Comparison and convergence studies are conducted with two non- linear finite element beam-column models. Both models admit physical and geometric nonlinearities.

The models differ in their respective strain field representations. Model 2 permits a linear longitudinal and transverse strain variation. The strain field of model 3 can vary linearly in the transverse direction and quadratically in the longitudinal direction.

The system model and solution process used with both element models are identical. The system model is defined by an energy function. The equilibrium path of the system is determined at discrete load levels. At each load level, application of the solution process yields the new equilibrium configuration of the system.

Four test problems are selected to determine the relative merits of models 2 and 3. The models are tested over the entire range of response.

The tests indicate that both models predict accurate responses throughout the range of response. Mesh refinement is seen to increase the accuracy of the models' response predictions.

The models achieve identical response predictions in the linearly elastic range. The inelastic response predictions of model 2 are seen to be more accurate than those of model 3. Model 2 is computationally more efficient than model 3 over the entire range of response.

The results of the studies reveal that throughout the range of response, the efficiency of response of model 2 is greater than that of model 3.