MECHANICAL ANALYSIS OF SAND AND GRAVEL The methods of mechanical analysis now in common use were first used at the Lawrence Experiment Station about 1890. A description of them, as then developed, appeared in the Report of the Massachusetts State Board of Health for the year 1892. In thirty years practical experience has accumulated, committee reports have been presented, and some improvements in procedure have been made, but there has been no change in basic method. Information as to the sizes of the sand-grains is obtained by mechanical analysis. The sand is separated by sieves into portions having grains of definite sizes, and from the weights of the several portions the relative quantities of grains of any size are computed. SAMPLES There are numerous ways of taking sand samples. A sample to represent the run of a sand bank can be taken by drawing the point of a shovel up a vertical face so as to take a small portion of the material as it goes. A number of such moist samples are placed in a pail and thoroughly mixed. Small portions of sand from a washing machine may be put in a pail at regular intervals and mixed in the same way. In sampling sand in a car or barge a long butter tester is convenient for getting samples representing a considerable depth. The procedure must be varied to meet the requirements of the business in hand. Samples are best kept in their natural moist condition in handling and shipping, as there is then no tendency to separate into portions of unequalled sized grains. If the sand by any accident becomes dry such separation is inevitable, and it is then difficult to get a representative portion for analysis. Quartering, as in assaying, is frequently the only method that can be used. Eight ounce wide necked glass bottles are convenient for samples that are to be kept and compared. Small cloth money bags can be sent by mail or express without breaking, and the description can be written on the bag with a soft pencil. They should not be used where there is very fine material or dust. Small pastebroard containers are convenient for samples not to be permanently kept. Duplicate samples for comparison after obtaining the results of analysis are often useful. From half a pound to a pound is enough for a sample of filter sand, but for gravels and sand containing large particles larger quantities are necessary. WEIGHING A small "prescription" scale weighing 100 grams and sensitive to one-tenth gram is convenient, rapid and sufficiently accurate. A quantity of moist sand, such that after drying there will be approximately 100 grams, or such other quantity as is to be used for analysis, is put in a porcelain dish and dried. After drying, the sample is transferred to the scale pan for weighing. If it weighs slightly over the desired amount the excess may be removed and the weight adjusted without appreciable change in the composition of the sand. Otherwise the exact amount weighed is used whatever it may be. After sifting, the various portions of particles, beginning with the finest, are transferred from the bottom pan and from the various sieves to pieces of smooth paper to which the grains do not adhere and in turn are put upon the scale pan and weighed. The weight after each addition is entered on an appropriate record sheet. When all the portions have been brought back the original weight should be reached. If 100 grams were taken the final result should check within 0.2 gram. If a greater shortage is indicated there is some defect in the procedure or the sand was not thoroughly dried. Per cents of the original weights other than 100 grams, are computed by slide rule. DRYING The sample in the porcelain dish must be completely dried before the analysis proceeds. A close regulation of heat is not necessary. Exposure to sunshine in a hot dry climate may be sufficient. A steam radiator or an electric heater will serve or a kerosene stove may be used. With ordinary sands no damage is done if the temperature is carried considerably above the boiling point and this will insure immediate and complete drying; but the sand should not be baked. SIFTING The dried sand is put in the top of the nest of sieves, each fitting into the one below with a cover at the top and a pan at the bottom. The sieves may be shaken by hand for occasional field tests, but generally a shaking machine is employed. The machine frequently used is geared so that 150 turns of the handle give 700 single shakes to the sieves. This is enough ordinarily. The speed must be such that the sand is drawn back and forth over the wire cloth. If it is too slow the necessary movement will not take place. If it is too fast the grains will be thrown back and forth and will not pass the meshes. An experienced operator can tell by the sound what is going on. Test runs with the cover off will be helpful to a beginner. The sifting machine may be attached to an electric motor or other power. In this case it is essential that the proper speed be maintained; and a counter is desirable to show when the necessary amount of shaking has been given. It is sometimes stated that the process should be continued until no more sand passes. This is incorrect. The shaking should be sufficient to get the bulk of the sand that will pass through the openings, and it should be uniform to give comparable results. A few larger holes must be expected in every sieve, and if the shaking is continued long enough everything smaller than these accidental holes will pass. In ordinary work double the amount of shaking will not make more than a few tenths difference in the per cent passing; and this may be taken as a reliable criterion of sufficiency of shaking. The amount of shaking to be used in practice should be the same as that used when the sieves are rated as described below. CLEANING SIEVES After shaking, the material is emptied to a sheet of paper, and the sieve, while bottom side up, is briskly struck with the hand several times to loosen grains caught in the wire mesh. A mesh that is square cleans more easily than one that is longer in one direction; and that is the reason for insisting on a square weave in the wire cloth. Sometimes pressure with a soft rubber on the reverse side will help. Not all particles can ever be