|
Designing and Analysing Laboratory Experiments and Plant Trials
Minerals engineers are often required to do experiments and to analyse the results from those experiments. They may range from simple laboratory tests to major plant trials lasting several weeks and costing hundreds of thousands of dollars. Examples include:
- Laboratory trials of a new flocculant
- Laboratory grinding and flotation tests of new ores
- Pilot plant tests for flow sheet design
- Plant trials of a new flotation reagent
- Plant trials of a new circuit configuration or item of equipment
In each case, data are collected to allow some decision(s) to be made. It is important to arrive at the right decision, in the shortest possible time and at the lowest possible cost. This is often difficult to achieve because mineral processing data are usually imprecise and, especially in the case of plant data, subject to uncontrolled trends, cycles and variations, which make comparisons difficult. The figure below shows the daily gold recovery from a copper-gold concentrator. An improvement of 1% in recovery would be well worthwhile, but very difficult to detect against a daily variation of over 20%.
This course introduces the minerals engineer to simple statistical methods which allow wise decisions to be made in the face of this kind of uncertainty. Topics covered include:
- The nature of error
- Precision and accuracy
- Confidence limits
- Comparing samples
- The t-test and the F-test
- Sample size
- The chi-squared test and contingency tables
- The Analysis of Variance (ANOVA)
- Sampling schemes in chemical analysis
- Regression and comparison of two regression lines
- Design of experiments
- The randomised block design
- Factorial experiments
- Conducting plant trials
A full set of notes is provided, plus Excel spreadsheets on disk for many of the methods discussed. A wide range of numerical examples taken from real mineral processing case studies is used to illustrate the methods described. Tutorials allow course attendees to develop their analytical skills; questions and answers provide a library of additional case studies for future reference.
The course is intended for metallurgists, chemists and other professionals, technical staff and students concerned with the planning and analysis of laboratory experiments, assay data and plant trials will find the course a valuable and relevant addition to their skills. Knowledge of elementary statistics is useful but not essential. Much more important is a sense of enquiry and a real desire to run better, more decisive and more cost-effective experiments and plant trials. A sense of humour always helps.
Students are encouraged to bring along case studies and problems with which the group can work.
The Course Leader: Professor Tim Napier-Munn has been making sense of mineral processing data for 30 years. He has extensive experience in mineral industry statistics and experimental design, both as a practitioner and as a teacher. He has been giving the course for over 25 years to practicing engineers and to undergraduate and postgraduate students on five continents. He also consults to mining companies and reagent suppliers in the design and analysis of plant trials, and has published a number of papers on the topic.
Tim Napier-Munn is a minerals engineer with bachelors and PhD degrees from Imperial College, London, and a Masters degree from the University of the Witwatersrand, Johannesburg. He worked for De Beers in South Africa for many years, latterly as Manager of the Diamond Research Laboratory Mines Division. He lectured at the Royal School of Mines in London and, in 1985, joined the Julius Kruttschnitt Mineral Research Centre in Brisbane from where he retired as Director in 2004. He now works in project development for the University of Queensland's Sustainable Minerals Institute. He is a Fellow of both the Institution of Mining and Metallurgy (London) and the Australasian Institute of Mining and Metallurgy.
| 
