A New Rotation-Free Isogeometric Thin Shell Formulation and A Corresponding Continuity Constraint For Patch Boundaries

Available online at www.sciencedirect.
com
ScienceDirect
Comput. Methods Appl. Mech. Engrg. 316 (2017) 43–83

www.elsevier.com/locate/cma
A new rotation-free isogeometric thin shell formulation and a

corresponding continuity constraint for patch boundaries
Thang X. Duong, Farshad Roohbakhshan, Roger A. Sauer ∗
Aachen Institute for Advanced Study in Computational Engineering Science (AICES), RWTH Aachen University, Templergraben 55, 52056
Aachen, Germany
Available online 27 April 2016
Abstract
This paper presents a general non-linear computational formulation for rotation-free thin shells based on isogeometric finite
elements. It is a displacement-based formulation that admits general material models. The formulation allows for a wide range
of constitutive laws, including both shell models that are extracted from existing 3D continua using numerical integration and
those that are directly formulated in 2D manifold form, like the Koiter, Canham and Helfrich models. Further, a unified approach
to enforce the G 1 -continuity between patches, fix the angle between surface folds, enforce symmetry conditions and prescribe
rotational Dirichlet boundary conditions, is presented using penalty and Lagrange multiplier methods. The formulation is fully
described in the natural curvilinear coordinate system of the finite element description, which facilitates an efficient computational
implementation. It contains existing isogeometric thin shell formulations as special cases. Several classical numerical benchmark
examples are considered to demonstrate the robustness and accuracy of the proposed formulation. The presented constitutive
models, in particular the simple mixed Koiter model that does not require any thickness integration, show excellent performance,
even for large deformations.
⃝c 2016 Elsevier B.V. All rights reserved.
Keywords: Nonlinear shell theory; Kirchhoff-Love shells; Rotation-free shells; Isogeometric analysis; C 1 -continuity; Nonlinear finite element
methods
List of important symbols
1 Identity tensor in R3
a Determinant of matrix [aαβ ]
A Determinant of matrix [Aαβ ]
aα Co-variant tangent vectors of surface S at point x; α = 1, 2
Aα Co-variant tangent vectors of surface S0 at point X; α = 1, 2
aα,β Parametric derivative of aα w.r.t. ξ β
∗ Corresponding author.
E-mail address: sauer@aices.rwth-aachen.de (R.A. Sauer).
http://dx.doi.org/10.1016/j.cma.2016.04.008
0045-7825/⃝ c 2016 Elsevier B.V. All rights reserved.
44 T.X. Duong et al. / Comput. Methods Appl. Mech. Engrg. 316 (2017) 43–83
aα;β Co-variant derivative of aα w.r.t. ξ β

aαβ Co-variant metric tensor components of surface S at point x
Aαβ Co-variant metric tensor components of surface S0 at point X
a αβγ δ Contra-variant components of the derivative of a αβ w.r.t. aγ δ
b Determinant of matrix [bαβ ]
B Determinant of matrix [Bαβ ]
b Curvature tensor of surface S at point x
b0 Curvature tensor of surface S0 at point X
B Left Cauchy–Green tensor of the shell mid-surface
bαβ Co-variant curvature tensor components of surface S at point x
Bαβ Co-variant curvature tensor components of surface S0 at point X
b̃αβ Contra-variant components of the adjugate tensor of bαβ
B Matrix of the coefficients of the Bernstein polynomials for element Ω e
c Bending stiffness
cαβγ δ Contra-variant components of the derivative of τ αβ w.r.t. aγ δ
C Right Cauchy–Green tensor of the shell mid-surface
C̃ Right Cauchy–Green tensor of a 3D continuum
C∗ Right Cauchy–Green tensor of a shell layer
Ce Bézier extraction operator for finite element Ω e
d Shell director vector
d αβγ δ Contra-variant components of the derivative of τ αβ w.r.t. bγ δ
δ... Variation of ...
ϵ Penalty parameter
E Green–Lagrange strain tensor of the shell mid-surface
αβ
eαβγ δ Contra-variant components of the derivative of M0 w.r.t. aγ δ
f Prescribed surface loads
fα In-plane components of f
αβ
f αβγ δ Contra-variant components of the derivative of M0 w.r.t. bγ δ
fe• Discretized finite element force vector
F Deformation gradient of the shell mid-surface
F̃ Deformation gradient of a 3D continuum
g Determinant of matrix [gαβ ]
G Determinant of matrix [G αβ ]
gα , g3 Current tangent and normal vectors of a shell layer; α = 1, 2
Gα , G3 Reference tangent vectors and normal of a shell layer; α = 1, 2
gc G 1 -continuity and symmetry constraints
gαβ Co-variant components of the metric tensor of S ∗
G αβ Co-variant components of the metric tensor of S0∗
G ext External virtual work
G int Internal virtual work
γ
Γαβ Christoffel symbols of the second kind
H Mean curvature of surface S at x
H0 Mean curvature of surface S0 at X
i Identity tensor in S
I Identity tensor in S0
I1 First invariant of C
I˜1 First invariant of C̃
J Surface area change
J˜ Volume change of a 3D continuum
T.X. Duong et al. / Comput. Methods Appl. Mech. Engrg. 316 (2017) 43–83 45
Ja Jacobian of the mapping P → S

JA Jacobian of the mapping P → S0
ke• Finite element tangent matrices
κ Gaussian curvature of surface S at x
κ Pull-back of the curvature tensor b
K Relative curvature tensor
K Surface bulk modulus
K̃ Bulk modulus of 3D continua
L Patch boundary on which edge rotation conditions are prescribed
λi Principal stretches of surface S at x
µ Surface shear modulus
µ̃ Shear modulus of 3D continua
µ0 Moment tensor in the reference configuration
µ Moment tensor in the current configuration
αβ
M αβ , M0 Contra-variant bending moment components
n, ñ Surface normals of S at x
N, Ñ Surface normals of S0 at X
N Array of NURBS-based shape functions
N̂e Array of B-spline basis functions in terms of the Bernstein polynomials
N̂ Ae B-spline basis function of the Ath control point; A = 1, . . . , n
N αβ Total, contra-variant components of σ
ν unit normal on ∂S
ξ Out-of-plane coordinate
ξα In-plane coordinates; α = 1, 2
p External pressure
P Parametric domain spanned by ξ 1 and ξ 2
P Shell material point
ΠL Potential of the constraint used to enforce edge rotation conditions
ϕ Deformation map of surface S
q Lagrange multiplier for the continuity constraint
ρ Areal density of surface S
S Current configuration of the shell surface
S0 Reference configuration of the shell surface
S∗ Current configuration of a shell layer
S0∗ Reference configuration of a shell layer
∂S Boundary of S
S Second Piola–Kirchhoff stress tensor of the shell
S̃ Second Piola–Kirchhoff stress tensor of a 3D continuum
Sα Contra-variant, out-of-plane shear stress components
σ Cauchy stress tensor of the shell
σ̃ Cauchy stress tensor of a 3D continuum
σ αβ Stretch related, contra-variant components of σ
t Current shell thickness
T Reference shell thickness
T Traction acting on a cut ∂S normal to ν
Tα Traction acting on a cut ∂S normal to aα
τ Unit direction along a surface boundary
τ̃ Kirchhoff stress tensor of a 3D continuum
τ αβ Contra-variant components of the Kirchhoff stress tensor of the shell
τ̃ αβ In-plane components of τ̃
v Velocity, i.e. the material time derivative of x
V Space for admissible variations δx

wA NURBS weight of the Ath control point (=FE ˆ node); A = 1, . . . , n
W Strain energy density function per reference area
W̃ Strain energy density function per reference volume
x Current position of a surface point on S
X Initial position of x on the reference surface S0
x̃ Current position of a material point of a 3D continuum
X̃ Reference position of a material point of a 3D continuum
xe Array of all nodal positions for finite element Ω e
Xe Array of all nodal positions for finite element Ω0e
Ωe Current configuration of finite element e
Ω0e Reference configuration of finite element e
1. Introduction
This work presents a new rotation-free isogeometric finite element formulation for general shell structures. The
focus here is on solids, even though the formulation generally also applies to liquid shells. The formulation is
based on [1], who provide a theoretical framework for Kirchhoff–Love shells under large deformations and nonlinear
material behavior suitable for both solid and liquid shells.
From the computational point of view, among the existing shell theories, the rotation-free Kirchhoff–Love shell
theory is attractive since it only requires displacement degrees of freedom in order to describe the shell behavior. The
necessity of at least C 1 -continuity across shell elements and their boundaries is the main reason why this formulation
is not widely used in practical finite element analysis. Although various efforts have been made for imposing C 1 -
continuity on Lagrange elements (see e.g. [2–5] and references therein), the proposed computational formulations
are usually either expensive or difficult to implement. Due to this cost and complexity, finite shell elements derived
from Reissner–Mindlin theory, which require only C 0 -continuity but need additional rotational degrees of freedom,
are more widely used [6–11]. It is worth noting that there are some other formulations that are different from the
above prevailing approaches, like extended rotation-free shells including transverse shear effects [12], rotation-free
thin shells with subdivision finite elements [13–16], meshfree Kirchhoff–Love shells [17] and discontinuous Galerkin
method for Kirchhoff–Love shells [18,19].
Isogeometric analysis (IGA), initially introduced by Hughes et al. [20], has become a promising tool for the
computational modeling of shells. For instance, [21–26] study various Reissner–Mindlin shells with isogeometric
analysis. Uhm and Youn [27] introduce a Reissner–Mindlin shell described by T-splines. The hierarchic family of
isogeometric shell elements presented by Echter et al. [28] includes 3-parameter (Kirchhoff–Love), 5-parameter
(Reissner–Mindlin) and 7-parameter (three-dimensional shell) models. Solid-shell elements based on isogeometric
NURBS are investigated by Bouclier et al. [29,30], Hosseini et al. [31,32], Bouclier et al. [33], Du et al. [34]. The
shell formulation of Benson et al. [35] blends both Kirchhoff–Love and Reissner–Mindlin theories.
Particularly for rotation-free thin shells, which are the focus of this research, isogeometric analysis is a great help.
This is due to the fact that the IGA discretization can provide smoothness of any order across elements, allowing
an efficient yet accurate surface description, which is suitable for thin shell structures. The first work on combining
IGA with Kirchhoff–Love theory is presented by Kiendl et al. [36]. Later, Kiendl et al. [37] use the bending strip
method to impose the C 1 -continuity of Kirchhoff–Love shell structures comprised of multiple patches. Nguyen-Thanh
et al. [38] then propose PHT-splines for rotation-free shells. The approach is tested for a linear shell formulation and it
demonstrates the advantage of providing C 1 -continuity for thin shells. In the rotation-free shell formulation suggested
by Benson et al. [39], the Kirchhoff–Love assumptions are satisfied only at discrete points, so that the required
continuity can be lower. Nagy et al. [40] propose an isogeometric design framework for composite Kirchhoff–Love
shells with anisotropic material behavior. Goyal et al. [41] investigate the dynamics of Kirchhoff–Love shells
discretized by NURBS. Nguyen-Thanh et al. [42] propose an extended isogeometric element formulation for the
analysis of through-the-thickness cracks in thin shell structures based on Kirchhoff–Love theory. Deng et al. [43]
suggest a rotation-free shell formulation equipped with a damage model. For thin biological membranes, a thin shell
formulation is developed by Tepole et al. [44]. Riffnaller-Schiefer et al. [45] present a discretization of Kirchhoff–Love
thin shells based on a subdivision algorithm. Weak Dirichlet boundary conditions of isogeometric rotation-free thin
shells are considered by Guo and Ruess [46]. Lei et al. [47] introduce a penalty and a static condensation method to
enforce the C 0 /G 1 -continuity for NURBS-based meshes with multiple patches. Recently, isogeometric collocation
methods have been also introduced for Kirchhoff–Love and Reissner–Mindlin plates as an alternative for isogeometric
Galerkin approaches [48,49].
Recently, Kiendl et al. [50] extended the proposed formulation of Kiendl et al. [36] to non-linear material models.
By using numerical integration over the thickness of 3D continua, the extended formulation admits arbitrary nonlinear
material models. However, the formulation of Kiendl et al. [50] leads to special relations for the membrane stresses
and bending moments. In general, such a (fixed) relation is rather restricted. An example is cell membranes composed
of lipid bilayers (see e.g. [1]). In this case, the material behaves like a fluid in the in-plane direction, i.e. without
elastic resistance, while in the out-of-plane direction the material behaves like a solid with elastic resistance. Hence,
the membrane and bending response may range from fully decoupled to very complicated relations. Additionally, for
fluid materials, it is desired to include other conditions such as area incompressibility and stabilization techniques.
Therefore, a further extension of the formulation of Kiendl et al. [50] is needed.
Besides, in computation of thin shells, it is beneficial to accommodate material models that are directly constructed
in surface strain energy form, like the Koiter model [51], Canham model [52] or Helfrich model [53]. In these models,
contrary to some of the approaches mentioned above, no numerical integration is required such that the computational
time reduces drastically.
In this paper, we develop a general nonlinear IGA thin shell formulation. The surface formulation, presented in Sec-
tion 2.6, admits any (non)linear material law with arbitrary relation between bending and membrane behavior, while
the models of [36–38,28,54,26] are based on linear 2D stress–strain relationships. The 3D formulation, presented in
Section 3, can be reduced to [39,50] as special cases. However, the model of Benson et al. [39] is restricted to single
patch shells and the continuity constraint of Kiendl et al. [50] is different than our constraint. Further, our approach
to model symmetry and clamped boundaries is distinct from all the existing rotation-free IGA shell formulations. In
summary, our formulation contains the following new items:
• The bending and membrane response can be flexibly defined at the constitutive level.
• It admits both constitutive laws obtained by numerical integration over the thickness of 3D material models and
those constructed directly in surface energy form.
• It can thus be used for both solid and liquid shells.
• It is fully described in curvilinear coordinates, which includes the constitutive laws, FE weak form and
corresponding FE matrices.
• It includes a consistent treatment for the application of boundary moments.
• It includes an efficient finite element implementation of the formulation.
• It includes a unified treatment of edge rotation conditions such as the G 1 -continuity between patches, symmetry
conditions and rotational boundary conditions.
The remaining part of this paper is organized as follows: Section 2 summarizes the theory of rotation-free thin shells,
including the kinematics, balance laws, strong and weak forms of the governing equations, as well as remarks on
constitutive laws. In Section 3, we present a concise and systematic procedure to extract shell constitutive relations
from existing 3D material models. Section 4 discusses the finite element discretization as well as the treatment
of symmetry, surface folds and G 1 -continuity constraints for multi-patch NURBS. Several linear and nonlinear
benchmark tests are presented in Section 5 to illustrate the capabilities of the new model. These examples consider
solid shells. Liquid shells will be presented in future work [55]. Section 6 concludes the paper.
2. Summary of rotation-free thin shell theory
This section summarizes the theory of nonlinear shells in the unified framework presented in [1] and references
therein. Here, the shells are treated mathematically as 2D manifolds. Later, in Section 3, the kinematics and
constitutive formulations are defined in a 3D setting considering different shell layers through the thickness. These
two different approaches are schematically illustrated in Fig. 1.
Fig. 1. Mapping between parameter domain P, reference surface S0 and current surface S of a Kirchhoff–Love shell. The boundaries of the
physical shell are shown by solid red lines. A shell layer is denoted by S ∗ and S0∗ in the current and reference configuration, respectively, and
is shown by dashed blue lines. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this
article.)
2.1. Thin shell kinematics and deformation
As shown in Fig. 1, let the mid-surface S of a thin shell be described by a general mapping
x = x(ξ α ), α = 1, 2, (1)
where x denotes the surface position in 3D space and ξ α are curvilinear coordinates embedded in the surface. They can
be associated with a parameter domain P. Differential geometry allows us to determine the co-variant tangent vectors
aα = ∂x/∂ξ α on S, their dual vectors aα defined by aα · aβ = δβα , and the surface normal n = (a1 × a2 )/∥a1 × a2 ∥,
so that the metric tensor with co-variant components aαβ = aα · aβ and contra-variant components [a αβ ] = [aαβ ]−1
of the first fundamental form are defined. The relation between area elements on S and the parameter domain is
da = Ja dξ 1 dξ 2 with Ja = det aαβ .
The bases {a1 , a2 , n} and {a1 , a2 , n} define the usual identity tensor 1 in R3 as 1 = i + n ⊗ n with i = aα ⊗ aα =
a ⊗ aα . The components of the curvature tensor b = bαβ aα ⊗ aβ are given by the Gauss–Weingarten equation
α
bαβ = n · aα,β = −n,β · aα , (2)

