Assess Status and Potential Improvements To Accuracy of On-Line Gauges

An ACARP commissioned study into process control in coal preparation plants in 1993 suggested that there exists widespread dissatisfaction within the Australian industry that such units have not lived up to their early promise, and have failed to deliver the expected accuracies and savings in laboratory costs. The current study was initiated in response to this feedback, and is a joint effort under the direction of C Clarkson & Associates Pty. Ltd., with a major sub-contract to the JKMRC, and consulting advice from a number of CSIRO staff.

The project comprises two major components:

1. A comprehensive industry survey of the current state of the art involving detailed telephone surveys to all users of on-line gauges and eight site audits. With a 94% response rate, the results of this survey can be considered to fairly represent the true industry situation.
2. A detailed site trial of PGNAA (Coalscan 9500) and DUET (Coalscan 2600) installations in tandem, to identify and quantify sources of error in sampling, laboratory analysis, and the gauges.

Ash Gauge User Survey
The final report on the ash gauge user survey includes summaries of responses on some sixteen issues which are presented graphically for the most common installed gauges at the time of publishing (1995): Coalscan Models 1500, 3500 and 4500. The implications of these responses are discussed, and there is an overview of the findings in relation to each gauge type. Although experience with other units was limited, some user comments are presented.

Two major conclusions were that in most cases, gauge accuracy has not lived up to user expectations, but that
user satisfaction is, on average, moderate-to-high.

Major Case Study: PGNAA and DUET at South Blackwater
Comprehensive statistical analysis of the results from a series of experimental campaigns was used to assess the performance of a PGNAA gauge (Coalscan 9500) and a DUET gauge (Coalscan 2600) in the rail loadout system at South Blackwater Coal Ltd. Three objectives were achieved:

A quantitative estimate of gauge performance for a range of parameters (ash, moisture, specific energy, sulphur, elemental analysis) was provided over a range of coal sources and types (washed coking product, washed thermal product, raw open cut and raw underground). The performance of the automatic sampling system, manual sampling and laboratory preparation and testing procedures was measured using a variety of statistical tools and methods including replicate analysis, Grubbs analysis, and Gy's sampling principles. The test work and subsequent analysis provided a case study for use of the various statistical techniques and methods for assessing gauge and sampling accuracy, and optimising gauge calibration.

Key conclusions were that the gauge achieved a 1SD error of 0.32% ash on coking coal, and 1.29% ash on the full range of washed and raw coals. With allowance made for calibration and identifiable electronic errors, these errors could be as low as 0.23% ash and 0.67% ash respectively. For moisture, the gauge achieved a 1SD error of 0.18% for coking coal, approximately half that of the automatic and manual samples. Moisture results for thermal coal were compromised by a replaced moisture electronics board part-way through the test work, but for the subset of tests before the replacement, the error was 0.40% moisture.

Project Deliverables
The major deliverables of the project discussed thus far are in the form of technical reports on each project phase:

1. Report on Ash Gauge User Survey
2. Audits of Ash Monitor Installations
3. Major Audit Report: Results of South Blackwater Test Work

As the project advanced, a great deal of information was amassed regarding sampling, statistics, and gauge calibration. To enable efficient transfer of this knowledge base to ash gauge users within the coal industry, the following documents have also been produced:

4. Guide to Sampling and Statistics (an introductory guide)
5. Principles of Sampling and Statistics (a detailed textbook)
6. Guide to Gauge Selection and Calibration

he project has also seen the development of a series of statistically sound software tools to assist site technical staff in assessing sampling system and gauge performance, and developing optimum calibration maintenance procedures. These tools are Microsoft Excel based, and include a User's Manual complete with worked examples from the project results:

7. Sampling Statistics ToolkitGrubbs Estimator

• Variograms
• Sampling Error
• Heterogeneity
• DUET Gauge Theoretical Model
• DUET Gauge Calibration
• Statistical Function Library

8. Sampling Statistics Toolkit User Manual

The complete set of raw data and calculations from the South Blackwater test work are also included in Microsoft Excel workbooks:

9. Raw Data and Calculations Spreadsheets

To ease distribution, the project deliverables listed above are available on a single CD ROM, with all reports and documents in Microsoft Word format. Hard-copies of documents are also available, and are adequate for those wishing only to review the findings and conclusions. Those wishing to use the Sampling Statistics Toilkit will need to purchase the CD.