International Concrete Abstracts Portal

The usage of a high speed digital computer in the investigation of long concrete columns as integral parts of building frames is outlined. Extensive use of the computer was made in both the interpretation of data obtained in physical testing and in analytical studies utilizing idealized mathematical models. Numerical procedures were facilitated by development of a rapid and versatile method for obtaining the relationship between axial load, bending moment, and curvature for rectangular reinforced concrete members. A general program is presented which simulates the behavior of a rectangular frame by use of the method of successive approximations and with predictor and corrector functions based on the axial load-moment-curvature characteristics of both the column and its restraining frame members. The method recognizes the nonlinear characteristics of the problem and considers inelastic action, axial load effects, and the varying reduction instiffness of reinforced concrete members. Verification of many of the analytical procedures was obtained in a series of tests of isolated eccentrically loaded long columns under statically determinate load conditions. A series of tests of columns as integral parts of frames indicated that the analytical procedure can predict the mode of failure and type of long column action to be expected. Quantitative accuracy was shown to be reasonable with major discrepancies directly attributable to shortcomings in the failure criteria postulated for reinforced concrete sections. The analytical procedure showed itself to be a promising tool available for further exploration of long column behavior.

The aim of this work is to predict both the average value and the variance of the creep deflection of reinforced concrete beams under sustained loads. Two quite distinct problems emerge, the determination of a probabilistic model to predict the creep behavior of a concrete prism under axial compression, and the introduction of this description of material behavior into an analysis of the bending of a beam under an arbitrary vertical loading. The model of the creep mechanism of concrete is a simplified version of an earlier model suggested by one of the authors. Stochastic processes, namely varieties of the Markov birth process, are employed to represent both the viscous flow of the cement paste and the delayed-elastic effects caused by fluids -- water and viscous paste-initially trapped within the elastic skeleton of crystals and aggregate. In a manner similar to that developed by another of the authors for the bending of homogeneous beams of stochastically viscoelastic material, the bending of a reinforced concrete beam is formulated. The creep response of a unit length of concrete to a unit stress is assumedto be a stochastic process of the type presented in the first part of the paper. These arguments lead to the desired results, formulas which predict the mean and variance of the deflection of any point on the beam at any time. In addition, spatial and temporal covariance functions are obtained; the latter permits the engineer to take advantage of an early observation of the creep deflection to alter his prediction of later deflections and to reduce the variance of these predictions.

The chairman of this symposium has asked the writer to prepare a general and critical discussion of inelastic reinforced concrete design. This forced him to study more thanadozenof the preprinted papers, in considerable detail, in an attempt to assess the present state of knowledge and of the art.

The behavior of long restrained concrete columns as part of a building frame is much more complicated than that of long hinged concrete columns under eccentric load. A theoretical analysis for determining the critical column length for long hinged concrete columns has been derived previously by the writer. A method for determining the critical column length for long concrete column as part of a box frame is presented here. A long concrete column may buckle laterally as the critical section of the column reaches material failure; but the material failure of a column cannot be used as the criterion to determine the criticalcolumn length. Plastic hinges may be developed in a frame, but a long column may become unstable without developing plastic hinges. An analog computer was used as a tool to determine the critical column lengthfor the following reasons: (1) The problems involve differential equations which are particularly suitable for analog computer solutions (involving typically about 30 sec of computer time for a solution of adequate design accuracy); (2) the plotter, which is a standard unit of the computer, will plot the column or beam deflection curves on graph paper for visual reference; (3) the programmer can more readily make designdecisions by selection of proper constants for each preliminary trail of the problem. Concrete columns, subjected to eccentric loads at the ends will deflect laterally. As the columndeflects laterally the column moment along the column length will be increased by an amount equal to the product of column load and lateral displacement. This increment of moment becomes very important for the analysis of long columns. As the column deflects laterally, cracks will usually appear at the convex side of the column near the region of maximum moment. The error in using a constant EI (modulus of elasticity x moment of inertia) approximation to determine critical column length may be of substance. In considering variable E and I along the deflected column, moment versus edge-strain relationships was derived for a given column with a given column load. A nonlinear second order differential equation can then be obtained from each moment versus edge-strain curve. An analog computer was used to solve the differential equation and the column deflection curves and angle of rotation curves were plotted on graph paper by the computer plotter for a given column with given column load P. For any given values of end moment ME and the column load P, the critical column length for eccentrically loaded hinged column can be easily determined from the column deflection curves. The long column as part of a symmetrical box frame was further studied. It is assumed that all joints are rigid and that the joints do not move laterally. The end rotation 0E of the column must be equal to the end rotation of the beam, and the end moment ME of the column must equal to the end moment of the beam. For a given box frame with given column and beam loads, the critical column height can be determined. It is found that the co-tangency criterion for determining the critical column length for eccentrically loaded hinged column is not always applicable for determining the critical column length for restrained column.

With discussion by Peter R. Barnard, George Pincus, Charles A. Rich, and Gerald Sturman, Surendra P. Shah, and George Winter. Inelastic behavior of concrete was studied by direct observations of internal microcracking. Thin slices were made from strained specimens and internal cracks were examined by X-ray and microscope techniques. Bond cracks at the interface between coarse aggregates and mortar, exist in concrete even before any load is applied. Analytical and experimental studies showed that tensile stresses are present at the mortar-aggregate interface because of volume changes of mortar and may be partly responsible for bond cracks in virgin concrete. These bond cracks begin to propagate noticeably at applied compression stresses of one-quarter to one-third of the ultimate strength. At this level the stress-strain curve begins to deviate from a straight line. At about 70% to 90% of ultimate strength cracks through mortar begin to increase noticeably and bridge between bond cracks to form a continuous crack pattern. Upon further load increase this condition eventually leads to a descending stress-strain curve and failure. Other investigators have noted that in that same load range, the volume of concrete begins to increase rather than decrease. An hypothesis explaining this volume expansion and propagation of bond cracks in terms of shear bond strength of the interface and microcracking has been presented. In order to investigate the influence of flexural strain gradients, microcracking and the stress-strain relation of eccentrically loaded specimens were compared with those of concentrically loaded specimens, It was found that a flexural strain gradient definitely retards microcracking, especially mortar cracking as compared to cracking at the same strain in axial compression. The stress-strain curve for eccentric compression, which was computed by an experimental-statistical approach was found to differ materially from that for concentric compression. The peak of the flexural curve was located at a strain about 50% larger and at a stress about 20% larger than the peak of the curve for concentric compression. Structural implications of these findings are briefly examined.