Characterisation of Pressure (Hyperbaric) Filter Performance

Hyperbaric disc filter technology has been used to dewater fine coal concentrates in Eastern Europe and China for over 30 years. A pilot scale investigation was undertaken in 2016, which examined the suitability of this technology to dewater Australian fine coal flotation concentrates. This earlier work was part of the previous stage of this project which identified that the hyperbaric disc filter technology could produce a filtercake product of around 14% (a.r.) free moistures, and around 8% (a.r.) if steam was used to supplement the pressure filtration process. These moisture levels were achieved for on a - 0.25 mm coking coal flotation concentrate. Parallel to the pilot testing program, laboratory bench testing was undertaken on the same feed material using an established vendor testing procedure. An important finding from stage 1 was that it verified that laboratory bench testing analysis the test rig results were similar therefore enabling the hyperbaric filtration technology performance on a pilot scale and by extension for a commercial installation to be estimated from the laboratory bench testing analysis.

The aim of this stage of this project was to use this established correlation between laboratory and pilot scale testing outcomes from the earlier work, to develop correlations by the analyse of 9 individual flotation concentrate samples from across Queensland and New South Wales for their response to laboratory hyperbaric filtration. Multivariable linear regression analysis was then performed on the data which enabled the development of model equations that can be used to size a hyperbaric filter and estimate the expected total moistures based on simple coal quality information.

Model equations are presented in the report for porosity, cake resistance, solids filtration rate, cake free moisture and air rate. The correlation coefficient (R2) values for each of the model equations are also provided in the report.

The developed equations were validated with additional sources of data, as well as with vendor sizing data to compare the accuracy of the model equations. It was found that there was a good agreeance between the methods.

Limitations of the model equations include:

• Feed solids percent assumed to be thickened to be within 31.5 - 33.3% w/w;
• Average solids RD assumed to be within 1.30 - 1.45 RD;
• No safety factors have been applied to the model equations;
• Limit the porosity equation to output values between 0.25 and 0.50; and
• Particle sizing to be similar to tested samples (i.e., fine).