Renewable Energy 36 (2011) 866e871

Contents lists available at ScienceDirect

Renewable Energy
journal homepage: www.elsevier.com/locate/renene

Technical Note

A cost and performance comparison of LRTM and VI for the manufacture

of large scale wind turbine blades
J.R. Hutchinson, P.J. Schubel*, N.A. Warrior
Polymer Composites Group, Division of Materials, Mechanics and Structures, Faculty of Engineering, University of Nottingham, University Park, Nottingham,
NG7 2RD, United Kingdom

a r t i c l e i n f o a b s t r a c t

Article history: Light resin transfer moulding (LRTM) has been developed as an alternative to vacuum infusion (VI) but
Received 17 February 2010 a direct comparison between the two processes is needed to quantify any advantages. This paper uses
Accepted 30 July 2010 a technical cost model and an empirical study to show the potential nancial and performance benets
Available online 15 September 2010
of LRTM for manufacture of a generic 40 m wind turbine blade shell. The use of LRTM when compared to
VI demonstrated a possible 3% cost saving, improved dimensional stability (5.5%), and reductions in resin
Keywords:
wastage (3%) and infusion time (25%). A decrease in internal void formation (0.9%) resulted in an increase
LRTM
in mechanical performance (<4%) for LRTM moulded parts.
Wind energy
Blades
2010 Elsevier Ltd. All rights reserved.
Cost modelling
Mould design

1. Introduction moulding (LRTM) has been developed as a cheap alternative to

Wind energy is a clean, renewable alternative to nite fossil
fuels, and is a potential source for our growing energy needs. Total
global capacity increased by 28.8% in 2008 and, using current
trends, the worlds wind energy production is predicted to nearly
triple in the next ve years [1]. At present, around 20% of a wind
turbines total cost comes from the manufacture of its blades [2].
Developments in airfoil design and materials and growth in infra-
structure are allowing blades to continue to grow both in size and
cost. In order for wind energy to stay economically competitive
manufacturing technology needs to evolve accordingly.
Vacuum infusion (VI) is currently one of the most popular
methods for production of large scale wind turbine blades. It
requires only one unmatched mould half allowing lower cost
tooling compared to resin transfer moulding (RTM). Operational
costs however are higher due to the large amounts of consumables
needed; bagging lm, infusion medium, sealant tape, peel ply,
breather cloth, line-gate wrap and peripheral infusion braid
(optional). Signicant part thickness variation, dry-spots and
varying bre volume fractions are also inherent to VI mouldings, all
of which can result in poor mechanical performance. The process
also requires at least some operative presence during infusion to
watch for, and manage, any localised lock-off. Light resin transfer
* Corresponding author.
E-mail address: peter.schubel@nottingham.ac.uk (P.J. Schubel).
current resin infusion methods such as RTM and VI.
1.1. The LRTM process
The concept uses a rigid lower mould and a semi-exible,
composite upper mould supported by a steel frame to give it
sufcient stiffness (Fig. 1). A vacuum is drawn in the tool cavity,
pulling the composite shell down onto the bottom mould and
sealing the tool. Resin can be injected at any pressure up to atmo-
spheric; above this and the upper mould would lift. Compared to
RTM and VI, LRTM offers potential savings in its low-cost tooling
and lack of consumables. It also has the possibility of better tool and
part thickness control than VI.
2. Technical cost modelling
A detailed technical cost analysis has been conducted on
the shells of a generic 40 m wind turbine blade. The cost model
has been developed in-house and is based on an event-driven
process; a sequence of events is dened for each process, each of
which incurs a certain amount of raw materials and labour and
may use capital equipment and tooling [3]. The methodology and
calculated cost from this model have been veried by an inde-
pendent aerospace company and wind energy consultancy
company respectively.
No detailed consideration was given to plant layout in this
work, and no allowance is made for insurance, selling costs,

0960-1481/$ e see front matter 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.renene.2010.07.025
 J.R. Hutchinson et al. / Renewable Energy 36 (2011) 866e871
Fig. 1. Schematic of the LRTM process.
administration, transport, R&D etc. The results presented are
dependent on a number of assumptions:

Component program lifetime e 3 years (tooling costs are
amortised over the duration of the program)

