Thanks for the comments so far on the turbulence modelling of strong vortex driven flows.

I have to simulate the behaviour of an industrial cyclone which operates with ferrochrome particles in the size range 1-50 micron at a temperature of 340 degC. The cyclone is too efficient and removes nearly all the "coarse" particles and passes only the very fine particles which clogg the bagfilter plant. It is requested to simulate the current behaviour of the cyclone and suggest design changes in order to pass more "coarse" particles through the cyclone.

As such a two phase flow simulation is very computer intensive, I want to carefully consider which modifications to test.

From my limited understanding of cyclone behaviour, which operates strongly on centrifugal forces on particles, I thought of considering the following modifications in order to pass more "coarse" particles through the cyclone:

a) Decrease the inlet velocity which will decrease the centrifugal forces on the particles and allow more coarse particles to pass. This is however not very practical in the plant setup.

b) Reduce the number of spiral paths taken by the air and particles by numerically testing the effect on an cone which would "shorten" the taper part of the cyclone.

c) Shorten the vertical outlet gas pipe which protrude inside the cyclone to enable the "short circuit" of inlet particles directly to the outlet gas pipe. However, I have conducted serveral experiments using a mini-classifier and this option proofed not to be very successfull.

Regards Günther Hasse

I would like to introduce a good reference, which is very classical but my favorite one for cyclone design.

Title : New Design Approach Boosts Cyclone Efficiency Author : Wolfgang H. Koch & William Licht Citation : CHem. Eng., Vol. 7, November, 1977, pp.81-88

According to above mentioned paper, the most dominant parameter for cyclone efficiency is inlet velocity. I do not think that 'decrease inlet velocity' is so difficult in the plant setup. In general, cyclone inlet is reduced from the fairly large upstream pipe(or duct). So, simply speaking, do not reduce so much. You might have refractory problem to modify exisiting plant. However, the cost or effort should be accepted to get your purpose.

As a personal point of view, it is best to widen inlet width(not to tangential side but to center side). It has two effects, one is to reduce inlet velocity and the other is to deviate some of inlet flow from tangential entry, both has the effect to reduce centrifugal force.

Other parameters for decreasing(or increasing) cyclone efficiency can be found in the above reference.

Sincerely, Jinwook

(1). Based on the angular momentum conservation principle, m*r*Vt(r)=constant. (2). Based on the mass conservation, rho*Vt(inlet)*Area(inlet)=constant. (3). We are assuming that at the inlet, the velocity is Vt(inlet)=U, where U is velocity at the exit of the duct (connected to the inlet). (4). So, from the inlet duct side, you have rho*U*Area=constant. It is o.k. to keep U and Area fixed. So the mass flow rate will be the same. (5). Vt(inlet)=U is only one of the conditions you can use. You can use other relationship such as Vt(inlet)=U*cos(theta), where the theta is the angle between the U and Vt(inlet). (6). Since based on the angular momentum conservation of (1), you have determined that Vt(r) is responsible for the performance of the device. You can change the theta angle to reduce the Vt(inlet). Currently, we are assuming that theta is zero. (7). So, the design solution is: install a movable guide vane (vanes) at the inlet (duct exit), so that you can control the theta angle. In this way, the inlet tangential velocity Vt(inlet)=U*cos(theta) can be reduced and controlled by varying the theta angle. (8). You can also write a program to feed the computer with the desirable performance as a function of theta angle to control the guide vane angle theta. Then we can call this "Computer-Controlled-Optimum-Cyclone-Device". (If you decide to file a patent application, put my theory and my name on it also.) Good luck.

I think that John's idea is very good. He's idea is one step above than mine which was posted couple of days ago at this site.

For your reference, swirl number in the cyclone, which controls flow pattern in the cyclone, is defined as

S_N = (3.14/4) * De * Do / A_inlet

Where, Do and De are body and vortex finder(outlet) diameters, respectively.

If you adopt John's idea, the modified swirl number is approximately :

S_N_modified = S_N * cos(theta) : same definition as John's definition for theta.

I think that you can successfully acheive your purpose by adopting John's idea or mine, or both.

If you successfully acheive your purpose in the future, please post the result at this site. I really hope to know the reslut of the appication.

Sincerely, Jinwook