understanding and running an effective qa/qc program · understanding and running an effective...
TRANSCRIPT
Understanding and Running an Effective
QA/QC Program
Mineral Resource ReviewSt. John’s, Newfoundland & Labrador
November 4, 2015
2
Agenda
Introduction Fundamentals of a QA/QC Program Selecting a Laboratory Blanks Duplicates Standard Reference Material Questions
3Introduction
Todd McCracken, P. Geo. 24 years in the Mining Industry 12 years in operations 6 years in exploration 6 years consulting Companies include: WSP, Tetra Tech, Vale, Cameco, Royal Oak,
Hudbay
Experience in: Gold, Silver, Nickel Sulphides, Platinum Group Elements, Copper,
Lead, Zinc, Molybdenum, Vanadium, Uranium, Li-Rare Earth, and the list continues to grow
Creating Resource Estimates for past 12 years
Designing and monitoring QA/QC programs for past 15 years
FUNDAMENTALS OF A QA/QC PROGRAM
5Fundamentals
Definitions
Accuracy: the degree to which an analysis or mean of analyses approach a “true” concentration
Precision: an estimate of the reproducibility of the sample and analytical system
6Fundamentals
Accurate Precise Accurate & Precise
Comparison of Accuracy and Precision
Source: After Smee and Associates
7Fundamentals
Definitions
Blank sample: a sample that is barren of the key elements being analysed by the laboratory; results should be below detection limit for the element in question
Duplicate sample: a sample with the same spatial relationship as the original sample
Standard reference materials: material manufactured by an independent company, or in-house, to create a sample with known analytical results
8Fundamentals
What are the typical perceptions of QA/QC?
Only do it because we have to
Waste of project money
No time to do that
We can do it later
9Fundamentals
Why do we need to understand the
QA/QC Program and have it run in an
effective manner
10Fundamentals
It is difference between this…
11Fundamentals
And this…
12Fundamentals
Or this…
13Fundamentals
Or this.
ANALYTICAL QA/QC: SELECTING AN ANALYTICAL LABORATORY
15Analytical QA/QC – Lab Selection
Analytical programs should involve two or more laboratories: one acting as the primary laboratory for each project, and one or more additional laboratories serving as the check laboratories
16Analytical QA/QC – Lab Selection
The primary laboratory receives the majority of the analytical work for that project
The check laboratory receives only the check assay work
An umpire laboratory is one used by two parties to form a consensus on the grade of material, such as a shipped ore or concentrate
This term is also sometimes used to describe a third laboratory used to determine which of two disagreeing laboratories is likely to be correct
17Analytical QA/QC – Lab Selection
Primary laboratory selection is based upon a number of factors, including
Quality of work
Price
Convenience
Service− not necessarily in that order
18Analytical QA/QC – Lab Selection
The primary laboratory must provide results that are sufficiently timely, precise, and accurate for the project’s needs at a competitive cost
Precision is typically not as good as that offered by umpire laboratories, but must be sufficient for use in ore reserve estimates
Accuracy should be sufficient for use in resource modeling without involving correction factors
19Analytical QA/QC – Lab Selection
Submit a number of standard reference materials to each of the candidates
At least 5 or 6 different materials should be submitted
Minimum of at least 2 different submissions should be made on different days
The number of samples of each material should be varied within each submission
Total number of samples submitted should be equal when the results of all submissions are combined
The assay procedure selected should be the one being considered for work and similar with all candidates
20Analytical QA/QC – Lab Selection
How do you score the labs performance based on the results received?
21Analytical QA/QC – Lab Selection
Need to compare and rank accuracy and precision Rating accuracy The laboratory with the mean value closest to the certified value
ranks first in accuracy, the second-nearest is ranked second, and so on
Ranks are easily assigned by calculating the absolute value of the difference between the laboratory’s mean and the certified value, then sorting by this calculated field
Rating precision The standard deviation or variance obtained by each laboratory
on each reference material can be calculated
22Analytical QA/QC – Lab Selection
23Analytical QA/QC – Lab Selection
Check laboratory Not only should it check the accuracy of the analytical results, it
should check the following
Pulp fineness− every 10th sample submitted for check assay− using 10 or 20 g, check percentage of material passing
through a specified sieve size− discard material after use
Order and condition of the sample transmittal− undamaged packets− legibly marked− correct order− accompanied by a sample transmittal sheet
ANALYTICAL QA/QC: BLANKS
25Analytical QA/QC - Blanks
Each laboratory will have a standard preparation procedure, which typically includes blowing compressed air through the equipment or running silica through the equipment at a set frequency
Purpose of the blank is to monitor cross-contamination or missed sequencing during the sample preparation or analysis
26Analytical QA/QC - Blanks
Blanks are typically of a material harder than the host rock of the target mineralization
Blanks should be devoid of the elements undergoing analysis
Should as closely resemble the primary sample material in size
River rocks Slate rejects
27Analytical QA/QC - Blanks
Drill core blanks are most effective at masking the intention of the sample