Direct labour e 11.09/hour

Indirect labour factor e 100%

Interest e 7.5%

Building cost e 72.90/m2/per annum

Capital equipment residual value e 5%

Capital equipment installation e 10%

Capital equipment auxiliary equipment cost e 15%

Capital equipment is depreciated over 10 years
The component area and perimeter for the generic 40 m blade
shell were calculated as 270 m2 and 192 m respectively. Both area
and perimeter values are used throughout the model to account
for material usage and oor space rental within each different
manufacturing process. Basic material costs used in the model are
obtained from a variety of sources. Materials and prices are based
on the majority sales within the wind turbine blade manufacturing
industry for that particular type of product and are assumed to be
volume prices where available. Wastage is set for each material
individually.
Analysing the VI and LRTM cost centres with standard produc-
tion parameters set for the UK manufacturing environment
provides a clear indication that materials and labour make up the
majority of blade manufacturing costs (Fig. 2). Materials and labour
costs remain independent of the production volume and account
for 51% and 41% respectively for the VI process of total shell
production costs (based on 1000 parts per annum (ppa)). However,
a materials cost saving of 12% is seen for LRTM over that of VI due to
the reduction in consumables and material waste levels. Tooling
Fig. 2. Three volume levels split by cost centres for the 40 m shells.
867
costs are shown to be considerable for LRTM at low volume levels
(230 ppa) due to the matched moulds and accounts for 14% of the
total shell manufacture cost. However, this cost is reduced to 9% at
higher volumes. The additional cost of tooling has been shown to be
offset by materials savings as early as 154 ppa (Fig. 3). However, due
to the need to include additional tooling and labour to meet higher
production volumes (seen as step changes in Fig. 3), the true
advantages of LRTM are not fully realised until 279 ppa. Where
after, total cost savings of around 3% are seen (based on 1000 ppa).
3. Experimental methods
3.1. Tooling and manufacture method
It would not have been possible or practical to carry out this
study on a full 40 m blade tool due to cost restraints, so a small at
tool with a 700 mm by 700 mm square mould surface was
commissioned (Fig. 4). Although this allowed a direct comparison
of the two processes, the simplistic part geometry and scale meant
it was not possible to assess the practical challenges of lifting and
positioning a composite upper mould for a 40 m LRTM tool. Labour
and resource costs associated with this would need to be compared
to those spent on the task of ensuring a vacuum tight seal around
the part periphery for a VI moulding of the same sized blade.
Currently no computational model exists specically for the LRTM
process. Future work on this study aims to develop such a model in
order to verify the experimental ndings published here and to
allow simulation of a full-scale 40 m wind turbine blade moulding.
The LRTM composite shell was constructed from random E-glass
mat to provide increased exibility and reduce the cost. The cavity
vacuum was contained within two o-ring edge seals and a periph-
eral scavenger seal (Fig. 4), and part thickness was dened by
a 4 mm silicon picture frame. Resin was injected centrally into
a 400 mm silicon line-gate which had an omega cross-section to aid
resin ow. This meant the resin had to travel a maximum of
350 mm, compared to a 40 m blade tool where typical line-gate
branching necessitates a 1500 mm resin ow. Thin strips of bleeder
material were placed on the top face of the picture frame in each
corner to create four corner vents. The bottom mould was water
pre-heated to 80  C and the resin was pre-heated to 25  C in the
homogeniser to give it the desired viscosity of 290 mPas. The resin
injection was automated by a Hyperject 3 control unit supplied by
Magnum Venus Plastech. An electronic transducer measured cavity
pressure, ensuring it never exceeded atmospheric. This pressure
reading was automatically fed back into the control unit, which
would alter the position of the resin injection valve accordingly.
VI injections were carried out in the heated bottom mould of the
LRTM tool. The reinforcement was covered in a nylon peel ply
(Aerovac A100FLOGREEN e 85 gsm), resin diffusion layer (Aerovac
VI1), and vacuum bag material (Aerovac VACFILM450V248) and
sealed using mastic tape (Aerovac LTS90B). All of these were
process consumables, and had to be replaced for each moulding.
The vacuum ports were located in the four corners of the tool to
represent the four corner vents used in the LRTM process. Resin was
injected into a central 400 mm infusion wrap line-gate (Aerovac
INFUIONSWRAP12), maintaining the 350 mm resin ow length.
3.2. Materials
This study employed a wind energy grade two-part epoxy-based
resin system specically designed for this application by Hexcel
Composites. The reinforcement, supplied by OCV, was a wind
energy grade unidirectional E-glass fabric with an aerial weight of
1265 gm2 and an integrated random-bre coating to aid resin
dispersion.
868 J.R. Hutchinson et al. / Renewable Energy 36 (2011) 866e871

Fig. 3. Effects of production volume on the manufacture of a VI and LRTM shell.

Fig. 4. Commissioned LRTM tool (left) and composite upper mould with seals (right).

