors, in which case it would require being thicker than the parent laminate. The patch may be: (1) formed over the parent laminate from pre-preg tape cut to shape and then co cured with the adhesive, (2) procured in layers and bonded to the parent laminate during the repair with interleaved layers of adhesive, or (3) preformed to shape and then bonded to the parent material in a subsequent operation. This last option produces the best patch properties; however, since the preformed patch is not compliant, serious fitting problems may arise on curved surfaces. Determination of the patch stiffness is such that the stiffness of the sublaminate and patch leads to a critical buckling load greater than the applied design allowable load [1]. Or more simply, ensuring that the stiffness of the patch is equivalent to the undamaged stiffness, calculated by: E h E p = lam Eq. 4 tp EXPERIMENTAL CAMPAIGN An extensive experimental campaign has been conducted to find out the applicability of the repair methods selected. To this end, more than 100 static tension tests and 60 fatigue tests have been carried out on Glass/Epoxy coupons. The coupon dimensions have been selected so that the scarf repair could be applied, which resulted in using longer coupons than the usual 250mm specimens, as shown in Fig. 2. Composite material coupons in final dimensions, including the tabs, were delivered by either Gamesa Eolica SA or LM Glasfiber A/S and were tested either at CRES or WMC, according to the test program. Plate lamination sequence is [(±45/0)4/±45], where the 0• reinforcement coincides with the axis along the length of the coupon. Strain measurement was performed on a back to back configuration by either strain gauges or clip gauges to monitor both stiffness and eventually exhibited bending strains. Testing configuration as used at CRES is shown on the left in Fig. 3 and in a little more detail as used at WMC on the right of the same figure.
25
55
500
55
610
6.57 2
Fig. 2 Coupon dimensions Section dimensions have been measured at least at five locations along the gauge length for all specimens, due to their longer than standard length. It should be noted that a large thickness variation was observed on the repaired coupons. Due to the larger than the standard length of the repaired coupons, tests on coupons of the reference material (without repair) with the same geometry (and testing conditions) were conducted, so as to attain comparable results, eliminating any geometrical affect.
To investigate the effect of the repair depth, two batches of coupons were tested for each repair method selected; the one with the repair at 1/3 of the specimen thickness and the second with the repair at 2/3. Scarf repair and plug/patch repair was tested. For the investigation of the best scarf slope coupons with repair slopes 1:25, 1:40, 1:50, 1:75 and 1:100 were tested. Moreover, to examine the effect of the material form in the repair, repaired coupons with prepregs were also tested using the scarf repair with slopes 1:50 and 1:75. These results were compared with the respective ones with using liquid resin as an alternative to the prepreg. Additionally, as an alternative to the pure scarf repair with slope 1:50, an additional layer was used as a cover of the repair laminate in the form of tape. Carbon/Epoxy coupons with repair similar to the last one have also previously been tested and compared to those without an additional layer in [3]. From the static test results the scarf repair with a slope of 1:50 was selected for fatigue testing and both manufacturers delivered specimens.
