Selection and Sizing of Autogenous and Semi-Autogenous Mills. Pilot Plant Testwork Basis, страница 5

Figure 2   Secondary mill: calculated specific power consumption vs new feed sizing (from pilot mill primary circuit product Barratt et alia 1999)

For any project, it is important to recognize the differences between industrial-scale and pilot-scale mill circuit operation on the size distribution of the primary circuit product The transfer size (T₈₀), top size, and the percentage of final product in this stream are all affected by the screen aperture and conditions in the primary mill, more particularly in SAG milling with variations in ball charge volume, total mill charge volume, top ball size, and mill speed. The limitations that are prevalent in pilot milling are usually logistic: maximum ball charge volume; pebble port size, discharge screen aperture vs successful pumping of screen undersize (which is fine enough for transfer to a secondary grinding circuit), top size in the mill feed, and fixed mill speed. All of these limitations can be removed in industrial-scale operation and usually are (e.g., 15% v/v balls vs 12% v/v; 19 mm trommel screen aperture vs 9 or 12 mm pilot screen aperture). Generally, coarser apertures for grates and screens will naturally produce a coarser product, whereas ball charge size and composition in the larger mill can sometimes compensate for that. The length: diameter ratio in the larger mill is usually higher than in the pilot mill and, coupled with the vastly higher population of balls, can contribute to a finer product. All of these issues create a balance so that, invariably, the transfer size is not too different from that produced in the pilot mill, especially if the pilot screen aperture (e.g., 12 mm) is not constrained by the logististics of pumping the undersize to a secondary grinding circuit

Another factor which could influence scale-up from pilot plant test results is the comparison of impact forces and residence time hi a pilot mill with those in a full-scale production mill; i.e., "Can the specific power consumption which is derived from pilot plant testwork be relied upon for scale-up in the same manner as that calculated for secondary ball milling using Bond work indices and Rowland's efficiency factors (Rowland 1982)7" The answer, surprisingly, is yes in spite of the fact that the forces involved are proportional to mill diameter and the residence time is proportional to the differences in D:L ratio and inversely proportional to the ratio of the square root of the diameters. The production mill will generally see a coarser top size in the feed and have a shorter residence time than the pilot mill, so the comparison has to be based on the rate of breakage vs applied energy in different size classes. This, of course, is the principle that forms the basis of the Ж Tech comminution simulations (JK SimMet).

Scale-up from pebble milling tests follows the same principles as those for primary milling; i.e., accepting the measured specific power consumption to the required product size distribution and adding a designated contingency. Again, this contingency is dependent upon the stability of the tests and their duration, as well as the stability of pebble consumption rates to satisfy design criteria.

Bench-Scale Testwork Basis

Prediction of the specific power consumption for primary and secondary grinding from bench-scale testwork is, in many respects, still an art and is dependent upon the methodologies that are used. In the context of selecting and sizing semi-autogenous mills for basic engineering and equipment purchase, power-based models are commonly used in situations for which pilot plant testwork is not possible or practical.

In the absence of pilot plant testwork, it is the authors' opinion that power-based models should only be used to predict specific power consumption for autogenous grinding if they can be supported by a database of criteria which is specific to rock strength and known commercial autogenous grinding operations. Such a database would include results from the autogenous media competency test (AMCT), which is available from Amdel (Siddall 1996), analysis of rock strength data (unconfined compressive strength, either measured or scaled up from point load indices, coupled with fracture frequency), breakage rates by size class from Ж Tech drop weight tests and simulations, and Bond low energy impact work indices by size class as part of the AMCT. The database would make a distinction between operations with lower mill speed, high aspect mills which favour impact and those with higher mill speed, low aspect mills which grind more by abrasion (Powell 2002).