In a subsea facility, the pipelines will be surrounded by sea water at 4◦C, which will actively cool the production fluid inside. If temperatures drop below a critical value, water and gas will form unwanted solids called hydrates. Dead legs are inactive parts of production pipelines occupied by stagnant hydrocarbons. These areas pose a major hydrate formation risk, and needs to be insulated based on a prior heat transfer analysis. If a dead leg contains access points for e.g remotely operated vehicles, these areas need to be kept uninsulated, and will act as cold spots. Due to internal natural convection, these cold spots will potentially influence the temperatures throughout the system, and it is therefore crucial to predict the degree of influence. In this master thesis, experimental and numerical heat transfer analysis was conducted on a T-shaped plexiglass pipe, representing a production header with a vertical dead-leg. The header was insulated, while the dead-leg was uninsulated and carried a cold spot on top. In the first of two experimental phases, water was circulated through the header at constant flow rate, mimicking steady state production. In the second phase, the flow was enclosed and the water was cooled down over a period of 3 hours. During both phases, internal and external temperatures were measured with RTD s and thermocouples respectively, while velocities in the dead leg were measured using PIV. It was shown how the mean velocity field rotated periodically in a clockwise and counter-clockwise manner during both phases. A numerical model was created in Workbench, and simulations were carried out using RANS with a k − ω SST formulation in CFX. Temperatures were correctly predicted for 3 hours of cool down, by modelling the cold spot as an isothermal wall, even though simulations failed to recreate the periodic mean velocity field observed in the experiment.