where a comma denotes the parametric derivative aα,β = ∂aα /∂ξ β . With the definition of the Christoffel symbol
γ γ
Γαβ = aγ · aα,β , the so-called co-variant derivative is defined as aα;β := aα,β − Γαβ aγ = (n ⊗ n) aα,β . The mean and
Gaussian curvature of S can be computed from the first and second invariants of the curvature tensor b, respectively,
as
1 1 1
H := tr b = bαα = a αβ bαβ , (3)
2 2 2
and
b
κ := det b = , (4)
a
where
a = det[aαβ ], b = det[bαβ ]. (5)
The variations of the above quantities such as δaαβ , δbαβ ,δa αβ , δbαβ ,
δ H and δκ can be found e.g. in [1].
To characterize the deformation of S due to loading, the initially undeformed configuration is chosen as a reference
configuration and is denoted by S0 . It is described by the mapping X = X(ξ α ), from which we have Aα = ∂X/∂ξ α ,
Aαβ = Aα · Aβ , [Aαβ ] = [Aαβ ]−1 , Aα = Aαβ Aβ , N = (A1 × A2 )/∥A1 × A2 ∥, the initial identity tensor
I := Aα ⊗ Aα = Aα ⊗ Aα , where 1 = I + N ⊗ N, and the initial curvature tensor b0 := Bαβ Aα ⊗ Aβ .
Having the definition of aαβ and bαβ , the deformation map x = ϕ(X) is fully characterized by the two following
quantities:
1. The surface deformation gradient, which is defined as
F := aα ⊗ Aα . (6)
Accordingly, the right Cauchy–Green surface tensor, C = FT F = aαβ Aα ⊗ Aβ , its inverse C−1 = a αβ Aα ⊗ Aβ ,
and the left Cauchy–Green surface tensor, B = FFT = Aαβ aα ⊗ aβ , are defined. They have the two invariants
I1 = C : I = B : i = Aαβ aαβ and J 2 = det C = det B. Therefore, similar to the volumetric case, the surface
Green–Lagrange strain tensor can be defined as
1 1
E = E αβ Aα ⊗ Aβ := (C − I) = (aαβ − Aαβ ) Aα ⊗ Aβ , (7)
2 2
which represents the change of the metric tensor due to the surface deformation.
2. The symmetric relative curvature tensor, which is defined as
K = K αβ Aα ⊗ Aβ := κ − b0 = (bαβ − Bαβ ) Aα ⊗ Aβ , (8)

with
κ := FT b F = bαβ Aα ⊗ Aβ , (9)
which furnishes the change of the curvature tensor field, following the terminology of Steigmann [56].
2.2. Stress and moment tensors
In order to define stresses at a material point x of the shell surface S, the shell is cut into two at x by the parametrized
curve C(s).1 Further, let τ := ∂x/∂s be the unit tangent and ν := τ × n be the unit normal of C at x. Then, one can
define the traction vector T = T α να and the moment vector M = Mα να on each side of C(s) at x, where να = ν · aα .
According to Cauchy’s theorem, these vectors can be linearly mapped to the normal ν by second order tensors as
T := σ T ν,
(10)
M := µT ν,
where
σ := N αβ aα ⊗ aβ + S α aα ⊗ n,
(11)
µ := −M αβ aα ⊗ aβ
are the Cauchy stress and moment tensors, respectively.
It should be noted that the Cauchy stress σ in Eq. (11)(1) is generally not symmetric. Furthermore, as shown
e.g. in [1], a suitable work conjugation (per reference area) arising in the presented theory is the quartet
αβ
Ẇint := τ αβ ȧαβ /2 + M0 ḃαβ , (12)
where Ẇint is the local power density and we have defined
τ αβ := J σ αβ ,
αβ (13)
M0 := J M αβ .
Further,
σ αβ := N αβ − bγβ M γ α (14)
are the components of the membrane stress responding to the change of the metric tensor.2
1 s denotes the arc length such that ds = ∥dx∥.

2 Note that σ αβ ̸= aα · σ aβ , instead aα · σ aβ = N αβ .
We also note that the mathematical quantities introduced in Eqs. (10) and (11) stem from the Cosserat theory for
thin shells [56]. As shown e.g. by Steigmann [57] and Sauer and Duong [1], these quantities can be related to the
effective traction t and physical moment m transmitted across C(s). Namely for m, the relation is given by
m := n × M = m ν ν + m τ τ , (15)
where
m ν := M αβ να τβ ,
(16)
m τ := −M αβ να νβ ,
are the normal and tangential bending moment components along C(s). For the effective traction t, one can show
that [56,1]
t := T − (m ν n)′ , (17)
where (•)′ denotes the derivative with respect to s.
2.3. Balance laws
To derive the governing equations, the surface S is subjected to the prescribed body force f = f α aα + p n on S
and the boundary conditions
u = ū on ∂u S,
n = n̄ on ∂n S,
(18)
t = t̄ on ∂t S,
m τ = m̄ τ on ∂m S,
where ū is a prescribed displacement; n̄ is a prescribed rotation, t̄ := t¯α aα is a prescribed boundary traction and m̄ τ
is a prescribed bending moment. The equilibrium of the shell is then governed by the balance of linear momentum
together with mass conservation, which gives
α
T;α + f = ρ v̇ ∀x ∈ S, (19)
while the balance of angular momentum leads to the condition that σ αβ , defined in Eq. (14), is symmetric and the
shear stress is related to the bending moment via
βα
S α = −M ;β . (20)
We note that M αβ is also symmetric but N αβ is generally not symmetric.
2.4. Weak form
The weak form of Eq. (19) can be obtained by contracting Eq. (19) with the admissible variation δx ∈ V and
integrating over S, (see [1]). This results in
G in + G int − G ext = 0 ∀δx ∈ V, (21)

where

G in = δx · ρ0 v̇ dA,
S0   
1 αβ αβ
G int = δaαβ τ αβ dA + δbαβ M0 dA = δ E αβ τ αβ dA + δ K αβ M0 dA, (22)
2
S0  S0  S0 S0
δx · f da + δx · t ds + δn · m τ ν ds + [δx · m ν n .

G ext =
S ∂t S ∂m S
Remark 1. Here, the last term in G ext is the virtual work of the point load m ν n at corners of ∂m S (in case m ν =
̸ 0),
and the second last term in G ext denotes the virtual work of the moment m τ ν. It is worth noting that this is the
consistent way to apply bending moments on the boundary of Kirchhoff–Love shells.
2.5. Linearization of the weak form
For solving the nonlinear equation (21) by the Newton–Raphson method, its linearization is needed. For the kinetic
virtual work, one gets

∆G in = δx · ρ0 ∆v̇ dA, (23)
S0
in which ∆v̇ depends on the time integration scheme used. For dead loading of f , t and m τ ν, one finds3

m τ δaα · ν β n ⊗ aα + ν α aβ ⊗ n ∆aβ ds
 
∆G ext = (24)
∂S
and for the internal virtual work term
 
1 1 1 1
∆G int = cαβγ δ δaαβ ∆aγ δ + d αβγ δ δaαβ ∆bγ δ + τ αβ ∆δaαβ
S0 2 2 2 2
1 αβ

+ eαβγ δ δbαβ ∆aγ δ + f αβγ δ δbαβ ∆bγ δ + M0 ∆δbαβ dA, (25)
2
where the material tangent matrices are defined as
∂τ αβ ∂τ αβ
cαβγ δ := 2 , d αβγ δ := ,
∂aγ δ ∂bγ δ
αβ αβ (26)
αβγ δ ∂ M0 αβγ δ ∂ M0
e := 2 , f := .
∂aγ δ ∂bγ δ
They are given in [1] for various material models. Here it is noted that d αβγ δ = eγ δαβ possess minor symmetries,
while cαβγ δ and f αβγ δ possess both minor and major symmetries.
2.6. Constitution
In this framework, σ αβ and M αβ are determined from constitutive relations for stretching and bending. For
hyperelastic shells, we assume there exists a stored energy function in the form
W = W (E, K), (27)
where E and K are defined in Eqs. (7) and (8). The Koiter model (see Eq. (35)) is a simple example of this form. It
is easily seen that W can be equivalently expressed as a function of C and κ, or as a function of aαβ and bαβ , since
Aα is constant. If the material has symmetries, e.g. isotropy, the stored energy function can also be expressed as a
function of the invariants I1 , J , H and κ, defined in Section 2.1 (see e.g. [57]). Thus, the following functional forms
are equivalent to Eq. (27).
W = W (C, κ) = W (aαβ , bαβ ) = W (I1 , J, H, κ). (28)
Given the stored energy function W , the common argument by Coleman and Noll [58] leads to the constitutive
equations (see [1])
∂W
τ αβ = 2 ,
∂aαβ
∂W (29)
αβ
M0 = .
∂bαβ
3 For live m ν, see Eqs. (127) and (128).

τ
Remark 2. The constitutive equations (29) are defined in curvilinear coordinates. Therefore, they give the
components of the classical second Piola–Kirchhoff stress and moment
∂ W (E, K) ∂W
S := =2 Aα ⊗ Aβ =: τ αβ Aα ⊗ Aβ ,
∂E ∂aαβ
(30)
∂ W (E, K) ∂W αβ
µ0 := − =− Aα ⊗ Aβ =: −M0 Aα ⊗ Aβ ,
∂K ∂bαβ
of the surface manifold. Alternatively, they give the components of the Kirchhoff stress and moment defined by
pushing forward S and µ0 as
F S FT := τ αβ aα ⊗ aβ ,
αβ (31)
F µ0 FT := −M0 aα ⊗ aβ = J µ.
Here, we note that ∂(•)

∂X in Eq. (30) denotes the derivative of a scalar-valued function by an arbitrary second order
tensor X (see e.g. [59]).
Remark 3. In this framework, the stress σ αβ and moment M αβ are both constitutively determined from a surface
stored energy function W . The advantage of this setup is that it can accept general constitutive relations, which
implies that the relation between the membrane and bending response can be flexibly realized at the constitutive level
and it is not restricted within the formulation. The stretching and bending response may range from fully decoupled,
like in the Koiter material model (Eq. (35)), to a coupled relation, such as in the Helfrich material model (see e.g. [1]).
Due to this flexibility, the presented formulation is suitable for both solid and liquid shells and is also convenient for
adding kinematic constraints (e.g. area constraint) or stabilization potentials [55].
Remark 4. In the presented formulation, we note that the definition of the shell thickness is not needed for solving
Eq. (21), which is often referred to as a “zero-thickness” formulation. Consequently, the stress and moment in Eq. (29)
can be computed without defining a thickness. However, this does not imply that thickness effects have been neglected
or approximated. Instead, they are in some sense hidden in the constitutive law of Eq. (27). If desired, they can be
determined as noted in Remark 5. But this connection is not a requirement as it is in the case of constitutive laws
derived from 3D material models (see Section 3.4).
In the following, we consider some example constitutive models suitable for solid shells to demonstrate the flexibility
of the formulation. The application of the formulation to liquid shells is considered in future work [55].
2.6.1. Initially planar shells

For initially planar shells, one can consider the model of Canham [52] for the bending contribution, while for the
membrane contribution, the stretching response can be modeled with a nonlinear Neo-Hookean law. This gives
Λ 2 µ
(J − 1 − 2 ln J ) + (I1 − 2 − 2 ln J ) + c J 2 H 2 − κ ,
 
W = (32)
4 2
where c, Λ and µ are 2D material constants. From Eqs. (29) and (32), we find the stress
Λ 2  
τ αβ = J − 1 a αβ + µ Aαβ − a αβ + c J (2H 2 + κ) a αβ − 4c J H bαβ ,

(33)
2
and the moment
αβ
M0 = c J bαβ , (34)
which is linear w.r.t. the curvature. The tangents for this are given in [1]. We note that since Eq. (32) is given in
surface strain energy form, the stress and moment can be computed directly from Eqs. (33) and (34) without needing
any thickness integration.
2.6.2. Initially curved solid shell

For initially curved shells, the surface strain energy model proposed by Koiter [51,60] can be considered. It is
defined in tensor notation as
1 1
W (E, K) = E : C : E + K : F : K, (35)
2 2
with the constant fourth order tensors
C = Λ I ⊙ I + 2 µ (I ⊗ I)s ,
T2 (36)
F = C.
12
Since I ⊙ I : X = (tr X) I and (I ⊗ I)s : X = sym(X), with X being an arbitrary second order tensor, it follows from
Eqs. (35) and (36) that
τ αβ = Λ tr E Aαβ + 2 µ E αβ ,
αβ T2 (37)
M0 = (Λ tr K Aαβ + 2 µ K αβ ),
12
where K αβ := Aαγ K γ δ Aδβ , E αβ := Aαγ E γ δ Aδβ , and
tr K = K αβ Aαβ , tr E = E αβ Aαβ . (38)
From Eq. (37), we further find d αβγ δ = eαβγ δ = 0 and
cαβγ δ = Λ Aαβ Aγ δ + µ Aαγ Aβδ + Aαδ Aβγ ,

