Professional Documents
Culture Documents
Incremental Forming of Sandwich Panels:, J.M. Allwood, M. Landert
Incremental Forming of Sandwich Panels:, J.M. Allwood, M. Landert
a r t i c l e i n f o a b s t r a c t
Article history: Applications of sandwich panels as 3D shells are limited by the high costs of tooling required
Received 28 June 2007 for conventional forming operations. This paper presents an investigation into whether
Received in revised form incremental sheet forming (ISF) would be mechanically feasible alternative means to form
24 September 2007 sandwich panels. Process feasibility is assessed through examination of failure modes, thin-
Accepted 5 November 2007 ning and surface quality after application of ISF to various sandwich panel designs. It is
shown that ISF can be applied to sandwich panels which have ductile and largely incom-
pressible cores. For a formable sandwich panel, the influence of ISF on the panel and a metal
Keywords: sheet of comparable plastic bending moment were evaluated. Similar trends in tool forces
Incremental sheet forming were observed, the sine law had similar accuracy in predicting thinning and the through-
Sandwich panels thickness deformation was similar for both materials. It is shown that 3D sandwich shells
with aluminium foam cores can be produced by using ISF to form a precursor expand-
able material and that ISF can be used to form impressions on one surface of metal foam
core sandwich panels. Failure mode maps are proposed as a future means to represent the
boundaries of safe forming regions.
© 2007 Elsevier B.V. All rights reserved.
∗
Corresponding author. Tel.: +44 1223 764628; fax: +44 1223 338076.
E-mail address: kpj21@cam.ac.uk (K.P. Jackson).
0924-0136/$ – see front matter © 2007 Elsevier B.V. All rights reserved.
doi:10.1016/j.jmatprotec.2007.11.117
j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 2 0 4 ( 2 0 0 8 ) 290–303 291
ever, a recent study by Contorno et al. (2006) showed that A detailed investigation of the accuracy of the sine law
an aluminium/aluminium foam/aluminium sandwich panel across the profile of cones of various wall angles formed by
(Alulight® ) of thickness 13.5 mm could be formed by draw SPIF was carried out by Young and Jeswiet in 2004, which
bending to depth of 13 mm without damage. Bending tests showed that the wall thickness is higher than the sine law
have given insight into the deformation and failure modes prediction over the first 10–15 mm and reduces to approxi-
of these materials, where strips have been bent until failure mately 0.05–0.2 mm thinner than the sine law prediction over
against a cylinder of specified radius, enabling failure mode the remainder of the profile. In 2005, Ambrogio et al. (2005)
maps to be plotted against dimensionless ratios of core thick- similarly found that the wall thickness across a cone of con-
ness to span length and faceplate thickness to span length. stant wall angle formed by SPIF is higher than the sine law
McCormack et al. (2001) reported failure mode maps of face prediction close to the perimeter of the sheet and approxi-
yielding, face wrinkling, core failure and indentation in three- mately 0.2 mm lower than the sine law prediction over the
point bending. Bart-Smith (2001) similarly measured failure remainder of the profile. However, in both these studies the
modes of face yielding, core shear or indentation of thin metal sine law prediction was based on the wall angle of the tool
foam core sandwich panels in three-point bending. Chen et path, not the measured wall angle of the sheet, which is likely
al. (2001) measured collapse mechanisms in four-point bend- to reduce the accuracy of the sine law prediction.
ing as core shear, indentation or face yield. A more successful FE modeling has given a more detailed insight into the
method of creating intricate 3D sandwich shells with metal deformation mechanics through the thickness of the sheet.
foam cores is to use conventional sheet metal forming meth- Sawada et al. (1999) used a 2D numerical model to demonstrate
ods to form a compacted precursor material which can later the sheet experiences bending and stretching in the plane per-
be inflated into a metal foam core sandwich panel by heating, pendicular to the tool direction. Bambach et al. (2003) used a
as explained by Banhart (2001). An example of an application full 3D FE model to numerically predict that stretching and
given by Stöbener et al. is a frontal crash energy absorber for thinning occurs in the plane perpendicular to the tool direc-
a suburban railcar (Stöbener et al., 2003). tion, although, in contrast to Sawada et al., they also predicted
that shear in this plane dominates in TPIF. In the plane parallel
to the tool direction, it was predicted that shear strains occur,
2.2. The mechanics of incremental sheet forming
increasing with decreasing forming head diameter, which is
contradictory with the plane strain assumption.
