NACA-TN-1051

Abstract

INTRODUCTION <br>In recent years the need for more rational means of design and analysis of connections has been emphasized by the exacting requirements of modern airplane construction. The methods of joint-analysis are far more antiquated than those employed f o r other parts of the aircraft structure. Improved methods for predicting joint strength offer a means of reducing weight if they are adapted to make more efficient use of all connectors within a joint. From the production viewpoint, Jenkins (reference 1) has shown that approximately 50 percent of the total cost of the all-metal airplane frame is due t o connecting the various components of the structure and that the cost of riveting end bolting constitutes between 60 and 90 percent of the total cost of connections. These conditions suggest a promising field for investigation. <br>The well-established methods f o r computing rivet or bolt loads are based on assumptions derived from ultimate-strength tests of a number of riveted joints. Rivets or bolts of the Same size were thought to carry equal loads because the ultimate strength of the tested joint was approximately equal to the strength determined from the ultimate strength of a single rivet multiplied by the number of rivets in the joint. The fact that rivets or bolts in a structural joint do not generally carry equal loads in the elastic range was recognized as early as 1867 (.reference 2). Batho (reference 3) demonstrated that a riveted joint is a statically-indeterminate structural system and that the rivet loads may be obtained by the principle of least work. Hrennikoff (reference 2) developed equations for rivet loads for a limited number of joint arrangements from a consideration of the deformations of plates and rivets. Posner (reference 4) developed a general bolt- o r rivet-load equation for lap joints based on the deformations of the plates in tension and in bearing under the bolts or rivets Several investigators have made use of equations derived by means similar to those just mentioned hut have obtained factors for rivet behavior by tests of various joint arrangements (references 1, 2, pp. 464-469, and 3). Some investigators have dealt with butt joints (references 2, pp. 464-69, and 5), some with lap joints (references 1 and 4, and some with both (references 3, 2, 5, and 6). A brief history of early investigations is given in reference 2, pp. 474-484. <br>The present paper deals with tile problem of load distribution among the bolts of symmetrical butt joints. Tests were conducted to determine experimentally, both within and above the elastic range, the manner in which load was distributed among the various bolts. The test specimens were doubly symmetrical two- and three-bolt joints made of 24S-T aluminum-alloy plates joined by a single line of &#188; inch alloy-steel bolts. Reference to a joint having a certain number of bolts means that the total joint load is imposed on that number of bolts. Analytical expressions, based on elastic action of' the joint components, are given whereby the bolt loads may be computed and the experimental and analytical results are compared.The important question of joint action above the limit of elastic behavior is most readily treated from the standpoint of empiricism. Principles, based on the test results, are suggested, but such principles can be extended for general application only when appropriate empirical data are available.