Structural engineers are introduced to the slope-deflection method through two fundamental constants: 4EI/L and 2EI/L. These values underpin textbook treatments of moment distribution, stiffness matrix assembly, and portal-frame analysis. They are elegant, simple, and, within the context of pure bending, mathematically exact.
However, these constants are not comprehensive.
When a column is subjected to significant axial compression, these constants, which represent moment-fixity or end-restraint values, begin to diminish. The member's resistance to end rotation decreases progressively as the load increases. At the Euler critical load, Pcr = π²EI/L², both constants approach negative infinity, resulting in a complete loss of bending resistance. This phenomenon, known as buckling, is characterized not by sudden material failure but by a gradual and mathematically predictable reduction in rotational stiffness.
| Constants 4 and 2 are not universal truths — they are the P = 0 limiting cases of the stability functions Sii and Sij. For any member under significant axial compression, they must be replaced. |
The stability functions Sii and Sij precisely characterize this degradation. Originating from Euler's 1744 column equation and later formalized in matrix form by Livesley and Chandler in 1955, these functions provide the closed-form exact solution to the differential equation governing a beam-column subjected to bending and compression. For any load level, they quantify the rotational stiffness retained by a compressed member and indicate when a frame approaches elastic buckling.
Despite their significance, stability functions are largely absent from undergraduate curricula and receive only limited attention in graduate structural analysis texts. The geometric stiffness matrix [KG], commonly implemented in commercial finite element software, is frequently used as a substitute. However, it represents only the leading-order Taylor series approximation of the exact stability-function stiffness matrix. While this approximation is adequate at low load levels (P/Pe < 0.10), it becomes increasingly unconservative at moderate to high load levels (P/Pe > 0.30), overestimating lateral stiffness by 30 to 190 percent and failing to capture the sign reversal of Sij, which fundamentally alters moment distribution patterns.
This white paper presents a unified and mathematically exact framework for the stability analysis of compressed frame members. The exposition is structured into three self-contained sections.
Part I — Theoretical Foundations (Chapters 1–3) establishes the governing differential equation of a beam-column, derives the stability functions Sii and Sij in closed form, and verifies them against classical limits (c→0, c→π). No approximations are introduced.
Part II — Numerical Behavior (Chapters 4–6) applies the functions to a reference H-400×400 steel column, quantifies the load-by-load degradation of Sii and Sij, exposes the precision sensitivity of hand calculations, and measures the error introduced by [KG] at every P/Pe level.
Part III — Design Application (Chapters 7–9) assembles the theory into a complete frame analysis: a portal frame worked example, the KG-based P-Delta decomposition and its limitations, and the generalized eigenvalue problem det[KE + λKG] = 0 for elastic frame buckling.
Upon completion of this white paper, readers will be able to:
All numerical examples use a single reference member: a KS standard H-400×400×13×21 wide-flange column (EI = 1.365×10⁵ kN·m², Pe = 53,870 kN). Every computed value can be verified independently. The portal frame example in Chapter 7 extends the same section into a complete structural system.
The 280-year progression from Euler's 1744 column equation to contemporary eigenvalue solvers fundamentally centers on the development of stability functions. This paper systematically retraces that evolution with complete mathematical transparency. The objective is to enable readers to understand and critically assess software outputs encountered in professional practice from first principles.