Previous research into the mechanics of ISF has focused on
The influences of process parameters on tool forces in
three approaches: experimental measurement of strains and
ISF have given further insight into the deformation mechan-
displacements of the surfaces of sheets; the use of the finite
ics. Duflou et al. (2005) found that total tool force in SPIF
element (FE) method for prediction of strains through the
increases with vertical pitch, tool diameter and wall angle.
thickness; and measurement of tool forces. This previous
Bologa et al. (2005) measured the influence of sheet thick-
research, which although insightful is sometimes limited or
ness, vertical pitch and diameter on vertical and horizontal
contradictory, is reviewed below.
force and concluded that the sheet thickness had the most
Measurements and numerical simulations of both SPIF and
significant influence on vertical force. Trends in tool force
TPIF have shown that, for a conventional spiral tool path
throughout a given forming process have also been measured.
along straight or gently curved sides, material does not move
Duflou et al. (2005) observed for SPIF and Jeswiet et al. (2005b)
significantly in the direction parallel to the plane of the unde-
observed for both SPIF and TPIF that tool force increases,
formed sheet, but moves mainly normal to this plane. Hence,
sometimes to a peak, and then stabilises to a steady-state
strains on the surface of the sheet are zero or negligible par-
value whilst forming a product with a constant wall angle.
allel to the tool direction and positive perpendicular to the
Iseki (2001) measured tool force around the first few laps
tool direction, and these directions correspond to the minor
of square shell and found that the tool force peaks in the
and major directions of surface strain, respectively. This was
corners.
first measured for SPIF of circular and elliptical frustums by
Iseki in 1993 (Iseki et al., 1993). A similar result was later mea-
2.3. Scope of present research
sured experimentally by Park and Kim (2003) for pyramids
formed by both SPIF and TPIF with a central support post.
This paper is a first report in the journal literature on ISF of
Near to uniaxial strains have subsequently been measured for
sandwich panels and builds on previous work by the authors:
a straight sided pyramid formed by TPIF by Bambach et al.
the Ph.D. thesis of Landert (2006) and a conference paper by
(2003), a truncated cone formed by SPIF from various materials
Jackson et al. (2007). Two themes are investigated in this paper.
by Fratini et al. (2004), and for various shapes formed by SPIF by
Firstly, the mechanical feasibility of ISF of various formable
Jeswiet and Young (2005). Hence it has been widely accepted
sandwich panel designs has been evaluated in order to exam-
that the deformation created by ISF is largely one of plane
ine the link between the construction of the panel and the
strain.
forces applied in ISF (Section 3). Secondly, trials are conducted
The above result brings about an approximate relationship
to examine whether existing insights into ISF mechanics in
between the wall thickness after forming (t1 ) with the wall
sheet metal transfer to sandwich panels (Section 4). Ideas
angle (˛) and the original wall thickness (t0 ) known as the sine
for future work in this area are proposed in Section 5. The
law (1).
experiments reported in this paper were performed on the
specialised ISF rig at the University of Cambridge designed by
t1 = t0 sin ˛ (1) Allwood et al. (2005) (Fig. 2).
j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 2 0 4 ( 2 0 0 8 ) 290–303 293
MS/PP/MS Sollight® (Arcelor) 2.0 1.44 Polypropylene 0.28 Mild steel Hot co-laminating
Al/PP/Al Hylite® (Corus) 1.2 0.8 Polypropylene 0.2 AA5182 Not available
SS/SS Hybrix® (Lamera) 1.2 0.8 Stainless steel 0.2 Stainless steel Epoxy resin
fibre/SS fibre
Al/Al Alulight® (Alulight) 13.5 9.5 Al alloy foam 2 AA5251-H22 Epoxy adhesive
foam/Al AlMg1Si0.6
Fig. 3 – Design of tool paths to test mechanical feasibility of ISF of sandwich panels: (a) line and (b) spiral.