loosened, but usually the number that remain stuck in the meshes are not enough to affect appreciably either the present analysis or the size of separation of the sieve in subsequent work. AMOUNT OF SAND TO BE USED With sieves about 5 inches in diameter, 100 grams is the greatest amount to be used. This amount of bank sands and filter sand having uniformity coefficients (a term to be presently defined) of 2.5 and over is to be used. Where most of the grains are small and uniform in size, 50 grams or less should be used. With very fine and even grained sands, 20, or even 10 grams in special cases, will be better. In case of doubt the matter should be tested either by putting the same sample through the sieves a second time, giving it double the amount of shaking, or by repeating the analysis with a smaller quantity. If the results deviate from those first reached, the matter must be followed until it is certain that final results have been reached. An experienced operator can usually tell by the looks of a sample when it is desirable to use less than 100 grams, and with care to reduce the quantity when needed, the standard amount of shaking is almost always sufficient. Another way is to say that when more than 20 grams are found on any one sieve the quantity should be reduced. CEMENTING Sand sometimes contains clay or other cementing material that cakes on drying and will not break up in sifting. Slight cementing may be broken by a rubber pestle used in the porcelain dish before sifting. Cementing is determined by inspection. Good filter sand does not cement. Materials that cement strongly call for special methods which will not be discussed here. PLOTTING THE RESULTS The cumulative or mass diagram is best used for plotting, and the analysis should always be stated in terms of it. The analysis may be tabulated under the two headings: Size of separation Per cent by weight of sample finer than size The method of reporting the results of analysis as the separate per cents by weight between sieves or between other limits should never be used. Various kinds of cross section paper have been used. The kind of paper is not of first importance, but a well selected arrangement will facilitate the work and the accuracy of the results. Three kinds of work sheets deserve notice. 1. Logarithmic The scales are usually not the same in the two directions. This gives a good representation of the finest particles, but is less satisfactory for the coarser ones. 2. Mixed logarithmic In this the size of grain scale is logarithmic all the way and the per cent by weight scale is logarithmic up to 30 per cent and arithmetic above. The two scales are computed so that the differentials are the same at the point of change, and there is then no roughness in the resulting plotting. This form of plotting has been most generally used. It has all the advantages of the logarithmic scale and gives a better representation of the coarser particles. 3. Logarithmic probability paper It is an interesting discovery that sand analyses plot well on this paper devised for other purposes. Plottings on it show with all necessary accuracy both coarse and fine extremes. If this paper had been available when sand analyses were first made it is not likely that any other form of plotting would have been used. Special plotting sheets have been frequently made for a particular set of sieves in which vertical lines are drawn to represent each sieve in the set. This involves the expense of making drawings and lithographs and other printed copies for each set of sieves. Such special sheets facilitate plotting, and reduce somewhat the likelihood of errors, and are to be recommended whenever much work is to be done. Sheets should be of paper that will blue print; and the entire record of the analysis can be entered on the margins so that a blue print copy shows all. A plotting of a representative sand analysis on such a sheet of mixed logarithmic paper and of the same analysis on logarithmic probability paper are shown in figures 21 and 22. Some mechanical filter plant operators have found it useful, in addition to the above procedures, to plot sand analysis findings in a manner similar to those used in plotting test borings, in order to show the sizes of sands in various vertical sections of a filter. SIEVES Sieves are made from carefully selected brass wire cloth having as nearly as possible square and even sized meshes. The frames are of metal, fitting one to another and the last to a pan at the bottom so that there will be no loss of material. The wire cloth may be bought in large pieces, carefully inspected, and a number of sets of sieves made as nearly as possible alike from these pieces. A serial number is stamped on each sieve in small figures for permanent record. After the size of separation is determined, as described below, that size should also be stamped on it but in larger and more conspicuous figures. The dimensions of wire cloth recommended by the Bureau of Standards are shown in the Jour. Amer. Water Works Assn., 1918, page 346. It has not been found possible to get a number of sieves that will give exactly the same size of separation. Even among sieves made from the same piece of wire cloth there will be variations. A great deal of time and ingenuity has been spent in trying to make the sieves interchangeable. An approxima 10 160 2 13 4 5 6 7 8 i 0.16 1.7 2.1 0.20 3.2 4.0 0.29 8.0 9.9 2 13 4 567891 0.33 11.0 13.6 Diameter in Millimeters tion to this result can be accomplished, but it is not necessary and if accomplished would make no material change in present methods or results. With the mass system of plotting the line resulting from plotting the analysis of a given sample made by any rated sieves should be the same within reasonable error limits as that resulting from the use of any other rated set. The number of sieves to be used, and the relation between their sizes of separation have been frequently discussed. The requirements vary according to the kind of sand and the purposes for which it is to be used. In examining bank sands and sands for sand filters having uniformity coefficients of 2.5 and over, a set of sieves in which there is a ratio of approximately 2 to 1 be JTTT 20 Percent by Weight 100 90 |