Face sampling is the cutting of pieces of ore and rock from exposed faces of ore and waste. The faces may be natural outcrops or faces exposed in surface trenches and pits or in the backs, walls, and floors of development openings and stopes. Face samples may be taken:
- by cutting grooves or channels of uniform width and depth across the face or sections of the face or
- by picking off small pieces all over the face, more or less at random.
The former method is the more systematic, and although ordinarily it is the one that is employed when accuracy is desired, pick sampling may sometimes be as accurate, or more so, depending on conditions.
Whichever method is used, the face should be properly prepared before actual taking of the samples is begun. Good preparation involves thorough cleaning of the face to remove adhering particles of dirt and ore minerals, and frequently some smoothing off is required to remove loose slivers and lumps or sharp projections of solid rock and ore. Washing with water under pressure generally will remove adhering particles, but in some instances this may have to be supplemented by scrubbing with a stiff brush, particularly if the face is a floor.
In channel sampling, the object is to cut a channel of uniform width and depth across the face in order to obtain in the samples equal weights of material from equal lengths cut. This objective can only be approximated in actual practice, but when a large number of samples is taken, errors usually are compensating if each sample is so taken as to approach the ideal as nearly as possible.
The weight required from each inch or foot of channel will depend on the uniformity of mineralization, the size of the mineral particles, and the richness of the ore; the less erratic the mineralization, the smaller the particle size, and the lower the grade the smaller the weight necessary to give a satisfactory sample. The length of each sample cut is carefully measured and recorded. Usual practice is to take about 1 to 2 pounds per linear foot of cut, although in very hard ground it may be difficult to reach this minimum. Sometimes considerably larger weights are required to obtain reliable samples.
In hard ground channel samples usually are cut by hand with hammer and moil, but sometimes with moils fitted to a light hammer drill. In soft ground they may be taken with a sample pick. The channel should be marked on the face with a crayon or lamp black before starting to cut the sample.
Various devices have been employed to catch the cuttings, such as bags held open by a metal ring and fitted with a handle, but the usual practice is to spread a tightly woven, smooth-faced square of canvas or other material on the ground below the point being sampled. In cutting samples from hard ground, it is often difficult to catch flying chips in this manner, and material belonging in the sample may be lost unless great care is exercised. Obviously, this loss may be reduced by using larger squares of canvas.
In channeling bedded or banded deposits, the long dimension of the channel cuts should be at right angles to the banding (see fig. 7, A), and usually it is desirable to sample each band or bed separately unless the hardness and the mineralization are uniform. If there are bands of decidedly different hardness or toughness, the material from each band should constitute a separate sample, because it is practically impossible to obtain equal weights per unit of length from each band. With erratic mineralization, it is usually best to limit the lengths of individual samples to 4 or 5 feet. In sampling gold ores, the usual practice is to exclude all particles of visible gold from the sample.
In sampling drifts, crosscuts, and raises, it is common practice to cut a channel sample or set of channel samples at regular intervals of 5, 10, or 25 feet along the opening, depending on the uniformity of the ore. A regular measured interval between sampling points should be used to eliminate the human equation in selecting them. Incidentally, a regular interval simplifies calculation of the grade of ore reserves later on.
Pick sampling consists of chipping off pieces from points well-scattered over the face. The ideal method would be to divide the face into measured squares and take a chip from each common corner,
but such a refinement probably is rarely justified since in any event it is very difficult to obtain equal weights of material from the different points. In banded faces, pieces may be cut at points as shown in figure 7, B, the chips from each band constituting a separate sample.
Although chip sampling is not usually considered as scientific and accurate as channel sampling, it may give very reliable results in some ores.
Accuracy Of Face Samples
Channel sampling probably is the method most widely employed in mine-examination work, and it may be inferred from this that, broadly speaking, experience has demonstrated it to be the most accurate under the widest range of conditions. There are exceptions to this, however, as at Miami, where it has already been noted that channel samples return assays that average 13 percent too high, whereas stoper-drill samples are more accurate.
A few typical examples of face-sampling practice and the results obtained are summarized below.
At the Hollinger mine, Timmins, Ontario, the grade of ore reserves is estimated principally from assays of channel samples. The gold ore occurs in vein systems related to definite fractures within a main shear zone in schisted lavas intruded by masses of quartz porphyry. A crew of 45 men in charge of a head sampler is employed in underground sampling. A sampler will cut an average of 40 to 45 feet of channel in a shift. Routine channel samples are cut from all development and stope faces after each round, with check samples across the backs of development headings and slashes when required. The samples are cut with hammer and moil and are caught on canvas; about 1 pound of sample is obtained per linear foot of channel. Samples are cut normal to the banding; and each separate band, whether it be ore or waste, is sampled separately, its length being measured for proper weighting when results are combined. Over-breaks into waste in the stopes are sampled likewise, to provide a basis for estimating the average grade of all rock broken. All erratic high assays are cut arbitrarily to $50 before they are entered in the calculations. A high degree of precision is indicated by the experience over a long period of years, as shown by the following figures:
Tons milled………………………………………………………………27, 500. 000
Value as estimated from sampling……………………………………………………$208, 071, 740
Value actually in mill feed…………………………………………………$206, 465, 707
Total………………………………………………………………….$1, 606, 033
Per ton……………………………………………………………………$0. 06
Channel samples from development headings average 12 to 20 percent higher than the ore actually stoped, but this is due to dilution in stoping, the fact that the drifts usually are kept in the best ore as far as possible, and to the deliberate mining of some ore of marginal grade.
