energies
Article
Technologies of Wind Turbine Blade Repair:
Practical Comparison
Leon Mishnaevsky, Jr. 1, * , Nicolai Frost-Jensen Johansen 1 , Anthony Fraisse 1 , Søren Fæster 1 , Thomas Jensen 2
and Brian Bendixen 2






Citation: Mishnaevsky, L., Jr.;
Frost-Jensen Johansen, N.; Fraisse, A.;
Fæster, S.; Jensen, T.; Bendixen, B.
Technologies of Wind Turbine Blade
Repair: Practical Comparison.
Energies 2022, 15, 1767. https://
doi.org/10.3390/en15051767
Academic Editors: Davide Astolfi
and Andrzej Bielecki
Received: 2 February 2022
Accepted: 25 February 2022
Published: 27 February 2022
Publisher’s Note: MDPI stays neutral
with regard to jurisdictional claims in
published maps and institutional affil-
iations.
Copyright: © 2022 by the authors.
Licensee MDPI, Basel, Switzerland.
This article is an open access article
distributed under the terms and
conditions of the Creative Commons
Attribution (CC BY) license (https://
creativecommons.org/licenses/by/
4.0/).
Energies 2022, 15, 1767. https://doi.org/10.3390/en15051767
1 Department of Wind Energy, Risø Campus, Technical University of Denmark, Frederiksborgvej 399,
4000 Roskilde, Denmark; nijoh@dtu.dk (N.F.-J.J.); antf@dtu.dk (A.F.); sfni@dtu.dk (S.F.)
2 Danish Blade Service Aps, Bavnevej 10 B, 6580 Kolding, Denmark; tj@danishbladeservice.dk (T.J.);
bb@danishbladeservice.com (B.B.)
* Correspondence: lemi@dtu.dk
Abstract: Maintenance and repair of wind turbines contribute to the higher costs of wind energy. In
this paper, various technologies of structural repair of damaged and broken wind turbine blades
are compared. The composite plates, mimicking damaged blade parts, were damaged and repaired,
using various available curing and bonding technologies. Technologies of repair with hand layup
lamination, vacuum repair with hand layup and infusion, ultraviolet repair and high temperature
thermal curing were compared. The repaired samples were tested under tensile static and fatigue tests,
and subject to microscopic X-ray investigations. It was observed that both the strength of the repaired
structures and the porosity depend on the repair technology used. Vacuum-based technologies lead
to relatively stiff and lower-strength repaired plates, while ultraviolet-curing technologies lead to
average stiffness and high strength. High-temperature vacuum curing leads to the highest maximum
stress. Hand layup (both vacuum and without vacuum) leads to high post-repair porosity in the
adhesive and scarf, while vacuum infusion leads to low porosity. Fatigue lifetime generally follows
the trend of porosity. There exist risks of micro-damaging the parent laminate and the formation of
residual stresses in the repaired structure.
Keywords: wind energy; wind turbine blades; maintenance; repair
1. Introduction
Expansion of renewable energy is one of the key elements of the global strategy to
mitigate global warming. For the next years and decades, a large expansion of wind energy
is planned. The global wind-power market is expected to reach 69,700 megawatts by
2027 [1]. The European Commission raised the EU’s renewable energy target from 32% to
40% by 2030, seeking to build 451 GW of wind-power capacity by 2030, up from 180 GW
today. The USA plans a massive expansion of wind farms off the coasts. The Departments of
the Interior, Energy and Commerce committed to a shared goal of generating 30 gigawatts
of offshore wind in the U.S. by 2030 [2].
However, the operation and maintenance (O&M) of wind turbines remain important
factors, influencing the wind energy price. For instance, onshore wind farm operators spent
around USD 15 billion on O&M services in 2019, of them, 57% on unplanned repairs [3,4].
The total repair budgets in Europe increased from 2019 to 2020 from USD 4.7–8.6 million,
and unplanned repairs also increased slightly in Europe (2%) and even by 10% in the
United States [5].
Furthermore, wind turbines installed at the beginning of 20th century are becoming
old now. The number of WTs reaching 20 years of operation annually, for instance, in Spain
and Germany will be of the order of 800 per year in the next decade [6]. Ageing fleets lead
to increased repair requirements.
https://www.mdpi.com/journal/energies
Energies 2022, 15, 1767 2 of 17

Wind turbine service companies and wind-park owners face a large choice of tech-
nologies for the blade repair. In this work, the authors sought to compare and evaluate
available blade repair technologies, using the materials testing and microscopy analysis
of repaired samples. The authors sought to provide the service companies and end users
with necessary information for the comparing and selecting of the repair technologies,
and, on the other hand, to give the adhesive and equipment developers feedback on the
performance of their products in quasi-real conditions.

2. Repair Costs and Technologies: State of the Art

With an estimated 700,000 blades in operation globally, there are, on average, 3800 incidents
of blade failure each year [7]. The defect types can be classified as surface defects (e.g.,
leading edge erosion), nonstructural defects (e.g., matrix defects, delamination), and struc-
tural defects (with failed fibers). In [8], it was reported that 6.2 minor repairs, 1.1 major
repairs, and 0.43 major replacements are required on average per turbine per year. The
major repairs are carried out when both laminates, including fibers, are damaged.
Different repair technologies can be used, depending on the type of defects, e.g., tapes
and shells (for erosion repair, surface defects), filling, sealing, resin injection (for matrix
cracks, debonding), and plug/patch and scarf repair for structural damage. Damaged
regions are removed by grinding, and a patch is attached to the parent laminate [9]. A
detailed overview of wind turbine (WT) blade repair technologies is given in [9].
In [10], a survey of damage mechanisms and frequencies of wind turbine blades was
carried out among service companies and wind turbine owners. Among other things, it
was observed that current repair technologies are far from perfect, since the next repair can
be required 4–6 years after the first repair.
The quality of blade repair is very important for the post-repair time of wind turbine
blades [9,11]. Low quality repair can lead to an eccentric load path, bending in the patch,
and stresses in the adhesive and composite. One of the common defects in composites and
repaired composites are voids. Voids can form during curing by the mechanical trapping
of air, and the diffusion of dissolved water and gases in the resin [11,12]. Katnam and
colleagues [11] reported that “micro-voids are frequently seen in adhesive bonds cured at
elevated temperatures” and showed that the tensile strength depends on the micro-voids.
Voids create local stress concentrations [13–15].
According to [7], the costs of unscheduled maintenance activities, and the repair of
failures, can be estimated as:

