Flows in the Upper Region of Recovery Boilers, 1991 Forest Products Symposium Proceedings

The upper portion of the recovery boilers built in the US has changed dramatically in the late 80’s. Use of screens as the first heating surfaces ceased and the typical boiler size increased. This prompted Ahlstrom Machinery to conduct a study in the flow and heat transfer of the upper portion of recovery boilers, that included three MSc theses. Two of these studies concentrated on modelling the upper portion of a recovery boiler. Similar studies had been performed earlier, but they concerned boilers of only half the size and of different superheater configuration.

In the present studies, experiments with a 1:15 plexiglas water model were carried out. To verify experimental results, the same experiments were modelled using a 3D CFD code called FLUENT. Most runs used a grid of 33*37*20 = 24420. The grid size used in the computational model should have been finer, but a compromise between convergence rate and the time frame for the study had to be made.

The pressure drops in the physical model and computational model were compared. The pressure drop predicted by the FLUENT model were higher. This was probably due to the fact that the superheater panels were modelled using only one element row.

One of the main problems with the upper section of the recovery boiler is the modelling of the flow recirculation zone above the nose. When comparing results obtained with the Plexiglas water model to 3D flow simulation, some of the known deficiencies of k-e turbulence modelling were;

1. As the turbulence intensity and the dissipation rate change in the recirculation zone, the true velocities are difficult to obtain.

2. Modelling the flow with the FLUENT code without internals (superheaters) gave better results than with internals.

3. The recirculation zone was smaller in the k-c model than in the physical model.

Of major interest in these studies was the minimum allowable nose free flow area. Free flow area ratios were varied from 0.4 to 0.6 (distance nose -to- front wall divided by distance front-to-back wall) and a rounded nose was compared to a bullnose. One result of trying to squeeze the superheaters closer to the front wall is that the recirculation flow above the bullnose will severely lower the superheater heat transfer.