 
T 2 αβγ δ (39)
f αβγ δ = c .
12
Remark 5. The set of parameters Λ and µ can be determined in different ways. Firstly, they can be determined directly
from experiments, i.e. without explicitly considering the thickness. Secondly, they may be obtained by analytical
integration over the thickness of the simple 3D Saint Venant–Kirchhoff model (see e.g. [51] and Section 3.4). In this
case, µ and Λ are given by
2 Λ̃ µ̃
Λ := T , µ := T µ̃, (40)
Λ̃ + 2 µ̃
where Λ̃ and µ̃ are the classical 3D Lamé constants in linear elasticity. Thirdly, they can be determined by numerical
integration over the thickness of a general 3D material model.
Remark 6. Note that, in Eq. (35), the stretching and the bending behavior can be specified separately. For instance,
the first term can be replaced by a nonlinear Neo-Hooke model (see Eq. (32)), i.e.
Λ 2 µ 1
W = (J − 1 − 2 ln J ) + (I1 − 2 − 2 ln J ) + K : F : K, (41)
4 2 2
αβ
in order to capture large membrane strains. In this case τ αβ follows from the front part of Eq. (33), while M0 is given
by Eq. (37)(2).
3. Shell constitution derived from 3D constitutive laws
Provided the displacement across the shell thickness complies with the Kirchhoff–Love assumption, a constitutive
relation W = W (E, K) for shells can also be extracted, without approximation, from classical three-dimensional
constitutive models of the form W̃ = W̃ (C̃), where C̃ is the right Cauchy–Green tensor for 3D continua. This
procedure corresponds to a projection of 3D models onto surface S, or to an extraction of 2D surface models out
of 3D ones. This approach goes back to [61–64,50], and it is presented here to show how it relates to our formulation.
3.1. Extraction procedure
As shown in Fig. 1, a shell material point P can be described w.r.t. the mid-surface in the reference configuration
as [11]
X̃(ξ α , ξ0 ) = X(ξ α ) + ξ0 N(ξ α ), (42)

and in the current configuration as
x̃(ξ α , ξ ) = x(ξ α ) + ξ d(ξ α ), (43)

where ξ0 ∈ [−T /2, T /2] is the thickness coordinate of the shell and d is the director vector which has three unknown
components in general. For Kirchhoff–Love theory, which is considered here, d := n and ξ = λ3 ξ0 , where λ3 denotes
the stretch in the normal direction. Thus, the tangent vectors at P are expressed w.r.t. the basis formed by the tangent
vectors on the mid-surface as
gα := x̃,α = aα − ξ bαγ aγ ,
g3 := x̃,ξ = d,
(44)
Gα := X̃,α = Aα − ξ0 Bαγ Aγ ,
G3 := X̃,ξ0 = N = G3 ,
and the metric tensors at P can also be expressed in terms of the metric tensors on the mid-surface as
gαβ := gα · gβ = ga aαβ + gb bαβ ,

g αβ := gα · gβ = g a a αβ + g b bαβ ,
(45)
G αβ := Gα · Gβ = G A Aαβ + G B Bαβ ,
G αβ := Gα · Gβ = G A Aαβ + G B B αβ ,
where gα = g αβ gβ , [g αβ ] = [gαβ ]−1 and likewise Gα . Due to Eq. (44), the coefficients in Eq. (45) are found to be
ga := 1 − ξ 2 κ, gb := −2 ξ + 2 H ξ 2 ,
(46)
G A := 1 − ξ02 κ0 , G B := −2 ξ0 + 2 H0 ξ02 ,
and
g a := s −2 (ga + 2 H gb ), g b := −s −2 gb ,
(47)
G A := s0−2 (G A + 2 H0 G B ), G B := −s0−2 G B .
Here,
g/a = 1 − 2 H ξ + κ ξ 2 , G/A = 1 − 2 H0 ξ0 + κ0 ξ02 ,

 
s := s0 := (48)
denote the so-called shifters, which are the determinants of the shifting tensors i − ξ b and I − ξ0 B, respectively, and
we have defined
G := det[G αβ ], A := det[Aαβ ], B := det[Bαβ ], g := det[gαβ ]. (49)
From Eqs. (46), (47), and (48) we thus find the variations (considering ξ fixed)
δga = ξ 2 κ a γ δ δaγ δ − ξ 2 b̃γ δ δbγ δ ,

(50)
δgb = −ξ 2 bγ δ δaγ δ + ξ 2 a γ δ δbγ δ ,
where b̃αβ := 2 H a αβ − bαβ . With these we get

δgαβ = ga δaαβ + gb δbαβ + aαβ δga + bαβ δgb . (51)
Further, we can represent the right 3D Cauchy–Green tensor w.r.t. the basis {G1 , G2 , N} as (e.g. see [11])
C̃ = C̃αβ Gα ⊗ Gβ + C̃α3 Gα ⊗ N + N ⊗ Gα + C̃33 N ⊗ N.

 
(52)
We note here that Gα accounts for the surface stretch due to the initial shell curvature (see Eq. (44)(3)) and the basis
{G1 , G2 , N} on a shell layer at ξ defines the usual identity tensor in R3 as
1 = I + N ⊗ N, I := Gα ⊗ Gα = Aα ⊗ Aα . (53)
Further, the 3D Kirchhoff stress tensor can be written as
τ̃ = τ̃ αβ gα ⊗ gβ + τ̃ α3 gα ⊗ g3 + g3 ⊗ gα + τ̃ 33 g3 ⊗ g3 .
 
(54)
For the Kirchhoff–Love shell we have C̃αβ = gαβ , while C̃α3 = gα3 = gα · g3 = 0 and C̃33 = λ23 . Thus the variation
of C̃ is
δ C̃ = δC∗ + 2 λ3 δλ3 N ⊗ N, C∗ := gαβ Gα ⊗ Gβ . (55)
With this, the total strain energy of the shell can be expressed w.r.t. the mid-surface as
   T   T
2 2
W= W̃ dV = W̃ dξ0 dG = s0 W̃ dξ0 dA, (56)
Ω0 S0 − T2 S0 − T2
where dG is the infinitesimal area element at layer ξ along the thickness, and dA is the infinitesimal area element on
the mid-surface. Thus, the surface energy of the shell (per unit area) follows from the thickness integration
 T
2
W = s0 W̃ dξ0 = W (aαβ , bαβ ), (57)
− T2
and its variation reads

 T  T  
2 2 1 αβ
δW = s0 δ W̃ dξ0 = s0 τ̃ δgαβ + τ̃ 33 λ3 δλ3 dξ0 , (58)
− T2 − T2 2
where
∂ W̃ 1 ∂ W̃
τ̃ αβ := 2 ; τ̃ 33 := . (59)
∂gαβ λ3 ∂λ3
Here, similar to Remark 2, τ̃ αβ are the components of either the second Piola–Kirchhoff stress tensor S̃ or the
Kirchhoff stress τ̃ , as
τ̃ αβ = gα · τ̃ gβ = Gα · S̃ Gβ . (60)
For shells, since it is common to assume a plane stress state, we have the condition
1 ∂ W̃
τ̃ 33 = = 0. (61)
λ3 ∂λ3
Substituting Eqs. (51) and (61) into Eq. (58) we get
 T
1 2  
δW = s0 ga τ̃ αβ + εa ξ 2 κ a αβ − εb ξ 2 bαβ dξ0 δaαβ ,
2 − T2
 T
1 2  
+ s0 gb τ̃ αβ − εa ξ 2 b̃αβ + εb ξ 2 a αβ dξ0 δbαβ , (62)
2 − T2
with εa := τ̃ αβ aαβ and εb := τ̃ αβ bαβ . Since

∂W ∂W
δW = δaαβ + δbαβ , (63)
∂aαβ ∂bαβ
we find
T
∂W

2
 
τ αβ = 2 = s0 ga τ̃ αβ + εa ξ 2 κ a αβ − εb ξ 2 bαβ dξ0 ,
∂aαβ − T2
T (64)
∂W

αβ 1 2

αβ 2 αβ αβ

M0 = = s0 gb τ̃ − εa ξ b̃ + εb ξ a 2
dξ0 .
∂bαβ 2 − T2
It should be noted here that s0 , ga , gb , εa and εb are all functions of ξ0 .
Remark 7. So far, Eq. (64) is exact, since we have not made any approximations apart from the Kirchhoff–Love
hypothesis, i.e. d = n, and the plane stress assumption. For a curved thin shell (T ≪ R, where T is the thickness and
R is the radius of curvature) one may approximate the variations δga ≈ 0 and δgb ≈ 0 in Eq. (51). In this case the
two rear terms in Eq. (64) vanish so that
 T
2
τ αβ = s0 (1 − ξ 2 κ) τ̃ αβ dξ0 ,
− T2
 T (65)
αβ 2
αβ
M0 = s0 (−ξ + H ξ ) τ̃ 2
dξ0 .
− T2
This case will be examined in the examples in Section 5. If only the leading terms in front of τ̃ αβ in Eq. (65) are
considered, it is further simplified into
 T
2
αβ
τ = τ̃ αβ dξ0 ,
− T2
 T (66)
αβ 2
M0 =− ξ τ̃ αβ dξ0 .
− T2
These expressions are commonly used in shell and plate formulations, see e.g. [50,38]. Here the sign convention for
αβ
M0 follows [1,56].
Remark 8. The stretch through the thickness, λ3 , can be determined in various ways. It can be included in the
formulation as a separate degree of freedom. Another common and straightforward approach is to derive it from
the plane stress condition (61). The resulting equation is usually nonlinear and can be solved numerically (see e.g.
[61–63,50]), or analytically for some special cases (e.g. see Section 3.2 for a Neo-Hookean model). In particular, if
incompressible models are used, i.e. J˜ = 1, then λ3 can be determined analytically (e.g. see Section 3.3). Sometimes
also λ3 ≈ 1 is assumed. In this case, condition (61) is generally not satisfied anymore. Instead τ 33 should be treated
as the Lagrange multiplier to the constraint λ3 = 1.
Remark 9. For an efficient implementation, we consider that τ̃ αβ is expressible in the form

τ̃ αβ = τ̃g g αβ + τ̃G G αβ . (67)
Substituting this into Eq. (64), and taking into account Eq. (45), we get the surface stress and moment written in the
form
τ αβ = τa a αβ + τb bαβ + τ A Aαβ + τ B B αβ ,
αβ (68)
M0 = Ma0 a αβ + Mb0 bαβ + M A0 Aαβ + M B0 B αβ ,
so that only the (scalar) coefficients need to be computed by thickness integration as
 T  T
2 2
 
τa = s0 ga τ̃g g + εa ξ κ dξ0 , τ A =
a 2
s0 ga τ̃G G A dξ0 ,
− T2 − T2
 T  T (69)
2 2
 
τb = s0 ga τ̃g g − εb ξ b 2
dξ0 , τB = s0 ga τ̃G G dξ0 ,B
− T2 − T2
and
 T  T
1 2
  1 2
Ma0 = s0 gb τ̃g g + (εb − 2 H εa ) ξ
a 2
dξ0 , M A0 = s0 gb τ̃G G A dξ0 ,
2 − T2 2 − T2
 T  T
(70)
1 2
  1 2
Mb0 = s0 gb τ̃g g + εa ξ
b 2
dξ0 , M B0 = s0 gb τ̃G G dξ0 .
B
2 − T2 2 − T2
The linearization then follows the representation of Eq. (141) with the scalar coefficients computed by integration
over the thickness.
Remark 10. We note that since the integrands in Eq. (64) and even in its reduced forms (65), (66), are rather complex
w.r.t. ξ , we need numerical integration in general. Analytical integration may only be possible for special cases,
such as the Saint Venant–Kirchhoff constitutive model discussed in Section 3.4. Thus, it is obvious that the models
constructed in 2D manifold form, such as the Koiter (Section 2.6.2), Canham (Section 2.6.1) and Helfrich constitutive
models (see e.g. [55]), are much more efficient than models requiring numerical thickness integration.
In the following, we will present the extraction example for Neo-Hookean materials to demonstrate the procedure. We
will also show that by analytical integration through the thickness of the 3D Saint Venant–Kirchhoff strain energy, the
Koiter model is recovered.
3.2. Extraction example: compressible Neo-Hookean materials
The classical 3D Neo-Hookean strain-energy per reference volume is considered in the form
Λ̃ µ̃
W̃ ( I˜1 , J˜) = ( J˜2 − 1 − 2 ln J˜) + ( I˜1 − 3 − 2 ln J˜), (71)
4 2
where I˜1 and J˜ are the invariants of the 3D Cauchy–Green tensor C̃. Due to Eq. (52), they can also be expressed in
terms of the kinematic quantities on the mid-surface, i.e.

I˜1 := C̃ : 1 = I1∗ + λ23 , J˜ := det C̃ = J ∗ λ3 , (72)
where
√

αβ g
I1∗ ∗
:= C : I = gαβ G , ∗
J := det C∗ = (73)
G
denote the invariants of C∗ defined in Eq. (55). Therefore, their variations are
δ I˜1 = δ I1∗ + 2 λ3 δλ3 , δ I1∗ = G αβ δgαβ , (74)

and
J ∗ αβ
δ J˜ = λ3 δ J ∗ + J ∗ δλ3 , δJ∗ = g δgαβ . (75)
2
λ3 can be determined using the plane stress condition Eq. (61),
∂ W̃
   
Λ̃ 1 1
= J ∗2 λ3 − + µ̃ λ3 − = 0, (76)
∂λ3 2 λ3 λ3
which is solvable analytically w.r.t. λ3 as
Λ̃ + 2 µ̃
λ23 = . (77)
Λ̃ J ∗2 + 2 µ̃
From Eq. (59)(1) we then find
Λ̃ + 2 µ̃
τ̃ αβ = µ̃ G αβ − µ̃ g αβ . (78)
Λ̃ J ∗2 + 2 µ̃
The stress in Eq (78) is then substituted into Eqs. (64), (65) or (66). Numerical integration is still required to evaluate
those. The corresponding linearized quantities and tangent matrices are provided in Appendix C.
3.3. Extraction example: incompressible Neo-Hookean materials
The Neo-Hookean strain-energy per reference volume in this case is defined as

µ̃ ˜
W̃ ( I˜1 , J˜, p) =
( I1 − 3) + p (1 − J˜), (79)
2
where p is the Lagrange multiplier associated with the volume constraint. From Eq. (79), we thus find
∂ W̃ (gαβ , λ3 , p)
τ̃ αβ = 2 = µ̃G αβ − p J˜ g αβ . (80)
∂gαβ
Further, the plane stress condition implies ∂ W̃ /∂λ3 = 0 and the incompressibility constraint requires λ3 := 1/J ∗ ,
which together allow us to determine the Lagrange multiplier analytically as
µ̃
p= . (81)
J ∗2
Substituting this into Eq. (80) we obtain
 
αβ αβ 1 αβ
τ̃ = µ̃ G − ∗2 g , (82)
J
for Eqs. (64) or (65), (66). Numerical integration is also required here.
3.4. Extraction example: Saint Venant–Kirchhoff model
We now consider the Saint Venant–Kirchhoff model,