CCM = ∑ cD TD + Ctrans + Mctrep

(1)
NF

where cD —costs of downtime per hour, TD —duration of downtime, NF —number of failures

per year, and Crep1 —costs of single repair. The single repair costs can be calculated:

Crep1 = Ctrans + Cequip + Clabor (2)

where Ctrans —costs of transportation, Cequip —costs of equipment and materials, and Clabor —
costs of field labor. Clabor = Mtrep c, where M—size of team, c—hourly costs, and trep —time
of repair. The technology-related parameters include duration of work tw , size of team M,
equipment and materials costs Cequip , and the “output parameter” amount of repairs NF (or
inversely, time between repairs, 1/NF ).
There are various methods to improve the repair technology, taking into account
Equation (1): reduce time of repair and size of team, increase quality of each repair, and
reduce the required number of repairs. One interesting direction is to reduce the time of
curing by using fast curing polymers or ultraviolet (UV). The various repair and curing
technologies are associated with various initial investments, but also can lead to different
qualities of repaired structures. In this paper, the quality and properties of composite
structures, repaired with the use of different techniques, is investigated.
Energies 2022, 15, 1767
Energies 2022, 15, 1767 3 of 17 3 of

3. Concept of 3. Concept of Experiments

Experiments
In this
In this study, the ideastudy, the idea
of testing of testing
of several of several
available availablerepair
composite composite repair technologies
technologies is
realized. Laminate plates were damaged, and then repaired, using
realized. Laminate plates were damaged, and then repaired, using various repair technolo- various repair techno
gies, availableogies,
on theavailable on the
market. The market.
repair The repair
of blades of blades
was carried by awas carried
repair by a repair
technician, who technicia
has many years who has many years
of experience in theof experience
practical in the
repair practicaland
of offshore repair of offshore
onshore and onshore win
wind turbines
turbinesof(to
(to prevent influence prevent
human influence
errors, of humancaused
or variability errors, by
or variability caused
different skill by different
levels). Then, skill le
els). Then,
the repaired plates were the repaired
tested plates
(tensile were
static andtested (tensile
fatigue), and static and fatigue),
the microscopic and theofmicroscop
analysis
analysis of
structures of repaired structures
samples wasofcarried
repaired samples
out. Figurewas carried
1 shows a out.
schemaFigure
of a1 damaged
shows a schema of
damaged blade and
blade and a photo of a repaired blade. a photo of a repaired blade.

(a) (b)

(c)
Figure
Figure 1. Schema 1. Schema
of damaged of damaged
plate plate
(a), photos (a), photos
of plate of of
(b) and plate (b) and
repair of repair
procedure (c).procedure (c).

In the experiments,
In the six different repair
experiments, technologies
six different were used, namely:
repair technologies were used, namely:
- - Hand layup lamination (traditional repair technology,
Hand layup lamination (traditional repair technology, by putting resin byon
putting resin on the lam
the laminate
nate and hand
and hand rolling afterward) (HL),rolling afterward) (HL),
- -
Vacuum repair Vacuum
with hand repair
layupwith hand
(V1): layup
after (V1): is
the resin after
putthe
on resin is put on
the laminate andthevacuum
laminate and va
is applied, to uum is the
get all applied,
resintothrough
get all the
theresin through
laminate andthe
to laminate and
get all the airtoout
getof
allthe
the air out
laminate, the laminate,
- -
Vacuum repair Vacuum repair with
with infusion (V2):infusion (V2): dry
dry laminate laminate
is put on theisrepair
put onandthe vacuum
repair and is vacuum
applied to pullapplied to pull
the resin the resin
through through the
the laminate laminate
with vacuum with vacuum pressure,
pressure,
- Ultraviolet
- repair
Ultraviolet (UVh1): repair (UVh1):
handheld, handheld,
portable, portable, UV
continuous continuous
spectrum UVcuring,
spectrum curing,
- - Ultraviolet repair (UV2):
Ultraviolet repair (UV2): stationary device, stationary device,
- - High-temperature
High-temperature thermal curing thermal
(HTC). curing (HTC).
Table 1 shows the evaluation of the time/labour efforts of using the different repa
Table 1 shows the evaluation of the time/labour efforts of using the different repair
technologies in real conditions, made by the service team. Table 2 shows estimations
technologies in real conditions, made by the service team. Table 2 shows estimations of
temperatures and duration of curing for different technologies.
temperatures and duration of curing for different technologies.
Several laminate plates (60 cm × 60 cm, with five layers, thickness 4 mm) were take
Several laminate plates (60 cm × 60 cm, with five layers, thickness 4 mm) were taken,
and a three-layer-deep crack was introduced in the center of the plate. The damaged si
and a three-layer-deep crack was introduced in the center of the plate. The damaged site
was sanded out (with the same sand tool and angle in all cases) and repaired. The laminates
were from glass fibers, and Bodonex 900 matrix resin, Biax 600: Orientation: +/−45. Table 3
shows the adhesives and laminates used in each technology.
Energies 2022, 15, 1767 4 of 17

Table 1. Evaluations of efforts of different repair technologies made by technician team.

Time/Cost Temperature
Technology Time for Repair Materials Costs
for Training of Curing
High-temperature thermal curing (HTC) 165 1 day training 120 degrees
Ultraviolet repair, handheld (UV1) 120 1 day training €40,000 UV light
Hand layup lamination (HL) 330 1 day training €20,000 80 degrees
Vacuum repair with hand layup (V1) 615 2 days training €30,000 80 degrees
Vacuum repair with infusion (V2) 375 2 days training €30,000 80 degrees

Table 2. Overview: Temperatures and duration of curing.

