Hwangecolliery.co.zw

HWANGE COLLIERY COMPANY LIMITED
QUALITY ASSURANCE SERVICES
Revision N°: 1 Issue Date: 15 June 2010 Revision No.:
Issue date:

Reviewed by:

Signature:

Designation:
Quality Control Chemist

Approved by:

Signature:

Designation:
Quality Assurance Superintendent

Revision N°: 1 Issue Date: 15 June 2010 The test method outlines different sampling methods of coal and coke as well as handling such samples. Sampling is a word commonly used in our everyday life, both at the workplace and outside. Consciously and or subconsciously, this is an activity we carry out in our every day life so as to come to some decision in relation to problems that we encounter. For example, during the process of preparing a meal we take samples of the food and taste it with our mouths so as to determine whether it will be palatable to those who will consume it. The way we sample in our ordinary daily lives may not be formalised, written down or standardised and thus it could be subject to personal judgment and preferences. This sampling method is aimed at formalising sampling of coal and coke at Hwange according to the International Standard ISO 1988-1975(E)/ERRAT. Sampling according to this standard ensures uniformity of sampling practices and procedures between interested parties, e.g. suppliers and customers. Personal judgements and preferences are eliminated such that independent samples obtained by the different parties will of necessity be the same in characteristics if the standard is followed by all. The Quality Control Chemist is responsible for ensuring implementation and maintenance of this test method and the Samplers for carrying out the method. Approved signatories are responsible for approval of records generated as a result of this test method 4.0 DESCRIPTION OF ITEMS 4.1 REAGENTS AND MATERIALS 4.2 EQUIPMENT AND APPARATUS a. Scoops or ladle 4.3 SAMPLE(S) 4.3.1 Importance of sampling The objective of sampling is to collect a portion of coal or coke that will serve for the determination of the quality characteristics of the coal or coke consignment concerned. A definite number of portions (increments) distributed throughout the whole consignment are collected to constitute a representative sample. In commercial practice the sample derived from the definite number of increments should be representative of the consignment in question, if proper sampling standard procedures are used. Thus when the sample is analysed at the laboratory for its physical and chemical quality characteristics, the quality characteristics so obtained will be deemed to represent the consignment from which the sample was obtained. It should be apparent at this stage that the quality characteristics obtained from samples yield important information for decision making in the following areas. a. Suitability of input coal to various plant in order to produce specific products. b. Process control changes in order to produce required products. c. Product certification d. Product pricing e. Arbitration in case of dispute between supplier and customer Useful quality characteristics for the above areas can be obtained if and only if the essential condition of sampling is met. The condition is such that the bulk of the coal and coke to be sampled should be exposed such that all parts are equally accessible and have equal chance of being included in the sample. Revision N°: 1 Issue Date: 15 June 2010 4.3.2 Types of sampling The various types of sampling differ in the way increments are spaced over the whole consignment. There are three types, namely systematic, random and stratified random sampling. In the majority, Hwange Colliery Co. uses systematic sampling and thus it is the only type of sampling which will warrant discussion here. Routinely, the other types of sampling are more expensive to administer. Systematic sampling spaces increments evenly in time or position over the unit of consignment. Cyclic variation in coal or coke quality can pose serious flaws in this type of sampling i.e. sampling frequencies could coincide with specific cycles of good or bad quality resulting in biased sampling. Thus care should be taken to avoid such coincidences and therefore the elimination of bias. 4.3.3 Bias in sampling Bias in sampling manifests itself when samples yield quality characteristics that are persistently too high or low. Bias is caused by the following factors; a. Selection of an increment from only one side of a consignment e.g. the middle or side of a belt. When coal and coke is being moved by conveyors or stockpiled, it naturally segregates according to size as shown on figure I and II respectively. b. Poorly designed sampling equipment whose dimensions of entry are too small resulting in the exclusion of larger pieces of coal and coke. 4.3.4 Avoidance of bias in sampling a. The minimum dimension of entry of the sampling equipment must be related to the upper nominal size of the coal or coke. b. The mass of an increment must be related to the upper nominal size of coal or coke. c. Change the shape of the sampling equipment. d. Change the sampling points. e. Change sampling system. 