Books
Matrix Structural Analysis is the stiffness method you already run
McGuire, Gallagher, and Ziemian's Matrix Structural Analysis — what the second edition teaches, what to skip if you only need linear elastic frames, and why OpenSees is not a different theory.
Recommendation
Recommended for
- engineers who use frame analysis software and need to know what it assembles
- readers implementing or checking elastic models in OpenSees or similar
- graduate structural analysis after a first stiffness-method course
Not recommended for
- a first statics or strength-of-materials course
- continuum finite-element research (solids, plates as a specialty)
- readers who want a steel-connection detailing manual
What problem this book solves
Every linear frame program you have used — including OpenSees with elastic beam-column elements — is the direct stiffness method with a solver and a postprocessor. McGuire, Gallagher, and Ziemian wrote the book that still explains that method as engineering, not as a GUI. The second edition (John Wiley & Sons, 2000; later reprints exist, including a 2015 print associated with MASTAN2) adds geometric nonlinearity and the solution of nonlinear equilibrium at the level a frame analyst needs, which is why it survived the undergraduate “matrix methods” paperback.
If you cannot write the 12x12 (or 6x6) elastic stiffness of a prismatic member in local coordinates and transform it, you are not checking the model. You are hoping.
Who it is for
Structural engineers who analyze frames and trusses for a living, and students in a second structural-analysis course. Especially useful if you are about to trust P-Delta, notional loads, or a “nonlinear static” pushover whose geometric stiffness you have never seen on paper. Not written for architects, and not a substitute for a steel or concrete design specification.
Required background
A first course in structural analysis (determinacy, moment distribution or an equivalent, at least qualitative member behavior). Linear algebra at the level of solving K u = P and knowing what a singular stiffness matrix means physically (mechanism or missing restraint). Strength of materials: axial, bending, and shear deformation ideas. Dynamics is not required here; that is Chopra. This book stops at forming and solving the static (or sequentially static) equations.
Chapter map
The second edition keeps the first edition’s aim — matrix methods as used in practice — and then does the thing most undergrad texts refuse: geometric nonlinearity for framed structures.
You get kinematics and stiffness of truss and beam-column members, global assembly, boundary conditions, and the usual recovery of member forces. Virtual work is not decoration; it is how consistent loads and the later nonlinear terms stay honest. The nonlinear chapters treat equilibrium on the deformed geometry and the solution of the resulting nonlinear equations (incremental/iterative ideas, not a research monograph on arc-length variants).
MASTAN2, developed with Ziemian, is the companion computational environment historically packaged with the text. Use it as a check against a hand-assembled two-member frame, then against OpenSees on the same elastic system. If the three disagree, the bug is yours.
Best chapters
Member stiffness and transformation until they are mechanical, not memorized. Then assembly and restraints until a rigid-body mode is something you can provoke on purpose. Then geometric stiffness and P-Delta in the second-edition nonlinear material, because that is where commercial “second-order” buttons hide.
What to skip, depending on the goal
- First elastic OpenSees model. Do not skip local stiffness, transformation, and assembly. You may defer geometric nonlinearity and nonlinear solution until the linear elastic model matches a textbook portal frame.
- Steel designer who only needs code P-Delta. Still read the geometric-stiffness chapter once. Then your specification’s notional-load and B1/B2 (or equivalent) clauses will look like approximations with names, not magic. You can defer mixed interpolation and research-level beam-column elements.
- Continuum FEM (solids, shells as a career). This is the wrong primary book. Use it for frames, then Bathe, Hughes, or Zienkiewicz for continua. The FEM chapter in Chopra is also only an introduction.
- Dynamics. Mass, damping, and time integration are out of scope. Pair this review with Chopra after K is trustworthy.
Theory quality
High. The book is careful about work theorems, coordinate transformations, and the difference between material and geometric nonlinearity. It will not inflate a 2D beam into a brick element. That restraint is a virtue.
Practical value
High for anyone who signs frame analysis. Practical here means: you can reconstruct what the solver did, you can explain a warning about an unstable structure, and you can tell a P-Delta run from a material-nonlinear run. Low if you expected connection design, slab punching, or BIM workflows.
Programming and computational value
High relative to most civil textbooks. The stiffness method is an algorithm. Implementing a 2D truss and then a 2D frame in a small script is the correct homework, even in 2026. OpenSees is the production-shaped version of the same assembly. A language model can emit Python that looks like a stiffness method and still transpose the rotation matrix. Check against a two-member hand solution before you trust either the LLM or your own code.
How it compares
- McGuire, Gallagher, and Ziemian versus Kassimali or Hibbeler matrix chapters. Those undergraduate chapters get you to K u = P. This book is what you read when that is the starting line.
- Przemieniecki, Theory of Matrix Structural Analysis. Classic, more aerospace/continuum flavor. Keep as a second reference.
- Bathe, Finite Element Procedures. The general FEM bible. Overkill as the first frame book; essential if you leave line elements.
- Chopra, Dynamics of Structures. Dynamics after this, not instead of this. OpenSees elastic models need both: McGuire for K, Chopra for M, modal analysis, and time integration.
Final recommendation
If you run frame software, this is not optional reading in the long run. Use the second edition to make the stiffness method boring, then add geometric nonlinearity before you trust second-order flags. Validate with a tiny elastic OpenSees model of a structure you have already solved by hand. The related dynamics book on this site is Chopra; the related programming lesson is that the solver is the method, not a chat transcript.