CFD model to be used for feed distributor

R.Sripriya

We are trying to solve biased feed distribution in our washeries using CFD techniques. At our coal washeries, the run-of mine coal after passing though a series of crushers is crushed to -13 mm. The -13+0.5 mm is treated in Dense media cyclones and the -0.5 mm in froth flotation cells. the number of dense media cyclones depends on the amount of coal to be treated. In the present case which we are studying we have two banks of 4 cyclones to which the coal gets distributed from a feed distributor. The feed distributor is basically a cylinder into which the coal falls from the top. there are two tangential water inlets which carry the coal and distribute it through eight vertical cylindrical outlets at the bottom. the outlets are placed at 45 degree from each other.the main problem has been the distribution of the coal water slurry through the outlets. the flow through the distributor is turbulent and the vol% of coal in the slurry is 15%. the coal coming out of the outlets vary in quantity and also in the size distribution of the coal they carry. as a result there is variation in the quantity and size distribution of coal which the cyclones treat , thus affecting their efficiency. We have taken up the problem of identifying the causes for variations in feed quantity and size coming out of the distributor and suggest solutions for the same. We would like to try different alternative designs and study the variation in flow. we are trying to simulate the flow of water and coal at the plant conditions using fluent and see the distribution. We have made the geometry in gambit, meshed it and exported the mesh to fluent. We need your inputs on the following What kind of a turbulence model should we use. We have come across a paper where the vof model is used followed by DPM for particle trajectory tracking We tried the standard kE model and specified the backflow turbulent intensity and hydraulic dia. We have noticed that there is a reverse flow in the outlets. The outlets look more inlets than outlets with more than 95% of the flow being upwards., When we try using the k,E options then this problem disappears. Why is this so? As I have mentioned the size range through the distrbutor is very huge. Is it possible to study the flow of whole size range at a time or has it got to be done for small size ranges separately as in the case of hydrocyclones. If it has to be done for individual size ranges separately then how do we take care of the ununiform size distribution

R.Sripriya