4. Mechanical testing
The modulus and UTS of the mouldings were determined using
a dual column, hydraulic Instron mechanical tester. All tests were
performed according to BS EN ISO 527-4, and each test was
repeated ve times to obtain an accurate average and prove
repeatability. Precision epoxy was used to tab the specimens to
ensure correct loading.
The exural modulus and strength of the mouldings were
calculated from a 3-point-bend test in compliance with BS EN ISO
178:2003, repeated ve times to detect and eliminate the effect of
anomalous specimens.
The only signicant error in these tests came from the limits of
the equipment, which gave a maximum experimental error of 1%
for the strength results and 2% for the modulus values. The Ins-
tron has a load accuracy of 0.5%, and the extensometer has an
error of 0.25 mm on the gauge length (0.5%) and 0.25% on the
reading.
4.1. Results and discussions
Table 1 shows the LRTM laminates had marginally greater
tensile and exural strengths, while VI produced slightly stiffer
mouldings. The differences in strength and modulus between the
two processes, taking into account the experimental error, were
minimal; between 0 and 4%. The standard deviations showed
results varied within each test, up to a maximum of 11.2% for the VI
exural strength. Samples were taken from various areas of the

Table 1
Mechanical properties for LRTM and VI specimens.

Tensile Standard Tensile Standard Flexural Standard Flexural Standard

strength (MPa) deviation (%) modulus (GPa) deviation (%) strength (MPa) deviation (%) modulus (GPa) deviation (%)
VI 733.9 0.6 38.5 9.8 790.2 11.2 31.2 4.2
LRTM 760.0 9.7 36.7 0.9 832.1 3.7 28.3 8.7
 J.R. Hutchinson et al. / Renewable Energy 36 (2011) 866e871 869

Table 2 therefore VF, of a VI moulding is limited only by the fabric perme-

Fibre volume fractions and laminate thicknesses for each process. ability and compressibility, and the pressure and viscosity of the
Experimental Standard Average Standard Maximum resin [4]. The semi-rigid state of the LRTM composite shell however
bre volume deviation laminate deviation variance (%) gives the tool a minimum cavity height regardless of the physical
fraction (%) thickness (mm) (%) properties of the reinforcement and resin. Despite giving a lower
LRTM 0.46 2.9 3.89 5.9 13.2 average VF, this extra mould stiffness did result in greater stability
VI 0.49 11.1 3.38 11.5 25.7
and control for the LRTM method as shown by the standard devi-
ation of the results.
It has been shown that the distribution of local thickness and VF
laminates, so the deviation can be attributed to the different sample of a VI moulding is dependent on the distance from the inlet [5],[6].
thicknesses, a result of non-uniform bre volume fractions and The plots of the thickness measurements for both laminates (Fig. 5)
possible edge effects during sample preparation. At best, LRTM show LRTM (left) follows a similar trend. The highest part thickness
mouldings performed around 4% better mechanically than VI, and can be seen in the area where the resin was injected into and along
at worst, their performances were equal. the central line-gate, with local thickness decreasing towards the
part edge. The VI plot (right) follows a similar pattern to some
5. Physical properties degree, but the increased variability has resulted in a number of
non-uniform peaks and troughs. A non-commercial analytical
The shape and surface of a wind turbine blades aerofoil deter- model for the resin ow in the LRTM process was not derived in this
mines how much power can be generated, so the processing study, but will be left for a future effort.
accuracy is crucial. By analysing the physical characteristics and Correia et al. [5] showed the VI uid pressure eld, and as
structure of the mouldings, the differences between the two pro- a result volume fraction and thickness, to be scalable. For a 40 m
cessing methods can be better understood and quantied. blade, LRTM could see reduced dimensional stability due to the
greater tool width, particularly at the root, allowing the centre of
5.1. Fibre volume fraction and laminate thickness the composite tool to ex more relative to the cavity edge. A stiffer
composite shell could be required, possibly employing some
The bre volume fraction (VF) gives an indication of mechanical directional bre architecture.
performance as well as the level of compaction and dimensional
accuracy. A burn-off test was conducted as specied by BS EN 5.2. Void content
ISO 1172:1999 and was repeated eight times for each process
from various locations in the laminates. The thicknesses of the Voids arise from air entrapment within the resin and can result
two nished parts were also measured. A deep-throat Vernier in crack propagation and premature failure. Knowing and
micrometer, rated with an accuracy of 0.02 mm was used to controlling the void content in a moulded component is critical in
measure the laminate thickness at multiple points. Each moulding predicting how and when the part might fail. It has been shown
was divided into a 10  10 grid and a thickness reading was taken in that a 2% increase in void content can reduce exural strength by
the middle of each square. 20% and exural modulus by 10% [7] e two properties which are
The VI process produced a much thinner laminate with a 3% crucial to wind turbine blades during operation.
higher average VF than LRTM (Table 2). The more compliant draping A void analysis was conducted to BS EN ISO 7822:1999 using
properties of the bagging lm means the part thickness, and a statistical counting method. Eight samples were taken from

Fig. 5. Final part thickness for LRTM (left) and VI (right) mouldings.
870 J.R. Hutchinson et al. / Renewable Energy 36 (2011) 866e871

Fig. 6. 5 magnied view of worst case void for VI (top) and LRTM (bottom).