Grinding—
Heating
Apply Vacuum, Measure
Technology Blanket, UV Light Total in Min Laminate
Minutes Patch—Cut
Minutes
Reinforcement
Ultraviolet repair
60
(UV2)
High-temperature
60 30 45 30 165
thermal curing (HTC)
Ultraviolet repair,
60 30 30 120
handheld (UV1)
Hand layup
60 30 240 330
lamination (HL)
Vacuum repair with
60 30 45 480 615
hand layup (V1)
Vacuum repair with
60 30 45 240 375
infusion (V2)

Table 3. Adhesives, laminates, and resins used.

Amount of Repair
How Is It Cured Adhesive Used Laminate Used
Product Used
Number of layers: 3. Size of
Ultraviolet repair, Heat Blanket 80 Vinyl Ester 200 mL resin and
layers xy
handheld (UV1) degrees, 4 h Epoxy Resin 200 g glass fiber
20 × 20–30 × 30–40 × 40
Number of layers 3. Size of
Hand layup layers xy 300 mL mixing bag
Heat blanket 4 h SIKA Sikadur 30
lamination (HL) 20 × 20–30 × 30–40 × 40. and 200 g glass fiber.
Type of GF: Biax 600
Number of layers: 3. Size of
Vacuum repair with layers xy 200 mL and 200 g
Heat blanket, 4 h Gurrit Ampreg 31
hand layup (V1) 20 × 20–30 × 30–40 × 40. glass fiber
Type of GF: Biax 600
Number of layers: 3. Size of Used 200 mL.
Vacuum repair with layers xy 1 kg repair kit or mix
Heat blanket, 4 h Bodotex 900
infusion (V2) 20 × 20–30 × 30–40 × 40. Type by weight and 200 g
of GF: Biax 600 glass fiber
Number of layers: 4
High-temperature Size of layers:
Heat blanket, 28 min - 200 g glass fiber
thermal curing (HTC) 20 × 20–27 × 27–33 × 33–40 × 40
Type of GF: Biax 450
 with infusion Heat blanket, 4 h Bodotex 900 xy 20 × 20–30 × 30–40 × 40. Type of mix by weight
(V2) GF: Biax 600 and 200 g glass
fiber
High- Number of layers: 4
temperature Heat blanket, 28 Size of layers: 20 × 20–27 × 27–33 ×
- 200 g glass fiber
Energies 2022, 15, 1767 thermal curing min 33–40 × 40 5 of 17
(HTC) Type of GF: Biax 450.

4.4.Experimental
ExperimentalTesting
TestingProgram
Program
Allsamples,
All samples,including
includingthetheunrepaired
unrepairedreference
referenceplate,
plate,were
weretested
testedininuniaxial
uniaxialtension
tension
until failure. From each plate, four samples were cut from the center on the scarf
until failure. From each plate, four samples were cut from the center on the scarf repair, repair,
with the dimensions 300 × 25 mm. Due to the large elongation, sample
with the dimensions 300 × 25 mm. Due to the large elongation, sample length had to length had tobe
be
reduced to not cover the entire repair but only the center of patches. Figure 2
reduced to not cover the entire repair but only the center of patches. Figure 2 shows how shows how
thesamples
the sampleswerewerecut
cutfrom
fromthe
therepaired
repairedlaminates.
laminates.Samples
Sampleswere
wereheld
heldusing
usingmechanical
mechanical
jaws,with
jaws, withconstant
constanttest
testspeed
speedofof22mm/min
mm/min applied.
applied.

Figure2.2.Photo
Figure Photoof
ofrepaired
repairedplate,
plate,and
andgeometry
geometryofofsamples
samplestotobe
becut
cutout.
out.

Extensometers were mounted on each side of the sample at the center of the repair, to
measure the strain between 0–5%. Above ≈ 5% the strain was measured using machine
displacement.
Figure 3 shows the typical stress-strain curves of tested samples.
Table 4 shows the mechanical properties of repaired samples, determined in the tests.
Figure 4 shows the comparison of Young moduli and maximum failure stresses for different
tested cases, normalized by reference/unrepaired case values. Figure 5 shows also locations
of cracks for some samples (vacuum repair with hand layup, hand layup lamination and
handheld UV curing technology).
Based on the studies, the following conclusions can be drawn. All the technologies
provide relatively good quality of repair under static loading. Technologies with application
of vacuum lead to stiffer repaired structures, and also to lower maximum stresses. High-
temperature curing leads to softer structures, but with higher maximum stress. Ultraviolet
curing technologies ensure average stiffness and high strength.
Energies 2022, 15, 1767 6 of 18

Extensometers were mounted on each side of the sample at the center of the repair,
to measure the strain between 0 − 5%. Above ≈ 5% the strain was measured using ma-
Energies 2022, 15, 1767 6 of 17
chine displacement.
Figure 3 shows the typical stress-strain curves of tested samples.

Figure3.3.Typical
Figure Typicalstress-strain
stress-straincurves
curvesof
ofthe
thespecimens.
specimens. Here:
Here: HL—hand
HL—hand layup
layup lamination,
lamination, HTC—
HTC—
high-temperaturethermal
high-temperature thermalcuring,
curing,UV1—ultraviolet
UV1—ultravioletrepair
repairwith
withhandheld,
handheld,portable,
portable,continuous
continuousUV
UV
spectrum curing, UV2—ultraviolet repair with stationary device, V1—vacuum repair
spectrum curing, UV2—ultraviolet repair with stationary device, V1—vacuum repair with hand with hand
layup, V2—vacuum repair with infusion, and Ref—reference, unrepaired sample.
layup, V2—vacuum repair with infusion, and Ref—reference, unrepaired sample.