At the Teck-Hughes mine in the Kirkland Lake district, Ontario, all development faces, drift backs and crosscuts, and stope breasts and backs are channel-sampled. High-grade streaks and bands of different hardness are sampled separately, as are bands of waste. Test holes are drilled into the walls, and the sludges from 2-foot sections are collected and assayed. The channel samples over a large area have averaged about 10 percent lower than the actual value. Visible gold is rejected from the samples before assaying.
At the Wright-Hargreaves mine in the Kirkland Lake district all faces are pick-sampled after each round is blasted. Pick samples are taken (fig. 7, B) across each band, the material from each band constituting a separate sample. The width of each band is measured normal to its dip, and the widths are used for weighting the values in estimating the average grade of the face. This method has been found to check very closely with results from regular channel samples.
Bradley has given some results of moil samples of gold ores from the Alaska-Juneau mine in Alaska.
Methods of sampling at the Lucky Tiger mine, Sonora, Mexico, have been discussed by Mishler and Budrow. The vein occurs along fractures in volcanic rock and the ore is narrow (1.1 to 1.9 feet, average). Stoping width averaged 3.4 feet. The ore was a high-grade silver ore with some gold associated with sphalerite, galena, pyrite, chalcopyrite, tetrahedrite, and stromeyerite in a gangue of kaolinized or silicified rhyolite.
All drifts, raises, and winzes were sampled at 5-foot intervals, each sample across the back was divided into as many samples as there were varieties of ore and waste, and the exact width of each was noted. Backs of stopes were sampled in the same manner. Following are some results of sampling along the Tiger vein. Odd-numbered samples were on one side of the drift and even-numbered samples on the opposite side.
Over a period of 14 years, estimates of ore reserves based on sampling averaged 34.0 ounces silver per ton, whereas the ore mined during the same period averaged 37.1 ounces.
The following is abstracted from a discussion of sampling at the United Verde mine by Quayle. Faces usually are sampled either by chipping with a pick or, less commonly, by cutting channels. The chips are taken in four lines across the face for the width of the sample. The average sample weighs about 4 pounds, and individual pieces have a maximum diameter of 1 inch. A sample covers a width of 5 feet, and the height averages 7 feet. A stope sample of a face after a 6-foot round has been blasted represents about 20 tons of ore.
Sampling errors vary with the type of ore. In massive sulfides, the ore mineralization is uniform, and the error is very low—the samples average about 2 percent too high. In schist and porphyry ores, the occurrence of the ore minerals is very erratic, and the sampling error has ranged from 8 to 20 percent. For 1928 the average error for all classes of ore between the stopes and the smelter was 5 percent.
The foregoing examples indicate roughly the degree of accuracy that may be achieved by face sampling. Unfortunately, the published data on the accuracy of this method of sampling are limited. There are many articles describing face-sampling practice, but in most instances the accuracy is not indicated either because the samples are used for stoping-control purposes or for direction of development work only, and waste faces, rock from which is mixed with ore going to the mill, are not sampled. Thus, the dilution in stoping is not measurable, and there is no check between the mine and mill sampling. In other words, the rock going to the mill is not the same as that sampled in the mine. Many base-metal ores are readily distinguishable from the waste material by their appearance, and no cut samples are taken in the stopes. Where ore and waste cannot be distinguished readily, as is the case with most gold ores, many oxidized base ores, and some pyritic copper ores, channel or pick sampling provides a very satisfactory method of stoping control and is employed widely for this purpose; but unless the waste faces as well as the ore faces broken in stoping are sampled, no check between mine and mill sampling is possible.
In calculating the grade of ore reserves, the engineer often must base his estimates solely or largely upon face-sampling of development openings and must make allowances for dilution in stoping and for other factors.
At the Ground Hog mine channel samples are cut with hammer and moil 3 inches wide and about 1 inch deep across the vein at 5-foot intervals. The raises are not sampled. Experience indicates that estimates based on these samples are too high in grade and too low in tonnage, because the sublevels are driven close to the hanging wall, where the grade of the ore is highest, whereas in stoping considerable low-grade ore is mined from the footwall side. About 25 percent more ore is mined than is estimated, and the grade is correspondingly lower. The ore contains silver, copper, lead, and zinc.
At the Verde Central mine, face samples are taken of all drifts and raises as they advance. Channels are cut across the vein at the back of the drift at 10-foot intervals, and raises are sampled in the same manner. Channel samples are taken by hand with moil and chisel or with a stoper machine. Dilution factors employed in estimating grade of ore depend on experience and usually are greater for narrow than for wide stopes.
In one stope 5 feet wide the factor was 20 percent, whereas in another 20 feet wide and having good walls it was 5 percent. In one 15 feet wide, wherein andesite dikes made up about 10 percent of the area, a factor of 25 percent was employed. Chalcopyrite is the chief valuable mineral, and it occurs with quartz and pyrite in lenses in a larger mineralized zone.
At the Vipond mine, where gold occurs in quartz veins and irregular masses of mineralized schist impregnated with quartz, channel samples cut from each working face in development headings and stopes were used for control of stoping and direction of development, whereas grab samples from muck piles were used as a basis for ore-reserve estimates.
At Elkoro, development headings were sampled after each round by cutting channels across the faces or backs. In stopes, cuts comprising three to five samples were made across the entire stope at intervals of 5 feet throughout its length. The grade, as estimated from the development samples, averaged about 17 percent high, owing largely to dilution in stoping. The valuable metals are gold and silver, which occur in veins with sulfides, quartz, and adularia.
In general, face samples may be said to afford a method of sampling within the range of accuracy required for control of stoping operations and direction of development headings, and when carefully done give a basis for estimating grades of ore reserves when proper allowance is made for dilution in stoping and for other experience factors.