4.3.5 Sampling Situations The most reliable reference method of sampling is the stopped belt. This method ensures that the samples are free from bias. In general, this method is used for checking other methods. Routinely, this method is not used because of its costly labour requirements, tedious procedure, unproductive conveyer down time and associated safety risks. The next satisfactory method involves taking increments of cross sections of a moving stream. This is the commonly used sampling method in Hwange. A cross sectional increment is taken from a moving stream at the point of discharge (i.e. at transfer points). Representativeness of the increment is ensured by cutting through the stream in a uniform movement without overfilling the sampling cup during the operation. Sampling from stationery consignments e.g. wagons or stockpiles presents problems of inaccessibility coupled with tendencies of segregation according to size and thus biased samples could be obtained. Use of appropriate sampling equipment such as mechanical augers (improves accessibility) and taking of more increments than usual would of necessity minimise bias in this sampling situation. 4.3.6 Types of samples Two types of samples can be collected i.e. a. General analysis sample b. Moisture/physical determination sample. However for Hwange, common samples are generally collected for both analysis and moisture determination except for larger sized coke and coal, where separate samples are collected. 4.3.7 Sampling points Sampling points for coal and coke are illustrated on Exhibit I & II in the appendices and attachment section. It was indicated earlier on that such tests are a tedious and an expensive process. In principle a sampling method is tested against one or more methods known to be unbiased. Equal number of Revision N°: 1 Issue Date: 15 June 2010 increment are collected for each method from the same coal or coke.
The differences between the results are statistically analysed. The collection of increments or samples is
spread over at least three consignments (i.e. there could be variation between consignments).
The comparison methods used are stated below.
a. Reference method known to be unbiased e.g. stopped belt.
b. If (a) is not practicable, test for relative bias of a specific sampling point to another sampling
point.
Generally more than one comparison should be used since agreement of just two methods is not
conclusive in declaring that the possibility of bias does not exist. The quality characteristics or
variables in coal and coke that are normally tested are ash and moisture. Size analysis comparisons
lead to misleading results i.e. breakages between sampling points does occur, i.e. refer to (b).It should be
noted that no sampling scheme can yield the correct results on every occasion. Thus the test can not
establish that there is no bias but only that it is not likely to be have bias of more than a certain magnitude
i.e. minimum detectable bias. In carrying out this exercise, increments or samples can be used depending
on the circumstances. If the sampling point are close enough in terms of space and time, individual
increments can be compared. The timing of the increment should be such that the part of the coal stream
is being sampled. If it is not possible to compare increments, samples constituted from a reasonable
number of increments are compared. Generally the number of increments per sample should be such that
the precision of single sample is + 1%. The number of samples or increments required is obtained from
the Table IV in the appendices and attachments section.
5.2 Sampling from a stream of coal
Moving stream
As indicated earlier on sampling at Hwange involves in the majority sampling from a moving stream
at a point of discharge (ie conveyor transfer points) using cups. The initial number of increments for
an ash/moisture for consignment up to 1000 tonnes are given in table VII below.
Table A
Initial number of increment for sampling from a stream of coal or coke.
For consignments greater than 1000 tonnes, the number of increments indicated on table A are multiplied by the following factor. Mass in consignment in tonnes 10001 The following aspects should be adhered to during sampling. a. Increments should be taken from the whole width and thickness of the coal/coke stream in a single operation. If this is not possible as in streams M-Block, WP11, WS26 a procedure discussed later of sampling from wide streams should be used. b. The sampler should be able to reach the whole cross section of stream in safety and without undue strain in order to avoid biased sampling ie exclusion of other parts. c. Intervals of sampling between successive increments should not coincide with cycles of any specific quality. d. Mass of increments should be proportional to the flow density of the coal or coke. e. Avoid taking increments at the beginning, and end of the flow. Revision N°: 1 Issue Date: 15 June 2010 f. Increments should be collected at normal load.
Stopped belt
Some methods of sampling tend to collect too many of either large or small particles and thus
introducing elements of bias. The above method ensures that all particles are collected and thus
eliminating the element of bias. It is therefore used as a reference method.
The above method can only be instituted if and only if proper safety procedures and communication
between samplers and production personnel, have been ensured. A suitable frame is placed on the