1
W̃ (C̃) := Ẽ : C̃ : Ẽ, C̃ := Λ̃ 1 ⊙ 1 + 2 µ̃ (1 ⊗ 1)s . (83)
2
In this case, the Koiter model of Eq. (35) can be recovered by using Eq. (57) and Eq. (61) together with the following
ingredients:
• s0 ≈ 1, ξ ≈ ξ0 , and only the leading terms w.r.t. ξ in Eq. (46) and Eq. (47) are taken into account, so that Eq. (83)
reduces to
1
W̃ (C∗ ) := E∗ : C∗ : E∗ , C∗ := Λ̃ I ⊙ I + 2 µ̃ (I ⊗ I)s , (84)
2
where
E∗ := E + ξ K. (85)
• The bending and stretching response is considered fully decoupled, so that all the mixed products, such as tr E tr K
and tr (EK), are disregarded in the strain energy function. Eq. (84) then further reduces to
Λ̃ Λ̃
W̃ (E, K, ξ ) = (tr E)2 + µ̃ tr (E2 ) + ξ 2 (tr K)2 + ξ 2 µ̃ tr (K 2 ). (86)
2 2
Integrating the above energy over the thickness [−T /2, T /2] gives
T T3 T3
W (E, K, T ) = Λ̃ (tr E)2 + T µ̃ tr (E2 ) + Λ̃ (tr K)2 + µ̃ tr (K 2 ). (87)
2 24 12
• Finally, the plane stress assumption is used throughout the thickness. Accordingly, the 3D Lamé constant Λ̃ in
Eq. (87) is replaced by its plane-stress counterpart (2 Λ̃ µ̃)/(Λ̃ + 2 µ̃) (see e.g. [65]), which results in the Koiter
model of Eqs. (35) and (40).
Alternatively [60], the Koiter model can also be derived systematically as the leading-order model from the Taylor
expansion of W with the aid of Leibniz’ rule for small thickness.
4. FE discretization
In this section, we first discuss the discretization of the weak form (21) and its linearization on the basis of IGA
to obtain FE forces and tangent matrices. We will then discuss the continuity constraint between patches, patch folds,
symmetry and rotational Dirichlet boundary conditions.
4.1. C 1 -continuous discretization
In order to solve Eq. (21) by the finite element method, the surface S is discretized using the isogeometric analysis
technique proposed by Hughes et al. [20]. Thanks to the Bézier extraction operator Ce introduced by Borden et al. [66],
the usual finite element structure is recovered for NURBS basis function {N A }nA=1 , where n is the number of control
points defining an element e. Namely,
w A N̂ Ae (ξ, η)
N A (ξ, η) = n , (88)
w A N̂ Ae (ξ, η)

A=1
where { N̂ Ae }nA=1 is the B-spline basis function expressed in terms of Bernstein polynomials as
N̂e (ξ, η) = Ceξ B(ξ ) ⊗ Ceη B(η), (89)
with N̂ Ae to be the corresponding entries of matrix N̂e . For T-splines, the construction of the isoparametric element
based on the Bézier extraction operator can be found in [67].
4.2. FE approximation
The geometry within an undeformed element Ω0e and deformed element Ω e is interpolated from the positions of
control points Xe and xe , respectively, as
X = N Xe , x = N xe , (90)
where N(ξ ) := [N1 1, N2 1, . . . , Nn 1] is defined based on the NURBS shape functions of Eq. (88). It follows that
δx = N δxe ,
aα = N,α xe ,
δaα = N,α δxe , (91)
aα,β = N,αβ xe ,
aα;β = Ñ;αβ xe ,
where N,α (ξ ) := [N1,α 1, N2,α 1, . . . , Nn,α 1], N,αβ (ξ ) := [N1,αβ 1, N2,αβ 1, . . . , Nn,αβ 1], N A,α = ∂ N A /∂ξ α ,
N A,αβ = ∂ 2 N A /(∂ξ α ∂ξ β ) (A = 1, . . . , n) and4
γ
Ñ;αβ := N,αβ − Γαβ N,γ . (92)
Inserting those expressions into the formulas for δaαβ and δbαβ [1] gives
δaαβ = δxTe NT,α N,β + NT,β N,α xe ,

 
(93)
4 A tilde is included in Ñ
;αβ to distinguish it from N,αβ .
and
δbαβ = δxTe ÑT;αβ n. (94)
4.3. FE force vectors
Substituting Eqs. (93) and (94) into Eq. (21) gives

n el

G ein + G eint − G eext = 0
 
∀δxe ∈ V, (95)
e=1
where n el is the number of elements and


G in = δxe fin , with fin :=
e T e e
NT ρ0 v̇ dA. (96)
Ω0e
Similarly,
G eint = δxTe feintτ + feintM ,

 
(97)
αβ
with the internal FE force vectors due to the membrane stress τ αβ and the bending moment M0

feintτ := τ αβ NT,α aβ dA (98)
Ω0e
and

αβ
feintM := M0 ÑT;αβ n dA. (99)
Ω0e
Here the symmetry of τ αβ has been exploited. The external virtual work follows as [68,1]
G eext = δxTe feext0 + feext p + feextt + feextm ,

 
(100)
where the external FE force vectors are

feext0 := NT f0 dA,
Ω0e

feext p := NT p n da,
Ωe
 (101)
feextt := NT t ds,
∂t Ω e

feextm := − NT,α ν α m τ n ds.
∂m Ωe
Here f0 is a constant surface force and p is an external pressure acting always normal to S [68]. t is the effective
boundary traction of Eq. (17) and both ν α and m τ are defined in Section 2.2. We note that, in the following sections,
the inertia term is neglected, i.e. ρ0 v̇ = 0. The corresponding tangent matrices can be found in Appendix A.
4.4. Edge rotation conditions
This section presents a general approach to describe different rotation conditions. These conditions are required
due to the fact that the presented formulation has only displacement degrees of freedom as unknowns. As shown in
Fig. 2, the G 1 -continuity constraint is required for multi-patch NURBS in order to transfer moments. Additionally,
Fig. 2. Edge rotation conditions: (a) G 1 -continuity constraint, (b) fixed surface folds (e.g. V-shapes and L-shapes), (c) symmetry (or clamping)
constraints, (d) symmetry constraint at a kink, (e) rotational Dirichlet boundary condition. Surface edge L, shown by a filled circle, is perpendicular
to the plane and parallel to the inward pointing unit direction τ .
other rotation conditions may be needed such as fixed surface fold constraints, symmetry (or clamping) constraints,
symmetry constraints at a kink and rotational Dirichlet boundary conditions.
There are various methods to enforce the continuity between patches, such as using T-Splines [69], the bending strip
method [37] and the Mortar method for non-conforming patches [70]. Alternatively, Nitsche’s method as described
in [71,54] can also be used. Recently, Lei et al. [47] applied static condensation and a penalty method to enforce
G 1 -continuity of adjacent patches based on subdivision algorithms.
Here, we introduce a general approach that can systematically enforce different edge rotation conditions, including
G 1 -continuity of adjacent patches, within the framework of a curvilinear coordinate system. The presented approach
has new features compared to existing formulations. For instance, our approach is conceptually and technically
different from the method of Lei et al. [47]. First and foremost, the presented method is not only restricted to G 1 -
continuity of adjacent patches of NURBS-based models but it also includes other constraints as mentioned above.
Second, here the constraint is enforced by systematically adding a potential to the weak form (Eq. (21)). In [47], the
G 1 -continuity is enforced by constraining the position of control points in the vicinity of the shared edge so that the
adjacent NURBS surfaces have the same tangent plane at the intersection points. This leads to a completely different
constraint equation.
Like the bending strip method, we focus on conforming meshes, where the control points of adjacent patches
coincide at their interface. For nonconforming meshes, the proposed method should be modified or alternative
methods, like Nitsche’s or Mortar methods, can be used. Compared to the bending strip method [37], the proposed
method requires only line integration instead of surface integration. For nonlinear problems, the implementation of
Nitsche’s method (e.g. [54]) becomes more complex as it requires the tractions and their variations on the interface,
which depend on the type of constitutive equations. Our formulation is independent of the choice of material model.
Further, as shown in Sections 4.4.1 and 4.4.2, our constraint equation, enforced by both the penalty and Lagrange
multiplier methods, gives an exact transmission of both traction and moment across the interface.
In order to model all five cases depicted in Fig. 2, we first consider the constraint equation on the surface edge L
gc := cos α0 − cos α = 0, ∀x ∈ L, (102)
where
cos α0 := N · Ñ, cos α := n · ñ. (103)
Here, N is the surface normal of a considered patch and Ñ is the surface normal of a neighboring patch. They are
defined in the initial configuration. Similarly, n and ñ are the corresponding quantities in the deformed configuration.
As shown in Fig. 2(a), in the case of G 1 -continuity between two patches, we have N · Ñ = 1. Similarly for fixed
surface folds, N · Ñ = cos α0 , where α0 is the angle between the normal of the patches (see Fig. 2(b)). In the case of
symmetry, ñ = Ñ is the (fixed) normal vector of the symmetry plane (see Fig. 2(c)–(d)). Furthermore, the constraint
equation (102) can be used to apply rotational Dirichlet boundary conditions, by setting N · Ñ = 1, and ñ is then the
prescribed normal direction.
The constraint equation (102) can uniquely realize angles within [0, π]. However, Eq. (102) becomes non-unique
if α and α0 go beyond this range. Therefore, an additional constraint equation in the form of a sine function is
considered,
gs := sin α0 − sin α = 0, ∀x ∈ L (104)

with
sin α0 := N × Ñ · τ 0 , sin α := n × ñ · τ ,
   
(105)
where τ and τ 0 are the unit directions of interfaces L and L0 (see Section 2.2). Here, Eq. (105) implies that α is
measured from n to ñ, and likewise for α0 . Together, Eqs. (102) and (104) uniquely define any physical angles of α0
and α ∈ [0, 2π ]. One can easily show that they are equivalent to the system of equations
g¯c := 1 − cos(α − α0 ) = 0,
(106)
g¯s := sin(α − α0 ) = 0.
It is also worth noting that an initial NURBS mesh imported from CAD programs is usually given with G 1 -continuity
instead of C 1 -continuity between patches. In this case, a strict C 1 -continuity enforcement without a mesh modification
will affect the FE solution. Thus, a G 1 -continuity constraint is more practical in such a case. In the following, we
present a penalty and a Lagrange multiplier method for enforcing constraints (102) and (104).
4.4.1. Penalty method for edge rotations

The system of constraints (102) and (104) can be enforced by the penalty formulation
ϵ 2
 
gc + gs2 dS = ϵ 1 − c0 cos α − s0 sin α dS,
  
ΠL = (107)
L0 2 L0
which is a locally convex function w.r.t. α and α0 . Thus, for the Newton–Raphson method, the existence of a unique
solution is guaranteed, provided that |α − α0 | < π. Here ϵ is the penalty parameter and s0 := sin α0 and c0 := cos α0 .
Taking the variation of the above equation yields

δΠL = − ϵ δτ · θ + δn · d̃ + δ ñ · d dS,
 
(108)
L0
where we have defined
θ := s0 n × ñ, d̃ := c0 ñ + s0 ν̃, d := c0 n + s0 ν (109)
and ν̃ := τ̃ × ñ = −τ × ñ. The linearization of the above equation can be found in Appendix D. We note that, for
some applications, the variation of τ and ñ in Eq. (108) vanishes since they are considered fixed.
From Eq. (108), the bending moments m τ = m τ̃ = ϵ sin(α0 − α) can be identified. This implies that the bending
moment is transmitted across the interface exactly.
Remark 11. If α0 = 0 (for the cases in Fig. 2(a) and (e)), the integrand in Eq. (107) reduces to ϵ (1 − cos α) such that
ΠL becomes
ϵ

ΠL = (n − ñ) · (n − ñ) dS, (110)
L0 2
enforcing the constraint
gL := (n − ñ) = 0. (111)
4.4.2. Lagrange multiplier method for edge rotations

Alternatively, the Lagrange multiplier approach can be used to enforce the system of continuity constraints (106).
In this case, we consider the constraint potential

 
ΠL = q ḡc + ḡs dS, (112)
L0
where q is the Lagrange multiplier. This potential guarantees unique solutions of α and α0 ∈ [0, 2π ] provided that
|α − α0 | < π/4. Furthermore, within this range, the gradient of gL := ḡc + ḡs w.r.t α and α0 is non-zero, which is the
necessary condition to have a solution for the Lagrange multiplier q (see e.g. [72]).
Taking the variation of Eq. (112) yields

 
δΠL = gL δq dS − q δτ · θ + δn · d̃ + δ ñ · d dS,
 
(113)
L0 L0
where θ , d and d̃ are now defined as

θ := (s0 − c0 ) n × ñ,
d := (s0 + c0 ) n + (s0 − c0 ) ν, (114)
d̃ := (s0 + c0 ) ñ + (s0 − c0 ) ν̃.
Here, we find the bending moment m τ = m τ̃ = −q, which is also transmitted exactly across the interface. The
linearization of Eq. (113) can be found in Appendix D.
5. Numerical examples
In this section, the performance of the proposed shell formulation is illustrated by several benchmark examples,
considering both linear and non-linear problems. The computational results are verified by available reference
solutions.
5.1. Linear problems
For the linear problems discussed below, we will consider two material models: the Koiter shell material model of
Eq. (35), and
 the projected shell material model obtained from the numerical integration of Eqs. (65) and (78), where
Λ̃ = E ν/ (1 + ν)(1 − 2 ν) and µ̃ = E/ 2 (1 + ν) . For the sake of simplicity, hereinafter we denote the former as
  
the Koiter model and the latter as the projected model. Furthermore, all physical quantities are introduced in terms of
unit length L 0 and unit stress E 0 .
5.1.1. Pinching of a hemisphere
A hemisphere pinched by two pairs of diametrically opposite forces F = 2 E 0 L 20 on the equator is examined in
this example. The model parameters are adopted from [73] as E = 6.825 × 107 E 0 , ν = 0.3, R = 10.0 L 0 and
T = 0.04 L 0 . Due to the symmetry, 1/4 of the hemisphere is modeled as shown in Fig. 3(a). Here, the symmetry
boundary conditions are applied on the Y = 0 and X = 0 planes by the penalty method of Eq. (107) with the penalty
parameter ϵ = 800 E L 20 and the rigid body motions are restricted by fixing the top control point.
The benchmark reference solution for the radial displacement under the point loads is 0.0924 L 0 [73–75]. As
shown in Fig. 3, the numerical results converge to the reference solution as the mesh is refined and the NURBS order
is increased. Here, the radial displacement of the force is normalized by the reference solution. As can be observed
in Fig. 3(c)–(d), for linear elastic deformations, both the Koiter and the projected model are identical. This is also the
case for several other problems examined here. In the following, only the results of one of the models are reported if
the differences are negligible.
5.1.2. Simply supported plate under sinusoidal pressure loading
As a second example, we analyze a plate with size L × L = 12 × 12 L 20 , thickness T = 0.375 L 0 , Young’s modulus
E = 4.8×105 E 0 , Poisson’s ratio ν = 0.38, subjected to sinusoidal pressure p(x, y) = p0 sin(π x/L) sin(π y/L) E 0 .
According to Navier’s solution [76], the maximum deflection is at the middle point and given by
p0 L 2
wmax = , (115)
4 π4 D
where D := E T 3 /12 (1 − ν 2 ) is the flexural rigidity of the plate. The setup of the computational model is shown
in Fig. 4(a). Only 1/4 of the plate is modeled using symmetry boundary conditions enforced by the penalty method
of Section 4.4.1 with the penalty parameter ϵ = 10−2 n p−1 E L 20 , where p is the NURBS order and n is the number
of elements in each direction. Fig. 4(b) shows the deformed plate with the Koiter model and Fig. 4(c) shows the
convergence of the computational solution to the analytical one as the mesh is refined. For the comparison, the vertical
displacement at the center of the plate is normalized by the analytical solution given in Eq. (115). The corresponding
relative error is shown in Fig. 4(d). As expected, more accuracy is gained by increasing the NURBS order.
0.08
a b 0.06
0.04
0.02
-0.02
-0.04
-0.06
-0.08
c d
Fig. 3. Pinching of a hemisphere: (a) Undeformed configuration with boundary conditions. Here, the blue curves denote the symmetry lines. (b)
Deformed configuration (scaled 50 times), colored by radial displacement. Radial displacement at the point load vs. mesh refinement for (c) the
Koiter and (d) the projected shell model, considering various NURBS orders. (For interpretation of the references to color in this figure legend, the
reader is referred to the web version of this article.)
5.1.3. Pinching of a cylinder

