Composite Structures 220 (2019) 699–707

Contents lists available at ScienceDirect

Composite Structures
journal homepage: www.elsevier.com/locate/compstruct

Robotic filament winding: An innovative technology to manufacture T

complex shape structural parts
Luca Sorrentinoa, , Emmanouel Anamaterosb, Costanzo Bellinia, Luigi Carrinoc,
⁎

Giuseppe Corcioneb, Antonio Leoneb, Gianluca Parisb

a
Department of Civil and Mechanical Engineering, University of Cassino and Southern Lazio, via G. Di Biasio, 43, 03043 Cassino, Italy
b
Leonardo Helicopters, Via G. Agusta, 520, 21017 Cascina Costa di Samarate, Italy
c
Department of Chemical, Materials and Production Engineering, University of Naples Federico II, Piazzale V. Tecchio 80, 80125 Napoli, Italy

ARTICLE INFO ABSTRACT

Keywords: The robotic filament winding (RFW) technology consists of an industrial robot that is equipped with a feed and
Polymer-matrix composites (PMCs) deposition systems; they are suitable for placing a tape along the stress directions the part is subjected to during
Structural composites operation. RFW cell replaces the human operator that is the only system that has been previously used to
Mechanical properties manufacture complex shape parts. It represents an advantage as regards process control, repeatability and
Complex shape
manufacturing times. The present work focuses on the family of structural parts whose shape may be obtained as
Filament winding
a result of a full section sweeping along a closed non auto-intersecting 3D curve. The aim of the paper is to
present a design methodology for the forming process of complex shape structural parts by RFW, in order to
obtain structural homogeneity and fibre content uniformity, and so to demonstrate the potential of an innovative
robotic cell for manufacturing these parts.

1. Introduction Afterwards, a significant improvement in the quality of the products

was possible thanks to the development of automatic systems and ad-
Filament winding (FW) was born in the 50s as a production tech- vanced processing units, which allowed a greater control of the process
nology for composite materials and it underwent a considerable evo- parameters, a better precision and a higher repeatability of the manu-
lution over time. The process consists in winding resin-impregnated factured parts; in the scientific literature it is possible to find different
continuous fibres, coming from single coils, over a rotating mandrel by studies aimed at the optimization of the physical and mechanical
means of a delivery eye that moves parallel to the axis of the mandrel. characteristics of the produced parts as a function of the process
The impregnation can take place just before winding (wet winding) or parameters [11–19]. Researchers are currently interested in beha-
prepreg materials can be used (dry winding); the resin can be added vioural analysis of both the damage caused by high and low velocity
after winding process too [1,2]. The cure phase of the material, the impacts and the fatigue strength of fibre reinforced polymer and/or
possible removal of the mandrel and the finishing operations complete hybrid structures [20–23].
the process. The advantage of the FW technology lies in the ability to Since the 90s, in parallel to the study and optimization of axial-
manage the process parameters, as tape tension and fibres impregna- symmetrical structures, the interest was born for creating complex-
tion, and to vary the winding angle in a programmed and automated shaped composite products, such as T, L, S and Y joints, that are shapes
way during the fibre deposition, in order to optimize the stratification infeasible with traditional FW systems, characterized by 2 degree of
as a function of the maximum stress directions of the part during op- freedom. This need, and more generally the feasibility of complex
eration, and so of the finished product structural properties. The most geometries, became the object of study of some researchers, who led to
used manufacturing systems present a two degree of freedom actuated the development of innovative robotic cells with more than 2 degree of
mechanism as structure; this configuration, called “traditional FW”, is freedom [24–27]. In general, a robotic filament winding system is
directed towards the manufacturing of axisymmetric geometry parts. Its composed of an industrial robot that is equipped with a feed and de-
application is mainly limited to the manufacturing of pressure tanks position system for placing fibres impregnated with resin along the
and tubular parts [3–10]. major stresses directions of complex shape parts. Robotic filament

⁎
Corresponding author.
E-mail address: [email protected] (L. Sorrentino).

