Volume 07 - Powder Metal Technologies and Applications Part 4 pot

Fig. 1 Sampling spears The spear is thrust into the powder with the inner chamber closed off, and when in position the outer tube is rotated to allow powder to fall into the inner chamber. When the chamber is full, the inner tube is turned to the closed position and the spear is withdrawn. Possible segregation throughout the bed may be investigated with type 3, an average value for the length of the spear with type 2, and a spot sample with type 3. Frequently the spears are vibrated to facilitate filling, and this can lead to an unrepresentative quantity of fines entering the sample volume. The sampling chambers also have a tendency to jam if coarse particles are present, because they can get lodged between the inner and outer chambers. Coning and Quartering. In industry it is common practice to sample small heaps by coning and quartering. The powder is formed into a heap, which is first flattened at the top and then separated into four equal segments with a sharp- edged board or shovel. The segments are drawn apart and, frequently, two opposite quadrants are recombined and the operation is repeated until a small enough sample has been generated. This practice is based on the assumption that the heap is symmetrical, and since this is rarely so, the withdrawn sample is usually nonrepresentative. Sampling from Trucks or Wagons. Consignment sampling is carried out on a single consignment (e.g., a truck or wagon load). In sampling from a truck or a wagon it is recommended that eight samples be extracted (Ref 1). No increment should be taken at less than 12 in. below the surface; this avoids the surface layer, in which segregation can have occurred due to vibration (Fig. 2). Care needs to be taken to prevent powder sliding down the slope created due to removal of surface material. Fig. 2 Sampling points for a wagon or a truck Sampling Stored Free-Flowing Material. It is practically impossible to representatively sample stationary free- flowing powder because of the severe segregation that has almost certainly occurred. There is only one sound piece of advice to give regarding sampling such material: Don't! If there is no alternative, several samples should be taken and analyzed separately so that an estimate can be made of the degree of segregation. Sampling from Containers. Suppose an analysis is required from several tons of material that is available in bags or small containers. Several of these containers should be selected systematically or, preferably, using a table of random numbers. The recommended number of samples depends on the number of containers (Table 1). The whole of each bag should then be sampled using a full stream or Vezin-type splitter so that the golden rules of sampling are obeyed. This is the only way to obtain a representative sample from each bag. Where this is not possible a sampling thief may be used. It is preferable to obtain a sample as the containers are being filled or emptied. Table 1 Recommended number of containers to be sampled from a packaged lot Number of containers in lot Number of containers to be sampled 1-5 all 6-11 5 12-20 6 21-35 7 36-60 8 61-99 9 100-149 10 150-199 11 200-299 12 300-399 13 Note: For every additional 100 containers, one additional container should be sampled. Source: ISO 3954 Reference cited in this section 1. "Sieve Analysis of Granular Mineral Surfacing for Asphalt Roofing and Shingles," D 451- 63, American Society for Testing and Materials, 1963 Sampling and Classification of Powders Terence Allen Sampling Flowing Streams Most powder systems are transported at some time during their manufacture as flowing streams: Hoppers are emptied by screw or belt conveyors, powders are transferred to bagging operations by screw or pneumatic conveyors, and many solids are transported through pipes. A general rule in all sampling is that whenever possible, the sample should be taken while the powder is in motion. This is usually easy with continuous processes; with consignment sampling it may be possible during filling and emptying of storage containers. Sampling from a Conveyor Belt. When a sample is to be collected from a conveyor belt, the best position for collecting the increments is where the material falls in a stream from the end of the belt. If access at such a point is not possible, the sample must be collected from the belt. The whole of the powder on a short length of the belt must be collected. The particles at the edge of the belt may not be the same as those at the center, and particles at the top of the belt may not be the same as those at the bottom. If the belt can be stopped, the sample may be collected by inserting into the stream a frame consisting of two parallel plates shaped to fit the belt; the whole of the material between the plates is then swept out. A scoop can be used to scoop out an increment, but this operation can be hazardous if the belt is moving. When sampling from a continuous stream, the sampling may be continuous or intermittent. In continuous sampling, a portion of the flowing stream is split off and frequently further divided subsequently. In intermittent sampling, the whole stream is taken for many short increments of time at fixed time intervals. These increments are usually compounded and samples for analysis are taken from this gross sample. Continuous sampling is deprecated because if there is segregation on the belt, the extracted sample may not be representative. It is common practice, in sampling from a blender, to extract three samples: the first after the blender has been emptying for a few minutes, the second when the blender is half empty, and the third when the blender is almost empty. Note