Sampling grain with accuracy
September 01, 1995
by Teresa Acklin
Quality differences in bulk grain require effective sample collection.
BY ROBIN WILKIN
Some variation exists within almost all bulk commodities, and the amount and distribution often is of fundamental economic importance. Attempts to estimate this variation center on collecting and assessing samples.
But good sampling can be negated by bad assessment, and a non-representative sample can mean hours of expensive testing that provides nonsense results.
A number of fundamental requirements must be met when sampling grain. First, a small sample, taken in isolation, provides very little information about the quality of the batch. Any effective method must involve the collection of several samples from different points.
Second, the method of collecting samples may influence the results of analysis; agreement between buyers and sellers is more likely if the sampling method has been assessed and agreed on in advance.
The quality characteristics of grains, oilseeds and feedstuffs tend to fall into two patterns of distribution. For grain, moisture and weight are more likely to be similar throughout a batch, unless the batch is made up from a number of different parcels. But other grain characteristics, such as sprout damage, fine material and insects, are almost certain to be unevenly distributed.
The aim of any sampling system must be to obtain a representative sample from the bulk. Meeting this aim requires the collection of several incremental samples from different points. (See examples on page 27.) These samples can be combined or, better still, assessed individually.
Once results are obtained from a series of samples, the data can be used for statistical analysis, which will allow better prediction of the range of quality within the bulk. It will even allow the number of samples needed for specific degrees of accuracy to be calculated. (See article on page 26.) In practice, the collection of sampling data can be done in one of several ways. For example, when a store is being filled with a large number of different truckloads of grain, results obtained by testing a representative sample from each load can be collected and collated. This will give the best estimate of grain quality and the variations within the bulk.
When good data are required on individual truckloads of grain, a greater number of sub-samples must be taken from different parts of the load and then either combined or assessed separately. There are a number of recognized approaches to grain sampling that follow these principles.
In some jurisdictions and in some circumstances, sampling and testing methods are regulated or fall under certain guidelines. The European Union, for example, uses “ISO 950, Sampling Bulk Cereals and Pulses,” while the U.S. grain industry relies on either the formal system administered by the U.S. Department of Agriculture or on legislation enacted by individual states.
Even so, the rules or guidelines can be limited in scope or open to interpretation, giving rise to variations within the grain and grain processing industries.
Sampling guidelines can be used to design standard operating procedures that will produce consistent samples. However, the operator must always be aware that using a “better method” that gives “more accurate” samples can lead to problems unless there is dialogue between buyers and sellers.
All equipment must be clean and in good condition. Proper maintenance is very important in ensuring consistency in sampling. For example, if air leaks develop in automatic probe systems or the opening of a hand spear becomes damaged, the consistency of the samples may be affected. Standard operating procedures should also specify what equipment should be used for specific tasks and how it is to be operated. Staff must be properly trained in the use of equipment.
Samples must be identified so that the results can be related to the batch from which the samples were collected. Some quality parameters, such as moisture content, will change with time, so these parameters should be determined as soon as possible after collection.
When rapid analysis is not possible, samples must be stored in sealed containers, with minimum free air space, and held under cool conditions. It is worthwhile testing the effects of sample storage by checking some samples immediately, and then again after storage. This routine should be part of standard quality assurance procedures.
Labeling and storage of samples are of paramount importance when samples may be needed at a later date to help resolve a dispute. Try to establish a formal system to archive representative sub-samples that can be used to confirm quality at a later date. However, great care is needed to prevent these samples from changing during prolonged storage, and features such as infestation cannot be preserved.
When grain is stored in a bin, deep silo or flat store, it may be necessary to sample it for one of two reasons: to estimate the quality or to check its condition in relation to continued storage. It is very difficult to collect representative samples from a large bulk of grain in either a flat store or a deep silo. Better results will always be obtained by sampling grain as it enters the store or as it is loaded out.
Stored grain can be sampled using hand spears or a portable vacuum sampler, and sampling methods typically are governed by official standards. For instance, grain sampling standards in the European Union recommend collecting samples, each weighing 5 kilograms, at the rate of about 10 samples per 100 tonnes.
Samples must be taken at different depths as well as different positions. This suggests that 100 samples with a total weight of 50 kg must be collected from 1,000 tonnes bulk. In practice, sampling rates are often lower, and therefore the samples may not be representative. The problem is compounded because it is difficult to collect samples at depths greater than 2 meters with hand spears.
The only method for very deep silos is to turn the grain and collect from the flow. This technique may require a properly designed diverter sampler; U.S. research indicates uncontrolled sampling of grain flows produces biased results. The overall conclusion is that it is very difficult to obtain a true picture of grain quality by sampling static bulks.
