You are on page 1of 12

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/296333691

Design of Shape Morphing Cellular Structures and Their Actuation Methods

Conference Paper · October 2015


DOI: 10.13140/RG.2.1.4530.1527

CITATIONS READS

4 1,192

2 authors:

Robin Neville Fabrizio Scarpa


University of Bristol University of Bristol
9 PUBLICATIONS   342 CITATIONS    570 PUBLICATIONS   14,605 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Piezoelectric properties via atomistic finite element method View project

MEDYNA - https://medyna2017.sciencesconf.org/ View project

All content following this page was uploaded by Robin Neville on 29 February 2016.

The user has requested enhancement of the downloaded file.


ICAST2015: 26th International Conference on Adaptive Structures and
Technologies
October 14-16th, 2015, Kobe, Japan

ICAST2015 #035

Design of Shape Morphing Cellular Structures and Their


Actuation Methods

Robin M. Neville1*, Fabrizio Scarpa1

1 Advanced Composites Centre for Innovation and Science (ACCIS), University of Bristol, Queen's
Building, University Walk, Bristol BS8 1TR, United Kingdom
* rn7318@bristol.ac.uk

Abstract

Kirigami techniques are used to create a new type of “open” honeycomb geometry with shape
morphing characteristics. Cables are embedded in the structure to actuate the shape morphing. We
present techniques for designing the manufactured shape of the honeycomb, and an analytical model
to predict the deformed shapes.

1 INTRODUCTION

In this work we present a family of novel honeycombs produced using Kirigami techniques.
Kirigami is the ancient Japanese art of cutting and folding paper. This technique can be used to form a
2D sheet material into a 3D cellular structure. By applying these techniques to sheet materials such as
composites and thermoplastics, engineering cellular structures wth complex shapes can be produced.
The idea of engineering Kirigami was first explored by Nojima and Saito [1]. Saito, Pellegrino and
Nojima developed the mathematical framework linking the 2D cutting pattern to the 3D cellular
structure, which enables the creation of specially shaped honeycombs without requiring any
machining [2]. They also present a method for applying Kirigami to fibre-reinforced composites [2].
This work builds on that of Nojima and Saito, and introduces the concept of an “open”
honeycomb (see Figure 1). We name this architecture “open” honeycomb because it lacks a closed cell
topology (we follow the naming conventions of foams, which are also called “open” or “closed”
depending on their cell configuration). The folds in the structure give it significantly different
mechanical behaviour compared to closed honeycombs. The folds give rise to two new parameters –
fold angle 𝛼 and fold stiffness 𝑘. In past work we have characterised the effect of 𝛼 and 𝑘 on the
mechanical behaviour of open honeycombs, and found that the folds give the structure reduced
density, significant anisotropy, and variable stiffness [3], [4].
In this work we exploit two key features which arise from using the Kirigami process.
1. The folds can also function as “hinges” which allow the structure to flex in the Y-
direction, while retaining stiffness in the X-direction. This can be expoited to create a
morphing or deployable structure.
2. By including circles in the 2D cutting pattern, holes are produced in the walls of the 3D
structure. We can then thread cables through these holes to actuate the structure.
The honeycomb studied in this work is an open honeycomb specifically designed for morphing
(see Figure 2). We call this a “bridged” honeycomb because of the bridges which link adjacent rows of
cells. These bridges are intended to reduce the stress on the folds (by providing two 90° folds instead
of one 180° fold), and also to provide a flat surface for potential skin attachment.

1
ICAST2015: 26th International Conference on Adaptive Structures and
Technologies
October 14-16th, 2015, Kobe, Japan

In this work we present the following:


1 A method for designing useful shape open honeycombs. This is very similar to Saito’s
method [2], but is a bit more complicated because there are more variables, and more
room for design.
2 A method for modelling the structure as a series of rigid linkages, which can be used to
predict the deformed shape of the structure in response to a cable load.

Figure 1. Traditional honeycomb (left) and open honeycomb (right). The folds in the open
honeycomb are highlighted yellow.

Figure 2. The “bridged” honeycomb used in this work.