https://doi.org/10.1016/j.compstruct.2019.04.055
Received 15 January 2019; Received in revised form 6 March 2019; Accepted 8 April 2019
Available online 09 April 2019
0263-8223/ © 2019 Elsevier Ltd. All rights reserved.
L. Sorrentino, et al. Composite Structures 220 (2019) 699–707

winding (RFW) cell replaces the human operator, that represented the a careful design of the forming equipment is appropriate in order to
only system previously used to manufacture complex shape parts. RFW guarantee a uniform material compaction at the end of the poly-
represents an advantage as regards process control, repeatability and merization process.
manufacturing times [28,29]. Therefore, at present, the robotic cells are
used for manufacturing parts with complex geometry, even with con- 2.1. Planning of manufacturing process
cave surfaces, but only with a hollow section. For the feasibility of parts
with concavity, the adopted solution is to stratify the coils of fibre in a In order to manufacture composite structural complex parts by
direction transverse to the local curvature, in order to avoid fibre guarantying their quality in relation to damage acceptability and good
bridging, that occurs in the presence of concavities and it consists in the mechanical properties, to consider three main elements is essential: i)
formation of bridges by the fibres not adherent to the mandrel shape. forming tools (forming mould, preforming systems and curing mould);
However, the proposed solution allows to realize only “tubular” parts ii) winding trajectory; iii) stratification sequence.
subject to hydrostatic loads, as T, L, S and Y joints, therefore it is not
possible to produce “full section” structural parts. These systems re- 2.1.1. Design of forming tools
present the birth of a new filament winding concept, a new more For this class of parts, once the geometry and the related quantity of
flexible and versatile technology that is called Robotic Filament material to be deposited in the mould have been deliberated, to define
Winding (RFW) and it is the evolution of Traditional Filament Winding the geometry of the forming moulds is opportune because it strongly
(TFW): the basic concept remains the same, but it is extended to the characterizes the quality of the final product. These equipments can be
generation of complex trajectories. The aim is to generate winding subdivided into 4 main elements, as reported in Fig. 2: i) the central
trajectory along geodesic and non-geodesic paths, free of slippage and body, that represents the element on which all the auxiliary systems
bridging, in order to guarantee the stability of the tape on a mandrel and the tool for the subsequent polymerization are connected; ii) the
[30–36]. side walls, which guarantee the tape stability during the winding on the
The present work focuses on the family of structural parts whose mould; iii) the preforming system, driven by pneumatic cylinders,
shape may be obtained as a result of a full section sweeping along a which assure the localized pre-compaction for each tape layer on the
closed non auto-intersecting 3D curve. In previous studies, the cell mould to avoid the “bridging” effect, that causes a lack of adhesion
structure [37] and the potential in terms of managing the process between the material to be laid and the already deposited one, as de-
parameters and the realization of complex trajectories in space [38–42] noted in Fig. 3; iv) the curing tools, consisting of several sub-elements,
were presented. This work has the aim of presenting a design metho- which must guarantee both the homogeneity of the compaction pres-
dology for the forming process by RFW of complex shape structural sure and the geometric conformity during the polymerization process.
parts, presenting concave surfaces, in order to obtain structural Among the 4 above-mentioned elements, the preforming system is
homogeneity and fibre content uniformity, and so to demonstrate the of importance and innovation. As anticipated, it allows to eliminate the
potential of an innovative robotic cell for manufacturing these parts. “bridging” effect and to obtain a compaction degree on the concave
surfaces like that obtainable on flat or convex surfaces. Establishing the
2. Material and method correct dimensions of the preforming system and the compaction force,
driven by a pneumatic cylinder, is proper to achieve the desired com-
In the present work a family of complex shape structural parts was paction degree; for the purpose of the present study, the pressure was
considered; an example is shown in Fig. 1a. The considered part, that is determined according to the size and geometry of the preforming
named “fork” and it represents the structural element of a helicopter system and the tape tension.
component for assembling the blade to the rotor (Fig. 1b), is stressed
mainly by a unidirectional load due to the centrifugal force (Fc) of the 2.1.2. Design of winding trajectory
blade which stresses the fork radially with respect to the rotor axis of The path along which the tape has to be wound is called deposition
rotation. Therefore, in order to better utilize the mechanical char- path. Then, the trajectory along which the deposition head should move
acteristics of the composite material, the fibres must be positioned on in order to wind the tape along the defined deposition path needs to be
the mould along the load direction that will be generated inside the delineated. The winding trajectory of the deposition head is a set of
part. Given the geometric complexity of the part, the only technology points ordered in space; it represents the image of the points of the base
suitable for the stratification of fibres in these directions is the RFW. path, as shown in Fig. 4. The existence of a biunivocal relationship
Moreover, since these parts are structural ones with high mechan- between the points of the base path and those of the winding trajectory
ical performances, which require a high structural uniformity (with a is absolutely necessary. It is possible to generate the winding coils
high fibres/matrix volumetric ratio and low voids content), to carry out through extrusion along the base path of the cells belonging to the

Fig. 1. An example of complex shape part: a) fork; b) assembly of rotor and the helicopter blade.