Table Table 4 showsproperties

4. Mechanical the mechanical properties
of repaired samples.of repaired samples, determined in the tests.
Figure 4 shows the comparison of Young moduli and maximum failure stresses for differ-
ent tested E-Modulus
cases, normalized Max Stress caseLinear
by reference/unrepaired Strain
values. Figure Uniform
5 shows Strain
also lo-
Technology
cations of cracks for some [GPa]
samples (vacuum[MPa]repair with hand[%]
layup, hand layup [%]lamina-
tion and handheld
Reference plate UV curing technology).
12.6± 0.74 99.2± 1.10 0.8 ± 0.0098 12.3± 0.76
(non-repaired)
Based on the studies, the following conclusions can be drawn. All the technologies
provide
Vacuum relatively
repair withgood quality of repair under static loading. Technologies with applica-
12.2 ± 0.764 88.5 ± 10.45 0.7 ± 0.0434 14.9 ± 5.61
infusion
tion (V2)
of vacuum lead to stiffer repaired structures, and also to lower maximum stresses.
High-temperature
Vacuum repair with curing leads to softer structures, but with higher maximum stress. Ul-
11.8 ± 0.546 71.3 ± 3.45 0.6 ± 0.0049 15.0 ± 1.52
traviolet curing
hand layup (V1) technologies ensure average stiffness and high strength.
Hand layup
11.2 ± 0.409 89.5 ± 1.78 0.8 ± 0.0136 21.1 ± 2.025
lamination (HL)
Ultraviolet repair,
10.9 ± 0.227 97.3 ± 5.30 0.9 ± 0.033 16.9 ± 4.524
handheld (UVh1)
High-temperature
9.5 ± 0.396 113.7 ± 11.22 1.2 ± 0.096 11.6 ± 3.03
curing (HTC)
Ultraviolet repair
11.2 ± 0.325 108.9 ± 5.73 1.0 ± 0.057 10.6 ± 1.45
(UV2 (stationary))
 Ultraviolet repair,
10.9 ± 0.227 97.3 ± 5.30 0.9 ± 0.033 16.9 ± 4.524
handheld (UVh1)
High-temperature cur-
9.5 ± 0.396 113.7 ± 11.22 1.2 ± 0.096 11.6 ± 3.03
ing (HTC)
Energies 2022, 15, 1767 Ultraviolet repair (UV2 7 of 17
11.2 ± 0.325 108.9 ± 5.73 1.0 ± 0.057 10.6 ± 1.45
(stationary)

Energies 2022, 15, 1767 8 of 18

(a)

(b)
Figure 4. Comparison of different repair technologies, (a) relative (normalized by reference/unre-
Figure 4. Comparison of different repair technologies, (a) relative (normalized by reference/unrepaired
paired case) Young moduli, and (b) relative (normalized by unrepaired case) strengths of samples.
case) Young moduli, and (b) relative (normalized by unrepaired case) strengths of samples.
 (b)
Energies 2022, 15, 1767 8 of 17
Figure 4. Comparison of different repair technologies, (a) relative (normalized by reference/unre-
paired case) Young moduli, and (b) relative (normalized by unrepaired case) strengths of samples.

Energies 2022, 15, 1767 9 of 18

(a)

(b)

(c)
Figure Locations of
Figure 5. Locations offailure
failurefor
forsome
some samples:
samples: (a)(a)
V1,V1, vacuum
vacuum repair
repair withwith
handhand layup,
layup, (b)hand
(b) HL, HL,
hand
layuplayup lamination,
lamination, (c) UV1,
(c) UV1, handheld
handheld UV curing.
UV curing.

5. Fatigue
5. Fatigue Testing
Testing
At the
At the next
next stage,
stage, fatigue
fatigue testing
testing of
of repaired
repaired samples
samples was
was carried
carried out. The fatigue
out. The fatigue
specimens, measured 520 × 25 mm with the distance 420 mm between
specimens, measured 520 × 25 mm with the distance 420 mm between tabs, were cut fromtabs, were cut from
the repaired
the repaired samples.
samples.
The samples were
The samples weretested
testedunder
undercyclic loading
cyclic loading (load control
(load fatigue)
control at 0.5%
fatigue) strainstrain
at 0.5% level.
The tests were carried out at 3 Hz using load control. The samples were
level. The tests were carried out at 3 Hz using load control. The samples were tested totested to failure
and theand
failure cyclethecount
cyclewas
count recorded.
was recorded.
Figure 6 shows the common
Figure 6 shows the common strain strain level
level on
on aa peak
peak strain
strain basis
basis for
for all
all series.
series.
Vacuum repair with hand layup (V1) shows the lowest performance. Even hand
layup (HL) is better than vacuum repair with infusion (V2). UV2 (stationary UV system)
shows average lifetime, but high variability. UV1 (handheld UV curing) shows the best
lifetime. HTC shows average level.
Figure 7 shows locations of failure for some samples (UV1, V1, V2). It can be seen
Energies 2022,
Energies 15,15,
2022, 1767
1767 10 of 918of 17

Energies 2022, 15, 1767 10 of 18

Figure6.6.Cycle
Figure Cycle to
to failure
failure for
for fatigue
fatiguetesting
testingofofpeak
peakstrain
strain0.5%.
0.5%.

Vacuum repair with hand layup (V1) shows the lowest performance. Even hand layup
(HL) is better than vacuum repair with infusion (V2). UV2 (stationary UV system) shows
average lifetime, but high variability. UV1 (handheld UV curing) shows the best lifetime.
HTC shows average level.
Figure 7 shows locations(a)of failure for some samples (UV1, V1, V2). It can be seen that
while all the static samples failed in the middle of the sample (i.e., in the scarf/adhesive/parent
laminate “sandwich”), failure under fatigue loading started at different locations. the HL
and UV2 samples broke on scarf/laminate interface, while the UV1, V2, and V1 samples
were broken
Figure in the
6. Cycle middle
to failure of the testing
for fatigue samples, similar
of peak strainto0.5%.
the static samples.
(b)

(a)
(c)

(b)
(d)
Figure 7. Locations of failure for some samples: (a–c) UV1, V1, and V2 samples (failure in the middle
of the sample), and (d) UV2 sample (failure at the border scarf/parent laminate).