Next, we consider the pinched cylinder test with rigid diaphragms at its ends. It is designed to examine the
performance of shell elements in inextensional bending modes and complex membrane states [73]. The analytical
solution for this problem was introduced by Flügge [77] based on a double Fourier series, see Appendix E.
The parameters are adopted from [73] as E = 3 × 106 E 0 , ν = 0.3, R = 300 L 0 , L = 600 L 0 , T = 3 L 0 , where
E is Young’s modulus, ν is Poisson’s ratio; R, L and T are radius, length and thickness of the cylinder, respectively.
For the FE computation, 1/8 of the cylinder is modeled due to its symmetry as is shown in Fig. 5(a). The symmetry
boundary conditions are enforced by the penalty method discussed in Section 4.4.1. The penalty parameters used are
p−1 p−1
ϵ = 2 × 102 nl E L 20 for the axial symmetry, and ϵ = 2 × 102 n t E L 20 for the circumferential symmetry. Here,
n t and nl are the number of elements in circumferential and axial directions, respectively.
Two opposing pinching forces F = 1 E 0 L 20 are applied at the middle of cylinder. For the symmetric model, a
quarter of the force is applied. The deflection due to the loads is measured for comparison with the reference solution.
For 80×80 Fourier terms, the deflection under the point load is 1.82488×10−5 , which is the reference value commonly
used in the literature. However, for 8192 × 8192 Fourier terms the converged solution is 1.82715781 × 10−5 and for
214 × 214 Fourier terms, we obtain the solution of 1.82715797 × 10−5 .
Fig. 5(c) shows the convergence behavior of the computed displacement at the loading point w.r.t. the analytical
solution as the mesh is refined. From Fig. 5(d), we observe that the relative error only converges up to a certain point
and then seems to get stuck. This is due to several numerical and theoretical reasons. Firstly, although the NURBS
order can be increased for the patches, the patch boundaries are only G 1 -continuous according to Eq. (107). This
a b
c d
Fig. 4. Simply supported plate under sinusoidal pressure: (a) Initial configuration with boundary conditions, (b) deformed configuration (scaled
104 times) colored by the vertical displacement, (c) displacement of the plate center normalized w.r.t. the analytical solution and (d) relative error
of the displacement. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
affects higher order NURBS, especially at the boundaries around the singular point load. Secondly, the analytical
solution represents the concentrated load by a pressure distribution in the form of a Dirac delta function. Numerically,
this is represented by a Fourier series and thus the Gibbs effect can lead to a loss in numerical accuracy for high order
terms (see Appendix E). Therefore, the Fourier solution is truncated at some point and we cannot expect the error to go
down to machine precision as the mesh is refined. We note that, although both the results of the Koiter shell material
model of Eq. (35), and the projected shell material model of Eqs. (65) and (78) are identical, the computational time of
the former is roughly half of the latter (considering three quadrature points through the thickness). Thus, the presented
Koiter shell material model is more efficient for the same accuracy.
5.2. Nonlinear problems
In the following, several nonlinear test cases are presented to illustrate the robustness and accuracy of the proposed
shell formulation.
5.2.1. Pure bending of a flat strip

We first examine the pure bending of a flat strip with the dimensions S × L. The strip is fixed as is shown in
Fig. 6(a)–(d) in order to apply the moment M at the two ends of the strip. The Canham material model, Eq. (32), is
used. For this problem, the analytical solution is given by Sauer and Duong [1]. Accordingly, the strip is deformed
into a curved sheet with dimensions s × l = λ1 S × λ2 L with constant curvature κ1 that is linearly related to M
as
M = c κ1 . (116)
Fig. 5. Pinching of the cylinder with rigid end diaphragms: (a) Setup of the computation model, (b) deformed shell (scaled 106 times) colored by
the radial displacement, (c) normalized radial displacement at the point load and (d) error w.r.t. the reference solution as the mesh is refined. (For
interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
Since all the boundaries are free to move, we expect that the in-plane stress components vanish. This condition leads
to the stretches λ1 and λ2 in terms of the applied moment M as

2
M2 M2

λ2 = λ1 /a0 , a0 := + + 1, (117)
2µc 2µc
and

µ

λ1 = −µ̄ (a02 + 1) + µ̄2 (a02 + 1)2 + a02 (4 µ̄ + 1), µ̄ := . (118)
2Λ
To assess the FE computation, four different mesh schemes using quadratic NURBS are considered as is shown in
Fig. 6(a)–(d): 1. A single patch with regular mesh; 2. A single patch with skew mesh with distortion ratio r = 1/5;
3. A double patch with regular mesh; and 4. A double patch with skew mesh with the same distortion ratio. Here, the
mesh refinement is carried out by the knot insertion algorithm (see e.g [20]). In particular for the second scheme, the
skew mesh is obtained by distortion of the knot vectors as explained in detail in Appendix F.
For all mesh schemes, the material parameters µ = 10 L 2 /c, Λ = 5 L 2 /c and the bending moment M = 1 c/L are
applied (setting m τ = M in Eq. (101)(4)). For mesh schemes 3 and 4, the G 1 -continuity constraint is used between
the two patches (see Section 4.4.1) with ϵ = 10 000 (n S ×n L ) L/c, where n S and n L are the number of elements along
the length and width of the strip, respectively. Fig. 6(e) shows the deformed mesh and the mean curvature error for
mesh 4 × 8 of scheme 4. Fig. 6(f) shows the L 2 error of the displacement field for the four considered mesh schemes.
a e 0.06
0.05
0.04
0.03
b 0.02
0.01
f
c
Fig. 6. Pure bending of a flat strip considering: (a) Single patch with regular mesh, (b) single patch with skew mesh, (c) two patches with regular
mesh, and (d) two patches with skew mesh. (e) Deformed configuration for the mesh in (d). The color here shows the relative error of mean
curvature H . (f) L 2 error of u (Eq. (119)) w.r.t. mesh refinement. (For interpretation of the references to color in this figure legend, the reader is
referred to the web version of this article.)
Here, the L 2 error is defined as

 
h 1  2
∥u − u∥ L 2 := uh − u dA, (119)
S L S0
where uh is the displacement obtained from the FE analysis and u is the corresponding analytical quantity calculated
at each point on S. Here, u can be computed for any applied moment by using the parametrization described in [1].
The convergence observed in Fig. 6(f) verifies the presented finite element formulation. It also shows, that the
accuracy of the double patch meshes is of the same order as the single patch results. This indicates the effectiveness
of the penalty constraint presented in Section 4.4.
5.2.2. Pure bending of a folded strip

Next, we demonstrate the robustness of the edge rotation constraint presented in Section 4.4 for multiple patch
interfaces with kinks. For this purpose, we reconsider the pure bending test of the previous section, but here the mesh
consists of 8 patches and has a kink at 3/4 π L from the left end, as shown in Fig. 7(a). The material of the strip is the
same as the example in Section 5.2.1. The deformed configuration is shown in Fig. 7(b).
Here, the continuity between patch interfaces including both G 1 -continuity and the kink constraints are enforced
by the penalty method (PM) (107) and the Lagrange multiplier method (LM) (112). Further, for LM, we consider a
constant and a linear interpolation scheme of the Lagrange multiplier q denoted as N2Q0 and N2Q1c, respectively.
We note that at patch junctions, q can jump across interfaces as there is a change of patch pairings. In this case, N2Q0
can automatically capture the jump since q is interpolated discontinuously over the elements, whereas for N2Q1c, the
a c 10–1
10–2
0.04
0.035 10–3
L2 error of U
0.03
0.025 10–4
0.02
10–5
b 0.015
0.01 10–6
0.005
10–7
102 103 104
Number of elements nel
Fig. 7. Pure bending of a folded strip: (a) Initial FE configuration and boundary conditions (for S1 + S2 = π L, opening angle β0 = π/6, discretized
with quadratic NURBS elements); (b) current FE configuration and relative error of H w.r.t Hexact = 0.8 (1/L) for an imposed bending moment
M = 1.6 c/L; (c) L 2 error of u (Eq. (119)) w.r.t. mesh refinement, considering the penalty method (PM) and the Lagrange multiplier method (LM).
(For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
continuous interpolation of q would result in over-constraining, i.e. is LBB-unstable. However, this problem can be
simply solved by repeating the pressure degrees of freedom at patch junctions for each interface so that the jump can
be captured.
In order to assess the FE results, the L 2 error defined by Eq. (119) is computed. As the analytical solution, given
by Eqs. (116) and (117), is also valid for this problem, we can use the same procedure as in Section 5.2.1 to compute
u and uh . It only needs to be adapted to include the kink.
Fig. 7(c) shows the convergence of the computed L 2 error as the mesh is refined. With an appropriate choice of the
penalty parameter ϵ, the rate of convergence of PM can be achieved at the same order as for LM. Furthermore, the
rate of convergence here is also the same order as for bending of the flat strip (see Fig. 6). Besides, the accuracy of
PM approaches that of LM as ϵ is increased. However, note that the stiffness matrix becomes ill-conditioned if ϵ is
too high. For LM, we observe that both interpolation schemes N2Q0 and N2Q1c are robust and stable. It can be seen
that N2Q0 is as accurate as N2Q1c here.
5.2.3. Cantilever subjected to end shear forces

The large deflection of a cantilever beam, with dimensions L × W × T = 10 × 1 × 0.1 L 30 , due to shear traction
t̄ = F/W applied to its free end is computed. The forces at the nodes located on the free end are derived from
Eq. (101)(3). The fixed end is clamped by the penalty method (Section 4.4.1) with ϵ = 1000 E L 20 . The material
parameters are E = 1.2 × 106 E 0 and ν = 0.0 [78]. The beam is modeled with the Koiter shell material and is
discretized by 1 × 10 NURBS elements, considering both a regular mesh and a skew mesh with distortion ratio
r = 0.2 in y-direction. The applied maximum shear force is Fmax = 4 F0 with F0 = E I /L 2 and I = 1/12 W T 3 . As
shown in Fig. 8, considering 10 elements, cubic NURBS give very good results for both regular and skew meshes.
To examine the effects of both bending and large membrane strains, the same problem is studied with an extra
Dirichlet boundary condition applied at the beam tip imposing zero displacements in the y-direction. Here, since
the beam is considerably stretched, the in-plane membrane strains are significant. Instead of a shear force, a vertical
displacement is applied at the tip and the reaction forces are measured. In addition to the Koiter model and the
projected shell model, a mixed model that combines the bending part of a Koiter shell material with a Neo-Hookean
membrane formulation is considered (see Eq. (41)). As shown in Fig. 9, all three models predict similar results as long
as the bending effects are dominant. By increasing the membrane strains, the mixed formulation is very close to the
projected model while the full Koiter model has a different trend. This shows that the simple, more efficient mixed
Koiter model can accurately capture the full 3D model behavior.
5.2.4. Pinching of a hemisphere with a hole

In this example, we compute the large deformation of a hemisphere with a hole at the top, under two pairs of equally
large but opposing forces that are applied on the equator of the hemisphere as shown in Fig. 10. The parameters are
a b
c d
Fig. 8. Cantilever subjected to end shear force: (a) Undeformed configuration, (b) deformed configuration colored by the vertical displacement.
Horizontal (−v A ) and vertical (w A ) displacement at tip A for (c) a regular mesh with uniform element length and (d) a skew mesh as shown in (b).
The results are compared with [78]. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of
this article.)
a b
Fig. 9. Cantilever with prescribed tip displacement: (a) Vertical reaction force Fz vs. vertical displacement and (b) horizontal reaction force Fy vs.
vertical displacement.
a b
c d
Fig. 10. Pinching of a hemisphere with a hole at 18◦ : (a) Undeformed configuration with boundary conditions, (b)–(c) deformed configuration
colored by the radial displacement and (d) force vs. displacement of points A and B compared with the reference solution of Sze et al. [78]. (For
interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
extracted from [78]: R = 10 L 0 , T = 0.04 L 0 , E = 6.825 × 107 E 0 , ν = 0.3 and the point load Fmax = 400 E 0 L 20 .
The symmetry of the hemisphere is modeled by the penalty method with ϵ = 6 × 103 E L 20 (see Fig. 10(a)). The
Koiter shell model is used and the surface is meshed with 20 × 20 quadratic, cubic and quartic NURBS based finite
elements. As observed in Fig. 10(d), the computed results approach the reference solution as the NURBS order is
increased.
5.2.5. Pinching of a cylinder with end rigid diaphragms

We reconsider the pinched cylinder test of Section 5.1.3, but here with large deformations as shown in Fig. 11.
Here, the length, thickness and radius of the cylinder are L = 200 L 0 , T = 1.0 L 0 and R = 100 L 0 , respectively.
Both the Koiter and the projected shell material models are used with E = 30 × 103 E 0 and ν = 0.3 and they
give indistinguishable results. The point load Fmax = 12 × 103 E 0 L 20 is applied. Due to the symmetry, only 1/8
of the cylinder is modeled. The symmetry boundary conditions are enforced by the Lagrange multiplier method (see
Section 4.4). The cylinder is discretized by 50×50 quadratic NURBS finite elements. Fig. 11(f) shows good agreement
with the reference result of Sze et al. [78].
5.2.6. Spreading of a cylinder with free ends

In this test, a cylinder, with dimensions L × R × T = 10.35 × 4.953 × 0.094 L 30 , with open ends is pulled apart by
a pair of opposite forces up to Fmax = 40 × 103 E 0 L 20 , which are applied at the middle of the cylinder (see Fig. 12).
The Koiter material is used with E = 10.5 × 106 E 0 and ν = 0.3125. Here, the results of Sze et al. [78] are used
as a reference for comparison. Due to the symmetry, only 1/8 of the cylinder is modeled as shown in Fig. 12(a).
The cylinder is discretized by 20 × 20 NURBS finite elements. The symmetry boundary conditions are enforced by
a c d
b e
Fig. 11. Pinching of a cylinder with rigid end diaphragm: (a) Undeformed configuration with boundary conditions. Deformed configurations in
(b) 3D view, (c) y-axis view, (d) z-axis view. The color here denotes the radial displacement. (f) Force vs. displacement at points A and B compared
to the results of Sze et al. [78]. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this
article.)
the penalty method (see Section 4.4) with ϵ = 20 E L 30 /L along the axial edge and ϵ = 20 E L 30 /π R along the
circumferential edge. A good agreement with the reference results is also observed in Fig. 12(c).
6. Conclusion
A new unified FE formulation for rotation-free thin shells is presented. The formulation uses isogeometric analysis
to benefit from its high order continuity. We have also introduced a penalty and a Lagrange multiplier approach
to enforce continuity at patch boundaries. The approach can be used to model G 1 -continuity required for multi-
patch NURBS, fixed surface folds, symmetry (or clamping) constraints, symmetry constraints at a kink and rotational
Dirichlet boundary conditions. It is required in order to transfer moments at patch boundaries. We observed that
the penalty method provides a simple and very efficient implementation, yet still maintains sufficient accuracy. The
proposed formulation is tested by several numerical examples considering both linear as well as non-linear regimes.
The numerical results are verified either by available analytical solutions or reliable reference solutions. The results
demonstrate the robustness as well as good performance of the formulation.
Further, we have presented a detailed and systematic procedure to obtain shell material models based on through-
thickness integration of existing 3D constitutive laws. Besides that, our formulation is also designed to accept material
models given directly in surface energy form. Since there is no need for numerical integration over the thickness, such
material models are much more efficient. For the considered numerical examples, it turns out that these models,
particularly the mixed Koiter formulation, are equally accurate as the expensive integration models, even for large
deformations and stretches. Further, our formulation is presented fully in curvilinear coordinates. This allows for
a direct and efficient implementation, since no local coordinate transformation is needed. The formulation also
Fig. 12. Spreading of a cylinder with free ends: (a) Undeformed configuration with boundary conditions, (b) deformed configuration colored with
radial displacement, (c) force vs. displacement of points A, B and C compared to the results of Sze et al. [78]. (For interpretation of the references
to color in this figure legend, the reader is referred to the web version of this article.)
allows for a straightforward and efficient inclusion of material anisotropy [79] and stabilization schemes for liquid
shells [55].
Acknowledgments
Financial support from the German Research Foundation (DFG) through grant GSC 111 is gratefully
acknowledged. The authors also wish to thank Callum J. Corbett and Yannick A.D. Omar for their help.
Appendix A. FE tangent matrices
Using Eqs. (93) and (94), the terms in Eq. (25) become
1 1
cαβγ δ δaαβ ∆aγ δ = cαβγ δ δxTe NT,α (aβ ⊗ aγ ) N,δ ∆xe ,
2 2
1
d αβγ δ δaαβ ∆bγ δ = d αβγ δ δxTe NT,α (aβ ⊗ n) Ñ;γ δ ∆xe ,
2 (120)
αβγ δ 1 αβγ δ
e δbαβ ∆aγ δ = e δxTe ÑT;αβ (n ⊗ aγ ) N,δ ∆xe ,
2
f αβγ δ δbαβ ∆bγ δ = f αβγ δ δxTe ÑT;αβ (n ⊗ n) Ñ;γ δ ∆xe ,
and
1
τ αβ ∆δaαβ = δxTe NT,α τ αβ N,β ∆xe , (121)
2
 