700
L. Sorrentino, et al. Composite Structures 220 (2019) 699–707

Fig. 2. Forming tools: winding mould and curing mould.

tape in terms of alignment and compaction on the underlying layers of

material. The tension T applied to the tape produces a normal force P on
the deposition surface in accordance with the relation P = T/(r * l),
where r is the curvature radius of the mould and l the width of tape. It is
precisely the force P that compacts the material and develops frictional
forces that prevent relative movements of the fibres. This is valid for a
convex surface, while for concave surfaces a preforming system is re-
quired.
A constant tension value during winding is an indispensable con-
dition for achieving a homogeneous distribution of the compaction
degree throughout the part thickness, and it is ensured by a tensioning
system. In this work, an innovative integrated deposition with a ten-
sioning system was introduced [37]; as visible in Fig. 6, it consisted of
some subsystems (feed, tensioning, driving and deposition one)
mounted on a frame, connected by means of a flange on the robot end
effect in order to avoid the torsion of the tape during the continuous
Fig. 3. Tape deposition on concave surface: fibre bridging effect and its removal
winding. In order to guarantee a continuous unwinding of the tape from
by pneumatic preforming system.
the bobbin and not to lose tape tension, to plan an appropriate winding
path is opportune: the winding trajectory must be planned by offsetting
the mould surface, since it is characterized by a trajectory angle (θ)
higher than 90°. As reported in Fig. 7a, the trajectory angle (θ) is the
angle that the movement vector of the deposition system, which is
tangent to the winding trajectory, forms with the tape direction. If the
deposition system moves from A′1 to A′4 along the trajectory during
winding, the trajectory angle is higher than 90° in each of the 4 con-
sidered points of the winding trajectory. The trajectory angle assumes a
value of 90° when the winding trajectory is put at infinite distance from
the die. The value of the trajectory angle decreases with the increase of
the distance of the winding trajectory from the mould surface, as visible
in Fig. 7b, where α > β > γ, while the value of the trajectory angle
keeps constant along the same winding trajectory, as shown in Fig. 7a.
This involves a more continuous movement of the deposition system
without sudden change in head direction; thus, favouring to keep
Fig. 4. One of the wound coils: deposition path (blue line) and its winding constant the tension on the tape; more details are reported in Refs.
trajectory (red line). [38,39]. Therefore, for a complex shape part, like that represented in
Fig. 1, to offset the boundary curve of the winding die is enough; Fig. 4
straight section of the workpiece. To guarantee an accurate winding of shows one of the winding trajectories obtained by offsetting the closed
the tape on the support, the winding trajectory must be tangent to the path used to build the fork model.
tape trajectory along each contact point between the winding tape and Moreover, the deposition head and the robot arms should be moved
the already wound tapes, as shown in Fig. 5. on collision free trajectories, that is the free tape must not interfere with
During the winding process, impressing a tensile force, called the support and the whole environment. A control volume is generated
winding tension, on the tape is necessary for the correct laying of the to calculate the impacts, in order to keep the path outside the

701
L. Sorrentino, et al. Composite Structures 220 (2019) 699–707

Fig. 5. Winding die of the fork component with one of its winding trajectories.

Fig. 6. Deposition integrated system.

Fig. 7. Trajectory angle: a) along the same trajectory; b) for some trajectories at different distances from the mould.

702
L. Sorrentino, et al. Composite Structures 220 (2019) 699–707

Fig. 8. Simulation of a planning winding trajectory.

considered volume (see Fig. 8). The control volume is a positive offset
of the solid model related to part-support assembly.