that blenders sometimes have a "heel" of unmixed material that is first out of the blender. Powder from the whole cross section of the blender discharge steam should be collected for each sample. This practice should only be used after the mixing efficiency of the blender has been established for each product by taking multiple samples and analyzing these separately. Point Samplers. Samples can be extracted from the product stream by projecting a sample tube, containing a nozzle or orifice, into the flow. The particles impact the tube and fill the open cavity. The sampling head is out of the stream when not sampling Snorkel-type samplers are available for vertical or inclined applications and can be preprogrammed for sampling frequency. It is not possible to sample nonhomogeneous streams representatively with this type of device. With the auger-type sampler, a slot inside the process stream is rotated to capture a cross section of the process stream, which is then delivered into a sample container. This type of device does not collect a representative sample unless the stream is homogeneous, and it has the added disadvantage that it obstructs flow. Sampling from Falling Streams. In collecting from a falling stream of power, care should be taken to offset the effects of segregation. Each increment should be obtained by collecting the whole of the stream for a short time. Care must be taken in putting the sampler in and out of the stream. Figure 3 shows correct and incorrect ways of doing this. Unless the time during which the receiver is stationary in its receiving position is long compared with the time taken to insert and withdraw the sampler, the method shown in Fig. 3(a) will lead to an excess of coarse particles, because the surface region of the stream, usually rich in coarse particles, is sampled for a longer time than the rest of the stream. The method shown in Fig. 3(b) is not subject to this objection. If this method is not possible due to some obstruction, the ratio of stationary to moving time for the receiver should be made as large as possible. In many cases it is not possible to collect the whole of the stream as this would give too large an amount to be handled. The best procedure in this case is to pass a sample collector of the form shown in Fig. 3(c) through the stream. Fig. 3 Sampling from falling streams. (a) Bad sampling technique. (b) Good sampling technique. (c) Sampling procedure to be adopted for high mass flow rate The width of the receiver, b, should be chosen to give an acceptable weight of sample but must not be made so small that the biggest particles have any difficulty in entering the receiver. Particles that strike the edges of the receiver are likely to bounce out and not be collected, so that the effective width is (b-d), where d is the particle diameter. The effective width is therefore greater for small particles than for large ones. To reduce this error to an acceptable level, the ratio of receiver width to the diameter of the largest particle should be made as large as possible with a minimum value of 3:1. The depth, a, must be great enough to ensure that the receiver is never full of powder. If the receiver fills before it finishes its traverse through the powder, a wedge-shaped heap will form that is size selective. As more powder falls on top of the heap, the fine particles will percolate through the surface and be retained, whereas the coarse particles will roll down the sloping surface and be lost. The length of the receiver, c, should be sufficient to ensure that the full depth of the stream is collected. Stream Sampling Ladles. Powder may be manually withdrawn from a moving stream of powder using one of the several commercially available ladles. These are suitable for occasional use, but automatic on-line stream sampling samplers are preferred for frequent applications. Traverse Cutters. With large tonnages, samples taken from conveyors can represent large quantities of material that need to be further reduced. With the action shown in Fig. 4(a) and 4(b), uniform increments are withdrawn to give a representative sample, but with the action shown in Fig. 4(c), a biased sample results if the inner and outer arcs of the container are significantly different and the powder is segregated horizontally on the belt. Fig. 4 Traversing cutters. (a) Straight path action, in line. (b) Cross line. (c) Oscillating or swinging arc path Often, a traversing cutter is used as a primary sampler, and the extracted sample is further cut into a convenient quantity by a secondary sampling device. The secondary sampler must also conform with the golden rules of sampling. A traversing cutter is satisfactory for many applications, but it has limitations that restrict its use: • Although a traversing cutter is comparatively readily designed into a new plant, it is frequently difficult and expensive to retrofit an existing plant because of the space requirements. • The quantity of sample obtained is proportional to product flow rate, and this can be inconvenient when the plant flow rate is subject to wide variations. On the other hand, where the daily average of a plant is required, this is a necessary condition. • It is difficult to enclose the sampler to the extent required to prevent the escape of dust and fumes when handling dusty powders. Sampling Dusty Material. Figure 5 shows a sampler designed to sample a dusty material, sampling taking place only on the return stroke. This is suitable provided that the trough extends the whole length of the stream and does not overfill. The radial cutter or Vezin sampler shown in Fig. 6 is suitable for sample reduction. These samples vary in size from a 15 cm laboratory unit to a 152 cm