stationery belt so that it is in contact with belt across its full width. Large pieces obstructing the
insertion of the frame are pushed.
a. At the left side of frame, into the sample.
b. At the right side of frame, out of the sample
Sampling from wide streams
This procedure refers to manual sampling of wide streams such WPC 11, M/Block and WS26.
Mechanical or automatic sampling is best suited for wide and high capacity streams. If those are not
available the stream could be sampled as detailed below:
x x x x x x and etc
The following sampling scheme shows how increments can be taken in two parts. Up to five positions can
be sampled depending on the width of stream. Invert the sampling cup and move it into back of the
stream. Position it the right way up and quickly withdraw it from the stream in order to avoid filling the cup.
Great care should be taken when sampling such streams.
Sampling from stockpiles
The above method can only be used if the preferred procedure of sampling from a moving falling
streams during stocking or destocking can not be used. The initial number of increment for the up to
1000 tonnes are as indicated on Table B.
Table B
Initial number of increments for sampling from stockpiles.
For tonnages greater than 1000 adjustments of increments are made as indicated under sampling from streams. If a stockpile contains different coals or cokes piled in separate areas of the total pile, separate gross samples should be collected which are then prepared and analysed separately. The value of the required characteristics is determined on a weighted bias. In taking the increments, these should be evenly spaced over the surface or layers of the stockpile. Increments are obtained by digging holes to traverse the layers. For large stockpiles, it is necessary to work a plan indicating sampling position on the map of the area and to mark positions before sampling. In collecting each increment it should be ensured that the increment represents the coal/coke in the vicinity. Large particles should not be allowed to roll from or into the shovel used for sampling during increment extraction. Holes should be dug to different depths and from the bottom of each hole, an increment should be taken by shovel so that samples of approximately the same mass are collected from different layers ensure that no particles truckle down the sides into the hole. 6.0 QUALITY CONTROL AND ACCEPTANCE CRITERIA 6.1 Accuracy and Precision It is not possible to know with certainty the true values of the quality characteristic of any coal or coke consignments. Thus accuracy of sampling method may be difficult to determine since the true values are unknown. Accuracy relates to the closeness of results obtained from a sampling method and the true Revision N°: 1 Issue Date: 15 June 2010 value. Thus, precision of sampling is used instead. It relates to the closeness of results obtained from a sampling method. However if the sampling method is bias free precision is synonymous to accuracy. 6.2 Precision of sampling The reference standard of precision for coal or coke is + one tenth of the true ash or moisture value and +2% absolute for ash or moisture values less than or equal to 20% and greater than 20% respectively. The deviation from the true values represent errors due to sampling, sample preparation and analysis. Generally, of the total deviation 20% is due to sample preparation and analysis and the rest 80% is due to sampling. Again this underpins the importance of proper sampling standards otherwise the quality characteristics obtained will be far different from the true values. It has already been stressed that coal/coke tends to segregate according to sizes during transportation and storage and thus proper sampling procedures should be followed to avoid bias. Another factor which must also be recognised is that coal is not a homogenous material i.e. it is heterogeneous. Coal can contain various dissimilar materials as indicated below; a. True coal b. Impure coal i.e. black shale c. Containments such as fireclay and footwall sandstone. Typically the above are so dissimilar in nature and this results in coal whose quality is very variable. Thus if proper sampling procedures are not used, the results obtained will be worthless and no where near the true values. The adequate number of increments should be ascertained and replicate sampling used to check for sampling precision. 6.3 Handling and storage of samples The handling and therefore storage of increments / samples depends on the type of analysis or tests required. Above all, proper handling and storage ensures that samples are representative of their consignments i.e. they should not end up degraded or contaminated. The types of analysis are as follows; a. Moisture/physical analysis b. Chemical analysis. Common samples are routinely collected except for the following; a. Rounds b. Large Cobbles c. Wharf Coke d. Foundry Coke (+60mm) e. Export (40-80mm) coke For these samples, separate samples for moisture/physical and chemical analysis are collected. The physical/moisture analysis samples are stored and transported in metal bins in order to avoid breakages of coal and coke particles. More so, the samples/increments must be put quickly into the metal bins to avoid loss of moisture and contamination from dust in the atmosphere. Note that the metal bins should be