2.1.3. Design of stratification sequence

The design of the winding process consists in the definition of the
optimal stratification sequence of the tape on the winding mould, so as
to assure a constant reinforcement content and a correct orientation of
the fibres, which represent two conditions needed for ensuring the
structural resistance of the part. The particular part geometry requires a
change in the deposition direction of the tape, which has a rectangular
section with a base of 3.2 mm and a height of 0.18 mm. This is due to
the need to maintain stable the tape, along geodetic trajectories, on the
layering surfaces of the mould, which have different orientations in the
space, as shown in Fig. 9. In fact, the stratification surface is parallel to Fig. 10. Stratification sequence for a generic section of the part.
the plane YZ in the zone A, while to the XY plane in zone C. To better
understand the layering sequence, a generic section of the part should
be considered: the individual tapes are first layered next to each other,
on a plane parallel to the stratification surface, to form a layer, and
several successive layers are deposited along the growth direction,
which is orthogonal to the stratification plane, as shown in Fig. 10. As
reported in Fig. 11, in zone C each layer expands on a plane parallel to
the XY plane and the successive layers have an increase in the Z di-
rection, while in zone A there is a development of each layer on a plane
parallel to the plane YZ, with a growth of the subsequent layers along
the X axis. Therefore, the tape in zone A is rotated 90° around its axis
with respect to the one deposited in zone C, as can be seen from the
rotation of the local uv reference system, anchored to the tape section.
The tape twisting around its longitudinal axis is one of the factors to
keep under control for ensuring a high product quality. Therefore, it is
necessary to warrant that the deposition tool changes its orientation Fig. 11. Twisting of the tape.
during the winding phase, in order to allow the correct tape layering on
the two indicated planes and to locate the torsion point in a well de- mechanical performances: in the case of the present study, this area is
termined area. This area must be the least stressed of the whole com- represented by the B zone in Fig. 9.
ponent, as it represents a defect with a consequent reduction of the Once the winding phase is completed, the side walls are removed
and the curing tools are assembled on the central body for the sub-
sequent autoclave polymerization, as reported in Fig. 12.

2.2. Experimental tests to evaluate the parts quality

The quality of parts made of composite material is connected to the

presence of defects inside part volume or on its boundary surface. These
defects are inevitable, since manufacturing process is very complex;
they are due to the different manufacturing steps, such as winding, die
closing and curing.
This paragraph defines and describes the tests performed on the part
in order to evaluate its quality and, consequently, to indirectly verify
the characteristics of both the robotic cell and deposition system in
terms of repeatability and potentiality in ensuring conformity of the
Fig. 9. Stratification surfaces on the mould. manufactured parts to the project requirements. A preliminary visual

703
L. Sorrentino, et al. Composite Structures 220 (2019) 699–707

Fig. 12. a) part layered on its winding mould; b) curing tools mounted on the central body.

Table 1
Destructive tests made on the parts.
Spatial identification Test

B Density (ASTM D792)

D Resin and void content (ASTM D3171)
W Interlaminar shear strength (ASTM D2344)

Fig. 14. Overall view of a sectioned part.

Fig. 13. Position for specimen collection.

The fork was sectioned in 15 parts (S1 ÷ S15), as visible in Figs. 13
and 14; this sectioning was used to evaluate both the presence of por-
analysis was carried out on the manufactured parts, in order to quali- osity inside the part, by means of a visual inspection, and to obtain the
tatively evaluate the presence of defects. Subsequently, a quantitative samples for the mechanical and chemical tests. In particular:
quality assessment was made by means of destructive structural tests.
The visual analysis was carried out by examining, by means of an op- •S 3 was used for obtaining the specimens B1, B3, D1, D3 and W1;
tical microscope, the surfaces obtained from the different sections •S 4 was utilized for extracting the specimen W3;
executed on the part for collecting the samples for the destructive tests, •S 11 was used for obtaining the specimen W4;
as indicated in Table 1 and Fig. 13. The specimen surfaces were po- •S 12 was utilized for extracting the specimens B2, B4, D2, D4 and W2.
lished with abrasive cloth and observed under a stereoscopic micro-
scope. In addition, DSC (differential scanning calorimetry) tests were car-
For the quantitative evaluation of the parts, the structural test ried out on samples of material obtained from sections S3 and S4
methods and the minimum accepted or admissible values were those of (specimens E1 and E2), that represented the thinnest and the thickest
the reference standards in the aeronautical sector; the types of tests sections, respectively. The absence of residual polymerization enthalpy,
performed were: i) density; ii) resin and voids content; iii) interlaminar which was found at the end of the DSC test, was indicative of a com-
shear strength (ILSS); iv) differential scanning calorimetry. To perform plete cure of the examined parts.
the above mentioned tests, it was necessary to obtain specific samples
directly from predetermined zones of the parts, so as to be able to 3. Result and discussion
evaluate the structural uniformity on the whole parts. The specimens
were identified by a letter, as shown in Table 1, which identifies the test Some experimental tests were executed to evaluate the rules ex-
method according to ASTM standards, while the indicative positions in plained in the previous paragraph and to verify the quality of manu-
which the specimens were obtained are presented in Fig. 13 (B1, B2, B3, factured parts. In fact, structural tests carried out on produced parts are
B4, D1, D2, D3, D4, W1, W2, W3, W4). useful for assessing the whole process suitability [43]. The planned

