Degrees of freedom in solids versus shells and beams: a critical difference

In Finite Element Analysis (FEA) modelling, one of the deepest and hardest-to-spot mistakes for a novice analyst comes from assuming that every node in a model behaves the same way. It is very common to see an engineer clamp the face of a 3D solid by constraining all three rotations, or connect a 1D beam section directly to a 3D solid block, only to discover unstable results, disproportionate deformations or fatal solver errors afterwards.

The cause is not a software bug, but a fundamental kinematic mismatch in the degrees of freedom (DoF) of each family of finite elements. Understanding why solid nodes have no rotational stiffness, and how that affects constraints and mixed connections, is essential to guarantee the physical stability of an elastic model.

1. The kinematic mismatch: 1D, 2D and 3D face to face

A degree of freedom (DoF) represents an independent direction in which a node can translate or rotate. The mathematical formulation of an element's stiffness matrix strictly determines how many DoF each of its nodes has:

• 1D beam elements: 6 DoF per node (3 translations u_x, u_y, u_z and 3 rotations θ_x, θ_y, θ_z). They directly transmit axial forces, shear, bending moments and torsion through their end nodes.
• 2D shell elements: usually 5 or 6 DoF per node (3 translations + 2 or 3 rotations). The rotations let them resist out-of-plane bending and twisting.
• 3D solid elements (HEX bricks and TET tetrahedra): only 3 DoF per node, the three translations (u_x, u_y, u_z). The nodes of 3D solid elements have no rotational degrees of freedom.

You can go deeper into each family in the articles on trusses and beams, shell elements and 3D solids.

2. Why constraining rotations on a solid node has no effect

A frequent conceptual mistake in pre-processing is to select a row or a set of nodes on a 3D solid element and apply a boundary condition that imposes θ_x = 0, θ_y = 0, θ_z = 0.

The local stiffness matrix of a 3D solid is obtained by integrating the translational displacement field (u, v, w) over the element volume, so there is no term that relates a nodal moment to a nodal rotation at that point. In other words, for the equations of a solid node the rotational stiffness is zero:

If you ask the solver to constrain the rotations of a solid node, it will simply ignore the condition or raise an "inactive degree of freedom" warning: in that node's system of equations there is no θ variable to fix. It is the same issue we saw when discussing compatibility between elements and constraints.

3. The "artificial hinge" phenomenon (unwanted mechanism)

If the nodes of a 3D solid neither rotate nor transmit moments directly, how does a solid block of steel resist bending?

Bending is not produced by nodal rotation, but by a pair of opposing translational forces (tension in the upper fibres and compression in the lower ones) spread over several nodes through the thickness or section. A 1D beam, by contrast, transmits the moment directly through a single nodal rotation θ.

How is an involuntary "hinge" created?

This behaviour causes two serious methodological failures in practice:

1. A false clamp on a single layer of nodes: if you try to clamp a solid part by constraining the translations of a single row of nodes along an edge, the part is not clamped: it pivots freely about that line as if on a hinge (cylindrical joint), with rigid body motion or an unrealistically massive deflection.
2. Direct 1D-3D connection with no connector: if you join the end node of a beam (6 DoF) to a single node of a solid mesh (3 DoF), the beam transmits translational forces to the solid, but the solid node cannot absorb the bending moment. The beam pivots freely about the contact point like a pin or ball joint, and the moment continuity of the structure is lost.

4. Safe mixed connections: RBE2 and RBE3 couplings

To connect 1D beams or 2D shells to 3D solids without creating unwanted joints, you must use kinematic coupling or interpolation elements of the R-Type family (RBE2 and RBE3). They act as an interface that translates the 6 DoF of the master node (moments and rotations included) into a field of forces and translations applied to a set of solid nodes distributed over the contact area, or footprint:

• RBE3 (interpolation, recommended): transfers the force and moment of the beam node by distributing equivalent translations over the footprint of solid nodes through static equilibrium, without adding artificial stiffness to the component.
• RBE2 (rigid connection): ties the beam node to the solid nodes by imposing infinite stiffness. It simulates a perfectly rigid weld or an undeformable spreader plate, but must be used with care so as not to lock local deformation.

5. The golden rule for supporting and connecting each element type

To avoid unstable mechanisms or wrong over-constraints, the analyst should apply this practical guide according to the mesh topology:

The 3 golden rules of pre-processing

1. Never rely on a single solid node to resist moments: an isolated solid node cannot react to a rotation or a bending moment.
2. Use the thickness in solids: to clamp a solid without RBE connectors, fix the translations of at least two or three continuous layers of nodes through the depth, so that the tension/compression reaction couple can form.
3. Always distribute mixed connections: when coupling beams or shells to solids, project the DoF with an RBE3 onto a surface equivalent to the profile cross-section, instead of joining nodes at a single point.

Conclusion

Guaranteeing the kinematic compatibility of DoF at joints and supports prevents artificial hinges and ensures that the solver works on a numerically stable and physically coherent stiffness matrix.