commercial unit. Fig. 5 Full-stream trough sampler Fig. 6 Schematic of a primary and secondary syst em based on Denver Equipment Company's type C and Vezin samplers Diverter Valve Sampler. The diverter valve shown in Fig. 7 is suitable for online intermittent sampling when there is limited head room. It can also be operated manually. Fig. 7 Diverter valve sampler Sampling and Classification of Powders Terence Allen Sample Reduction The gross sample is frequently too large to be handled easily and may have to be reduced to a more convenient weight. Obviously, the method employed must conform with the two golden rules mentioned above. The amount of material to be handled is usually small enough that getting it in motion poses no great difficulty. There is a natural tendency to remove an aliquot with a scoop or spatula, and this must be avoided because it negates the effort involved in obtaining a representative sample from the bulk. Placing the material in a container and shaking it to obtain a good mix prior to extracting a scoop sample is not recommended. To obtain the best results, the material should be made as homogeneous as possible by premixing. It is common practice then to empty the material into a hopper, and this should be done with care. A homogeneous segregating powder will segregate when fed to a hopper from a central inlet, because in essence it is being poured into a heap. In a core flow hopper the central region (which, with improper feeding, will be rich in fines) empties first, followed by the material nearer the walls, which has an excess of coarse particles. The walls of the hopper should have steep sides (at least 70°) to ensure mass flow, and the hopper should be filled in such a way that size segregation does not occur. This can best be done by moving the pour point about so that the surface of the powder is more or less horizontal. Several sample-dividing devices are discussed briefly below. Scoop sampling consists of plunging a scoop into the powder and removing a sample. This method is particularly prone to error because the whole of the sample does not pass through the sampling device, and because the sample is taken from the surface, where it may not be representative of the mass. For powder in a container, it is usual to shake the sample prior to sampling in an attempt to achieve a good mix. However, the method of shaking can promote segregation. Coning and quartering consists of pouring the powder into a heap and relying on its radial symmetry to give identical samples when the heap is flattened and divided by a cross-shaped cutter. This method is no more accurate than scoop or thief sampling, which are simpler to carry out, but gross errors are to be expected. Coning and quartering should never be used with free-flowing powders. Table Sampling. In a sampling table the material is fed to the top of an inclined plane in which there is a series of holes. Prisms placed in the path of the stream break it into fractions. Some powder falls through the holes and is discarded, while the powder remaining on the plane passes on to the next row of holes and prisms, and more is removed, and so on. The powder reaching the bottom of the plane is the sample. The objection to this type of device is that it relies on the initial feed being uniformly distributed, and on a complete mixing after each separation, and in general these conditions are not achieved. As it relies on the removal of part of the stream sequentially, errors are compounded at each separation, hence its accuracy is low. Chute Splitting. The chute splitter consists of a V-shaped trough, along the bottom of which is a series of chutes that alternately feed two trays placed on either side of the trough. The material is repeatedly halved until a sample of the desired size is obtained. When carried out with great care this method can give satisfactory sample division, but it is particularly prone to operator error, which is detectable by unequal splitting of the sample. The above methods are all popular because the samplers contain no moving parts and are consequently inexpensive. Rotatory Sample Divider. The rotary sample divider conforms to the golden rules of sampling. The preferred method of using this device is to fill a mass flow hopper in such a way that segregation does not occur. The table is then set in motion and the hopper outlet is opened so that the powder falls into the collecting boxes. The use of a vibratory feeder is recommended to provide a constant flow rate. Several versions of this instrument are available, some of which were designed for free-flowing powders, some for dusty powders, and some for cohesive powders. They handle quantities from 40 L down to a few grams. Sampling and Classification of Powders Terence Allen Slurry Sampling Slurry process streams vary in flow rate, solids concentration, and particle size distribution. Any sampling technique must be able to cope with these variations without affecting the representativeness of the extracted sample. For batch sampling, automatic devices are available where a sampling slot traverses intermittently across a free-falling slurry. Unfortunately, it is difficult to improvise with this technique for continuous sampling, because such samplers introduce pulsating flow conditions into the system. Sampling and Classification of Powders Terence Allen Evaluation of Sampling Experimental tests of sampling techniques are compared in Table 2 as an example. Binary mixtures of coarse and fine sand (60:40 ratio) were examined (Ref 2) using various laboratory sampling techniques. In every case, 16 samples were examined to give the standard deviations shown in column 2. It may be deduced that very little confidence can be