704
L. Sorrentino, et al. Composite Structures 220 (2019) 699–707

Fig. 15. a) Robotic filament winding cell; b) Winding die: closing of preforming system on the tape.

shown in Fig. 17. As it can be seen from Fig. 18a, the presence of voids
can be detected on some sections, identifiable as darker areas, which
involve a certain porosity and they are indicative of a possible reduc-
tion in the material resistance. ImageJ software was used to quantify
the voids presence. In fact, to process the images of the sections was
possible by means of this software, that was useful for highlighting the
darker areas (Fig. 18b), and then defining the percentage portion of the
section occupied by these areas (Fig. 18c). From the visual examination
results it can be stated that the area percentage covered by voids and
porosity is between 1 and 2%, with a very low coefficient of variation
(CoV).1
As concerns structural investigations, the results of the different
tests mentioned in the materials and methods section are reported in
Fig. 16. Helicopter fork: final component after the curing process. Table 2, that describes the results of the tests for the determination of
the resin weight content Rw%, the composite material density ρc and the
void content Vc%. Comparing these results with minimal requirement
winding trajectories were implemented by means of a robotic cell for the structural parts, that is indicated in the table too, it can be de-
constituted by an anthropomorphic robot, a unique and innovative duced that the parts can be accepted.1 The void content values reported
device for tape deposition installed on the robot arm and a winding in Table 2 are slightly different from those determined by visual ex-
mould, as shown in Fig. 15 [36,39]. amination. This can be explained by the fact that the former are related
The winding device was designed and built on the basis of com- to the whole volume of a specimen, while the latter to a single section.
pactness, structural lightness, stiffness and functionality principles, in As regards the ILSS tests, the results obtained from experimental
order to guarantee the maximum dexterity of the robot, to minimize the tests (Fig. 19) state an average value of 60.78 MPa, with a standard
probability of crashes between the winding mould and the components deviation of 11.50 MPa and a CoV of 18.9%. The ILSS test results are
of the cell, and to improve the control of the process parameters for satisfactory since the obtained values are above the required one, that is
accuracy and repeatability. 40 MPa.1
The number of composite parts produced was equal to eight. The The technical data sheet of the tape considered in this work reports a
material used to manufacture the parts was carbon tape impregnated by nominal value of the resin density ρR equal to 1.25 g/cm3; therefore,
epoxy resin. The tape had a width of about 3 mm.1 For every produced average values of resin volume content equal to 34.34% and of fibre vo-
composite part more than 100 coils of tape were wound on the winding lume content equal to 63.22% are obtained taking into account the results
mould, in order to achieve the target of deposited tape, and a winding in Table 2 and the relations for calculating the volume resin content
tension higher than 50 N was used. Once wound, the parts were closed
R ·
Rv % = w % C and the volume fibre content Fv % = 100 RV % Vc %.1
between the winding mould and the curing tool; then, they were placed
R
Each of the values found in the individual measurements respected
inside an autoclave to carry out the curing process. The parts were the minimum required value for each kind of test, even in the most
cured at a temperature of about 130 °C and the cure length was 180 critical areas such as the concave ones; therefore, it can be concluded
mins. Finally, the parts were extracted from the die (Fig. 16) and they that the considerations made for the process design, concerning the
were subjected to both preliminary observations and structural tests in definition of the mould, the preforming system, the trajectory and the
order to identify the quality and so the acceptability of the parts. stratification sequence, were valid.
As regards the observations on the surfaces obtained from the part
cut, all the sections showed a non-critical, slight or absent porosity
(< 5%).1 Some examples of the obtained and analysed sections are 4. Conclusion

This paper presents a methodology for manufacturing a complex

1
More details cannot be given in this paper for confidentiality reasons. shape part by filament winding, using an innovative robotic cell, that is

705
L. Sorrentino, et al. Composite Structures 220 (2019) 699–707

Fig. 17. Surfaces obtained from part sectioning.

Fig. 18. Elaboration of a section by means of ImageJ software.

