Sieve analysis: how to read a grading table and a failing result

The lab sends back a sheet with nine sieves and a column of numbers, one of them circled. The site wants to know whether to stop the paver. The plant wants to know which deck to look at. Both answers are on it.

A sieve analysis reports, sieve by sieve, the percentage of the sample by weight that passed through. You read it by setting each figure against the lower and upper limit for that sieve in the specified grading. Inside both limits on every sieve is a pass. Outside on any one is a fail, and where it is outside is the useful part.

What percent passing means

The sample is dried, weighed and shaken through a stack of sieves, coarsest at the top. Percent passing a sieve is the weight of everything that got through it, divided by the weight of the whole sample. So 71 percent passing 22.4 mm means 71 percent of the sample, by weight, is finer than 22.4 mm. The method is IS 2386 (Part 1): 1963, the part that covers particle size and shape.

Each sieve is finer than the one above, so percent passing can only fall or stay level down the column. A figure that rises is an arithmetic mistake.

Retained, cumulative retained and passing

A lab sheet often carries three columns, and mixing them up is the commonest reading error. Percent retained is what stayed on that one sieve. Cumulative retained is the running total held on that sieve and all the coarser ones above. Percent passing is 100 minus the cumulative retained.

Say nothing stays on the 53 mm sieve and 3 percent of the sample stays on the 45 mm sieve, so 97 percent passes it. A further 26 percent stays on the 22.4 mm sieve. Cumulative retained at 22.4 mm is 3 plus 26, which is 29 percent, and percent passing is 71. MoRTH grading tables are written in percent passing, so that is the column to compare.

A grading envelope has two sides

A grading table gives two numbers for most sieves, such as 25 to 40 percent passing 4.75 mm. Plot the lower and upper limits against sieve size and you get two curves with a band between them. That band is the envelope. Below the lower limit on a sieve is too coarse at that size. Above the upper limit is too fine.

Passing is not the same as being safe. A blend one point inside the upper limit on three sieves will be outside on the next sample, because no two samples are identical. A plant aims for the middle of the band.

Reading a result against the WMM table

Here is the wet mix macadam envelope from the MoRTH Specifications for Road and Bridge Works, Fifth Revision (2013), Table 400-13, with a result beside it. The result is invented for this page to show the reading. It is not a lab report.

Read it from the top. The coarse end is sound: nothing oversize, and the 22.4 mm figure is in the middle of its band. From 4.75 mm downwards every figure is over the upper limit or within a point of it. That is not three separate failures. It is one: the blend carries too much of the fraction below 4.75 mm, and the excess shows on every sieve under it. The full clause is in WMM under MoRTH Clause 406.

What each kind of failure points to

Mesh wears at the edge of each opening, so apertures grow slowly and the product creeps coarser before a wire finally breaks. A torn panel is easier to spot: oversize appears overnight. What wears on a crusher covers mesh alongside jaw plates and liners. Blinding is the reverse fault. Damp fines plug the openings, and material that should have dropped through rides over with the larger size.

Sampling errors that look like plant faults

A stockpile is not uniform. As material falls from a conveyor the coarse stone rolls to the toe and the fines stay in the core, so a shovel from the edge reads coarse and one from the middle reads fine. Neither is the stack. A fair sample is taken from several points and depths, combined, mixed and reduced to test size by quartering or a riffle box.

Size matters too. In a small sample of coarse material a handful of 40 mm stones is a large share of the weight, and five or eight of them in the shovel moves the top of the column by several points. Dust clings to damp stone, so ask whether the fraction finer than 75 micron was found by washing. A dry-sieved figure for that sieve reads low.

How often grading is tested

MoRTH Table 900-3 sets the minimum: one grading test per 200 m3 of aggregate for WMM and one per 400 m3 for granular sub-base, whose six envelopes are in the GSB gradings under Clause 401. A plant that waits for those acceptance tests is running blind between results, so on our crushing plants samples are pulled at commissioning and at intervals through the run, and go to whichever lab the contract nominates. What comes off those decks is on the materials page.