2 KIRIGAMI MANUFACTURING

The Kirigami manufacturing process consists of three main steps:

1. Cutting: a pattern of slits and holes are cut into the sheet material (see Figure 3a,b,c). The
spacing and dimensions of the 2D cutting pattern directly affect the dimensions of the 3D
structure.
2. Corrugating: the sheet material is thermoformed between hexagonal tooling in a hot press.
The slits and holes align with the tool such that the slits describe half a hexagon (see Figure
3d,e). Different shaped tools can be used to produce different shaped cells.

2
ICAST2015: 26th International Conference on Adaptive Structures and
Technologies
October 14-16th, 2015, Kobe, Japan

3. Folding: the corrugated sheet is folded repeatedly back on itself such that the semi-
hexagonal slits open up into hexagonal holes (see Figure 3f,g). The result is an open
honeycomb. Normally one would bond together the adjacent cells to form a prismatic
hexagonal structure, but in this work we leave the structure un-bonded and allow the hinges
to flex.

The bridged honeycomb studied in this work is created using the same process, with a slightly
different cutting pattern. Figure 4 shows the location of the bridges on the cutting pattern and in the
3D structure.
PEEK Aptiv film (supplied by Victrex PLC [5]) is used throughout this work because it can be
quickly and easily thermoformed. A hot press was used to thermoform the PEEK at 200°C. A
Blackman & White Genesis 2100 ply cutter was used to accurately cut the slitting patterns.
It is worth noting that the sheet material itself is of little interest; the Kirigami process can be
applied to any sheet material which can be cut and reshaped.

Figure 3. The Kirigami manufacturing process. a) The flat PEEK sheet. b) The cutting patterns are
made using a ply cutter. c) The cut sheet with periodically spaced holes. d) The sheet is thermoformed
between hexagonal tooling in a hot press to create the hexagonal corrugations. e) The corrugated
sheet. f) The corrugated sheet is folded back on itself repeatedly. g) The resulting open honeycomb.

3
ICAST2015: 26th International Conference on Adaptive Structures and
Technologies
October 14-16th, 2015, Kobe, Japan

Figure 4. Cutting pattern (left) for bridged honeycomb (right). The bridges are highlighted.

3 GEOMETRY TAILORING

In this section the geometry of the bridged honeycomb is defined, and a method is presented for
designing the honeycomb with a specific shape. By changing the cutting pattern we can change the
shape of the honeycomb, but the cutting pattern must always line up with the tooling in the X-
direction, and the tooling dimensions are fixed. Because the X-dimensions of the honeycomb are
fixed, we consider only the shape of the honeycomb in the YZ plane. Figure 5a shows the bridged
honeycomb, and Figure 5b shows the bridged honeycomb viewed in the YZ plane, with the unit cell
highlighted.
It can be seen from Figure 5b that several of the vertices of the 3D unit cell share the same YZ
coordinates. This allows us to use a considerably simplified unit cell in the YZ plane, as shown in
Figure 5c (points J and K represent the locations of holes in the cell walls).
We now find the YZ coordinates of the points. We take point 𝐴 as our starting point and give it
coordinates 𝑌0 and 𝑍0 , as shown in equation (1). From here we find the coordinates of points B to K
using the dimensions shown in Figure 5.

𝐴 = [𝑌𝐴 , 𝑍𝐴 ] = [𝑌0 , 𝑍0 ] (1)


+(𝑐1 − 𝑏1 ) sin 𝛼1 +(𝑐1 − 𝑏1 ) cos 𝛼1
𝐵 = [𝑌𝐵 , 𝑍𝐵 ] = 𝐴 + [ 𝑑 sin(𝛼1 +𝛽1 ) , 𝑑1 cos(𝛼1 +𝛽1 ) ] (2)
− 1 −
sin 𝛽1 sin 𝛽1

𝐶 = [𝑌𝐶 , 𝑍𝐶 ] = 𝐵 + [𝑏1 sin 𝛼1 , 𝑏1 cos 𝛼1 ] (3)