Table 2 process design. First of all, the winding tool must be designed: it con-
Test results for weight resin content, density and void content. sists of several parts that must guarantee the tape stability and a correct
fibre compaction during both winding and curing processes. Then,
Resin content Composite density Void content
[weight %] [g/cm3] [volume %] winding trajectory must be defined, warranting the absence of tape
loosening and a continuous and harmonious movement of the deposi-
Average 28.53 1.50 2.43 tion head. These objectives can be achieved by offsetting the boundary
St. Dev. 1.84 0.03 0.55
curve of the winding mould, thus obtaining a high-quality process.
CoV 6.4 2.0 22.5
Requirement > 25.0 1.50 ± 5% < 5.0 Finally, the optimal stratification sequence must be defined, in order to
obtain the correct fibres orientation and a constant fibre content, re-
legating the fibre orientation change areas in the less stressed zones of
the part. This planning rules allows to minimize the empty spaces, the
wrinkling and the folds inside the composite part and therefore to in-
crease their quality.
To assess the suitability of the proposed methodology, a complex
shape part, belonging to aeronautic field, was chosen to be manu-
factured by robotic filament winding technology. The part quality was
evaluated by several tests, both qualitative and quantitative, to testify
the reliability of the manufacturing process. The qualitative tests con-
sisted in a preliminary observation of the surface of some sections, that
highlighted a low level of porosity in the manufactured parts. As con-
cerns the quantitative tests, they consisted in structural tests, as inter-
laminar shear strength and the determination of resin content, density
and void content. A resin content of about 27–29% was found in the
parts, while the density was between 1.48 g/cm3 and 1.53 g/cm3, the
void content ranged from 2.3% to 2.5% and the interlaminar shear
strength from 55.92 MPa to 64.05 MPa. All the tests pointed out a
quality and a strength higher than the minimum requirements for
aeronautic sector; in fact, the void content was about a half of that
Fig. 19. ILSS test setup. indicated by standard, while the shear strength was about 50% higher
than requested.
constituted by an anthropomorphic robot, a unique and innovative All these results allowed also to verify indirectly both the validity of
feeding device and a winding mould with integrated compaction the rule used to plan the winding trajectory and the flexibility of the
system. The methodology introduces three fundamental steps for robotic cell when 3D complex trajectories are realized.