impermeable and must also have well fitting lids. Bins and lids that have holes should never be used. The samples other than those mentioned above are common samples for both physical and chemical analysis. The sample increments collected at the sampling point are immediately placed into 22kg ammonium nitrate bags that have inner plastic bags. The bags are obtained from Opencast and only those which are not damaged or holed should be used. Once an increment has been placed into a bag, it must be sealed using a string to avoid loss of moisture and contamination. Samples/increments should only be stored under a roof. 6.4 Labelling of samples Once manageable increments have be collected into the ammonium nitrate bags, the sampler must ensure correctness of the sample labels. One tag label enclosed in a zip grip packrite plastic bag i.e. to avoid wetting of the label should be placed on top of the sample in the bag and the bag is immediately sealed with a string and another tag label is attached onto the outside of the bag using the same tying string. Both the inside and outside labels should reflect the following information; a. Date i.e. day, month and year. b. Name of the sample e.g. (20-60mm) LP Coke. c. Destination of the consignment being sampled e.g. Zimasco, Kwekwe d. Truck Number. Revision N°: 1 Issue Date: 15 June 2010 e. Initials of the sampler f. Number of increments Labelling shall only be done in pen and not in pencil. The sealed and labelled bags can then be moved from the sampling point to the appropriate concrete slab loading bay. The loading bay should of necessity have a roof above it. 6.5 Fundamentals of sampling a. Precision.
The international standard stipulates that the determined values ie ash or moisture on samples
derived from the same consignment are expected to be within the specified limits of precision 95
times out of a 100. Without bias in sampling the limits are spread uniformly around the true value.
b. Precision and number of increments.
The desired precision can be achieved by adjusting the number of increments. Generally, the more
the increments that are obtained the better the precision of the results. The reference standard of
precision has been discussed earlier on and thus the number of increments for all Hwange samples
indicated on table 1 are designed to meet that level of precision.
c. Replicate sampling
As indicated earlier on replicate sampling is used to test the level of precision obtainable from a
specific sampling scheme. This aspect of work will be carried from time to time under the auspices of
the Audit Samplers. The procedure entails collecting the same number of increments as usual but the
successive increments are placed in a number of different containers to give a number of replicate
subsamples. Each of these separate subsamples are then prepared and analysed for various quality
characteristics. However the procedure for replicate sampling differs depending on whether we are
dealing with a single/isolated or regular consignment. Hwange sampling generally involves regular
sampling. Thus replicate. Sampling for single/isolated consignments will be considered just briefly in
this procedure.
The procedure for single/isolated consignment replicate sampling involves the following steps;
a. Determine the initial number of increments, i.e. refer to Table I.
b. If the number of increments are not divisible by 6, or that the mass of the subsamples are not
sufficient for laboratory analysis, then the next multiple of six is used.
c. The increments are collected in the usual way with the first being collected into container 1,
then 2 and so on up to 6. The seventh increment is then placed into container 1, the 8th into
container 2 and the process is continued until the total number of increments have been collected.
d. The collected six subsamples are then prepared for analysis of the various quality characteristics.
e. The results of the quality characteristics are tabulated and analysed statistically.
The procedure for regular consignment replicate sampling is described as follows. Each sample is
collected in duplicate.
a. Obtain the numbers of increments from Table I
b. Increments are collected as usual with the first increment being placed in the container 1 and the
second into container 2, then the third is placed into container 1 and the forth into container 2. The
process is repeated until the required number of increment have been collected.
c. The process above is repeated for the next other nine consignment to yield 10 duplicate samples.
Note for example the days production of coking coal into the Opencast bunkers can be considered as
a consignment. Thus such an operation will
