- Title
- Towards an algorithm for the de-convolution of fractionation data
- Creator
- Mackellar, Jason
- Relation
- University of Newcastle Research Higher Degree Thesis
- Resource Type
- thesis
- Date
- 2024
- Description
- Masters Research - Master of Philosophy (MPhil)
- Description
- The production of size and density distributions of mineral samples is an important part of the mineral processing industry. Accurate estimates of distributions can significantly reduce costs. Long-standing methods for the generation of such distributions such as the "sink-float" method use hazardous liquids of high density with negative impacts on human health and the environment. The REFLUX^{TM} Classifier uses fluids in laminar flow and the hydrodynamic properties of particles to separate particles on the basis of size and density. It has been shown that non-toxic solutions such as Lithium Heteropolytungstates (LST) may be used as an effective solution in which to suspend particles in the Classifier and that under this regime, the Classifier does indeed perform effective separations. In this thesis we aim to quantify the observed separation properties of the Classifier. We first develop two differential models for the behaviour of the Classifier in batch mode. By comparing the output of the models to experimental data, we are able to discard one of these models. The remaining model is then used to develop an integral equation to describe the action of the Classifier on feeds of constant density and variable size. The inverse problem (whereby the integral operator is inverted to produce estimates of the size distribution of a feed) is shown to be ill-posed and regularization techniques are employed to deal with this. The predictions of size distributions generated from the inverse model are then compared to size distributions generated by physical sieving of laser sizing data, and under certain assumptions, a reasonable fit to experimental data is achieved.
- Subject
- classifier; particle separation; REFLUX; gravity separation
- Identifier
- http://hdl.handle.net/1959.13/1514798
- Identifier
- uon:56856
- Rights
- Copyright 2024 Jason Mackellar
- Language
- eng
- Full Text
- Hits: 566
- Visitors: 560
- Downloads: 6
| Thumbnail | File | Description | Size | Format | |||
|---|---|---|---|---|---|---|---|
| View Details Download | ATTACHMENT01 | Thesis | 74 MB | Adobe Acrobat PDF | View Details Download | ||
| View Details Download | ATTACHMENT02 | Abstract | 9 MB | Adobe Acrobat PDF | View Details Download |