Deformation due to ISF occurs by local indentation as opposed 3.1. Experimental design
to wide contact over the surface of a die, and according to
Zenkert (1995), sandwich panels are particularly susceptible Two simple tool paths were used to represent the least
to failure under this type of loading. The aim of this sec- mechanically demanding deformation that could occur in ISF
tion is therefore to evaluate the mechanical feasibility of ISF because if the sandwich panels fail under these conditions
of various sandwich panel designs. Four sandwich panels, they will also fail under the more demanding descending con-
representing the range of properties of formable sandwich tour or spiral tool paths typical of ISF. The first tool path is a
panels available, have been selected on which to evaluate straight line (Fig. 3a), representing the first deformation which
the mechanical feasibility of ISF. Two panels with metal face- occurs in any ISF process. The length of the line is 100 mm
plates and polymer cores have been selected: mild steel and tool speed is 10 mm/s. The second tool path is a spiral
faceplates with a polypropylene core (MS/PP/MS) (Sollight® in a horizontal plane with overlapping passes (Fig. 3b). This
manufactured by Arcelor), and aluminium faceplates with subjects the material to repeated passes, which is typical of
a polypropylene core (Al/PP/Al) (Hylite® manufactured by ISF, whilst maintaining a wall angle of zero. The pitch between
Corus). One sandwich panel with stainless steel faceplates successive laps is 0.1 mm, tool speed is 40 mm/s, and the outer
and a stainless steel fibre core (SS/SS fibre/SS) (Hybrix® man- diameter of the spiral is 40 mm. In both tests the tool radius
ufactured by Lamera) has been selected. The fourth sandwich is 10 mm, the size of the unsupported region of the sheet is
panel has aluminium faceplates and an aluminium foam core 140 mm × 140 mm and lubrication is Castrol Spheerol grease
Fig. 4 – Four failure modes of sandwich panels which may occur in ISF and required loading: (a) faceplate fracture, (b) core
shear failure, (c) local indentation and (d) delamination.
294 j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 2 0 4 ( 2 0 0 8 ) 290–303
Fig. 5 – Cross-sections of sandwich panels before and after deformation: (a) MS/PP/MS, (b) Al/PP/Al, (c) SS/SS fibre/SS and (d)
Al/Al foam/Al.
L-EP2. The depth of the path below the initial plane of the structural significance, but is of critical importance for appli-
sheet is 8 mm because this is sufficient to cause a localised cations such as car body panels.
plastic deformation under the indenter whilst accommodat- The only mechanical failure mode associated with sheet
ing non-localised elastic deformation over the rest of the metals in ISF is rupture, whereas 9 failure modes of sand-
sheet. wich panels have been identified by Zenkert (1995). These are:
The mechanical feasibility has been evaluated by observ- faceplate fracture, core shear failure, face wrinkling (inwards
ing three events which reduce panel performance: mechanical or outwards), buckling, shear crimpling, face dimpling and
failure, reduction in thickness and surface quality degrada- local indentation. Assuming that shear, in-plane tension and
tion. Mechanical failure implies that the faceplates and core local indentation are the only loadings that may occur in
are no longer able to co-operate so that bending stiffness ISF, Zenkert’s original list may be reduced to 3 modes most
or strength is reduced. Thickness reduction also results in a likely occur in ISF: (a) faceplate fracture, (b) core shear failure
reduction in bending stiffness, although the core and face- and (c) local indentation. In addition, delamination (d) may
plates can still co-operate. Surface quality degradation has no occur. These four failure modes are illustrated in Fig. 4 and
j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 2 0 4 ( 2 0 0 8 ) 290–303 295
Fig. 7 – Lower and upper surface of sandwich panels after spiral test: (a) MS/PP/MS, (b) Al/PP/Al, (c) SS/SS fibre/SS and (d)
Al/Al foam/Al.