∆δbαβ = − δxTe NT,γ (n ⊗ aγ ) Ñ;αβ + ÑT;αβ (aγ ⊗ n) N,γ + NT,γ a γ δ bαβ (n ⊗ n) N,δ ∆xe . (122)
Thus, the linearization of G eint yields
∆G eint = δxTe keτ τ + keτ M + keMτ + keM M + keτ + keM ∆xe ,

 
(123)
with the material stiffness matrices


e
kτ τ := cαβγ δ NT,α (aβ ⊗ aγ ) N,δ dA,
Ω0e

keτ M := d αβγ δ NT,α (aβ ⊗ n) Ñ;γ δ dA,
Ω0e
 (124)
αβγ δ
keMτ := e ÑT;αβ (n ⊗ aγ ) N,δ dA,
e
Ω0
keM M := f αβγ δ δxTe ÑT;αβ (n ⊗ n) Ñ;γ δ dA
Ω0e
and the geometric stiffness matrices
keM = keM1 + keM2 + (keM2 )T , (125)

with

αβ
keM1 := − bαβ M0 a γ δ NT,γ (n ⊗ n) N,δ dA,
Ω0e
 (126)
αβ
keM2 := − M0 NT,γ (n ⊗ aγ ) Ñ;αβ dA.
Ω0e
For an efficient implementation of these matrices see Appendix B. Similarly, the linearization of G eext
yields
∆G eext = δxTe keext p + keextm ∆xe ,

 
(127)
with the external FE tangent matrix5

 
keextm = m τ NT,α ν β n ⊗ aα + ν α aβ ⊗ n N,β ds − m τ ν α NT,α (n ⊗ aξ ) N,ξ ds,
 
(128)
∂m Ω e ∂m Ω e
which corresponds to feextm , in Eq. (101)(4). Here, ξ denotes the convective coordinate of the curve ∂m Ω e . For the
application of surface pressure through Eq. (101)(2), the corresponding keext p can be found e.g. in [68].
Appendix B. Efficient FE implementation
This section presents an efficient implementation of the above equations. First a general algorithm is suggested,
which can be used for a variety of models like for example the Koiter model (35) and the projected shell model (71).
Then a special implementation for models like the Canham model (32) is presented.
B.1. For general cases
For an efficient implementation, we arrange the tangent components cαβγ δ as

 1111
c1122 c1112

c
C := c2211 c2222 c2212  , (129)
c1211 c1222 c1212
5 The unsymmetric rear term in Eq. (128) disappears if m ds is constant, which corresponds to dead loading.
τ
and likewise d αβγ δ , eαβγ δ and f αβγ δ are arranged as D, E and F. Note that for model (35), C = CT , E = DT and
F = FT , leading to keMτ = keT τ M . The stress, moment, and curvature tensors are written in Voigt notation as
τ̂ := [τ 11 , τ 22 , τ 12 ]T ,
M̂0 := [M011 , M022 , M012 ]T , (130)
b̂ := [b11 , b22 , 2 b12 ] . T
Taking n as the number of control points per element, we define the (3n × 1) arrays
Laαβ := NT,α aβ ,
Lnα := NT,α n,
(131)
Gnαβ := ÑT;αβ n
and organize them into
L̂a = [La11 , La22 , La12 + La21 ],

(132)
Ĝn = [Gn11 , Gn22 , Gn12 + Gn21 ].
We thus obtain

feintτ = L̂a τ̂ dA,
Ω0e
 (133)
feintM = Ĝn M̂0 dA,
Ω0e
for the FE forces in Eq. (98) and (99) introduced in Section 4.3, and

keττ = L̂a C L̂aT dA,
Ω0e

keτ M = L̂a D ĜTn dA,
Ω0e
 (134)
keMτ = Ĝn E L̂aT dA,
Ω0e

keM M = Ĝn F ĜTn dA,
Ω0e
for the material stiffness matrices in Eq. (124). Similarly, for the geometric stiffness matrices in Eq. (126), we get
  
keM1 = − b M a 11 Ln1 LnT
1 + a 22 n nT
L2 2L + 2 a 12 n nT
L L
1 2 dA, b M := b̂T M̂0 ,
Ω0e
    (135)
keM2 =− Ln1 a1T + Ln2 a2T M011 Ñ;11 + M022 Ñ;22 +2 M012 Ñ;12 dA.
Ω0e
B.2. For particular cases
For some material models, the material tangents fit into the format
ĉαβγ δ = ĉaa a αβ a γ δ + ĉa a αβγ δ + ĉab a αβ bγ δ + ĉba bαβ a γ δ + ĉbb bαβ bγ δ , ĉ = c, d, e, f, (136)
with suitable definitions of coefficients ĉaa , ĉa , ĉab , ĉba and ĉbb . We can then obtain a more efficient implementation
of the possible contractions within Eqs. (124) and (125). Examples for this case are the Canham model (Section 2.6.1)
and Helfrich model [55]. The sequential computation is as follows.
Precompute the (3 × 3n) arrays
Ñαβ := a αγ Ñ;βγ ,
Ña := Ñαα , (137)
Ñb := bαβ Ñ;αβ
and the (3n × 1) vectors
Lαβ := NT,β aα , L̃αβ := Ñαβ T n,

La := Lαα , L̃a := ÑTa n,
Lb := NT,α bα , L̃b := ÑTb n, (138)
L̃αa := ÑTa aα , L̃αb := ÑTb aα ,
Lα := NT,α n, LA := NT,α Ãα (∗),
with bα := bαβ aβ and Ãα := Ãαβ aβ . The equation marked by (∗) is only needed for the stabilization schemes for
liquid shells [55] and can be skipped for solid shells. (Depending on the type of stabilization scheme used, either
αβ αβ
Ãαβ := Aαβ or Ãαβ := apre , where apre is the corresponding quantity at the previous load step.)
We then compute the (3n × 3n) arrays
N2a := a αβ NT,α N,β ,

N2A := Ãαβ NT,α N,β ,
(139)
N2b := bαβ NT,α N,β ,
L2 := a αβ Lα LTβ .
With
τ αβ = τa a αβ + τb bαβ + τA Ãαβ ,
αβ (140)
M0 = Ma0 a αβ + Mb0 bαβ ,
and
cαβγ δ = caa a αβ a γ δ + ca a αβγ δ + cab a αβ bγ δ + bαβ a γ δ + cbb bαβ bγ δ ,

 
d αβγ δ = daa a αβ a γ δ + da a αβγ δ + dab a αβ bγ δ + dba bαβ a γ δ + dbb bαβ bγ δ , (141)

αβγ δ αβ γδ αβγ δ
f = f aa a a + fa a ,
we thus get

feintτ τa La + τb Lb + τA La dA,
 
=
Ω0e
 (142)
feintM = Ma0 L̃a + Mb0 L̃b dA
 
Ω0e
and
  
keττ = c dA, keτ M = d dA = (keMτ )T , keM M = f dA, (143)
Ω0e Ω0e Ω0e
with
ca  β α T
c := caa La LTa + cab La LTb + Lb LTa + cbb Lb LTb − Lα Lβ + N2a − L2
  
2
d := daa La L̃a + dab La L̃b + dba Lb L̃a + dbb Lb L̃b − da Lβα L̃αβ T ,
T T T T (144)
f := f aa L̃a L̃Ta − f a L̃βα L̃αβ T ,

and
  
keτ = t dA, keM1 = m1 dA, keM2 = m2 dA (145)
Ω0e Ω0e Ω0e
with
t := τa N2a + τA N2A + τb N2b ,

m1 := − 2H Ma0 + (4H 2 − 2κ) Mb0 L2 ,
 
(146)
m2 := −Ma0 Lα L̃αa T − Mb0 Lα L̃αb T .
For further efficiency, the symmetric matrices c, f, t and m1 should only be computed in upper triangular form,
integrated and then reflected after integration (i.e. after the quadrature loop). Likewise keMτ and keM should be
determined from keτ M , keM1 and keM2 after quadrature.
Appendix C. Linearization of the projected shell model
In this section, the material tangents cαβγ δ , d αβγ δ , eαβγ δ and f αβγ δ that are needed for the linearization of the
projected shell model, i.e. Eqs. (65) and (78), are derived considering ξ ≈ ξ0 .
C.1. Linearization of g αβ
The linearization of g αβ gives
∆g αβ = g αβγ δ ∆gγ δ , (147)

where (see [1])
1  αγ βδ
g αβγ δ := − g g + g αδ g βδ .

(148)
2
Further, from Eq. (51) we have
ϵη ϵη
∆gγ δ = aγ δ ∆aϵη + bγ δ ∆bϵη , (149)
with
ϵη η
aγ δ := ga δγϵ δδ + κ ξ 2 aγ δ a ϵη − ξ 2 bγ δ bϵη ,
ϵη η
(150)
bγ δ := gb δγϵ δδ − ξ 2 aγ δ b̃ϵη + ξ 2 bγ δ a ϵη .
We thus obtain
αβγ δ αβγ δ
∆g αβ = ga ∆aγ δ + gb ∆bγ δ , (151)
with
αβγ δ αβγ δ
ga := g αβϵη aϵη
γδ
, gb := g αβϵη bϵη
γδ
. (152)
C.2. Linearization of τ̃ αβ
Considering that τ̃ αβ has the form (see e.g. Eq. (78))
τ̃ αβ = µ̃ G αβ + f (J ∗ ) g αβ , (153)
we have
f ′ J ∗ αβ γ δ
 
αβ
∆τ̃ = g g + f g αβγ δ ∆gγ δ . (154)
2
Taking into account Eq. (149), we obtain
1 αβγ δ
∆τ̃ αβ = c̃ ∆aγ δ + d̃ αβγ δ ∆bγ δ , (155)
2
with
∂ τ̃ αβ αβγ δ
c̃αβγ δ := 2 = J ∗ f ′ g αβ g ϵη aϵη
γδ
+ 2 f ga ,
∂aγ δ
(156)
αβγ δ ∂ τ̃ αβ J ∗ f ′ αβ ϵη γ δ αβγ δ
d̃ := = g g bϵη + 2 f gb .
∂bγ δ 2
αβ
C.3. Linearization of τ αβ and M0
From Eq. (65), we have6
∂τ αβ
 T
2
 
αβγ δ
c := 2 = s0 2 κ ξ 2 τ̃ αβ a γ δ + (1 − ξ 2 κ) c̃αβγ δ dξ,
∂aγ δ − T2
∂τ αβ
 T
2
 
d αβγ δ := = s0 −ξ 2 τ̃ αβ b̃γ δ + (1 − ξ 2 κ) d̃ αβγ δ dξ,
∂bγ δ − T2
(157)
αβ T
∂ M0

2
 
eαβγ δ := 2 = s0 −ξ 2 τ̃ αβ bγ δ − (ξ − H ξ 2 ) c̃αβγ δ dξ,
∂aγ δ − T2
αβ T
∂ M0
  
2 1 2 αβ γ δ
f αβγ δ := = s0 ξ τ̃ a − (ξ − H ξ 2 ) d̃ αβγ δ dξ.
∂bγ δ − T2 2
Appendix D. Linearization of edge rotation conditions
In this section, the linearization of the constraint equations, introduced in Section 4.4 for the edge rotation
conditions, is presented for both the penalty method and the Lagrange multiplier method. Simplifications are provided
in Remarks 13 and 14.
D.1. Linearization for the penalty method
For the penalty method Eq. (108), one has (e.g. see [1,80])
δn = Rα δaα , δ ñ = R̃α δ ãα , (158)
6 Note that due to the approximation assumed in Eq. (65), the tangents in Eq. (157) lose major symmetry, but they still have minor symmetry.
Therefore C ̸= CT , E ̸= DT and F ̸= FT in Appendix B.1. If Eq. (66) is used the symmetry is retained.
where
Rα := −aα ⊗ n, R̃α := −ãα ⊗ ñ. (159)
Along the patch interface, one can define the tangents τ and τ 0 as
aξ Aξ
τ := , τ 0 := , (160)
∥aξ ∥ ∥Aξ ∥
where aξ := N,ξ xe , Aξ := N,ξ Xe , and ξ denotes the convective coordinate along the interface. From this we find
δτ = Mξ δaξ , (161)
where
1 
Mξ := 1−τ ⊗τ .