Fig. 3 Test set-up (Left: CRES, Right: WMC) Static tests were carried out at stroke control mode with a constant stroke rate of 4mm/min due to the specimen length. During testing stroke, applied load and exhibited strain was continuously recorded. Fatigue tests have been conducted at load control mode at a stress ratio, R, equal to 0.1. The test frequency varied according to the findings of OPTIMAT BLADES project, so that the temperature of the coupon under test does not exceed 35°C during the fatigue loading (except just before failure). Thus the testing frequencies were in the range of 2-6Hz. RESULTS On Fig. 4 the effect of the slope of the scarf repair on the static strength of the coupons is presented, for both repair depth variants that were tested. The failure stress of the coupons has been normalized by use of the average static strength of the reference coupons. Therefore, it can be seen that there seems to be a limiting slope of the scarf, above which the strength of the repaired laminate is theoretically restored. Tests on coupons with a scarf slope of 1:40 were also conducted (but not shown on the figure), with similar results with the 1:25 slope. Thus, for the presented case, the limit is between 1:40 and 1:50. The theoretical justification behind that is that when this limiting slope is used then the shear stresses on the adhesive are lower than the shear strength of the adhesive and thus the repaired laminate does not fail due to failure of the adhesive under tension loading (but rather due to axial stresses on either the parent or the repair laminate). It should be noted, however, that the limiting slope depends on the properties of the adhesive and the parent laminate in the structure. That means that for each material system the limit is not the same. Nevertheless, for all systems tested during the current research it was found that a slope of 1:50 suffices. Moreover, it should be mentioned that due to the variation in thickness observed for the repaired coupons, all results have been normalized by use of a nominal thickness. Similar with the previous graph, on Fig. 5 the effect of the various repair configurations on the static strength is shown for repairs at the 1/3 of the coupon thickness. It is noted that the results shown for the scarf repair refer to a slope of 1:50. Similar results were obtained also for repairs at 2/3 of the thickness. From this graph the method that stands out is the scarf repair with the use of prepreg layers for the repair patch with a restored strength of over 90%
and scatter comparable with that of the reference coupons. But it should be made clear, that the method that can be used to arrive at the best results depends on the manufacturer and the material system used. As already mentioned, during this study, the manufacturers participating were free to use the repair method that suited their manufacturing procedures, without limitations on curing cycles, repair material, etc. Thus, they could experiment on repair techniques, comparing the repair result, such as the repair with liquid resin instead of prepreg. Scarf Repair at 2H/3
Scarf Repair at 1H/3 1:50
1:75
Reference
1:100
1.20
Strength / (Average strength of ref.)
Strength / (Average strength of ref.)
Reference
1.00 0.80 0.60 0.40 0.20 0.00
1:50
1:75
1:100
1:25
1.20 1.00
0.80 0.60 0.40
0.20 0.00
1
2
3
4
5
1
2
Coupon Number
3
4
5
Coupon Number
Fig. 4 Effect of the slope for scarf repairs on static strength (Left: Repair at 1H/3, Right: Repair at 2H/3) Reference Plug/patch Scarf - Liquid resin + tape
Strength / (Average strength of ref.)
Repair at 1/3 H (scarf slope 1:50)
Scarf - Prepreg Scarf - Liquid resin
1.20 1.00 0.80 0.60 0.40 0.20 0.00 1
2
3 Coupon Number
4
5
Fig. 5 Effect of different repair methods and material on the static strength of the coupons Regarding the exhibited stiffness, results are within 10% of the reference elasticity. It should be noted that the elasticity obtained is usually not reported for repaired coupons in the literature. Even the DOT/FAA report [3], which evaluates in detail the scarf repair on Carbon/Epoxy composite material, does not present strain response. Most of the studies concentrate on the strength obtained by the repair. However, in the case of repairing blade stiffness is also of interest since for large blades, at least for the current state of the art blades the driver force during design is stiffness and not strength. Nevertheless, although the stiffness of the repaired coupons was found similar to the parent structure, from the back to back strain readings it was also noticed that the stress-strain behavior of the repaired patch was different from that of the parent. This fact might imply a not so good transfer of stresses from the parent to the repair laminate, but be also due to the testing configuration, which uses shear for the load transfer from the testing machine to the specimen. For the fatigue tests, as already pointed out, it was decided to use scarf repair with a slope of 1:50. Fatigue results for both repair depths studied, are presented in Fig. 6 for the first material alternative. The experimental data show that the repair affects the slope of the S/N line, since at higher fatigue cycles both reference and repair coupons are quite close. Obviously the S/N curve for the repaired coupons lies lower than that of the reference, since during static experiments the strength of the repaired coupons was found to be in the order of 80% of the reference ones for this material combination. Moreover, the results for the deeper repair depth are inferior to that of the shallower, but both are quite close and no decisive conclusion can be made by these data only. Experimental data from the fatigue tests of coupons from the second material variant are shown in Fig. 7. For this case, the slope of the S/N curve seems also to be affected by the repair, but exactly opposite than for the former material. On the same graph the straight lines, indicate the stress corresponding to the strain on the blade at the extreme load case (4000 microstrain) and the working strain on the blade (2000 microstrain).