296 j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 2 0 4 ( 2 0 0 8 ) 290–303
Fig. 10 – Tool path used in comparison of vertical tool forces: (a) in x–y plane and (b) cross-section A–A in y–z plane through
apex.
175 mm × 175 mm. A right-handed co-ordinate set is defined measured and is shown on the typical lap in Fig. 11b. This
as (x, y, z) such that x and y are in the plane parallel to the is plotted against tool radius and vertical pitch for each of the
undeformed sheet and z is the normal direction downwards nine experiments on MS/PP/MS and aluminium in Fig. 12a and
(parallel to the tool axis). Lubrication was Castrol Spheerol L- b, respectively. There are similarities in trends in Fss for both
EP2 extreme pressure grease and tool speed was 40 mm/s for MS/PP/MS and aluminium, and these trends are consistent
each experiment. with previously reported results for sheet metals. Fig. 12 shows
Fig. 11a shows a time plot of force for the two materi- that Fss increases approximately linearly with tool radius for
als in one of the trials (pitch = 0.5 mm, tool radius = 10 mm). both materials. This is consistent with previous results for
The shape of this curve is similar in all 9 experiments, and sheet metal in SPIF measured by Bologa et al. (2005). Both
is consistent with previously reported results for sheet met- graphs also show increasing Fss with increasing vertical pitch,
als. Firstly, for both materials, the average vertical force on which has previously been reported for sheet metals by Duflou
each lap increases over the first few laps and then stabilises et al. (2005). Although trends in Fss are similar, the magnitude
to a steady state. Similar behaviour was previously reported of Fss for MS/PP/MS was on average 50% that of aluminium
for sheet metals by Duflou et al. (2005) and Jeswiet et al. (1.03 kN for MS/PP/MS vs. 2.01 kN for aluminium).
(2005b). Secondly, across any given side of the pyramid, tool
force increases towards the corners and then decreases. This 4.2. The accuracy of the sine law
was similarly observed for sheet metal by Iseki (2001). Tool
force throughout a typical lap in the steady state is shown in The accuracy of the sine law has been evaluated for both
Fig. 11b, which shows four peaks corresponding to the four materials for the truncated pyramids formed in Section 4.1 by
corners A, B, C, D of the lap. To allow comparison between comparing the wall thickness across the cross-sections pre-
the nine trials, the average minimum vertical force on each dicted by the sine law (1) to the measured wall thickness. This
side of the pyramid during the steady-state phase (Fss ) was was found with vernier callipers to an accuracy of ±0.01 mm
298 j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 2 0 4 ( 2 0 0 8 ) 290–303
across each of the corresponding pairs of marked points. Note Fig. 12 – The influence of tool radius and vertical pitch on
that the measured wall angle rather than the wall angle of the tool force (Fss ) in ISF of a 40◦ pyramid: (a) MS/PP/MS and (b)
tool path has been used for the sine law to ensure that the aluminium.
accuracy of the results is not affected by deviation between
the profile of the finished product and the profile of the tool
path. 20 mm between the inclined surfaces, using a vertical pitch of
Comparison of measured and predicted wall thickness is 0.5 mm, tool of radius 10 mm, tool speed of 40 mm/s and Cas-
given in Fig. 13. The sine law is shown to give a prediction trol Spheerol L-EP2 grease as the lubrication. The positions of
of wall thickness which follows the measured wall thickness the marks were re-measured with the CMM after deformation
to within ±0.2 mm across the cross-sections of both materi- and removal from the clamp.
