GB/T 2007.4-2008General rules for the sampling and sample preparation of minerals in bulk - Experimental methods for checking the precision and bias (English PDF)

4.2.1 Select reference method

4.2.1.1 The recognized reference sampling method is the stop-belt sampling method, that is, stop the loaded conveyor belt, from the specified position, in the ore flow on the stopped conveyor belt, take each sample set from the full cross-section; the intercepted length in the direction of the conveyor belt shall be greater than 3 times the nominal maximum particle size or at least 30 mm, whichever is larger. Take out all the ore in the intercepted area as a reference sample set.

Note: According to the site conditions and the requirements of 4.1.1, or the appropriate method can also be designated as the reference method through negotiation by the relevant parties.

4.2.1.2 When checking the sample preparation method, use the first method (riffle method) in GB/T 2007.2 as the reference method. When calibrating the reducer, take all discarded samples as reference samples.

4.2.3 Determine the quality characteristics of test

Generally, the main component of the ore is selected as the quality characteristic of the test. It may also choose particle size or other chemical and physical testing items as quality characteristics.

4.2.5 Determine the number of test groups required

The number of pre-tested groups K shall not be less than 20 groups. The required number of test groups is calculated and determined according to the

Where: Sd - The standard bias of the difference; K - The number of pairs of measured values.

4.4.5 Determine the number of pairs required

The required number of pairs of sample sets, the number of duplicate samples or the number of large samples are determined according to the precision of the sampling method, the uniformity of the ore, the maximum allowable bias value.

4.4.5.1 Make the K pairs of data into a list. Calculate the difference, average difference, standard bias of the difference according to formula (1), formula (2), formula (3).

4.4.5.2 Calculate the coefficient D according to formula (4):

Where: delta - Maximum allowable bias.

4.4.5.3 Check out the required number of pairs of measured values, nt, (corresponding to the D value) from Table 1.

When the state nt <= 20, calculate the value of t0 according to 4.4.6; perform an additional test when nt > 20.

5.1.3 Size of sample set

According to the maximum particle size of the ore, it shall be carried out in accordance with the requirements of GB/T 2007.1.

5.1.4 Sampling interval

Divide the batch by 2n to get the sampling interval (if it takes n sample sets, divide the batch size by n); discard the decimal.

5.1.5 Preparation of specimens and determination

The preparation of the specimen is carried out in accordance with the provisions of GB/T 2007.2. The determination is carried out in accordance with relevant standards.

5.2 Test method

5.2.1 Sampling

According to the needs of the checking, for the sampling methods to be checked, take sample in accordance with the relevant standards (systematic sampling method, stratified sampling method, secondary sampling method), to get the large samples A and B.

5.2.2 Sample preparation and determination

According to the purpose of the checking test, select the following method 1, method 2 or method 3 for sample preparation and determination.

Divide the large sample A and B, to obtain four specimens of A1, A2, B1, B2.

Conduct a double sample determination for each specimen. Method 1 can obtain the precision of sampling, sample preparation and determination (see Figure 1).

5.2.2.1 Method 1

C.1 Product name: Bauxite

C.2 Measurement item: Al2O3

C.3 Number of test batches: 15 batches.

C.4 Batch size: An average of 418 tons per batch. The minimum batch size is 320 t and the maximum batch size is 500 t. C.5 Maximum particle size: 150 mm. C.6 Sample set size: 15 kg.

C.7 The number of sample sets taken for each batch of alumina: 2n = 20 x 2 = 40

C.8 Sampling method: In the process of loading the ore on the ship, take sample sets from the grab surface at the calculated interval; place the odd number of sample sets in container A, to form a large sample A; place the even number of sample sets into another container B, to form a large sample B. Each of large samples A and B contains 20 sample sets.

C.9 Sample preparation: Crush the two large samples A and B respectively by a jaw crusher to less than 22.4 mm; after thorough mixing them, divide them into four groups of samples A1, A2 and B1, B2, about 150 kg each, using a riffle according to the Method 1 of 5.2.2.1. Then prepare four component analysis samples A1, A2, B1, B2 according to the procedures as specified in GB/T 2009.

C.10 Precision requirements: beta SPM = ±1.0%, beta S = ±0.6%

C.11 Data analysis (95% probability): 15 batches of test data are listed in Table C.1.

C.11.1 According to formula (6) and formula (7), calculate the average value and range of the double test; fill the results in the rows of Xij and R1 of Table C.1.

C.11.2 According to formula (8) and formula (9), calculate the average value and range of component analysis samples A1, A2 and B1, B2; fill the results in the rows and R2 of Table C.1.

C.11.3 According to formula (10) and formula (11), calculate the average value and range of large sample A and B; fill the results in the rows and R3 of Table C.1.

C.11.4 According to formula (12), formula (13), formula (14), calculate the average value,, of the range.

C.11.5 According to formula (15), formula (16), formula (17), calculate the control upper limit of R1, R2, R3:

Upper limit of R1: x 3.267 = 0.100 x 3.267 = 0.3267 Upper limit of R2: x 3.267 = 0.2636 x 3.267 = 0.8612 Upper limit of R3: x 3.267 = 0.323 x 3.267 = 1.055 Make a chart and mark all data in the R control chart (see Figure C.1).

Check whether the values of R1, R2, R3 are in the control state. After checking, all R3 are in the control state. The 15th batch of B1 component analysis sample's R1 value of 0.37 and the 9th batch of B2 large sample's R2 value of 1.43 are abnormal.

After checking, the 15th batch of B component analytical samples splashed during boiling before titration in a test during the determination process. Before preparing specimen of the 9th batch of B2 large samples, contamination occurs because the crusher was not cleaned, so the above two data can be discarded. After discarding, is 0.096, the upper control limit is:

0.096 × 3.267 = 0.314 is 0.223, the upper control limit is:

0.223 × 3.267 = 0.7285 After discarding, the remaining values of R1 are all less than 0.314; the values of R2 are all less than 0.7285. According to formulas (18), (19), (20), calculate the estimated values of determination, reduction, sampling standard bias: […]