706
L. Sorrentino, et al. Composite Structures 220 (2019) 699–707

Acknowledgements [17] Henninger F, Hoffmann J, Friedrich K. Thermoplastic filament winding with online-
impregnation. Part B. Experimental study of processing parameters. Compos Part A
Appl Sci Manuf 2002;33:1677–88. https://doi.org/10.1016/S1359-835X(02)
Special thanks to Dr Franco Natalizia, Dr Antonio Casini and Eng 00186-0.
Gianluca Conte of Leonardo Helicopters for the scientific and technical [18] Koussios S, Bergsma OK, Beukers A. Filament winding: kinematics, collision control
support. and process optimisation through application of dynamic programming. Compos
Part A Appl Sci Manuf 2006;37:2088–104. https://doi.org/10.1016/j.compositesa.
2005.12.003.
Funding sources [19] Lü H, Schlottermüller M, Himmel N, Schledjewski R. J Thermoplast Compos
2007;18:469–87. https://doi.org/10.1177/0892705705054409.
[20] Kim Y-S, Kim L-H, Park J-S. The effect of composite damage on fatigue life of the
This research did not receive any specific grant from funding high pressure vessel for natural gas vehicles. Compos Struct 2011;93:2963–8.
agencies in the public, commercial, or not-for-profit sectors. https://doi.org/10.1016/j.compstruct.2011.05.007.
[21] Han M-G, Chang S-H. Failure analysis of a Type III hydrogen pressure vessel under
impact loading induced by free fall. Compos Struct 2015;127:288–97. https://doi.
Data availability
org/10.1016/j.compstruct.2015.03.027.
[22] Ribeiro ML, Vandepitte D, Tita V. Experimental analysis of transverse impact
The raw/processed data required to reproduce these findings cannot loading on composite cylinders. Compos Struct 2015;133:547–63. https://doi.org/
be shared at this time due to legal reasons. 10.1016/j.compstruct.2015.07.088.
[23] Xu J, Ma Y, Zhang Q, Sugahara T, Yang Y, Hamada H. Crashworthiness of carbon
fiber hybrid composite tubes molded by filament winding. Compos Struct
Declarations of interest 2016;139:130–40. https://doi.org/10.1016/j.compstruct.2015.11.053.
[24] Saravanos DA, Lamancusa JS. Optimum structural design of robotic manipulators
with fiber reinforced composite materials. Comput Struct 1990;36:119–32. https://
None. doi.org/10.1016/0045-7949(90)90181-Z.
[25] Steiner KV. Development of a robotic filament winding workstation for complex
Appendix A. Supplementary data geometries. 35th Int SAMPE Symp Exhib Proceedings. 1990. p. 757–66.
[26] Markov L, Cheng RMH. Conceptual design of robotic filament winding complexes.
Mechatronics 1996;6:881–96. https://doi.org/10.1016/S0957-4158(96)00032-3.
Supplementary data to this article can be found online at https:// [27] Totaro G, De Nicola F. Recent advance on design and manufacturing of composite
doi.org/10.1016/j.compstruct.2019.04.055. anisogrid structures for space launchers. Acta Astronaut 2012;81:570–7. https://
doi.org/10.1016/j.actaastro.2012.07.012.
[28] Polini W, Sorrentino L. Comparison of winding cells with one and two rovings in
References robotized filament winding technology. SAE Trans 2003;112:331–9.
[29] Polini W, Sorrentino L. Design of deposition head trajectory for robotized filament
winding of complex shape parts. ASME. International Design Engineering Technical
[1] Miglioreno G, Henriksen TK. Structure, mechanism and deployables. In: Macdonald
Conferences and Computers and Information in Engineering Conference, Volume
M, Badescu V, editors. The international handbook of space technology. Berlin:
3d: 8th Design for Manufacturing Conference. 2004. p. 897–908. doi:10.1115/
Springer-Verlag; 2014. p. 197–248.
DETC2004-57774.
[2] Miaris A, Paessler M, Schledjewski R, Mitschang P. Modeling the impregnation
[30] Seereeram S, Wen JTY. An all-geodesic algorithm for filament winding of a T-
process of a siphon impregnation system during filament winding. Pressure Vessels
shaped form. IEEE Trans Ind Electron 1991;38:484–90. https://doi.org/10.1109/
Piping Div ASME 2011;6:79–87. https://doi.org/10.1115/PVP2011-57543.
41.107105.
[3] Park JS, Hong CS, Kim CG, Kim CU. Analysis of filament wound composite struc-
[31] Scholliers J, Van Brussel H. Computer-integrated filament winding: computer-in-
tures considering the change of winding angles through the thickness direction.
tegrated design, robotic filament winding and robotic quality control. Compos
Compos Struct 2002;55:63–71. https://doi.org/10.1016/S0263-8223(01)00137-4.
Manuf 1994;5:15–23. https://doi.org/10.1016/0956-7143(94)90015-9.
[4] Mertiny P, Ellyin F, Hothan A. An experimental investigation on the effect of multi-
[32] Chan S, Munro M, Fahim A. Accuracy-speed relationships of a robotic filament
angle filament winding on the strength of tubular composite structures. Compos Sci
winding cell. Robot Comput Integr Manuf 1996;12:3–13. https://doi.org/10.1016/
Technol 2004;64:1–9. https://doi.org/10.1016/S0266-3538(03)00198-2.
0736-5845(95)00025-9.
