The paper describes measurements on the heat transfer from iron and copper pipes of diameters ranging from to 3 inches, at various temperatures up to 90 deg. C. in air moving with velocities up to 30 ft. per sec. when the axis of the pipe is placed transverse to the stream of air. An iron pipe of external diameter inches was also tested when placed longitudinally along the direction of the air stream and subdivided into sections, the heat loss from various sections being measured. This loss is greatest at the section where the air first meets the pipe, and gradually falls off along the pipe. If the air is artificially made turbulent, the loss from all the sections shows an increase, the greatest change being shown by the first section. The heat loss from a longitudinal pipe is of the order of one-half that from a transverse pipe. A number of similar iron pipes of inches diameter were studied when arranged in banks in square formation. In this case the second layer of pipes met by the air loses more heat than the first layer, the third layer losing the same as the second. It is suggested that this is due to the eddy motion set up in the air when it passes the first layer, no further increase in the eddy motion occurring afterwards. In the diagonal formation with rows inches apart, the first section again loses least heat. In this case not only is the loss in the second row greater than that in the first, but there is a further increase in passing from the second to the third row, after which the coefficient is constant. When air impinging on a bank in staggered formation is artificially made turbulent, the heat loss from every row is increased to such an extent that the first row loses more heat under these conditions than the third row when there is no artificial turbulence. In addition, an investigation was carried out with pipes below 0 deg. C. If there is no ice or snow deposit on the pipes, the curve connecting H/kθ with Vd/ v is the same as for hot pipes. Here H is the heat loss per unit length, d the external pipe diameter, and θ the temperature excess, whilst k, V, and v are the conductivity, velocity, and kinematic viscosity of the air. Pipes below 0 deg. C. may become covered with a deposit of ice or snow, and in this case the interpretation of the results is less straightforward. If d and θ in the above formulae refer to the actual pipe diameter and temperature difference between the pipe and the air, then the smooth curve through the observations is again identical, within experimental error, with that for hot pipes. This can only be a coincidence, the insulating value of the deposit being compensated by the greater area presented to the wind, since the effective pipe diameter has become larger, and has a higher surface temperature than that of the metal. From the result for dry pipes, it would be expected that if d′ refers to the overall diameter, and θ′ represents the difference between the air and the outer surface of the deposit, this new curve would coincide with that for hot pipes. The experimental results show that this is so if the outer surface of the deposit is dry, but that if the conditions are such that water is dripping off, the heat transfer may be 30 per cent greater than the normal amount.
Read full abstract