(c) Residual Stresses in the Repaired Blades

6. Sample Deformation: Internal
In the static tests, it was observed that the repaired samples tend to bend and show
curvature after the tests. The hypothesis was formulated that the curvature of samples is
a result of residual stresses, caused by different heating in scarf and parent laminate dur-
(d)
Figure
Figure 7. 7. Locationsofoffailure
Locations failure for
for some
somesamples:
samples:(a–c) UV1,
(a–c) V1,V1,
UV1, andand
V2 V2
samples (failure
samples in theinmiddle
(failure the middle
of the sample), and (d) UV2 sample (failure at the border scarf/parent laminate).
of the sample), and (d) UV2 sample (failure at the border scarf/parent laminate).

6. Sample Deformation: Internal Residual Stresses in the Repaired Blades

In the static tests, it was observed that the repaired samples tend to bend and show
curvature after the tests. The hypothesis was formulated that the curvature of samples is
a result of residual stresses, caused by different heating in scarf and parent laminate dur-
 ing curing processes. Figure 8 shows several typical curved samples and a schema of po-
tential local stress concentration, caused by different thermal and mechanical stresses in
the repaired structure after curing. Figure 8c shows how the deformation in the substrate
laminate results in residual compressive stress in the laminate.
Energies 2022, 15, 1767 10 of 17
Curvatures of the samples were determined using the software IC Measure 2–0.026.
The inverse curvature diameter was measured from pictures, taken perpendicular to the
samples. The three-point circle was defined, and the diameter was calculated automati-
6.cally.
Sample
The Deformation:
curvature-tested Internal Residual
samples, plottedStresses in the
versus the Repaired Blades
performance relative to the refer-
enceIn(non-repaired)
the static tests, it was
case, areobserved
plotted inthat the 9.
Figure repaired samples tend to bend and show
curvature
Fromafter the9,tests.
Figure The observe
one can hypothesis was formulated
a correlation: that the
that highly curvature
bent samples of samples
correspond
isalso
a result of residual
to relatively stresses,
soft and toughcaused by different
(high maximum heating
stress) in scarf
repaired and parent laminate
structures.
during curing
While theprocesses. Figure 8isshows
sample bending several
a result typical
of large curved
strains samples
applied and
during thea destructive
schema of
potential local stress concentration, caused by different thermal and mechanical
testing, and such high strains cannot be observed in real life, this illustrates the stresses in
potential
the repaired structure after curing. Figure 8c shows how the deformation in the
for undesirable internal stresses, when repairs are applied to a potentially unknown sub- substrate
laminate
strate. results in residual compressive stress in the laminate.

(a) (b)

(c)
Figure 8. Curved samples (a) and schema of danger zones caused by residual stresses (b). The lower
Figure 8. Curved samples (a) and schema of danger zones caused by residual stresses (b). The lower
figure (c) shows the potential danger zone, which can appear in post-repair blade posts.
figure (c) shows the potential danger zone, which can appear in post-repair blade posts.

Curvatures of the samples were determined using the software IC Measure 2–0.026
(The Imaging Source Asia Co., Ltd., New Taipei City, Taiwan). The inverse curvature
diameter was measured from pictures, taken perpendicular to the samples. The three-point
circle was defined, and the diameter was calculated automatically. The curvature-tested
samples, plotted versus the performance relative to the reference (non-repaired) case, are
plotted in Figure 9.
From Figure 9, one can observe a correlation: that highly bent samples correspond
also to relatively soft and tough (high maximum stress) repaired structures.
While the sample bending is a result of large strains applied during the destructive
testing, and such high strains cannot be observed in real life, this illustrates the potential for
undesirable internal stresses, when repairs are applied to a potentially unknown substrate.
 Energies 2022, 15, 1767 12 of 18
Energies 2022, 15, 1767 11 of 17
Energies 2022, 15, 1767 12 of 18

Figure
Figure9.9.Curvature
Curvature of tested samples,
of tested samples,plotted
plottedversus
versusthe
the performance relative to the reference (non-
Figure 9. Curvature
repaired) case. of tested samples, plotted versusperformance relative
the performance to the
relative reference (non-
to the reference
repaired) case.
(non-repaired) case.

7.7.7.Microscopic
Microscopic Analysis
MicroscopicAnalysis
Analysis ofofMechanisms
of Mechanismsof
Mechanisms ofDamage
of Damageof
Damage of
of Repaired
Repaired
Repaired Blades
Blades
Blades
The
The tested
Thetested sampleswere
testedsamples wereinvestigated
were investigatedby
investigated byX-ray
by X-ray
X-ray tomography
tomography
tomography to
toto characterize
characterize the
thethe
characterize failure
failure
failure
mechanisms. The
mechanisms.
mechanisms. Thesamples
The sampleswere
samples werescanned
were scannedwith
scanned withaapolychromatic
with a polychromaticbeam
polychromatic beamwith
beam withenergies
with energiesup
energies up upto
to to
120
120 keV
keV using
using a Nikon 225
225 and
and aa Zeiss
Zeiss Xradia
Xradia 520
520 Versa scanner.
Versa Figure
scanner.
120 keV using a Nikon 225 and a Zeiss Xradia 520 Versa scanner. Figure 10 shows the 10
Figure shows
10 showsthe the
scanningsetup
scanning
scanning setupin
setup inthe two
thetwo scanners.
scanners.The
twoscanners. Thereconstructions
The reconstructions
reconstructions were performed
were
were performed
performed using a standard
using
using aastandard
standard
filteredback
filtered back projection
projection resulting
resulting in
inaavoxel
voxel size
sizeofof9.64 µm.
9.64 µm.
filtered back projection resulting in a voxel size of 9.64 µm.

