Numerical and analytical study of steady state and transient heat transfer in liquid filled dead legs

Abstract

A dead leg is an inactive part of a process pipe, and it has been the cause of serious incidents due to liquid reaching its threshold value (TV) for freezing. The existing industrial approach for the design of dead legs is based on rules of thumb and reference standards. Accordingly, a detailed understanding of the temperature development is not available. An objective of this thesis is therefore to develop a Computational Fluid Dynamic (CFD) model in OpenFOAM that can analyse such cases in a more comprehensive matter. This study is based on investigations of the existing theory and industrial approach. This is further used to develop a realistic CFD simulation. The results are obtained from two dead leg geometries. One geometry presents the dead leg as a Fin, while the other geometry presents the dead leg connected to a main pipe, creating a T-junction. The dead leg dimensions are based on rules of thumb, namely a diameter of 2" and a length of 0.5m. This gives a length to diameter ratio (L/D) of approximately 9. The results are obtained from three cases. Where the Fin geometry is related to Case 01 and the T-junction is related to Case 02 and Case 03. Case 01 and Case 03, are purely related to heat transfer, both steady-state and transient. The results of these two cases differs with approximately 10%, indicating that the fin analysis may be sufficient when the ALARP principle is taken into account. In addition, an analysis is performed to investigate the effect of wind, which is observed to have a significant effect on the time for water to reach TV. Case 02 introduces a flow field and turbulence in the T-junction. It is observed that for a range of velocities, u=1-2,86m/s, the circulation in the dead leg reach L/Di = 4-6 for the respective velocities. This results in a stagnant fluid in the end of the dead leg as the circulation cease to exist. This effect is reflected in the temperature development, the result implies almost no heat loss until L=Di = 4.5-6.8 where the temperature falls rapidly. Consequently, the effect of circulation implies that a design criterion may be established from further studies. Verification of the CFD-models has been performed by the use of grid independence test, existing theory and previous experimental results by Habib et al.