[5] Koussios S, Bergsma OK, Beukers A. Filament winding. Part 1: determination of the
[33] Scholliers J, Van Brussel H. Design and off-line programming of a robotic tape
wound body related parameters. Compos Part A Appl Sci Manuf 2004;35:181–95.
winding cell. Robot Comput Integr Manuf 1996;12:93–8. https://doi.org/10.1016/
https://doi.org/10.1016/j.compositesa.2003.10.003.
0736-5845(95)00021-6.
[6] Koussios S, Bergsma OK, Beukers A. Filament winding. Part 2: generic kinematic
[34] Li H, Liang Y, Bao H. CAM system for filament winding on elbows. J Mater Process
model and its solutions. Compos Part A Appl Sci Manuf 2004;35:197–212. https://
Technol 2005;161:491–6. https://doi.org/10.1016/j.jmatprotec.2004.07.093.
doi.org/10.1016/j.compositesa.2003.10.004.
[35] Fu J, Yun J, Jung Y. Filament winding path generation based on the inverse process
[7] Kim CU, Kang JH, Hong CS, Kim CG. Optimal design of filament wound structures
of stability analysis for non-axisymmetric mandrels. J Compos Mater
under internal pressure based on the semi-geodesic path algorithm. Compos Struct
2017;51:2989–3002. https://doi.org/10.1177/0021998316683238.
2005;67:443–52. https://doi.org/10.1016/j.compstruct.2004.02.003.
[36] Fu J, Yun J, Jung Y, Lee D. Generation of filament-winding paths for complex ax-
[8] Zu L, Koussios S, Beukers A. Shape optimization of filament wound articulated
isymmetric shapes based on the principal stress field. Compos Struct
pressure vessels based on non-geodesic trajectories. Compos Struct 2010;92:339–46.
2017;161:330–9. https://doi.org/10.1016/j.compstruct.2016.11.022.
https://doi.org/10.1016/j.compstruct.2009.08.013.
[37] Carrino L, Polini W, Sorrentino L. Modular structure of a new feed-deposition head
[9] Zu L, Koussios S, Beukers A. Design of filament-wound domes based on continuum
for a robotized filament winding cell. Compos Sci Technol 2003;63:2255–63.
theory and non-geodesic roving trajectories. Compos Part A Appl Sci Manuf
https://doi.org/10.1016/S0266-3538(03)00174-X.
2010;41:1312–20. https://doi.org/10.1016/j.compositesa.2010.05.015.
[38] Carrino L, Polini W, Sorrentino L. Method to evaluate winding trajectories in ro-
[10] Pässler M, Schledjewski R. Manufacturing hydrogen storage pressure vessels for
botized filament winding. J Compos Mater 2004;38:41–56. https://doi.org/10.
automotive applications: filament winding improvements. Sampe J 2007;43:8–12.
1177/0021998304038177.
[11] Lauke B, Friedrich K. Evaluation of processing parameters of thermoplastic com-
[39] Polini W, Sorrentino L. Influence of winding speed and winding trajectory on ten-
posites fabricated by filament winding. Compos Manuf 1993;4:93–101. https://doi.
sion in robotized filament winding of full section parts. Compos Sci Technol
org/10.1016/0956-7143(93)90076-K.
2005;65:1574–81. https://doi.org/10.1016/j.compscitech.2005.01.007.
[12] Cohen D. Influence of filament winding parameters on composite vessel quality and
[40] Polini W, Sorrentino L. Winding trajectory and winding time in robotized filament
strength. Compos Part A Appl Sci Manuf 1997;28A:1035–47. https://doi.org/10.
winding of asymmetric shape parts. J Compos Mater 2005;39:1391–411. https://
1016/S1359-835X(97)00073-0.
doi.org/10.1177/0021998305050431.
[13] Banerjee A, Sun L, Mantell SC, Cohen D. Model and experimental study of fiber
[41] Sorrentino L, Polini W, Carrino L, Anamateros E, Paris G. Robotized filament
motion in wet filament winding. Compos Part A Appl Sci Manuf 1998;29A:251–63.
winding of full section parts: comparison between two winding trajectory planning
https://doi.org/10.1016/S1359-835X(97)00091-2.
rules. Adv Compos Mater 2008;17:1–23. https://doi.org/10.1163/
[14] Imamura T, Kuroiwa T, Terashima K, Takemoto H. Design and tension control of
156855108X292648.
filament winding system. SMC ’99 Int Conf Syst Man, Cybern IEEE Conf
[42] Sorrentino L, Marchetti M, Bellini C, Delfini A, Del Sette F. Manufacture of high
Proceedings. 1999. p. 660–5. doi:10.1109/ICSMC.1999.825339.
performance isogrid structure by Robotic Filament Winding. Compos Struct
[15] Mertiny P, Ellyin F. Influence of the filament winding tension on physical and
2017;164:43–50. https://doi.org/10.1016/j.compstruct.2016.12.061.
mechanical properties of reinforced composites. Compos Part A Appl Sci Manuf
[43] Chinchan L, Shevtsov S, Soloviev A, Shevtsova V, Huang JP. Mechanical testing of
2002;33:1615–22. https://doi.org/10.1016/S1359-835X(02)00209-9.
polymeric composites for aircraft applications: Standards, requirements and lim-
[16] Henninger F, Friedrich K. Thermoplastic filament winding with online-impregna-
itations. Springer Proc Phys 2014;152:201–22. https://doi.org/10.1007/978-3-
tion. Part A: process technology and operating efficiency. Compos Part A Appl Sci
319-03749-3_17.
Manuf 2002;33:1479–86. https://doi.org/10.1016/S1359-835X(02)00135-5.

707