Sampling methods for truckloads also often are governed by official standards. E.U. sampling regulations, for instance, call for collection at five sampling points for trucks up to 15 tonnes and at as many as 11 points for trucks up to 50 tonnes; the points are specified as to depth and spacing.
One constraint to this technique is that it often requires someone to walk on top of the truck to collect samples, which can contravene health and safety regulations. An effective and safe approach to sampling trucks is to use an automatic probe, which can be programmable.
For wheat and barley, research in the U.K. indicates that the precise location of the sampling points is not important when automatic programmable probes are used. This is because the automatic system gives a mean value for quality that is not significantly different from the specific-point method.
Collecting six 1-kg samples from the truck tailgate during tipping will give a sample that can be used to estimate most quality parameters. However, research suggests that this method can be hazardous to personnel and will overestimate the amount of fine material in the load.
For large grains, such as maize, peas and oilseeds, it is common to use a standard sampling system consisting of an automatic probe or hand spear. The object is likely the same as when sampling wheat or barley, but the consistency of the products is sufficiently different to affect sampling.
Information from U.S. research shows that, with large grains, fine material will be distributed in a non-random manner when a truck is loaded from a spout or hopper. The highest concentrations of fines will be in the center and the lowest at the sides, and the differences are sufficient to have a commercial impact.
In the U.S., a system has been developed in which core samples are collected from points midway between the sides and center line of the load. No European research confirms this finding, but several U.S. states require different sampling methods for wheat than for other grains and oilseeds.
The best approach to sampling soybeans or maize, particularly if fine material is an important consideration, may be to collect at least six 1-kg samples per load. These samples should be arranged in two rows, each row being midway between the center and the side of the load.
Assessing these samples individually is the best practice, but they could be mixed and then divided to give a portion for assessment. It is very important to note that although this approach should give a good approximation of the true mean level of fine material, it is likely to produce different results from those obtained by using the ISO method.
No research data exist to validate sampling methods for very small seeds such as rapeseed. The best approach is to use the ISO method to collect the samples, but to assess each sample individually. Given data from several loads it should become apparent whether or not there are different results from different sample positions. Select the lowest number of sample points to give an answer within the required tolerance from the mean.
OTHER SAMPLING SITUATIONS.
It is often necessary to collect samples from moving flows of grain. However, health and safety regulations in some countries prohibit the direct manual collection of samples from conveyor belts, which restricts the options for sampling grain streams. Alteratives to direct, manual sampling include the use of “Pelican” scoops to collect samples of grain, or to install a mechanical diverter sampler or in-line sampler.
The only option for manual sampling would seem to be to use a “Pelican” sampler mounted on a pole. This device can be used to take sweeps across a stream of falling grain, but the slot in the sampler must be large enough to span the entire stream. The best approach to collecting representative samples would be to collect as many sub-samples as possible from a batch, mix these then divide down to give a working sample. The aim must be to collect samples at an appropriate rate (e.g., 1 kg per 20 tonnes) and to take sub-samples as often as possible from the flow.
If this approach to sampling grain flows is used, staff safety must be a paramount consideration. Staff must be prevented from falling into grain pits, etc., and they must be protected from inhalation of dust which will inevitably be associated with this operation.
In many parts of the world, grain flows are sampled by a mechanical device that passes though the grain flow at preset intervals and collects a sample.
The principle of these devices is simple: a scoop or spout is moved to intercept a stream of falling grain. However, in practice, achieving representative and repeatable samples presents difficulties.
A great deal of research has been done in the U.S. and Australia on the design and calibration of diverter or falling stream samples. The main conclusions seem to be that the design of the entry to the sampler is critical because the grain flow may be disrupted so that only the more dense particles of grain will enter the sampler.
Resistance to wear is also important, if the performance is to be consistent. Research shows that the use of a single tube or other static device is prone to blocking with straw and chaff. A good diverter sampler will intercept the whole of the flow in such a way as to collect a proportional volume from each segment of the stream.
Any automatic sampler should be calibrated by the manufacturer. This is usually done by adding specific contaminants to a batch of grain and comparing the sampler's results with theoretical values and perhaps with results given by other equipment.
Diverter or falling stream samplers offer considerable advantages at store or the premises of end users, where grain is moved through conveying systems at rates of 50 tonnes per hour or more. They will collect reliable, repeatable and representative samples safely and at low operating costs. Often, samples collected in this way can be passed directly to the laboratory for analysis.
Samplers can often be “tuned” to collect samples at different rates according to need. However, regular maintenance is essential if the characteristics of the sampler are to be preserved.
Robin Wilkin, an international adviser and consultant, worked for many years in the Central Science Laboratory of the U.K.'s Ministry of Agriculture, Fisheries and Food researching the control of stored product pests. This article is an excerpt from his manual “Grain, Seed and Feed: Sampling and Physical Characteristic Testing,” published by Samplex Ltd., U.K.