Table 3 V total volume of laminate (calculated using the average

Experimentally determined void content analysis.
thickness results)
Void content (%) Mean void size (mm2) Standard deviation (%) The results in Table 4 show that LRTM has a lower associated
LRTM 1.2 0.00019 0.31 resin loss. During VI injections, resin was drawn into the vacuum
VI 2.1 0.00026 0.47 lines before the part had been fully wet-out. This was less apparent
in the LRTM method, as the vacuum was bled into and contained
within the double o-ring seal, preventing resin from reaching the
Table 4 vacuum port.
Process resin wastage.

Resin Waste (% of Resin Injected) 6.2. Fill time

LRTM 4.57
VI 7.62
various locations on each part and potted in casting resin. The
surface was then polished and studied under a microscope. Image-
pro software was used to measure and count void content.
The voids in the VI samples were larger and more abundant
(Fig. 6 top). LRTM produced lower void content, smaller voids and
lower standard deviation (Table 3). This could account for the
higher mechanical performance of the LRTM mouldings.
Aside from material characteristics, ll time is governed by the
driving pressure of the resin. In VI this is equal to the atmospheric
pressure minus the residual pressure left inside the tool cavity by
the inefciency of the vacuum. In reality vacuum strengths rarely
exceed 90%. The automated LRTM injection unit negates all vacuum
associated losses using the cavity pressure transducer controlled
auto-sprue to ensure resin driving pressure is always at atmo-
spheric. This means ll times can be reduced by a minimum of 10%.
Injection times observed in this study show a 25% reduction
(Table 5).

7. Conclusions
6. Process properties
The technical cost modelling for the VI and LRTM processes has
6.1. Material wastage e resin loss
shown that for this particular analysis of a generic 40 m blade,
a potential cost saving of 3% can be achieved. The analysis has
Both processes allow net shape moulding but experience resin
assumed that labour levels remain the same for the two processes.
loss in injection and vacuum lines. The percentage resin loss was
However, in reality it would be anticipated that fewer personnel
quantied for each process using Eq. (1), by determining the mass
would be required to complete one LRTM moulding over that of VI
of resin injected, the nished laminates volume and its VF.
and setup time would also be reduced. This would further add to
mr  rr  V  1  VF the cost effectiveness of LRTM as a liquid moulding process.
Rwastage (1) Unfortunately, the labour and time savings were not able to be
mr
conrmed and therefore were not included in this analysis.
mr mass of injected resin Mechanically, LRTM mouldings performed up to 4% better than
rr density of resin those produced by VI, most likely due to their lower propensity for
internal void formation. LRTM also demonstrated its higher
dimensional stability, reproducibility and lower material wastage.
Table 5
A comparison of experimental injection times. The automated control unit reduced operator input, and resulted in
faster infusion times.
Average ll time (minutes)
Future work using a larger part with more representative
LRTM 8.3 geometry could help to further this comparison study and prove
VI 11.0
LRTMs performance. Overall this study has highlighted the
 J.R. Hutchinson et al. / Renewable Energy 36 (2011) 866e871
potential of LRTM as a method for producing cheaper wind turbine
blades, whilst maintaining and even improving part performance.
Acknowledgements
The authors of this paper would like to thank all the members of
the AIRPOWER consortium for their help. This project was/is co-
funded by the Technology Strategy Boards Collaborative Research
and Development programme, following an open competition.
The Technology Strategy Board is an executive body established
by the Government to drive innovation. It promotes and invests in
research, development and the exploitation of science, technology
and new ideas for the benet of business e increasing sustainable
economic growth in the UK and improving quality of life.
871
References
[1] Global Wind Energy Council. Global wind; 2008 [Report].
[2] http://www.compositesworld.com/articles/wind-turbine-blades-big-and-
getting-bigger.aspx, June 2008 e last accessed 20/04/2009.
[3] Schubel PJ. Technical cost modelling for a generic 45m wind turbine blade
produced by vacuum infusion (VI). Ren Energ 2010;35:183e98.
[4] Kessels JFA, Jonker AS, Akkerman R. Fully 2.5D ow modeling of resin infusion
under exible tooling using unstructured meshes and wet and dry compaction
properties. Composites: Part A 2007;38:51e60.
[5] Correia NC, Robitaille F, Long AC, Rudd CD, Simacek P, Advani SG. Analysis of
the vacuum infusion moulding process: I. Analytical formulation. Composites:
Part A 2005;36:1645e56.
[6] Andersson HM, Lundstrm TS, Gebart BR. Numerical model for vacuum infu-
sion manufacturing of polymer composites. Int J Numer Meth Heat Fluid Flow
2003;13(2e3):383e94.
[7] Ghiorse SR. Effect of void content on the mechanical properties of carbon/epoxy
laminates. SAMPE QUART 1993;1:54e9.