Welded tubular T-connections as used in offshore structures are investigated by means of three-dimensional photoelastic models. Photoelasticity is shown to provide good resolution even in areas of high stress gradients. Results verify work of previous investigators of steel models. The study also shows high stress concentrations at the intersection weld result from localized bending of the chord wall. This bending causes maximum stress to occur slightly outboard of the weld and not at the weld as popularly believed.
This is the first reported study of the application of three-dimensional photoe1asticity to elastic stress distributions on the surface of tubular connections. The study is restricted to T-connections since they provide a simple configuration on which to develop experimental techniques and, at the same time, generate usable data.
The term "welded tubular connection" refers to that portion of a structure where tubes are joined by fusion welding of the cross section of one tube (brace) to the undisturbed exterior surface of another tube (chord), Figure 1. Tubular connections are extensively used in the construction of offshore platforms. Lee in 1968 estimated that 2,000 fixed platforms had been installed in the Gulf of Mexico by the oil industry. The majority of these structures contain tubular connections.
Within the oil industry, two techniques have been used for investigating the structural integrity of tubular connections to assure safe designs; namely, theoretical studies and experimental tests. Theoretical solutions of elastic stress distributions have not gained widespread acceptance because 6f the simplifications necessary to obtain solutions due to the complexity of the problem (Bij1aard 1955a, 1955b, 1959). However, experimental analyses have been accepted and are responsible for the advances which have been made in the design of tubular connections. A complete survey of experimental analyses was provided by Toprac (1968, 1969) under the auspices of the Welding Research Council.
Most of the experimental analyses of tubular connections were performed on steel models. Stress determinations were obtained from strain gage measurements taken at selected points along the external surface of the model. Placement of strain gages on the areas of highest stress requires an a Driori knowledge of the stress distribution. Furthermore, considerable difficulty is experienced in placing the gages sufficiently close to the intersection weld to provide satisfactory strain resolution. Finally, the forces necessary to deflect even a moderately sized steel connection are large. Thus, experimental analysis of steel models is difficult and requires relatively 'large and rigid test jigs.
On the other hand, photoelasticity provides a continuous pictorial display of stress distributions. Areas of high stress gradients can be discerned easily from low stress areas without laborious analysis. Also, only small loads are required to adequately strain a plastic photoelastic model.