Figure 10. Scanning setup of three samples within the Nikon scanner (a) and the Zeiss Xradia 520
Versa scanner (b).
Figure10.
Figure 10. Scanning
Scanning setup
setup of
ofthree
threesamples
sampleswithin
withinthe
theNikon
Nikonscanner
scanner(a)(a)
and thethe
and Zeiss Xradia
Zeiss 520
Xradia
Versa scanner (b).
520 Versa scanner (b).
Energies 2022, 15, 1767 13 of 18
Energies 2022, 15, 1767 12 of 17

The reconstructions were segmented using a constant threshold to separate the black
The reconstructions were segmented using a constant threshold to separate the black
air bubbles from the grey epoxy and the white fibers. The histogram shows a clear dip
air bubbles from the grey epoxy and the white fibers. The histogram shows a clear dip
between the black air bubbles and the grey epoxy and this position was selected as the
between the black air bubbles and the grey epoxy and this position was selected as the
segmentation threshold. The segmentation was performed on a sub-volume consisting of
segmentation threshold. The segmentation was performed on a sub-volume consisting of
1108 × 68 × 129 pixels that was positioned within the repair.
1108 × 68 × 129 pixels that was positioned within the repair.
Figure 11 shows the distribution of the void content seen from two different direc-
Figure 11 shows the distribution of the void content seen from two different directions
tions (perpendicular to each other) and the shape (ellipsoid versus spherical) of the voids
(perpendicular to each other) and the shape (ellipsoid versus spherical) of the voids in the
in the repaired samples. The elongation of the segmented objects (voids) was determined
repaired samples. The elongation of the segmented objects (voids) was determined as the
as the ratio between the length and width of the bounding box containing the individual
ratio between the length and width of the bounding box containing the individual voids.
voids. This value will be in the interval [0,1] for elongated to spherical objects.
This value will be in the interval [0, 1] for elongated to spherical objects.

(a)

Figure 11. Cont.

Energies 2022,
Energies 15, 1767
2022, 15, 1767 1413ofof 18
17

(b)

(c)
Figure 11. Volume fraction of voids and shapes of voids (elongated versus spherical) for tested sam-
Figure 11. Volume fraction of voids and shapes of voids (elongated versus spherical) for tested
ples: (a) vacuum repair with hand layup/hand layup/handheld UV, (b) high-temperature cur-
samples: (a) vacuum repair with hand layup/hand layup/handheld UV, (b) high-temperature cur-
ing/stationary UV, and (c) hand layup/vacuum repair with infusion/vacuum repair with hand
ing/stationary UV, and (c) hand layup/vacuum repair with infusion/vacuum repair with hand layup.
layup.
The volume fraction of voids within the analyzed volume was calculated. In Figure 11
it canThe volume
be seen thatfraction
the vacuumof voids within
repair withthe handanalyzed volume
layup (V1) leadswasto acalculated.
rather highInporosityFigure
11 it can be seen that the vacuum repair with hand layup (V1)
in the adhesive, over 4.7%. High-temperature curing (HTC) and stationary UV curing leads to a rather high po-
rosity in the adhesive, over 4.7%. High-temperature curing (HTC)
(UV2) technologies ensure the porosity of the level of 0.5%, relatively low. UV handheld and stationary UV cur-
ing
(UV1) (UV2)leadstechnologies
to acceptable, ensure therelatively
but still porosity high,of theporosity
level ofof0.5%,
1.8%.relatively low. UV
High-temperature
handheld
curing (HTC) (UV1) leads
leads to to acceptable,
relatively equalbutdistribution
still relatively high, porosity
of elongated (i.e., of 1.8%.
fiber High-tem-
adjacent) and
perature curing (HTC) leads to relatively equal distribution of
spherical voids, while vacuum and hand layup lead to the prevalence of elongated, fiber elongated (i.e., fiber adja-
cent)
adjacent andvoids.
spherical voids, while vacuum and hand layup lead to the prevalence of elon-
gated,It fiber
is of adjacent voids.
interest that the V2 technology (vacuum repair with infusion) leads to very
It is of interest
low porosity (below 0.0002%). that the V2Suchtechnology (vacuumbetween
a big difference repair with
vacuum infusion) leads tocan
technologies very
be
low porosity
explained as (below
follows:0.0002%). Such a big
vacuum infusion is difference
“sucking” between
the resin vacuum
through technologies
the laminate/repair can be
explained
and, therefore, as follows:
draggingvacuum infusionout.
the porosities is “sucking”
When hand thelay-up
resin through
is carriedthe out,laminate/repair
the resin does
and, therefore,
not really dragging
flow, only the porosities
the operator appliesout.and When hand
traps air. Thelay-up
same is is the
carried
case for out,UV thecuring,
resin
does not really flow, only the operator applies and traps
which does not cause the resin travel, and thus can lead to more porosity. air. The same is the case for UV
curing, which does not cause the resin travel, and thus can lead to
To visualize the void shape distribution more clearly, the different voids were markedmore porosity.
with To visualize
different theFigure
colors. void 12 shape
shows distribution more clearly,
the void distribution, wherethe voidsdifferent
in resinvoids were
are marked
marked
in red, while with different
voids or colors.
air bubblesFigure 12 shows
near fibers arethemarked
void distribution,
in turquoise where color, voids in resin
for vacuum
are marked
repair in red,
with hand while
layup voids
(V1), handorlayup
air bubbles near (HL),
lamination fibersand
areUVmarked
curinginhandheld
turquoise(UVh1).color,
for
Many vacuumsmall airrepair withinhand
bubbles layup (V1),
the bundles, and hand layup
large air lamination
bubbles between(HL), bundles andhorizontally,
UV curing
handheld (UVh1). Many small air bubbles in the bundles, and large air bubbles between
 Energies2022,
Energies 2022,15,
15,1767
1767 15 of 17
14 of 18

bundles
form in V1horizontally,
technology.form
Handin layup
V1 technology.
laminationHand
(HL)layup
leadslamination (HL)air
to fewer small leads to fewer
bubbles in
small air bubbles in the bundles, and a few large and more spherical
the bundles, and a few large and more spherical air bubbles. Handheld UV technologyair bubbles.
Handheld
(UV1) leadsUVto technology (UV1)
both small and leads to both air
medium-sized small and medium-sized
bubbles airand
in the bundles, bubbles
largeinand
the
bundles,
more and large
spherical and more
air bubbles spherical
between air bubbles between bundles.
bundles.