als. The average errors in both cases are ±0.06 and ±0.09 mm, Fig. 14 shows straight lines adjoining corresponding points
respectively. The accuracy of the sine law for the sandwich on the upper and lower surfaces of the sheets after ISF in 3
panel is therefore comparable to that of the monolithic sheet planes: z–y, x–y and z–x. Error bars in the x direction are drawn,
metal and previous results presented by Young and Jeswiet whilst those in the y and z directions are not because they are
(2004) and Ambrogio et al. (2005). The error is sufficiently low negligible on the scales used for these axes. The lines adjoin-
for the sine law to be a useful tool for process design of ISF of ing corresponding points on the upper and lower surfaces have
sandwich panels by prediction of the initial thickness to give been corrected for initial misalignment to the vertical. It is
the required final wall thickness. important to note that these lines do not show the profile of
the deformation through the thickness; this is approximated
4.3. Displacement of upper and lower surfaces as a straight line.
Fig. 14 shows that the deformation of MS/PP/MS and
Both surfaces of one sheet of each material were marked aluminium are similar in the z–y plane and deformation corre-
at intervals of 5 mm along a straight line through the cen- sponds to previously reported results for sheet metals. In both
tre using a scribe. The positions of the scribed marks were cases, lines adjoining the upper and lower surfaces remain
measured before deformation using a co-ordinate measuring approximately perpendicular to the sheet surface and there is
machine (CMM) to an accuracy of ±0.15 mm, established by negligible translation of the surfaces in the y direction. This
repeatability tests. The sheets were then formed into a trun- implies that stretching is the dominant mechanism in this
cated pyramid, similar to Fig. 10 but with a plateau of width plane, whilst shear and bending are less significant. This is
j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 2 0 4 ( 2 0 0 8 ) 290–303 299
Fig. 13 – Comparison of wall thickness predicted by sine law to measured wall thickness and original wall thickness: (a)
MS/PP/MS and (b) aluminium.
consistent with the results previously found experimentally can accept a larger elastic deformation before permanently
by Allwood et al. (2007) and numerically by Sawada et al. (1999) changing shape. Through-thickness shear shown in MS/PP/MS
and Bambach et al. (2003). is small enough to be within the bounds of experimental error
The measurements in the x–y and z–x planes for both mate- across most of the cross-section, and therefore further work
rials show that there is translation of both upper and lower is required to confirm this result. A further useful insight into
surfaces in the direction of tool movement, with greater trans- the above results could be gained by observing the full profile
lation of the upper surface resulting in shear between the two of through-thickness deformation. This may be particularly
surfaces. This may be attributed to friction of the tool moving insightful for the sandwich panel, in which it is possible that
across the upper surface and is consistent with the analysis the deformation of the soft and ductile polypropylene core
of through-thickness shear reported by Allwood et al. (2007). does not follow a profile that smoothly interpolates between
To some extent this result is surprising, as the polypropy- the stiffer faceplates.
lene core is less strong than the steel faces, and it might
have been anticipated that the through-thickness shear stress
might cause the core and faceplates to delaminate. In fact, the 5. Discussion, future work and conclusions
sandwich panel has survived the shearing effect, although the
through-thickness shear measured for aluminium is greater The results of Section 3 show that sandwich panels with duc-
than that for MS/PP/MS—perhaps because the polypropylene tile and largely incompressible cores can be formed by ISF,
300 j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 2 0 4 ( 2 0 0 8 ) 290–303
Fig. 14 – Deformation of upper and lower surfaces of sheets in x–y plane, z–y plane and z–x plane: (a) MS/PP/MS and (b)
aluminium.
