follows. The subject of the presented research are large deformations of an arbitrarily
curved and twisted BE beam with an anisotropic solid cross section, without warping [1].
The nonlinear SR beam model has long been the main focus for researchers, partially
because its spatial discretization requires only C0-continuous basis functions, such as the
Lagrange polynomials. As the name suggests, the SR theory was founded by Reissner [2],
and later generalized by Simo [3], who conceived the term geometrically exact beam the-
ory. The main requirement of a geometrically exact formulation is that the relationship
between the configuration and the strain is consistent with the balance laws, regardless of
the magnitude of displacements and rotations. The adequate description of large rotations
is one of the principal challenges since these are not additive nor commutative and consti-
tute nonlinear manifolds. This issue has been a driving force for the formulation of various
algorithms for the parameterization and interpolation of rotation [4, 5, 6, 7, 8, 9, 10]. A
turning point in this development was the finding by Crisfield and Jeleni´c that the in-
terpolation of a rotation field between two configurations cannot preserve objectivity and
path-independence [11, 12]. The reason is that incremental material rotation vectors,
at different instances, do not belong to the same tangent space of the rotation manifold
[13]. An orthogonal interpolation scheme that is independent of the vector parameteriza-
tion of a rotation manifold is suggested in [11, 12] and several further strategies followed
[14, 15, 16].
Although the geometrically exact formulations represent the state-of-the-art in beam
modeling, their implementation is not straightforward and several alternatives exist, such
as the corotational and the Absolute Nodal Coordinate (ANC) approaches. The main idea
of the corotational formulations is to decompose the deformation into two parts. The first
part is due to large rotations and the second is the local part, measured with respect
to the local co-rotated frame. It resembles the strategy employed in [11, 12] and allows
accurate simulation of large deformations [17, 18, 19, 20]. The ANC method is, in essence,
a solid finite element for slender bodies. It is well-suited for the implementation of 3D
constitutive models, but has issues with engineering structural analysis, where integration
with respect to the cross-sectional area is required [21, 22].
The first BE beam models that are consistent with the geometrically exact theory are
[23] and [24]. Meier et al. have discussed the issues of objectivity and path-independence
in [25], and proposed an orthogonal interpolation scheme similar to that of Crisfield
an Jeleni´c [11, 12]. Membrane locking, contact, and reduced models are considered in
subsequent publications [26, 27], followed by a comprehensive review [1]. An efficient BE
beam formulation based on the Cartan frame was developed in [28], where the position
and the local frame are observed independently and subsequently related by the Lagrange
multipliers.
The emergence of the spline-based isogeometric analysis (IGA) [29] has led to the de-
velopment of a series of SR beam models [30, 31, 32, 33, 34, 35, 36, 37]. The formulation
[38] arguably represents the state-of-the-art since it employs extensible directors and mod-
els various couplings. One of the main features of IGA is the smoothness of utilized basis
functions, a property that benefits the BE beam due to its C1-continuity requirement.
The first IGA BE beam models were introduced by Greco et al. in [39, 40, 41, 42], while
the first nonlinear BE model was developed in [43]. Due to the reduction in number of
DOFs, in comparison with the SR model, multi-patch nonlinear analysis of BE beams has
received special attention [44, 45, 46, 47]. Invariance of the geometric stiffness matrix in
the frame of buckling analysis is considered in [48], while the effect of initial curvature on
2