(162)
∥aξ ∥
We thus have from Eq. (108)
δΠLe = δxTe fen + δ x̃Te feñ ,

(163)
∆δΠLe = δxTe kenn ∆xe + δxTe kenñ ∆x̃e + δ x̃Te (kenñ )T ∆xe + δ x̃Te keññ ∆x̃e ,
with
   
fen := − ϵ NT,ξ Mξ θ − NT,α d̃aα n dS, feñ := ϵ ÑT,α dãα ñ dS, (164)
Γ0e Γ0e
and
  
kenn := ϵ NT,α Qαβ N,β + NT,ξ Qξ ξ N,ξ − 2 s0 sym NT,α n ⊗ ñαa Mξ N,ξ dS,

Γ0e

keññ := ϵ ÑT,α Q̃αβ Ñ,β dS, (165)
Γ0e
  
kenñ := ϵ s0 NT,ξ Mξ nαã ⊗ ñ Ñ,β − NT,α â αβ n ⊗ ñ Ñ,β dS.
Γ0e
Here, we have defined

 
Qξ ξ := (τ · θ ) (1 − 3 τ ⊗ τ ) + θ ⊗ τ + τ ⊗ θ /∥aξ ∥2 ,
Qαβ := d̃n a αβ (n ⊗ n) + d̃aα Rβ + (d̃aα Rβ )T , (166)
Q̃αβ := dñ ã αβ (ñ ⊗ ñ) + dãα R̃β + (dãα R̃β )T ,
with d̃aα := d̃ · aα , d̃n := d̃ · n, dãα := d · ãα , dñ = d · ñ, nαã := n × ãα , ñαa := ñ × aα and â αβ := aα · (c0 ãβ − s0 τ × ãβ ).
Remark 12. Note that here N and Ñ denote the shape function arrays and should not be confused with N and Ñ. In
Eqs. (164) and (165), the shape functions and their derivatives with the indices α and β affect all control points of the
elements. Here the shape function array N,ξ is understood to have the same dimension as N,α – i.e. appropriate zeros
have to be added to N,ξ .
Remark 13. In case of α0 = 0, for Fig. 2(a) and (e), Eqs. (164) and (165) are further reduced to
 
fen := ϵ NT,α (n ⊗ ñ) aα dS, feñ := ϵ ÑT,α (ñ ⊗ n) ãα dS, (167)
Γ0e Γ0e
and
  
kenn := ϵ NT,α d̃n a αβ (n ⊗ n) − d̃aα (aβ ⊗ n) − d̃aβ (n ⊗ aα ) N,β dS,
e
Γ0
β
 
keññ := ϵ ÑT,α dñ ã αβ (ñ ⊗ ñ) − dãα (ãβ ⊗ ñ) − dã (ñ ⊗ ãα ) Ñ,β dS, (168)
e
Γ
0
kenñ := − ϵ NT,α (aα · ãβ ) (n ⊗ ñ) Ñ,β dS,

Γ0e
with d̃n = dñ := ñ · n, d̃aα := ñ · aα and dãα := n · ãα . In the case of Fig. 2(e), feñ and the related tangents are not
required.
D.2. Linearization for the Lagrange multiplier method
For the Lagrange multiplier method, using the interpolation q = Nq qe , from Eq. (113) we find
δΠLe = δxTe fen + δ x̃Te feñ + δqTe feq ,

∆δΠLe = δxTe kenn ∆xe + δ x̃Te keññ ∆x̃e + δxTe kenñ ∆x̃e + δ x̃Te (kenñ )T ∆xe (169)
+ δxTe kenq ∆qe + δqTe (kenq )T ∆xe + δ x̃Te kenq
˜ ∆qe + δqe (knq
T
˜ ) ∆xe .
e T ˜
As seen in Eq. (113), the last integral has the same form as for the penalty method, Eq. (108). Therefore, the quantities
fen , feñ , kenn , keññ , and kenñ are already given in Eqs. (164) and (165) with the substitution ϵ = 1, c0 =
ˆ q (s0 + c0 ),
s0 = ˆ q (s0 − c0 ); and d, d̃, θ are now given in Eq. (114). Additionally, we find

feq := NTq gL dS, (170)
Γ0e
and
   
kenq := − NT,α (Rα )T d̃ + NT,ξ Mξ θ Nq dS, kenq
˜ := − ÑT,α (R̃α )T d Nq dS. (171)
Γ0e Γ0e
Remark 14. Note that in the cases of symmetry constraints and rotational Dirichlet boundary conditions, ñ is given
and therefore Eqs. (164)(2), (165)(2)–(165)(3) and (171)(2) vanish.
Appendix E. Analytical solution of the pinched cylinder with rigid end diaphragms
This appendix provides the analytical solution following the approach of Flügge [77] for the pinched cylinder with
rigid diaphragms. It is based on the double Fourier representation of the displacement field and the applied loading.
Accordingly, if the shell deformation is described by u, v and w in the axial, circumferential and radial directions,
respectively, the equilibrium of the system satisfies the following system of partial differential equations [77]
1 − ν ∗∗ 1 + ν ′ ∗ 1 − ν ∗∗ 1 − ν ′ ∗∗ px R 2
  
′′
u +
 u + v +vw +k ′
u −w + ′′′
w =− ,


 2 2 2 2 d

 1 + ν ′∗ 1 − ν ′′ 3 − ν ′′ ∗ pφ R 2
  
3



 u + v ∗∗ + v + w∗ + k (1 − ν) v ′′ − w =− ,
2 2 2 2 d (172)
ν ν

 1 − ∗∗ 3 − ∗ ′′′′ ∗∗
νu +v +w+k
′ ∗ ′ ′′′
v + w + 2w
′′ ′′

u −u −


2 2



 2
 + w∗∗∗∗ + 2 w∗∗ + w = pr R ,



d
with k := T 2 /(12 R 2 ), d := (E T )/(1 − ν 2 ), where R and T are the radius and thickness of the cylinder, respectively,
ν is Poisson’s ratio, E is Young’s modulus and the partial derivatives are denoted by (•)′ := R ∂(•)/∂ x and
(•)∗ := ∂(•)/∂φ. Further, as shown by Bijlaard [81], if Nf concentrated radial loads are applied at the middle of
the cylinder and they are equally spaced along the circumferential direction, they can be expressed in terms of the
double Fourier series as
px = 0,


 pφ = 0,

∞  ∞ 
λx
 (173)

 r
 p = p r mn cos(N f m φ) sin ,
m=0 n=1
R
nπ R
where λ := L , L is the length of the cylinder,
P

 p = (−1)(n−1)/2 Nf , if m = 0 and n = 1, 3, 5, . . . ,
 r mn πR L


2P (174)
pr mn = (−1)(n−1)/2 Nf , if m ̸= 0 and n = 1, 3, 5, . . . ,
πR L




pr mn = 0, otherwise,
and P is the magnitude of the loads.
It should be noted that generally the representation of concentrated loads by a series such as Eq. (174) has numerical
deficiencies. The pr mn terms alternate in sign and do not decay as m → ∞. Although the series still converges in
theory, it does not converge numerically due to inaccuracies caused by the Gibbs effect.
In general, Eq. (173) can represent any symmetric pressure distribution. If the cylinder is pinched by two opposing
forces applied at the middle, the Fourier components can be found by setting Nf = 2, φ = 0 or φ = π and x = L/2.
The corresponding solution for Eq. (172) can be found by assuming the following ansatz for the displacement field
∞ 
∞
λx
 
u= u mn cos(Nf m φ) cos ,
m=0 n=0
R
∞  ∞
λx
  
v= u mn sin(Nf m φ) sin , (175)
m=1 n=1
R
∞  ∞
λx
  
w= u mn cos(Nf m φ) sin .
m=0 n=1
R
As seen, Eq. (175) satisfies the boundary conditions of the problem with rigid diaphragms or simple supports,
i.e. v = w = 0 if x = 0 or x = L. Plugging Eq. (175) into Eq. (172), the following system of equations is
obtained
 1−ν 2 2 1+ν
  

 λ 2
+ N f m (1 + k) u mn + − λ N f m vmn


 2 2

1−ν

   
+ −ν λ − k λ − 3
λ Nf m
2 2
wmn = 0,



2




1+ν 1−ν 2

    

λ Nf m u mn + Nf m + 2 2
λ (1 + 3 k) vmn

 −

 2 2
(176)
3−ν
 
k λ Nf m wmn = 0,
2

+ Nf m +





 2

 1−ν  3−ν 2
   
λ λ 3
λ 2 2
vmn


 −ν − k − N f m u mn + N f m + kλ N f m
2 2





 2
 + 1 + k (λ4 + 2 λ2 N 2 m 2 + N 4 m 4 − 2 N 2 m 2 + 1) w = pr mn R ,

  
f f f mn
d
which can be solved numerically for the unknown coefficients of the Fourier series (u mn , vmn and wmn ) in Eq. (175).
a b
Fig. 13. Distortion of a regular knot grid. (a) Regular knot grid and (b) skew knot grid. The control points are then recomputed using the knot
insertion algorithm (see e.g. Hughes et al. [20]).
Appendix F. Skew surface meshes
This appendix explains the procedure to create skew meshes out of regular meshes for a single patch, which is used
in Sections 5.2.1 and 5.2.3. Fig. 13(a) shows the knot grid of a regular rectangular mesh on a single patch. Assuming
the knot space has the dimensions L × S in general. The skew mesh is obtained by two steps:
Step 1. Distorting the regular knot grid by the mapping
Yd (s, η) = Y (ξ, η),