𝑑1 sin(𝛼1 +𝛽1 ) 𝑑1 cos(𝛼1 +𝛽1 )
𝐷 = [𝑌𝐷 , 𝑍𝐷 ] = 𝐶 + [ sin 𝛽1
, sin 𝛽1
] (4)

𝐸 = [𝑌𝐸 , 𝑍𝐸 ] = 𝐶 + [𝜒1 sin 𝜙1 , 𝜒1 cos 𝜙1 ] (5)


𝐹 = [𝑌𝐹 , 𝑍𝐹 ] = 𝐸 + [+𝑏2 sin 𝛼2 , 𝑏2 cos 𝛼2 ] (6)
𝑑2 sin(𝛼2 +𝛽2 ) 𝑑2 cos(𝛼2 +𝛽2 )
𝐺 = [𝑌𝐺 , 𝑍𝐺 ] = 𝐹 + [ sin(𝛽2 )
, sin(𝛽2 )
] (7)

𝐻 = [𝑌𝐻 , 𝑍𝐻 ] = 𝐺 + [−𝑐2 sin 𝛼2 , −𝑐2 cos 𝛼2 ] (8)

4
ICAST2015: 26th International Conference on Adaptive Structures and
Technologies
October 14-16th, 2015, Kobe, Japan

𝐼 = [𝑌𝐼 , 𝑍𝐼 ] = 𝐺 + [𝜒2 sin 𝜙2 , 𝜒2 cos 𝜙2 ] (9)

𝐽 = [𝑌𝐽 , 𝑍𝐽 ] = 𝐵 + [𝑒1 sin 𝛼1 , 𝑒1 cos 𝛼1 ] (10)


𝐾 = [𝑌𝐾 , 𝑍𝐾 ] = 𝐸 + [𝑒2 sin 𝛼2 , 𝑒2 cos 𝛼2 ] (11)
Having defined the dimensions of the unit cell, we can now begin to choose constraints to
design a honeycomb. 𝑒1 and 𝑒2 do not affect the cell shape (these specify the location of the holes),
and 𝑑1 and 𝑑2 are fixed by the tooling geometry, so there are 12 unknowns in equations (1)-(11).
We can generate 8 additional equations by constraining the vertices to lie on two functions 𝑓1
and 𝑓2 as shown in Figure 6. The functions 𝑓1 and 𝑓2 can be chosen to produce a useful shaped
component such as an aerofoil leading edge.
4 more constraints are required to solve the system of equations; the choice of these constraints
gives some freedom of design. In this paper we fix 𝛼1 , 𝛼2 , 𝜒1 , 𝜒2 to generate honeycombs with regular
geometry.
Once the YZ coordinates of the points have been found, the X coordinates are easily calculated
using the dimensions of the mould rods. Figure 7 shows an example honeycomb in both 2D and 3D.

Figure 5. 2D model of the open honeycomb unit cell. a) View of the open honeycomb showing the
axes directions. b) Side view of the YZ plane, with unit cell highlighted. c) Unit cell represented by
points and lines in the YZ plane, showing dimensions. Parts of the adjacent cells are shown in grey.
Points J and K represent locations of holes in the cell walls.

5
ICAST2015: 26th International Conference on Adaptive Structures and
Technologies
October 14-16th, 2015, Kobe, Japan

Figure 6. The equations generated by constraining points to lie on functions 𝑓1 and 𝑓2.

Figure 7. A honeycomb fit between two functions. Functions used were 𝑓1 = 0.005𝑦 2 ,
𝑓2 = 0.005𝑦 2 + 10. The four dimensions fixed manually were 𝛼1 = 15°, 𝛼2 = 165°,
𝜒1 = 𝜒2 = 2.5 𝑙 cos 𝜃 (where 𝑙 = 5𝑚𝑚, 𝜃 = 30° as determined by the mould rods).