Figure12.
Figure 12.Voids
Voidsininresin
resin(red)
(red)versus
versusvoids
voidsnear
nearfibers
fibers(blue),
(blue),for
forvacuum
vacuumrepair
repairwith
withhand
handlayup
layup
(V1), hand layup lamination (HL), and UV curing handheld (UV1).
(V1), hand layup lamination (HL), and UV curing handheld (UV1).

Furthermore,
Furthermore,the themicrocracks/defects
microcracks/defectswere wereidentified
identified in in
samples
samples by by
inspecting
inspecting virtual
vir-
cross-sections through the tomographic reconstructions of the repaired
tual cross-sections through the tomographic reconstructions of the repaired samples. Fig- samples. Figure 13
shows
ure 13 the cracks
shows the identified in samples,
cracks identified namely,namely,
in samples, in high-temperature
in high-temperaturecured samples (HTC)
cured samples
and
(HTC)in handheld UV cured
and in handheld UV samples (UV1). In(UV1).
cured samples the firstIncase (HTC),
the first casethe cracksthe
(HTC), were found
cracks in
were
the parent
found laminate.
in the In the second
parent laminate. case
In the (UV1),
second the(UV1),
case cracksthe were seenwere
cracks in scarf.
seen in scarf.
Based
Basedon onthese
theseinvestigations,
investigations,the thefollowing
followingconclusions
conclusionscan canbebedrawn:
drawn:the thevacuum
vacuum
hand
hand layup technology leads to the formation of a lot of voids in the repaired part,more
layup technology leads to the formation of a lot of voids in the repaired part, more
than
than all
all other technologies.However,
other technologies. However,the thevacuum
vacuum infusion
infusion technology
technology leads
leads to atovery
a very
low
low porosity.
porosity.
Comparing
Comparing thesethese observations
observations to to the
the observations
observations from from Section
section 55 (fatigue
(fatigue testing),
testing),
one
onecancansee
seethat
thatthe
thefatigue
fatiguelifetime
lifetimecorresponds
correspondsto tothe
thetrend
trendof ofporosity:
porosity:withwithaaviewviewonon
lifetime, vacuum repair with hand layup (showing very high porosity)
lifetime, vacuum repair with hand layup (showing very high porosity) shows the lowest shows the lowest
lifetime.
lifetime.Handheld
HandheldUV UVcuring
curingshows
showsthe thebest
bestperformance
performance(even (evenbetter
betterthan
thanindustrial
industrialUV UV
curing), what supposedly reflects the experience of the
curing), what supposedly reflects the experience of the technician. technician.
Energies 2022, 15,
Energies 2022, 15, 1767
1767 15of
16 of 17
18

Figure 13.
Figure 13. Cracks
Cracks identified
identified in
in samples
samples (a)
(a) HTC
HTC and
and (b)
(b) UV1.
UV1. The
The white
white dotted
dotted line
line indicates
indicates the
the
interface between the scarf and the repair, and the red arrows indicate cracks.
interface between the scarf and the repair, and the red arrows indicate cracks.

8. Conclusions
Technologies
Technologies for for the structural repair of damaged wind turbine blades are are compared,
compared,
with view
view ononstrength,
strength,stiffness,
stiffness,and
and porosity
porosity of repaired
of repaired laminates.
laminates. Six widely
Six widely used used
tech-
technologies
nologies werewere applied
applied to repair
to repair damaged
damaged laminates
laminates under
under comparable
comparable conditions,
conditions, andand
the
the resulting
resulting repaired
repaired samples
samples were
were tested
tested undertensile
under tensilestatic
staticand
andfatigue
fatiguetests,
tests, and
and subject
to microscopic X-ray investigations.
investigations.
Technologies
Technologies using the
using the application of vacuum lead to stiffer repaired structures, and
lower maximum stress. High-temperature
lower maximum stress. High-temperature curing leads
curing to softer
leads structures,
to softer but with
structures, buthigher
with
maximum stress. Ultraviolet curing technologies ensure average stiffness
higher maximum stress. Ultraviolet curing technologies ensure average stiffness and and high strength.
high
The vacuum hand layup technology leads to the formation of a lot of voids in the
strength.
repaired
The part,
vacuum more handthanlayup
all other technologies.
technology leads toThe thehand layupofwithout
formation a lot ofvacuum
voids inalso
the
creates long elongated voids, probably related to badly bonded fibers.
repaired part, more than all other technologies. The hand layup without vacuum also cre-
Several
ates long samples
elongated showprobably
voids, clear residual
relatedstresses
to badlyand were fibers.
bonded bent after testing. This can
meanSeveral
that some
samples show clear residual stresses and were stresses
repair technologies create internal/thermal bent after in testing.
laminates.ThisAlso,
can
in some cases, cracks in the “parent” composite, and microcracks in
mean that some repair technologies create internal/thermal stresses in laminates. Also, inscarf were observed
after
somerepair.
cases, cracks in the “parent” composite, and microcracks in scarf were observed after
Fatigue
repair. lifetime results follow the trend of porosity: vacuum hand layup technology
shows the lowest
Fatigue lifetime lifetime,
resultsand both UV
follow and high
the trend temperature
of porosity: vacuumcuring show
hand average
layup results.
technology
Handheld UV curing shows the best performance (even better than
shows the lowest lifetime, and both UV and high temperature curing show average re- industrial UV curing),
which supposedly reflects the experience of the technician. The UV2 (stationary UV system)
sults. Handheld UV curing shows the best performance (even better than industrial UV
shows an average lifetime, but high variability.
curing), which supposedly reflects the experience of the technician. The UV2 (stationary
This work is a basic study and comparison of the most commonly used repair tech-
UV system) shows an average lifetime, but high variability.
niques. The future directions should also include other more advanced technologies,
This work is a basic study and comparison of the most commonly used repair tech-
nanoengineered adhesives, comparison of various techniques of surface preparation, effect
niques. The future directions should also include other more advanced technologies,
of humidity and temperature on the repair techniques, and also more parametric studies
nanoengineered adhesives, comparison of various techniques of surface preparation, ef-
using large statistics (varying the same resins and temperatures of curing).
fect of humidity and temperature on the repair techniques, and also more parametric stud-
ies using large statistics (varying the same resins and temperatures of curing).
Energies 2022, 15, 1767 16 of 17