 
2∆S 4 (177)
 X d (ξ, η) = X (ξ, η) − η 1 − 2 ξ2 ,
L S
where (X, Y ) and (X d , Yd ) are the coordinates of the regular and distorted knot grids, respectively; ∆S = L/2 tan θ ,
where θ is the angle of the distorted mid-line w.r.t. the η-axis (see Fig. 13(b)). The distortion or skewness ratio, r , is
defined by
S1 − S2 2∆S L
r := = = tan θ, (178)
S1 + S2 S S
where r ∈ [0, 1].
Step 2. The positions of control points are recomputed with the corresponding distorted knot grid using the knot
insertion algorithm of Hughes et al. [20]. Here, the Bézier extraction operator is computed approximately at the center
point of the knot grid.
References
[1] R.A. Sauer, T.X. Duong, On the theoretical foundations of solid and liquid shells, Math. Mech. Solids (2015)
http://dx.doi.org/10.1177/1081286515594656.
[2] E. Oñate, F. Zárate, Rotation-free triangular plate and shell elements, Internat. J. Numer. Methods Engrg. 47 (1–3) (2000) 557–603.
[3] M. Brunet, F. Sabourin, Analysis of a rotation-free 4-node shell element, Internat. J. Numer. Methods Engrg. 66 (9) (2006) 1483–1510.
[4] H. Stolarski, A. Gilmanov, F. Sotiropoulos, Nonlinear rotation-free three-node shell finite element formulation, Internat. J. Numer. Methods
Engrg. 95 (9) (2013) 740–770.
[5] G. Munglani, R. Vetter, F. Wittel, H. Herrmann, Orthotropic rotation-free thin shell elements, Comput. Mech. 56 (5) (2015) 785–793.
[6] J.C. Simo, D.D. Fox, On a stress resultant geometrically exact shell model. Part I: Formulation and optimal parameterization, Comput.
Methods Appl. Mech. Engrg. 72 (1989) 267–304.
[7] J.C. Simo, D.D. Fox, M.S. Rifai, On a stress resultant geometrically exact shell model. Part III: Computational aspects of the nonlinear theory,
Comput. Meth. Appl. Mech. Engrg. 79 (1990) 21–70.
[8] M. Bischoff, E. Ramm, Shear deformable shell elements for large strains and rotations, Internat. J. Numer. Methods Engrg. 40 (23) (1997)
4427–4449.
[9] H.T.Y. Yang, S. Saigal, A. Masud, R.K. Kapania, A survey of recent shell finite elements, Internat. J. Numer. Methods Engrg. 47 (2000)
101–127.
[10] M. Bischoff, W.A. Wall, K.-U. Bletzinger, E. Ramm, Models and finite elements for thin-walled structures, in: E. Stein, R. de Borst, T.
J.R. Hughes (Eds.), Encyclopedia of Computational Mechanics, in: Solids and Structures, vol. 2, Wiley, 2004, (Chapter 3).
[11] P. Wriggers, Nonlinear Finite Element Methods, Springer, Berlin, 2008.
[12] F. Zárate, E. Oñate, Extended rotation-free shell triangles with transverse shear deformation effects, Comput. Mech. 49 (4) (2012) 487–503.
[13] F. Cirak, M. Ortiz, P. Schröder, Subdivision surfaces: a new paradigm for thin-shell finite-element analysis, Internat. J. Numer. Methods
Engrg. 47 (12) (2000) 2039–2072.
[14] F. Cirak, M. Ortiz, Fully C 1 -conforming subdivision elements for finite deformation thin-shell analysis, Internat. J. Numer. Methods Engrg.
51 (2001) 813–834.
[15] S. Green, G.M. Turkiyyah, A rotation-free quadrilateral thin shell subdivision finite element, Commun. Numer. Methods. Eng. 21 (12) (2005)
757–767.
[16] F. Cirak, Q. Long, Advances in subdivision finite elements for thin shells, in: P. De Mattos Pimenta, P. Wriggers (Eds.), New Trends in Thin
Structures: Formulation, Optimization and Coupled Problems, in: CISM International Centre for Mechanical Sciences, vol. 519, Springer,
Vienna, 2010, pp. 205–227.
[17] V. Ivannikov, C. Tiago, P. Pimenta, Meshless implementation of the geometrically exact Kirchhoff-Love shell theory, Internat. J. Numer.
Methods Engrg. 100 (1) (2014) 1–39.
[18] L. Noels, R. Radovitzky, A new discontinuous galerkin method for Kirchhoff-Love shells, Comput. Methods Appl. Mech. Engrg. 197 (33–40)
(2008) 2901–2929.
[19] G. Becker, C. Geuzaine, L. Noels, A one field full discontinuous galerkin method for Kirchhoff-Love shells applied to fracture mechanics,
Comput. Methods Appl. Mech. Engrg. 200 (45–46) (2011) 3223–3241.
[20] T.J.R. Hughes, J.A. Cottrell, Y. Bazilevs, Isogeometric analysis: CAD, finite elements, NURBS, exact geometry and mesh refinement, Comp.
Meth. Appl. Mech. Engrg. 194 (2005) 4135–4195.
[21] D. Benson, Y. Bazilevs, M. Hsu, T. Hughes, Isogeometric shell analysis: The Reissner-Mindlin shell, Comput. Methods Appl. Mech. Engrg.
199 (5–8) (2010) 276–289. Computational Geometry and Analysis.
[22] C.H. Thai, H. Nguyen-Xuan, N. Nguyen-Thanh, T.-H. Le, T. Nguyen-Thoi, T. Rabczuk, Static, free vibration, and buckling analysis of
laminated composite Reissner-Mindlin plates using nurbs-based isogeometric approach, Internat. J. Numer. Methods Engrg. 91 (6) (2012)
571–603.
[23] W. Dornisch, S. Klinkel, B. Simeon, Isogeometric Reissner-Mindlin shell analysis with exactly calculated director vectors, Comput. Methods
Appl. Mech. Engrg. 253 (2013) 491–504.
[24] W. Dornisch, S. Klinkel, Treatment of Reissner-Mindlin shells with kinks without the need for drilling rotation stabilization in an isogeometric
framework, Comput. Methods Appl. Mech. Engrg. 276 (2014) 35–66.
[25] P. Kang, S. Youn, Isogeometric analysis of topologically complex shell structures, Finite Elem. Anal. Des. 99 (2015) 68–81.
[26] Z. Lei, F. Gillot, L. Jezequel, Developments of the mixed grid isogeometric Reissner-Mindlin shell: Serendipity basis and modified reduced
quadrature, Eur. J. Mech. A Solids 54 (2015) 105–119.
[27] T.-K. Uhm, S.-K. Youn, T-spline finite element method for the analysis of shell structures, Internat. J. Numer. Methods Engrg. 80 (4) (2009)
507–536.
[28] R. Echter, B. Oesterle, M. Bischoff, A hierarchic family of isogeometric shell finite elements, Comput. Methods Appl. Mech. Engrg. 254
(2013) 170–180.
[29] R. Bouclier, T. Elguedj, A. Combescure, Efficient isogeometric NURBS–based solid-shell elements: Mixed formulation and method, Comput.
Methods Appl. Mech. Engrg. 267 (0) (2013) 86–110.
[30] R. Bouclier, T. Elguedj, A. Combescure, On the development of NURBS-based isogeometric solid shell elements: 2D problems and
preliminary extension to 3D, Comput. Mech. 52 (5) (2013) 1085–1112.
[31] S. Hosseini, J.J.C. Remmers, C.V. Verhoosel, R. de Borst, An isogeometric solid-like shell element for nonlinear analysis, Internat. J. Numer.
Methods Engrg. 95 (3) (2013) 238–256.
[32] S. Hosseini, J.J. Remmers, C.V. Verhoosel, R. de Borst, An isogeometric continuum shell element for non-linear analysis, Comput. Methods
Appl. Mech. Engrg. 271 (2014) 1–22.
[33] R. Bouclier, T. Elguedj, A. Combescure, An isogeometric locking-free NURBS-based solid-shell element for geometrically nonlinear analysis,
Internat. J. Numer. Methods Engrg. 101 (10) (2015) 774–808.
[34] X. Du, G. Zhao, W. Wang, Nitsche method for isogeometric analysis of Reissner-Mindlin plate with non-conforming multi-patches, Comput.
Aided Geom. Design 35–36 (2015) 121–136. Geometric Modeling and Processing 2015.
[35] D.J. Benson, S. Hartmann, Y. Bazilevs, M.-C. Hsu, T.J.R. Hughes, Blended isogeometric shells, Comp. Methods Appl. Mech. Engrg. 255
(2013) 133–146.
[36] J. Kiendl, K.-U. Bletzinger, J. Linhard, R. Wüchner, Isogeometric shell analysis with Kirchhoff–Love elements, Comput. Methods Appl.
Mech. Engrg. 198 (2009) 3902–3914.
[37] J. Kiendl, Y. Bazilevs, M.-C. Hsu, R. Wüchner, K.-U. Bletzinger, The bending strip method for isogeometric analysis of Kirchhoff–Love shell
structures comprised of multiple patches, Comput. Methods Appl. Mech. Engrg. 199 (37–40) (2010) 2403–2416.
[38] N. Nguyen-Thanh, J. Kiendl, H. Nguyen-Xuan, R. Wüchner, K.-U. Bletzinger, Y. Bazilevs, T. Rabczuk, Rotation free isogeometric thin shell
analysis using PHT-splines, Comput. Methods Appl. Mech. Engrg. 200 (47–48) (2011) 3410–3424.
[39] D.J. Benson, Y. Bazilevs, M.-C. Hsu, T.J.R. Hughes, A large deformation, rotation-free, isogeometric shell, Comp. Methods Appl. Mech.
Engrg. 200 (13–16) (2011) 1367–1378.
[40] A.P. Nagy, S.T. IJsselmuiden, M.M. Abdalla, Isogeometric design of anisotropic shells: Optimal form and material distribution, Comput.
[41] A. Goyal, M. Dörfel, B. Simeon, A. Vuong, Isogeometric shell discretizations for flexible multibody dynamics, Multibody Syst. Dyn. 30 (2)
(2013) 139–151.
[42] N. Nguyen-Thanh, N. Valizadeh, M. Nguyen, H. Nguyen-Xuan, X. Zhuang, P. Areias, G. Zi, Y. Bazilevs, L.D. Lorenzis, T. Rabczuk, An
extended isogeometric thin shell analysis based on Kirchhoff-Love theory, Comput. Methods Appl. Mech. Engrg. 284 (2015) 265–291.
Isogeometric Analysis Special Issue.
[43] X. Deng, A. Korobenko, J. Yan, Y. Bazilevs, Isogeometric analysis of continuum damage in rotation-free composite shells, Comput. Methods
Appl. Mech. Engrg. 284 (2015) 349–372. Isogeometric Analysis Special Issue.
[44] A.B. Tepole, H. Kabaria, K.-U. Bletzinger, E. Kuhl, Isogeometric Kirchhoff-Love shell formulations for biological membranes, Comput.
[45] A. Riffnaller-Schiefer, U. Augsdörfer, D. Fellner, Isogeometric shell analysis with NURBS compatible subdivision surfaces, Appl. Math.
Comput. 272 (1) (2016) 139–147. Subdivision, Geometric and Algebraic Methods, Isogeometric Analysis and Refinability.
[46] Y. Guo, M. Ruess, Weak Dirichlet boundary conditions for trimmed thin isogeometric shells, Comput. Math. Appl. 70 (7) (2015) 1425–1440.
High-Order Finite Element and Isogeometric Methods.
[47] Z. Lei, F. Gillot, L. Jezequel, A multiple patches connection method in isogeometric analysis, Appl. Math. Model. 39 (15) (2015) 4405–4420.
[48] J. Kiendl, F. Auricchio, L.B. da Veiga, C. Lovadina, A. Reali, Isogeometric collocation methods for the Reissner-Mindlin plate problem,
Comput. Methods Appl. Mech. Engrg. 284 (2015) 489–507. Isogeometric Analysis Special Issue.
[49] A. Reali, H. Gomez, An isogeometric collocation approach for Bernoulli—Euler beams and Kirchhoff plates, Comput. Methods Appl. Mech.
Engrg. 284 (2015) 623–636. Isogeometric Analysis Special Issue.
[50] J. Kiendl, M.-C. Hsu, M.C. Wu, A. Reali, Isogeometric Kirchhoff-Love shell formulations for general hyperelastic materials, Comput.
[51] P.G. Ciarlet, An introduction to differential geometry with applications to elasticity, J. Elasticity 78–79 (2005) 3–201.
[52] P.B. Canham, The minimum energy of bending as a possible explanation of the biconcave shape of the human red blood cell, J. Theoret. Biol.
26 (1970) 61–81.
[53] W. Helfrich, Elastic properties of lipid bilayers: Theory and possible experiments, Z. Naturforsch. 28c (1973) 693–703.
[54] Y. Guo, M. Ruess, Nitsche’s method for a coupling of isogeometric thin shells and blended shell structures, Comput. Methods Appl. Mech.
Engrg. 284 (2015) 881–905. Isogeometric Analysis Special Issue.
[55] R.A. Sauer, T.X. Duong, K.K. Mandadapu, D.J. Steigmann, A stabilized finite element formulation for liquid shells and its application to lipid
bilayers, 2016, http://arxiv.org/abs/1601.03907.
[56] D.J. Steigmann, On the relationship between the Cosserat and Kirchhoff-Love theories of elastic shells, Math. Mech. Solids 4 (1999) 275–288.
[57] D.J. Steigmann, Fluid films with curvature elasticity, Arch. Ration. Mech. Anal. 150 (1999) 127–152.
[58] B.D. Coleman, W. Noll, The thermodynamics of elastic materials with heat conduction and viscosity, Arch. Ration. Mech. Anal. 13 (1964)
167–178.
[59] M. Itskov, Tensor Algebra and Tensor Analysis for Engineers, second ed., Springer-Verlag, Berlin, Heidelberg, 2009.
[60] D. Steigmann, Koiter’s shell theory from the perspective of three-dimensional nonlinear elasticity, J. Elasticity 111 (1) (2013) 91–107.
[61] T.J. Hughes, E. Carnoy, Nonlinear finite element shell formulation accounting for large membrane strains, Comput. Methods Appl. Mech.
Engrg. 39 (1) (1983) 69–82.
[62] R. De Borst, The zero-normal-stress condition in plane-stress and shell elastoplasticity, Commun. Appl. Numer. Methods 7 (1) (1991) 29–33.
[63] E.N. Dvorkin, D. Pantuso, E.A. Repetto, A formulation of the mitc4 shell element for finite strain elasto-plastic analysis, Comput. Methods
Appl. Mech. Engrg. 125 (1) (1995) 17–40.
[64] S. Klinkel, S. Govindjee, Using finite strain 3d-material models in beam and shell elements, Eng. Comput. 19 (8) (2002) 902–921.
[65] K. Hackl, M. Goodarzi, Lecture Note: An Introduction to Linear Continuum Mechanics, Ruhr-University Bochum, 2010.
[66] M.J. Borden, M.A. Scott, J.A. Evans, T.J.R. Hughes, Isogeometric finite element data structures based on Bezier extraction of NURBS,
Internat. J. Numer. Methods Engrg. 87 (2011) 15–47.
[67] M.A. Scott, M.J. Borden, C.V. Verhoosel, T.W. Sederberg, T.J.R. Hughes, Isogeometric finite element data structures based on Bézier
extraction of T-splines, Internat. J. Numer. Methods Engrg. 88 (2) (2011) 126–156.
[68] R.A. Sauer, T.X. Duong, C.J. Corbett, A computational formulation for solid and liquid membranes based on curvilinear coordinates and
isogeometric finite elements, Comput. Methods Appl. Mech. Engrg. 271 (2014) 48–68.
[69] D. Schillinger, L. Dede, M.A. Scott, J.A. Evans, M.J. Borden, E. Rank, T.J.R. Hughes, An isogeometric design-through-analysis methodology
based on adaptive hierarchical refinement of nurbs, immersed boundary methods, and T-spline CAD surfaces, Comput. Methods Appl. Mech.
Engrg. 249 (2012) 116–150.
[70] W. Dornisch, Interpolation of rotations and coupling of patches in isogeometric Reissner–Mindlin shell analyis (Ph.D. thesis), RWTH Aachen
University, Aachen, Germany, 2015.
[71] V. Nguyen, P. Kerfriden, M. Brino, S. Bordas, E. Bonisoli, Nitsche’s method for two and three dimensional NURBS patch coupling, Comput.
Mech. 53 (6) (2014) 1163–1182.
[72] D. Bertsekas, Constrained Optimization and Lagrange Multiplier Methods, Academic Press, New York, 1982.
[73] T. Belytschko, H. Stolarski, W.K. Liu, N. Carpenter, J.S. Ong, Stress projection for membrane and shear locking in shell finite elements,
Comput. Methods Appl. Mech. Engrg. 51 (1–3) (1985) 221–258.
[74] R.H. Macneal, R.L. Harder, A proposed standard set of problems to test finite element accuracy, Finite Elem. Anal. Des. 1 (1) (1985) 3–20.
[75] L. Morley, A. Morris, Conflict Between Finite Elements and Shell Theory, Royal Aircraft Establishment, Great Britain, 1978.
[76] A.C. Ugural, Stresses in Beams, Plates, and Shells, CRC Press, Boca Raton, 2009.
[77] W. Flügge, Stresses in Shells, Springer-Verlag, 1962.
[78] K. Sze, X. Liu, S. Lo, Popular benchmark problems for geometric nonlinear analysis of shells, Finite Elem. Anal. Des. 40 (11) (2004)
1551–1569.
[79] F. Roohbakhshan, T.X. Duong, R.A. Sauer, A projection method to extract biological membrane models from 3D material models, J. Mech.
Behav. Biomed. Mater. 58 (2016) 90–104.
[80] R.A. Sauer, Stabilized finite element formulations for liquid membranes and their application to droplet contact, Internat. J. Numer. Methods
Fluids 75 (7) (2014) 519–545.
[81] P. Bijlaard, Stresses from local loadings in cylindrical pressure vessels, Weld. J. 33 (12) (1954) 615–623.

A New Rotation-Free Isogeometric Thin Shell Formulation and A Corresponding Continuity Constraint For Patch Boundaries

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

A New Rotation-Free Isogeometric Thin Shell Formulation and A Corresponding Continuity Constraint For Patch Boundaries

Uploaded by

Copyright:

Available Formats

Available online at www.sciencedirect.

Comput. Methods Appl. Mech. Engrg. 316 (2017) 43–83

A new rotation-free isogeometric thin shell formulation and a

Available online 27 April 2016

List of important symbols

aα;β Co-variant derivative of aα w.r.t. ξ β

Ja Jacobian of the mapping P → S

V Space for admissible variations δx

2. Summary of rotation-free thin shell theory

2.1. Thin shell kinematics and deformation

bαβ = n · aα,β = −n,β · aα , (2)

1. The surface deformation gradient, which is defined as

K = K αβ Aα ⊗ Aβ := κ − b0 = (bαβ − Bαβ ) Aα ⊗ Aβ , (8)

2.2. Stress and moment tensors

where Ẇint is the local power density and we have defined

1 s denotes the arc length such that ds = ∥dx∥.

2.3. Balance laws

We note that M αβ is also symmetric but N αβ is generally not symmetric.

2.4. Weak form

G in + G int − G ext = 0 ∀δx ∈ V, (21)

2.5. Linearization of the weak form

3 For live m ν, see Eqs. (127) and (128).

Here, we note that ∂(•)

2.6.1. Initially planar shells

2.6.2. Initially curved solid shell

tr K = K αβ Aαβ , tr E = E αβ Aαβ . (38)

From Eq. (37), we further find d αβγ δ = eαβγ δ = 0 and

cαβγ δ = Λ Aαβ Aγ δ + µ Aαγ Aβδ + Aαδ Aβγ ,

3. Shell constitution derived from 3D constitutive laws

3.1. Extraction procedure

X̃(ξ α , ξ0 ) = X(ξ α ) + ξ0 N(ξ α ), (42)

x̃(ξ α , ξ ) = x(ξ α ) + ξ d(ξ α ), (43)

gαβ := gα · gβ = ga aαβ + gb bαβ ,

g/a = 1 − 2 H ξ + κ ξ 2 , G/A = 1 − 2 H0 ξ0 + κ0 ξ02 ,

δga = ξ 2 κ a γ δ δaγ δ − ξ 2 b̃γ δ δbγ δ ,

where b̃αβ := 2 H a αβ − bαβ . With these we get

C̃ = C̃αβ Gα ⊗ Gβ + C̃α3 Gα ⊗ N + N ⊗ Gα + C̃33 N ⊗ N.

and its variation reads

with εa := τ̃ αβ aαβ and εb := τ̃ αβ bαβ . Since

It should be noted here that s0 , ga , gb , εa and εb are all functions of ξ0 .

Remark 9. For an efficient implementation, we consider that τ̃ αβ is expressible in the form

3.2. Extraction example: compressible Neo-Hookean materials

δ I˜1 = δ I1∗ + 2 λ3 δλ3 , δ I1∗ = G αβ δgαβ , (74)

From Eq. (59)(1) we then find

3.3. Extraction example: incompressible Neo-Hookean materials

The Neo-Hookean strain-energy per reference volume in this case is defined as

3.4. Extraction example: Saint Venant–Kirchhoff model

We now consider the Saint Venant–Kirchhoff model,

4.1. C 1 -continuous discretization

N̂e (ξ, η) = Ceξ B(ξ ) ⊗ Ceη B(η), (89)

δaαβ = δxTe NT,α N,β + NT,β N,α xe ,

4.3. FE force vectors

Substituting Eqs. (93) and (94) into Eq. (21) gives

where n el is the number of elements and

G eint = δxTe feintτ + feintM ,

G eext = δxTe feext0 + feext p + feextt + feextm ,

4.4. Edge rotation conditions

gc := cos α0 − cos α = 0, ∀x ∈ L, (102)

cos α0 := N · Ñ, cos α := n · ñ. (103)

gs := sin α0 − sin α = 0, ∀x ∈ L (104)

4.4.1. Penalty method for edge rotations

4.4.2. Lagrange multiplier method for edge rotations