6
ICAST2015: 26th International Conference on Adaptive Structures and
Technologies
October 14-16th, 2015, Kobe, Japan

4 MODELLING THE DEFORMED SHAPE

In this section we demonstrate a method to model the deformation of the structure in response to
cable tension. The primary deformation mechanism of the honeycomb is flexing of the folds. Because
the folds are all aligned with the X-direction, all the motion of the honeycomb is rotations in the YZ
plane. As a result, we can again consider a 2D unit cell.
We assume that the corrugated strips remain rigid, and that the structure deforms purely by
rotating about the folds (this is a reasonable assumption because the corrugated strips are much stiffer
than the folds). Given this assumption we can idealise the structure as a series of rigid linkages
(representing the corrugated strips) joined by rotational springs (representing the folds). Figure 8
shows the idealisation of the unit cell, with forces and moments.

Figure 8. The unit cell idealised as a series of rigid linkages. The large blue line represents the
cable with tension 𝑇. The cable exerts forces and moments at points 𝐽 and 𝐾 – these are shown in blue.
Internal fold forces are black. External reaction forces are red. For clarity, lengths and angles of each
beam are not shown. They are defined as shown in the bottom insert, with the angle measured positive
counterclockwise from the Y axis.

By considering each beam element starting from 𝐺𝐼 , and working backwards through the
structure, we can find the forces and moments at the folds 𝐴, 𝐶, 𝐸, 𝐺 in terms of the external forces and
the lengths and angles of each beam. We represent these in matrix form in equation (12) (𝑖 is the cell
number, and 𝑗 = 𝐴, 𝐶, 𝐸, 𝐺).
{𝑀𝑗 }𝑖 = [𝐶𝑀 ]𝑖 {𝐹𝑒𝑥𝑡 }i (12)
[𝐶𝑀 ] is a matrix of coefficients made up of cell dimensions. Treating each fold as a linear
rotational spring with relationship 𝛾 = 𝑀⁄𝑘, the fold deflections 𝛾𝑗 are given by:

7
ICAST2015: 26th International Conference on Adaptive Structures and
Technologies
October 14-16th, 2015, Kobe, Japan

{𝛾𝑗 }𝑖 = {1⁄𝑘𝑗 }𝑖 {𝑀𝑗 }𝑖


𝛾𝐴 1⁄𝑘𝐴 𝑀𝐴
𝛾𝐶 1⁄𝑘 𝐶 𝑀𝐶 (13)
{𝛾 } = { }
𝐸 1⁄𝑘 𝐸 𝑀𝐸
𝛾𝐺 𝑖 {1⁄𝑘𝐺 }𝑖 𝑀𝐺 𝑖
We substitute (12) into (13) to obtain:
{𝛾𝑗 }𝑖 = {1⁄𝑘𝑗 }𝑖 [𝐶𝑀 ]𝑖 {𝐹𝑒𝑥𝑡 }i (14)

We now have the deformation of each fold in terms of known parameters (stiffnesses {𝑘𝑗 }, cell
geometry in [𝐶𝑀 ], and external forces {𝐹𝑒𝑥𝑡 }). We can now find the absolute values of the fold angles
measured from the vertical. We denote undeformed dimensions with a superscript “0”, and deformed

