Cyclone Agglomeration

The separation of micro- and nanoscale particles from a gas stream is becoming increasingly important. The centrifugal forces in a cyclone are generally not suitable for separating such fine particles, but separation is possible if many of these particles form aggregates, "agglomerates". In the literature agglomeration is accepted as a separation mechanism for fine particles in sedimentation processes in general and to some extent in cyclones too. A certain degree of agglomeration happens in most cyclones and many researchers think that it is responsible for the `fish-hook` in the grade efficiency curve of a cyclone. On the one hand it improves total separation efficiency, but on the other hand it impairs classification. Only by gaining a better understanding of this effect can we manipulate it as required. It is for these reasons that we want to investigate agglomeration in a gas cyclone and the underlying mechanisms and forces in a two-year project. In the experimental part of the project we will use the existing cyclone test rig with the cyclones which were used in previous projects. Extensive PDA and LDA measurements are here available. We will try to collect agglomerates out of the flow, to position them on different substrates and analyse them by scanning electron microscopy (SEM). We will see how many elementary particles comprise an agglomerate, what the particle size distribution of the agglomerates and of the elementary particles is (some authors think large particles collect small particles). We will document at what place and frequency the agglomerates appear in the cyclone and whether they are of different types. We will try to describe the shape and the structure of the agglomerates as fractals. The shape and the structure of the particles can help to elucidate the mechanisms of agglomeration and to find out which kind of force makes the particles adhere to each other (long and/or short range forces). In the theoretical part, CFD-calculations will be performed with commercial software. These are required in order to confirm earlier LDA-measurements, and thus to determine location and direction of the probes and to select the proper suction velocity for isokinetic sampling. The main purpose of the CFD-calculation, however, is to simulate particle motion by an Euler-Lagrange approach. In this simulation we will incorporate a particle collision and an agglomeration model. From the knowledge about the properties of the agglomerates (see above) we can derive data which are helpful in selecting characteristic parameters, like density and maximum size of the agglomerates. Such data are needed to perform realistic calculations.

The separation of micro and nanoparticles from a gas flow is increasingly gaining importance. According to existing theories cyclones are not suitable for this. However, in practice such small particles (at least a fraction) can in fact be removed. The effective mechanism is agglomeration and this is not taken into account in existing theories. Agglomerates of up to several thousand small particles behave like large particles and can thus be separated. To date no theoretical or experimental proof has been provided for the occurrence of agglomeration. The goal here was thus to provide experimental proof of the existence of agglomerates as well as to develop a mathematical model for agglomerate formation in a cyclone. In the experiments conducted agglomerates were extracted from a cyclone and precipitated on a membrane micro filter. These agglomerates were classified according to three types: Type 1: A large particle of about 10 m diameter, referred in the Literature as a collector particle has attached to it a number of smaller particles of about 1 m diameter. Formations of this type are to be expected as they comply with Loffler`s agglomeration theory. Type 2: These agglomerates consist of small particles (= 3 m) which are lined up like a chain of beads. These formations may be up to 30 m long. Type 3: Type3 agglomerates are large, roundish formations, consisting of thousands of medium sized and small particles. Their diameter may be up to 30 m. Agglomerates of types 2 and 3 were unexpected. The mechanism of their formation is unknown. The experimental results prove the existence of agglomerates in a cyclone and support the hypothesis of small particle separation through agglomeration. For the mathematical modelling part the commercial CFD software FLUENT was extended to include models for wall contact and collision in flight according to Sommerfeld. The separation of the large particles could be predicted quite well. However, the separation of small particles could not be predicted satisfactorily. It seems that the model for agglomeration does not reflect the actual physical process of agglomeration in a cyclone. An explanation for this could be that nearly all particles are at the wall or close to it so that there is little chance for collision in flight. It may be that direct wall contact is involved in making two particles cling together.