and those of Section 4 show that the main established results ther tests on two sandwich panel materials showed that for a
about the effect of ISF on sheet metal apply in a similar manner vertical pitch of 0.5 mm with a tool radius of 10 mm the maxi-
to sandwich panels. mum wall angles (a widely used measure of ISF forming limits)
Section 4.1 demonstrated that the vertical tool force for ISF were approximately 50◦ and 40◦ on MS/PP/MS and Al/PP/Al,
of MS/PP/MS is approximately 50% that of an aluminium sheet respectively, and that the limiting failure mechanism in each
chosen to have a similar plastic bending moment. This result case was faceplate fracture (Fig. 15). This shows that the range
supports the observation in Section 4.3 that the deformation of geometries which can be formed in these materials is
of the sheet in the plane perpendicular to tool motion is largely slightly more limited than that of sheet metals, for which the
due to stretching rather than bending—the stretching stiffness limiting wall angle reported by Jeswiet et al. (2005a) for various
of the sandwich panel naturally being lower than that of the process conditions was between 40◦ (brass of initial thickness
metal. The lower tool force is advantageous in process design, 1 mm) and 78◦ (Al 3003-O initial thickness 2.1 mm). However,
as it will reduce tool wear, and allows a lighter weight machine it cannot be assumed that these are the limiting wall angles
design for equivalent precision. under all process conditions of ISF for these materials because
The work of this paper has demonstrated that ISF of sand- other failure mechanisms as illustrated in Fig. 4 may occur.
wich panels is feasible, but leaves open the question of how Failure mode maps have been used extensively to charac-
failure occurs in sandwich panels formed by this process. Fur- terise behaviour of sandwich panels: Petras and Sutcliffe (1999)
j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 2 0 4 ( 2 0 0 8 ) 290–303 301
Fig. 17 – ISF of an inflatable Al/Al foam/Al sandwich panel: (a) cross-section of precursor material before ISF and (b) an
inflated 3D sandwich shell from a precursor formed by ISF.
302 j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 2 0 4 ( 2 0 0 8 ) 290–303
Leszak, E., 1967. Apparatus and Process for Incremental Dieless Park, J.J., Kim, Y.H., 2003. Fundamental studies on the
Forming. U.S.P. Office. incremental sheet metal forming technique. Journal of
Markaki, A.E., Clyne, T.W., 2003a. Mechanics of thin ultra-light Materials Processing Technology 140, 447–453.
stainless steel sandwich sheet material Part I. Stiffness. Acta Petras, A., Sutcliffe, M.P.F., 1999. Failure mode maps for
Materialia 51, 1341–1350. honeycomb sandwich panels. Composite Structures 44,
Markaki, A.E., Clyne, T.W., 2003b. Mechanics of thin ultra-light 237–252.
stainless steel sandwich sheet material Part II. Resistance to Roux, P., 1960. Machines for Shaping Sheet Metal. U.S.P. Office.
delamination. Acta Materialia 51, 1351– Sawada, T., et al., 1999. Deformation analysis for stretch forming
1357. of sheet metal with CNC machine tools. In: Proceedings of the
McCormack, T.M., et al., 2001. Failure of sandwich beams with 6th ICTP, Advanced Technology of Plasticity.
metallic foam cores. International Journal of Solids and Somoyajulu, T.S.V., 2004. Vibration and formability characteristics
Structures 38, 4901–4920. of aluminium-polymer sandwich materials, PhD Thesis,
McKenna, L.W., Wohl, M.H., Woodbrey, J.C., 1981. New University of Michigan.
Light-Weight Materials for Vehicle Body Panels – Stöbener, K., et al., 2003. Composites based on metallic foams:
Aluminium/Nylon Laminates, vol. 800079. Society of phenomenology; production; properties and principles. In:
Automotive Engineers, Inc, pp. 506– International Conference “Advanced Metallic Materials”,
516. Smolenice, Slovakia.
Miller, W.K., 1981. Metal-Plastic Laminates for Vehicle Weight Young, D., Jeswiet, J., 2004. Wall thickness variations in
Reduction, vol. 800077. Society of Automotive Engineers, Inc, single-point incremental forming. Journal of Engineering
pp. 481–490. Manufacture 218, 1453–1459, Part B.
Mohr, D., 2005. On the role of shear strength in sandwich sheet Zenkert, D., 1995. In: E.M.A.S. Ltd. (Ed.), An Introduction to
forming. International Journal of Solids and Structures 42, Sandwich Construction. The Chameleon Press Ltd., London,
1491–1512. United Kingdom.