dimensions with a superscript “*”. We find the deformed fold angles {𝜃𝑗 }𝑖 = {𝛼1 , 𝜙1 , 𝛼2 , 𝜙2 }∗𝑖 by
0
adding the deformations {𝛾𝑗 } to the undeformed fold angles {𝜃𝑗 } = {𝛼1 , 𝜙1 , 𝛼2 , 𝜙2 }0𝑖 . The matrix [𝐶]
𝑖 𝑖
accounts for the fact that point 𝑗 experiences its deformation plus the deformations of the previous
points.
∗ 0
{𝜃𝑗 }𝑖 = {𝜃𝑗 }𝑖 + [𝐶]{𝛾𝑗 }
𝑖
𝛼1 ∗ 𝛼1 0
1 0 0 0 𝛾𝐴
𝜙1 𝜙1 1 1 0 0 𝛾𝐶 (15)
{𝛼 } = {𝛼 } + [ ]{ }
2 2 1 1 1 0 𝛾𝐸
𝜙2 𝑖 𝜙2 𝑖 1 1 1 1 𝛾𝐺 𝑖
This gives us the deformed fold angles in response to cable tension for one cell 𝑖. We now
expand this to multiple cells. Adjacent cells 𝑖 and 𝑖 + 1 are joined by point 𝐼𝑖 meeting point 𝐴𝑖+1 (see
the grey lines in Figure 8). We must account for moments and deformations accumulating throughout
the entire structure as well as through the cells. The fold forces and moments at point 𝐴𝑖+1 act on cell 𝑖
as the “external forces” at point 𝐼𝑖 . For the final cell in the chain, 𝑁, external forces at point 𝐼𝑁 will be
zero. Deformations are summed along the structure in a similar way to how they are summed along a
cell. Cell 𝑖 experiences its deformations plus those of previous cells. This is represented in matrix form
for 𝑁 cells:
∗ 0
{𝜃𝑗 }1 {𝜃𝑗 }1 [𝐶]{𝛾𝑗 } Σj {𝛾𝑗 }1
1 0 0 ⋯ 0
{𝜃𝑗 }2 {𝜃𝑗 }2 [𝐶]{𝛾𝑗 } 1 0 ⋯ 0 Σj {𝛾𝑗 }2
= + 2 +[ ]
⋮ ⋮ ⋮ ⋮ ⋮ ⋱ ⋮ ⋮
1 1 ⋯ 0 Σj {𝛾𝑗 }
{{𝜃𝑗 }𝑁 } {{𝜃𝑗 }𝑁 } {[𝐶]{𝛾𝑗 }𝑁 } { 𝑁} (16)
0 0 ⋯ 0
1 0 ⋯ 0
Where [ ]=𝐷
⋮ ⋮ ⋱ ⋮
1 1 ⋯ 0
We now have the deformed structure fold angles in terms of fixed geometry dimensions and
cable tension 𝑇. These can be substituted into equations (1)-(11) to find the deformed coordinates of
points 𝐴-𝐾. At this point we could input a value for 𝑇 and find the deformed geometry, but it would be
more useful to be able to specify a cable displacement 𝛿𝑐𝑎𝑏𝑙𝑒 , since the role of a morphing structure is
to assume a specific deflection. We must generate one equation so that we can specify 𝛿𝑐𝑎𝑏𝑙𝑒 and
solve for 𝑇. If we were to pull a section of cable with length 𝛿𝑐𝑎𝑏𝑙𝑒 out of the undeformed structure to
produce a deformed shape, the length of cable left inside the deformed structure would be given by:

8
ICAST2015: 26th International Conference on Adaptive Structures and
Technologies
October 14-16th, 2015, Kobe, Japan

𝐿∗𝑐𝑎𝑏𝑙𝑒 = 𝐿0𝑐𝑎𝑏𝑙𝑒 − 𝛿𝑐𝑎𝑏𝑙𝑒 (17)


Where 𝐿𝑐𝑎𝑏𝑙𝑒 is the length of cable inside the structure. We can find 𝐿𝑐𝑎𝑏𝑙𝑒 from the sum of the
distances between the various points 𝐽 and 𝐾 throughout the cells:
2:𝑁
𝐿𝑐𝑎𝑏𝑙𝑒 = (∑ |𝐾𝑖−1 𝐽𝑖 | + |𝐽𝑖 𝐾𝑖 |) + |𝐽1 𝐾1 | (18)
𝑖

We have 𝐽0 and 𝐾 0 because they are fixed by our choice of 𝑒1 and 𝑒2 , and we have 𝐽∗ and 𝐾 ∗ in
terms of 𝑇; thus we can specify a displacement 𝛿𝑐𝑎𝑏𝑙𝑒 and solve (17) for 𝑇. Figure 9 shows the
predicted displacement for a flat honeycomb compared to the displacement of a real honeycomb, and
shows that the model produces qualitatively similar results. From Figure 9 we can see differences in
the two deformed shapes – this is due to limitations of the model, which are discussed in detail in the
next section.

Figure 9. A qualitative comparison between predicted deflection (top) and actual deflection
(bottom) when subjected to a similar cable deflection. The transparent plot shows the undeformed
structure.

