Aggregate sampling and sieve analysis

Why aggregate sampling and sieve analysis are two halves of one job

The only way to know the real grading of the material a plant produces is to measure it, yet the accuracy of that measurement is limited by how representative the sample is, not by the resolution of the balance. Aggregate sampling and sieve analysis therefore has to be treated as a single chain: a badly taken sample gives a wrong answer even in a perfect laboratory, and that wrong answer then goes on to spoil the plant settings. The most common field mistake is a single bucket shovelled from the easy-to-reach flank of a stockpile. That flank is one of the least trustworthy places in the material, because coarse grains roll down the slope as the pile is built and collect at the bottom.

Where the sample comes from

The most reliable point is material in motion. Cuts taken at the discharge end of a belt conveyor with a container or a sampling scoop that crosses the full width of the stream are more representative than any sample lifted from a pile surface. If the belt can be stopped, the best method is to scrape off everything between two frames set a defined distance apart along the belt. When sampling from a running stream, one cut is not enough; several cuts taken at equal time intervals through the shift are combined into what is called a composite sample, which captures the variation over time.

Where a stockpile cannot be avoided, three rules rescue the job: take from at least five points at different heights and orientations, shovel the surface layer aside at each point and take from the material beneath, and combine all increments into one container. If the sample has to come from a truck body, level the surface and dig into separate zones of the bed.

Mass reduction by quartering

The composite sample arriving from site is usually many times the mass required for the test, and it cannot simply be scooped down. Two accepted routes exist for reduction: a riffle divider, or quartering. Quartering means coning and flattening the sample on a clean hard surface, splitting it into four equal quarters, discarding two opposite quarters, combining the remaining two, and repeating until the required mass is reached. When the material tends to segregate, using a sheet-metal or timber divider to separate the quarters is markedly more reliable than eyeballing the split with a shovel.

Test portion mass follows the maximum grain size

These figures can also be reproduced as roughly D divided by ten, squared, in kilograms; for 32 mm that arithmetic gives about 10 kg and for 16 mm about 2.6 kg. Where particle density sits clearly above the usual range, the mass is increased so that the test volume stays comparable. Cutting the test portion short is the most frequent deviation and it distorts the coarse end directly: on 32 mm material a 2 kg test portion lets the presence or absence of one single grain on the top sieve move the percentage by several points.

The limit on sieve loading

A second limit applies during the analysis itself: when sieving ends, the mass retained on any one sieve must not exceed the sieve area multiplied by the aperture and divided by two hundred. Exceed that and the upper sieve blinds, the finer fractions cannot pass, and the reported grading turns artificially coarse. The remedy is to sieve the portion in two or three batches instead of one.

Calculation and a typical result table

After sieving, the mass retained on each sieve is weighed, the fines in the pan are added, and the total is compared with the starting mass. If the loss is clearly more than one per cent, the analysis is repeated. The percentage passing is then computed for each sieve. On a 32 mm product tested with a 10 kg portion the result looks like this:

There is no single number to read out of a table like this; what gets read is the shape of the curve. In the example half the material lies below 16 mm and the fraction below 0.063 mm stays under half a per cent. Those two end values are what decide which application the product class actually suits.

How the result feeds back into plant settings

A sieve analysis is not a report, it is a setting instruction. If the curve comes out coarser than expected, the first place to look is the screen itself: torn or blinded media, inclination and stroke, and the feed load across the deck. If the fines fraction is higher than expected, the closed side setting of the tertiary stage and the fullness of its feed come under question. An unexpected dip in the middle fractions means the settings of two consecutive stages do not overlap. The effect of deck selection and aperture on the product curve is quickest to observe on a conventional vibrating screen, because changing a deck shifts the break points of the curve directly.

Keeping the record usable

Four items belong at the head of every analysis: the point the sample came from, the time and shift, which stages were running at that moment, and the source of the feed material. Without those four, curves cannot be compared. Two samples taken from the same stockpile a week apart will differ, and what allows the difference to be explained is not the curve but those four lines. Once the measuring discipline exists, sieve analysis becomes the shared language of mining engineering decisions from quarry planning through to product specification.

If you want to settle which sieve series and what sampling frequency suit your own product classes, we can map the measuring points onto your existing flow diagram together.