Author Contributions: Conceptualization, L.M.J.; methodology, L.M.J., A.F.; validation, A.F., N.F.-J.J.,
T.J., B.B.; formal analysis, N.F.-J.J.; investigation, L.M.J., S.F., N.F.-J.J.; data curation, N.F.-J.J., T.J., B.B.;
writing—original draft preparation, L.M.J.; writing—review and editing, L.M.J.; supervision, L.M.J.;
project administration, L.M.J.; funding acquisition, L.M.J. All authors have read and agreed to the
published version of the manuscript.
Funding: This research was funded by Innovation Foundation of Denmark in the framework of
the Grand Solutions project DURA-LEDGE, Durable leading edges for high tip speed wind turbine
blades, File nr.: 8055-00012A, and of the Ministry of Foreign Affairs of Denmark, in the framework of
Danida grant 19-M02-DTU “Maintenance and Repair Strategy for Wind Energy Development”.
Acknowledgments: The authors gratefully acknowledge the financial support of the Innovation
Foundation of Denmark in the framework of the Grand Solutions project DURALEDGE, Durable
leading edges for high tip speed wind turbine blades, File nr.: 8055-00012A, and of the Ministry
of Foreign Affairs of Denmark, in the framework of Danida grant 19-M02-DTU “Maintenance and
Repair Strategy for Wind Energy Development”. L.M.J. is grateful to Povl Brøndsted and Tom
Andersen for valuable discussions.
Conflicts of Interest: The authors declare no conflict of interest.

Abbreviations
c hourly costs,
cD costs of downtime per hour
Cequip costs of equipment and materials
Clabor costs of field labor
Crep1 costs of single repair
Ctrans costs of transportation
TD duration of downtime
M size of repair team
NF number of failures per year
trep time of repair
tw duration of work
HL hand layup lamination
HTC high-temperature thermal curing
UV1 ultraviolet repair with handheld, portable, continuous UV spectrum curing
UV2 ultraviolet repair with stationary device
V1 vacuum repair with hand layup
V2 vacuum repair with infusion

References
1. Global Wind Power Market Report 2021–2027; Global; Research and Markets: 2021. Available online: https://www.prnewswire.
com/news-releases/global-wind-power-market-report-2021-2027---wind-energy-research--development-projects-get-a-shot-
in-the-arm-leading-to-increase-in-patent-filing-301373291.html (accessed on 13 October 2021).
2. Nilsen, E. Biden Administration Announces Plans for Massive Expansion of Wind Farms off US Coasts. Available online:
https://www.wdsu.com/article/biden-administration-announces-plans-for-massive-expansion-of-wind-farms-off-us-
coasts/37954033# (accessed on 13 October 2021).
3. Unplanned Wind Turbine Repairs to Cost Industry $8 Billion+ in 2019. Available online: https://www.woodmac.com/press-
releases (accessed on 24 February 2022).
4. Ziegler, L.; Gonzalez, E.; Rubert, T.; Smolka, U.; Melero, J.J. Lifetime extension of onshore wind turbines: A review. Renew. Sustain.
Energy Rev. 2018, 82, 1261–1271. [CrossRef]
5. Liu, D. Where is the European Wind Power Market? And where is it going? In Proceedings of the Blades Global Conferences,
Online Conference, 11 March 2021.
6. Mishnaevsky, L., Jr. Offshore Wind Energy: Challenges and Perspectives, Green Technologies for Eurasia’s Sustainable Future.
Available online: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3990559 (accessed on 13 October 2021).
7. Mishnaevsky, L., Jr.; Thomsen, K. Costs of repair of wind turbine blades: Influence of technology aspects. Wind Energy
2020, 23, 2247–2255. [CrossRef]
8. Carroll, J.; McDonald, A.; McMillan, D. Failure rate, repair time and unscheduled O&M cost analysis of offshore wind turbines.
Wind Energy 2016, 19, 1107–1119.
Energies 2022, 15, 1767 17 of 17

9. Mishnaevsky, L., Jr. Repair of wind turbine blades: Review of methods and related computational mechanics problems. Renew.
Energy 2019, 140, 828–839. [CrossRef]
10. Boopathi, K.; Mishnaevsky, L., Jr.; Sumantraa, B.; Premkumar, S.A.; Thamodharan, K.; Balaraman, K. Balaraman, Failure
mechanisms of wind turbine blades in India: Climatic, regional and seasonal variability. Wind Energy 2021. [CrossRef]
11. Katnam, K.B.; Comer, A.J.; Roy, D.; da Silva, L.; Young, T. Composite Repair in Wind Turbine Blades: An Overview. J. Adhes.
2014, 91, 113–139. [CrossRef]
12. Eom, Y.; Boogh, L.; Michaud, V.; Sunderland, P.; Månson, J.-A.E. Dynamics of void formation upon curing of epoxy resin. In
Proceedings of the 12th International Conference on Composite Materials (ICCM 12), Paris, France, 5–9 July 1999.
13. Kardos, J.L.; Duduković, M.P.; Dave, R. Void Growth and Resin Transport During Processing of Thermosetting—Matrix
Composites. In Advances in Polymer Science 80; Springer: Berlin/Heidelberg, Germany, 1986; pp. 101–123.
14. Tang, J.-M.; Lee, W.I.; Springer, G.S. Effects of Cure Pressure on Resin Flow, Voids, and Mechanical Properties. J. Compos. Mater.
1987, 21, 421–440. [CrossRef]
15. Mischnaewski, L., III; Mishnaevsky, L., Jr. Structural repair of wind turbine blades: Computational model for the evaluation of
the effects of adhesive and patch properties on the repair quality. Wind Energy 2021, 24, 402–408. [CrossRef]