5 DISCUSSIONS & CONCLUSIONS

We have shown methods to design a morphing honeycomb with a functional shape, and predict
the deformation of the open honeycomb in response to cable loads. We have shown that the model
gives qualitatively sensible results.

9
ICAST2015: 26th International Conference on Adaptive Structures and
Technologies
October 14-16th, 2015, Kobe, Japan

5.1 Limitations of the model

The model does not take into account self-weight, and other factors.
These can be easily taken into account in future versions.
The model uses a very simple representation of the fold.
We based this on our observation that the honeycomb always recovers fully from all
deformations; this means that the folds are deforming elastically. However, this may not be the case
for all materials, and nonlinear stiffness behaviour could be implemented in the model in future.
The folds will repeatedly experience high local stresses which will make them susceptible to
breakage. This depends on the material and the thickness, and these will vary depending on the
application. We aim here to provide a model for the shape morphing; it is up to the designer to
characterise their chosen sheet material and investigate the behaviour of the folds.
The model assumes small displacements.
This means that it does not properly capture the effect of large displacements. The solution to
this is to break up a given displacement 𝛿𝑐𝑎𝑏𝑙𝑒 into small increments, and run the model over each of
these increments. This will then take into account the changing in angles and forces over the course of
a large deformation.
The curvature cannot be directly controlled.
Currently we can only control cable tension 𝑇 or displacement 𝛿𝑐𝑎𝑏𝑙𝑒 , and the deformed shape
depends purely on the geometry and the fold stiffnesses. The geometry is fixed by the desired
component shape, and the fold stiffnesses are fixed by choice of material. It is desirable to influence
the shape of the structure in some other way.

5.2 Potential improvements

Options for further curvature control.


Currently we are considering only a simple case with one cable. As shown in Figure 2, many
cables can be embedded in the structure. It is also feasible to put multiple knots in a cable, such that it
pulls on the structure in several places, and produces compound radii of curvature, as shown in Figure
10a.
It is desirable to limit how much the structure contracts under cable tension. By bonding the
structure to an axially stiff plate we could limit contraction and get pure bending (see Figure 10b).
If the cable inside the structure is pulled far enough, the structure will contract until the adjacent
strips touch one another; locking it in place. By controlling the length of the bridges between strips it
may be possible to tailor this collapsed configuration to a second useful shape (see Figure 10c).
Modelling the effect of a skin
A compliant outer skin could be modelled by spring elements between adjacent points on the
upper/lower surface.

6 ACKNOWLEDGMENTS

The Authors acknowledge the Engineering and Physical Sciences Research Council for
supporting this work through the Centre for Doctoral Training in Advanced Composites at the
University of Bristol, UK (Grant no. EP/G036772/1).

10
ICAST2015: 26th International Conference on Adaptive Structures and
Technologies
October 14-16th, 2015, Kobe, Japan

Figure 10. Possibilities for curvature control. a) Multiple radii of curvature produced by a cable
with multiple knots. b) An axially stiff skin limits contraction in the Y-direction. c) Bridges with
different lengths produce a curved shape when the honeycomb is fully collapsed.

7 REFERENCES

[1] T. Nojima and K. Saito, “Development of Newly Designed Ultra-Light Core Structures,” JSME Int. J.
Ser. A, vol. 49, no. 1, pp. 38–42, 2006.

[2] K. Saito, S. Pellegrino, and T. Nojima, “Manufacture of Arbitrary Cross-Section Composite Honeycomb
Cores Based on Origami Techniques,” J. Mech. Des., vol. 136, no. 5, p. 051011, Mar. 2014.

[3] R. M. Neville, A. Pirrera, and F. Scarpa, “Open shape morphing honeycombs through kirigami,” in
Proceedings of the ASME 2014 Conference on Smart Materials, Adaptive Structures and Intelligent
Systems (SMASIS2014), 2014.

[4] R. M. Neville, F. Scarpa, and A. Pirrera, “Shape morphing Kirigami mechanical metamaterials
(submitted to Scientific Reports),” 2015.

[5] Victrex PLC, “APTIV ® 2000 Series Films.”

11

View publication stats

You might also like