0.70 Reference Normalized Stress
0.60
H/3 2H/3
0.50 0.40 0.30 0.20 0.10 0.00 1000
10000
100000 Cycles to failure
1000000
10000000
Fig. 6 Fatigue results for different repair depth (scarf repair with slope 1:50) 0.90 reference repaired
Normalized Stress
0.80 0.70 0.60 0.50 0.40 0.30
blade extreme stress - 4000 microstrain
0.20
blade working stress - 2000 microstrain
0.10 0.00 10
100
1000
10000 100000 Cycles to failure
1000000
10000000
Fig. 7 Fatigue results of alternative material (scarf repair slope 1:50) Non Destructive Testing A selected number of specimens from each batch (reference, repaired at H/3 and repaired at 2H/3) were tested using the ultrasound technique, prior to any loading of the coupons. Purpose of this non-destructive testing was to see if there could be a clear distinction between reference and repaired coupons and to check whether the repair technique used left some defects that could be traced, as for example a delamination. To this end sensors with different characteristics were used, in order to select amongst them the most suitable. Sensor frequency ranged from 0.5 to 5MHz and their diameters from 6mm to 25mm. Sensors of different concept, e.g. contact transducers acting as sender and receiver at the same time, and dual element transducers were tested. Nevertheless, the combination of specimen geometry (which is rather narrow) and tested sensors did not produce any reliable results, therefore, it was decided that applying the ultrasound method for these coupons with the selected sensors was not adequate. CONCLUSIONS With the aim in the long run to reduce the cost of wind turbine blades, by extending their service life and cut down the rejection of blades during production, an investigation has been undertaken in the framework of a European Commission funded project (OPTIMAT BLADES) on the application of repair techniques for load carrying composite material blade parts. From the available techniques on repairing composite material parts two were found to stand out for application on wind turbine blade load carrying parts, namely the plug/patch and the scarf repair. An extensive experimental campaign was undertaken to verify theoretical predictions and select the most attractive (in terms of, among others, strength restoration, ease of application, possibility of on-site application) configuration of repair. To this end, tests on flat repaired Glass/Epoxy coupons were conducted under tensile static as well as fatigue loading. Test results showed that the scarf repair with a slope of at least 1:50 provides a restored static strength of nearly 95% with adequate material use. Tension-Tension fatigue results have been also judged as acceptable for this repair configuration. Within this work the stiffness was also investigated, since it is rather important for the performance of
the blade during operation. Thus, in terms of stiffness the attained elasticity was found comparable to the parent structure. In the near future the study will be continued by focusing on the one hand the behavior of the load carrying part under compressive stresses again both under static and fatigue loading, on the other hand on the effect of curvature on the restored strength of wind turbine blade parts. Other details of repair techniques applicable to composite material blades, such as curing cycles, environmental effects, etc. should also be investigated in order to arrive at a set of recommendations for the manufacturers leading to reliable repairs prolonging the operating life of the blades and accepted by accreditation bodies. ACKNOWLEDGEMENTS The research is supported by the European Commission Fifth Framework Programme, within the Energy, Environment and Sustainable Development Programme under contract number ENK6-CT2001-00552 (OPTIMAT BLADES). REFERENCES
[1] Heslehurst RB, Analysis and modeling of damage and repair of composite materials in aerospace, in Numerical analysis and Modelling of Composite Materials, ed. Bull JW, publ. by Blackie Academic & Professional, 1st edition, Chapman & Hall, 1996 [2] Hoskin BC, Baker AA, Composite Materials for Aircraft Structures, AIAA Educational Series, American Institute of Aeonautics and Astronautics, Inc., 1986 [3] FEDERAL AVIATION AUTHORITY, Repair of Composite Laminates, DOT/FAA/AR-00/46. Washington: Office of Aviation Research, 2000