placed in the first three techniques and that the rotary sample divider is so superior to all other methods that it should be used whenever possible. Table 2 Reliability of selected sampling methods using a 60:40 sand mixture Sampling technique Standard deviation, % Coning and quartering 6.81 Scoop sampling 5.14 Table sampling 2.09 Chute slitting 1.01 Rotary sample dividing 0.146 Random variation 0.075 Reference cited in this section 2. T. Allen, Particle Size Measurement, Chapman & Hall, 1997 Sampling and Classification of Powders Terence Allen Weight of Sample Required Gross Sample. Analyses are carried out on a sample extracted from the bulk, which, irrespective of the precautions taken, never represents the bulk exactly. The limiting (minimum) weight of the gross sample may be calculated, using a simple formula to give an error within predesignated limits, provided that the weight of the gross sample is much smaller than that of the bulk. The limiting weight is given by: (Eq 1) where M s is the limiting weight in grams, is the powder density in g · cm -3 , is the variance of the tolerated sample error, w 1 is the fractional mass of the coarsest size class being sampled, and is the arithmetic mean of the cubes of of the extreme diameter in the size class in cubic centimeters. This equation is applicable when the coarsest class covers a size range of not more than and and w 1 is less than 50% of the total sample. Table 3 gives sample values. Table 3 Minimum sample mass required for sampling from a stream of powder Upper sieve size, m Lower sieve size, m Mass % in class (100w 1 ) Sample weight required, g 600 420 0.1 37,500 420 300 2.5 474 300 212 19.2 14.9 212 150 35.6 1.32 Sample by Increments. For sampling a moving stream of powder, the gross sample is made up of increments. In this case the minimum incremental weight is given by: (Eq 2) where M i is the average mass of the increment, 0 is the average rate of flow, w 0 is the cutter width for a traversing cutter, and v 0 is the cutter velocity. If w 0 is too small, a biased sample deficient in coarse particles, results. For this reason w 0 should be at least 3d, where d is the diameter of the largest particle present in the bulk. ISO 3081 suggests a minimum incremental mass based on the maximum particle size in millimeters. These values are given in Table 4. Secondary samplers then reduce this to analytical quantities. Table 4 Minimum incremental mass required for sampling from a stream of powder Maximum particle size, mm Minimum mass of increment, kg 250-150 40.0 150-100 20.0 100-50 12.0 50-20 4.0 20-10 0.8 10-0 0.3 Gy (Ref 3) proposed an equation relating the standard deviation, which he calls the fundamental error F , to the sample size: (Eq 3) where W is the mass of the bulk; w = n is the mass of n increments, each of weight , that make up the sample; C is the heterogeneity constant for the material being sampled; and d is the diameter of the coarsest element. For the mining industry (Ref 3), Gy expressed the constant C in the form C = clfg where: (Eq 4) P is the investigated constant. is the true density of the material. l is the relative degree of homogeneity where for a random mixture l = 1, and for a perfect mixture l = 0. f is a shape factor assumed to be equal to 0.5 for irregular particles and 1 for regular particles. g is a measure of the width of the size distribution where g = 0.25 for a wide distribution and g = 0.75 for a narrow distribution (i.e., d max <2d min ). For the pharmaceutical industry, Deleuil (Ref 4) suggested C = 0.1lc with the coarsest size being replaced by the 95% size. For W w, Eq 4 can be written: (Eq 5) where = t( F/P) and t = 3 (99.9% confidence level) for total quality. • For d 95 = 100 m, = 1.5, P = 10 -3 (1000 ppm), = 0.2, l = 0.03 (random), and w = 1000 g. • For d 95 = 100 m, = 1.5, P = 0.05, = 0.05, l = 1 (homogeneous), and w = 4 g. • For d 95 = 20 m, = 1.5, P = 10 -4 (100 ppm), = 0.05, l = 0.03 (random), and w = 8000 g. [...]... of Particulate Matter: Theory and Practice, Elsevier, Amsterdam, 1982 4 M Deleuil, Powder Technology and Pharmaceutical Processes, Handbook of Powder Technology, Vol 9, D Chulia, M Deleuil, and Y Pourcelot, Ed., Elsevier, 19 94 Sampling and Classification of Powders Terence Allen Powder Classification Classification methods are used to obtain particular powder distributions or to exclude certain powder. .. the wall or forming vortices The so-called Coanda effect helps to maintain the flow around the bend for approximately 90°, and this is enhanced by the application of suction Fig 14 Principle of the cross-flow elbow classifier Reference cited in this section 5 D.F Kelsall and J.C.H McAdam, Trans Inst Chem Eng., Vol 41 , 1963, p 8 4- 9 4 Sampling and Classification of Powders Terence Allen Sieving Methods... be based on the median particle size (Table 5), but this neglects to take into account that the narrower the distribution, the smaller the sample required A more detailed table is presented in Ref 12 Table 5 Amount of sample required for 8 in diameter sieves Basis Particle density, g · cm-3 3.0 Median diameter of particle, mm >2 2-1 1-0 .5 0. 5-0 .25 0.2 5-0 .075 . lot Number of containers to be sampled 1-5 all 6-1 1 5 1 2-2 0 6 2 1-3 5 7 3 6-6 0 8 6 1-9 9 9 10 0-1 49 10 15 0-1 99 11 20 0-2 99 12 30 0-3 99 13 Note: For every additional 100 containers,. designed for free-flowing powders, some for dusty powders, and some for cohesive powders. They handle quantities from 40 L down to a few grams. Sampling and Classification of Powders Terence. Table 4 Minimum incremental mass required for sampling from a stream of powder Maximum particle size, mm Minimum mass of increment, kg 25 0-1 50 40 .0 15 0-1 00 20.0 10 0-5 0 12.0 5 0-2 0 4. 0