Matrix Stiffness Method

The Matrix Stiffness Method, often called the Direct Stiffness Method, is the computational engine behind nearly all modern structural analysis software. It provides a systematic, matrix-based procedure to determine the displacements, internal forces, and support reactions of complex structures, transforming what would be an intractable hand-calculation into a set of linear algebraic equations a computer can solve. Mastering this method is not just about learning a procedure; it’s about understanding the fundamental language of computer-aided structural engineering, forming the essential foundation for Finite Element Analysis (FEA).

The Building Blocks: Element Stiffness Matrices

The entire method is built upon the concept of an element stiffness matrix, denoted as $[k]$. This matrix fundamentally relates the forces at an element's nodes to the corresponding displacements. Its coefficients represent the force required at a specific degree of freedom to produce a unit displacement at another, while all other displacements are held at zero.

For a truss element (axial force only), which has two nodes and two degrees of freedom (axial displacement at each end), the stiffness matrix in its local coordinate system is straightforward. For an element with axial stiffness $EA/L$, where $E$ is the modulus of elasticity, $A$ is the cross-sectional area, and $L$ is the length, the local stiffness matrix is:

$$[k{]}_{truss}=\frac{EA}{L}\left[\begin{array}{cc}1& -1\\ -1& 1\end{array}\right]$$

This compact matrix states that a unit displacement at one end produces an axial force of $EA/L$ at that end and an equal but opposite force at the other end.

For a beam element (bending only), the complexity increases as it must relate moments and forces to translations and rotations. A 2D Euler-Bernoulli beam element has two nodes, each with a vertical displacement (translation) and a rotation, totaling four degrees of freedom. Its local stiffness matrix is a 4x4 matrix:

$$[k{]}_{beam}=\left[\begin{array}{cccc}\frac{12EI}{L^3}& \frac{6EI}{L^2}& -\frac{12EI}{L^3}& \frac{6EI}{L^2}\\ \frac{6EI}{L^2}& \frac{4EI}{L}& -\frac{6EI}{L^2}& \frac{2EI}{L}\\ -\frac{12EI}{L^3}& -\frac{6EI}{L^2}& \frac{12EI}{L^3}& -\frac{6EI}{L^2}\\ \frac{6EI}{L^2}& \frac{2EI}{L}& -\frac{6EI}{L^2}& \frac{4EI}{L}\end{array}\right]$$

Here, $I$ is the second moment of area. Row 1, for example, gives the shear force and moment at node 1 due to unit displacements at each of the four degrees of freedom.

Assembly: Constructing the Global Stiffness Matrix

Individual elements exist in their own local coordinate systems. To analyze the entire structure, we must assemble them into a unified global stiffness matrix, $[K]$. This process involves two key steps.

First, each element's stiffness matrix must be transformed from its local coordinates to the global coordinate system of the entire structure using a transformation matrix $[T]$. The global element stiffness matrix is computed as $[k{]}_g=[T{]}^T[k][T]$.

Second, these transformed matrices are added, or "assembled," into the master global stiffness matrix. This is done through a process called direct stiffness assembly. Each element contributes its coefficients to specific rows and columns in $[K]$ that correspond to its connected nodes and their global degrees of freedom. The size of $[K]$ is $n\times n$, where $n$ is the total number of unrestricted degrees of freedom in the structural model. The assembly process enforces compatibility, ensuring that connected elements share the same displacement at their common nodes.

Applying Boundary Conditions and Solving the System

The assembled global system represents the structure before it is connected to supports. The fundamental equation is $[K]\{D\}=\{F\}$, where $\{D\}$ is the global displacement vector and $\{F\}$ is the global force vector containing applied nodal loads.

However, $[K]$ in its assembled form is singular—it cannot be inverted because the structure is free to undergo rigid-body motion. Boundary conditions, which represent physical supports (pins, rollers, fixed bases), must be applied to remove this singularity. For a pinned support preventing translation, the corresponding displacement degree of freedom is set to zero. The most common and numerically stable technique is the penalty method or partitioning method.

In the partitioning approach, the equation is reorganized to separate known displacements (often zero at supports) from unknown displacements. The system reduces to: $$[K_{ff}]\{D_f\}=\{F_f\}-[K_{fr}]\{D_r\}$$ Here, subscripts $f$ and $r$ denote "free" (unknown) and "restrained" (known) degrees of freedom, respectively. $[K_{ff}]$ is now invertible. Solving this reduced system, typically via Gaussian elimination or similar matrix solvers, yields the vector of unknown displacements $\{D_f\}$.

Determining Member Forces and Support Reactions

Once the global displacements are known, the solution propagates back down to the element level. For each element, the displacements at its nodes in global coordinates are extracted and transformed back into the element's local coordinate system. The local member forces (axial force for trusses, shear and moment for beams) are then calculated using the original local stiffness matrix $[k]$: $$\{f\}=[k]\{d\}$$ where $\{d\}$ are the element's local displacements. This gives you the final internal forces.

Support reactions are calculated using the portion of the global equation that corresponded to the restrained degrees of freedom: $\{R\}=[K_{rf}]\{D_f\}+[K_{rr}]\{D_r\}-\{F_r\}$. This computes the forces the structure exerts on its supports, which are equal and opposite to the reactions.

Common Pitfalls

1. Incorrectly Applying Boundary Conditions: A frequent error is insufficiently restraining the structure, leaving it kinematically unstable (still singular), or over-restraining it, which introduces fictitious supports. Correction: Before assembly, perform a quick stability check. Ensure the model has the minimum number of restraints to prevent rigid-body translation and rotation in 2D or 3D space as required.

1. Ignoring or Miscalculating Coordinate Transformations: Assembling element matrices directly without transforming from local to global coordinates is a critical mistake for elements that are not aligned with the global axes. Correction: Always remember that assembly occurs in the global coordinate system. For any inclined member, you must compute and apply the transformation matrix $[T]$ to its local stiffness matrix before adding its contributions to $[K]$.

1. Confusing Local and Global Member Forces: Interpreting the forces from the solved global $\{F\}$ vector as member internal forces is incorrect. That vector contains applied loads and reactions. Correction: Member internal forces must be calculated in the element's local coordinate system using the procedure in the section above. The global solution provides displacements; internal forces are a subsequent, element-level calculation.

1. Poor Numerical Conditioning with Large Stiffness Differences: Using extremely large numbers (like $10^9$) to simulate fixed supports in the penalty method can lead to numerical rounding errors and an ill-conditioned matrix $[K]$. Correction: Use a partitioning approach if possible, or ensure consistent units (e.g., kN and meters) to keep matrix coefficients within a reasonable numerical range relative to each other.

Summary

• The Matrix Stiffness Method is a systematic, computer-friendly procedure for analyzing statically indeterminate structures by solving $[K]\{D\}=\{F\}$.
• It begins by formulating element stiffness matrices $[k]$ for each member type (truss, beam) that define their force-displacement relationship in local coordinates.
• These matrices are transformed to a global coordinate system and assembled into the global stiffness matrix $[K]$, which represents the entire unsupported structure.
• Boundary conditions are applied to $[K]$ to make it invertible, allowing you to solve for the unknown nodal displacements.
• Finally, displacements are used to compute local member forces and support reactions, completing the analysis and providing all necessary design information.