d. The ten duplicate sub samples are prepared for analysis of the required quality characteristics or variables. e. The results are tabulated and analysed statistically. An example will be given below for coking coal into the boat bunkers at the Opencast. The quality characteristics analysed for on the duplicate samples was the ash content. The results are as shown in the appendices section on Table I. 7.0 CALCULATIONS AND EXPRESSION OF RESULTS Revision N°: 1 Issue Date: 15 June 2010 8.0 ENVIRONMENTAL CONDITIONS Avoid bias when sampling. 9.0 SAFETY PRECAUTIONS Table I: Results of continuous duplicate sampling - at sampling point WPC B/B (WPC3)- Ash content % Mean = R/10 = 12.5 Mean difference between duplicates; D = T/10 = 0.53 Check the calculations as follows; a. P-Q = T = 5.3 = 250.3 The statistical analysis of the above data will be based on continuous sampling. Hwange samples all units as opposed to intermittent sampling were some of the units are not sampled). The statistical analysis is as follows; a. Rogue results: This refers to duplicate results whose difference is 3.5 times the mean difference. With this test such results are suspicious and thus imply mistakes in sampling or sample preparation. Such results are rejected. The results in table II show no such results and thus non will be rejected. b. Testing results. Number of units for which the stated precision is needed Revision N°: 1 Issue Date: 15 June 2010 From the above table, at the level of require precision of +0.50 corresponding to 10 units the expected D value is 1.30. Calculate the ratio; D = Expected mean difference between duplicates d Observed mean difference between duplicates The ratio is used to check whether the sampling is at the desired level as shown below: a. If D >2.0 therefore too many increments are being taken d b. If 0.67 < D < 2.0 therefore no evidence that the number increments is incorrect. d c. If D < 0.67 therefore too few increments are being taken d From our example D = 1.30 = 2.45 d 0.53 Therefore too if many increments are being taken, the above exercise is repeated over another nine days so as to verify whether the above results are stable. If the above conclusions repeat themselves, adjustment of the number of increments is carried out as indicated in the table IV below. Table III: Continuous sampling - Adjustment of the number of increments. The sampling frequency used to obtain the duplicate samples for the coking coal into the Opencast boat bunkers was every thirty minutes interval. Generally intervals of one hour have been used to meet the desired precision levels. A checking procedure is used to ensure that the conclusion of the previous two successive test have not changed.Every sampling point should be checked for sampling precision yearly but suspect areas should bechecked more frequently than this. If during the checking procedure the ratio D/d are outside the limit 2.0 and 0.67 the increments are adjusted and full duplicate sampling is resumed until two successive test of 10 give ratios D/d between 2.0 and 0.67. The checking procedure is then carried out at the stated frequency. Table II can be applied to any quality characteristic. For example, to obtain the phosphorus value in coal of + 0.005, look for a row which give values for a precision of + 0.05 % and divide by 100. Therefore if the determination of phosphorus value is required to an accuracy of + 0.005 for 20 units, the expected value is 0.018. As indicated earlier on, the sampling precision is not necessarily the sample as sampling accuracy. Sampling precision can be good even if there is biased sampling, this results in inaccurate sampling. It is only and only if bias has been removed that precision and accuracy mean the same thing. The method of testing for bias is shown in the PROCEDURE section 5.0. Revision N°: 1 Issue Date: 15 June 2010 Table IV Number of samples required from each point to test bias. B = Amount of bias to be detected A = Precision of a single sample/increment n = Minimum number of samples/increments required for each point. The above will be illustrated with following example; On a particular day the Opencast Mine was delivering coking coal from its boat bunkers directly into the No2 Processing Plant. It was decided to test the relative bias of sampling point WPC11 (Conveyor) to sampling point WS26 (Plant feed) over 3 days i.e. 1day equals 1consignment. The desired minimum amount of bias and precision were set each at 1%. The number of samples required from Table IV is 9 i.e. B/A = 0.5/0.5=1. Thus eight hourly increments derived from each point, for each shift were composited to give one sample over the 3days to give a total of 9 samples from each point. The samples so obtained were prepared and analysed and the results are as tabulated on Table V. Estimated bias = Z =f/n = -7.5/9 = -0.83. Variance of Z =Vz = 1/(n-1) [g - f x f / n] = 1/8 [9.98 - ( -7.5 x -7.5 / 9)] = 0.47 Its precision n= t ( Vz / n )½.It is obtained from the following Table VI. Revision N°: 1 Issue Date: 15 June 2010 Therefore n = t ( Vz / n ) = 2.31 ( 0.47/9 )½ = 0.65 Z is the estimated bias, if its sign is ignored (i.e. modulus) and if it is greater than n, it is therefore significant. In this example, modulus of Z is greater than n i.e. 0.83 > 0.65.Thus it has been established that bias is significant and it lies between 0.83 + 0.65. ie 0.18 and 1.48. Thus the original assertion of 1% bias is correct. Further samples are collected to obtain a more precise estimate of bias. Revision N°: 1 Issue Date: 15 June 2010 EXHIBIT I. Check sheet for Opencast samples: Date. EXHIBIT II. Check sheet for Raw coal samples near No2 Processing: Date. Revision N°: 1 Issue Date: 15 June 2010 EXHIBIT III. Check sheet for No2 Processing coal samples : Date. Dry and Wet Grades
Washed grades
12.0 REFERENCES ISO 13909-2, ISO 13909-3, ISO 13909-7, BS 1017: Part 1:1989

Source: http://hwangecolliery.co.zw/iso/qa/coalcoke/Sampling%20QATM%203.pdf