Commercial blanks are not recommended as the product has already been crushed and pulverized
Bags of decorative garden stone work well if no other material is available
Granite quarry Feldspar
28Analytical QA/QC - Blanks
Making drill core blanks From the drill logs, determine a rock type that is consistently at or
below detection limit for the elements is question Select several tens of metres of core of the desired rock type Avoid any occurrences of shearing, alteration, veining, or
mineralization Cut or split the entire core in half and place the material in pails Randomly select pieces from the pail to create a minimum of
10 separate 0.5 kg samples Properly tag and ship the removed sample for immediate analysis at
the laboratory Analyse for the key elements using the same procedure at the
primary laboratory. Failure of the batch is indicated if more than 20% of the samples
return assay values greater than 2 times the detection limit
Repeat the process to create additional blank material
29Analytical QA/QC - Blanks
Making field blanks Place up to 5 bags of the purchased garden stone into a cement mixer
or on a clean surface
Thoroughly mix the material together to ensure a homogeneous blend of the sample
Place approximately 0.5 kg of the garden stone into the same type of clean samples bags to be used during the drilling program
Place the sample bags into 5-gallon plastic pails for storage, removing every 10th sample for analysis
Properly tag and ship the removed sample for immediate analysis at the laboratory Analyse for the key elements using the same procedure at the
primary laboratory Failure of the batch is indicated if more than 20% of the samples
return assay values greater than 2 times the detection limit
Repeat the process to create additional blank material
30Analytical QA/QC - Blanks
Blank samples are typically inserted into the sample stream at a set interval (every 20th sample) when reviewing company QA/QC programs
Blank samples should actually be placed within or immediately after a mineralized interval if possible This may mean adjusting the insertion location
− can be difficult to track
31Analytical QA/QC - Blanks
Blank charting Important to note the detection limit for the element and the analysis
conducted
The failure line is typically plotted at 2 or 3 times the detection limit Failure if the blank values return 2 times detection limit 80% of
the time
Laboratories use 5 times the mean of the previous 2 or 3 samples as a guide
Review the assay certificate to see if any of the failures occur after a high-grade sample
32Analytical QA/QC - Blanks
33Analytical QA/QC - Blanks
• Is this chart an example of a major failure with this blank material?
• Analytical detection limit of 2 ppm (0.0002%).
• Mine average Zn grade is 2.86 %.
• Material is not “blank”.
• Maximum reading is <100 ppm (<0.01%).
ANALYTICAL QA/QC: DUPLICATES
35Analytical QA/QC - Duplicates
Duplicates A sample split from the original
sample (split from a bag of RC chips, or reject riffle-split sample)
Prepared and analysed as separate samples and issued with a unique sample number
36Analytical QA/QC - Duplicates
Field check sample (“field duplicate”) A second channel sample (chipped on
same site) or field split of diamond core (opposite quarters) or other sample material (riffle split of RC chips)
Prepared and analysed as separate samples and issued with a unique sample number
37Analytical QA/QC - Duplicates
Lab duplicate Taken from the reject sample (after
crushing and riffle splitting), but before grinding (about 300 g)
Prepared further and analysed as separate samples and issued with a unique sample number
The lab duplicate would measure the variance of sampling the initial field sample after crushing
38Analytical QA/QC - Duplicates
Second splits A second pulp portion (about 100 g at)
taken from the pulp produced from grinding (taken as would be normal for a routine sample)
Generally implemented routinely by all laboratories at a rate of about 1 in 20 as part of routine QC
Do not have a unique sample number
Second splits measure the variance of sampling the 300 g pulp
39Analytical QA/QC - Duplicates
Repeat / Replicate
Taken from the final sample pulp
Replicates are generally implemented routinely by all laboratories at a rate of about 1 in 20 as part of routine QC and do not have a unique sample number
Replicate analysis measures the variance of sampling the 100 g splits
40Analytical QA/QC - Duplicates
Rotary chips are the best representative sample split
Split core is not a true field duplicate
Splitting Core
Grade of A = B,
but ¼ core grade C ≠ D,
and grade of C ≠ A.
HETEROGENEOUS HOMOGENOUS
Grades equal in any quadrant
¼ core
½ core
A B
CD
41Analytical QA/QC - Duplicates
Coarse reject: resubmit a split of the reject Conduct internally, or have the lab generate
Pulp reject: resubmit pulp sample as a complete package and ask lab to return unused portion
OriginalSampleReject Pail
NewSampleReject
Empty into pail
Shake pail
Half of sample
placed in new sample
bag
Remaining sample returned to original sample bag
42Analytical QA/QC - Duplicates
Duplicate chart
Acceptable limits +/-20%
Typical duplicate chart
Keep scale on both axis the same
Can show sample swaps
Breaks down at low detection limits
43Analytical QA/QC - Duplicates
Mean difference duplicate chart
Duplicate results are averaged and the difference of duplicates
Failure line is plotted at +/- 20% difference limit
Chart makes it difficult to determine if samples are swapped
-3.000
-2.000
-1.000
0.000
1.000
2.000
3.000
0.00 0.50 1.00 1.50 2.00 2.50 3.00
Rela
tive
Diffe
renc
e of
Dup
licat
es
Mean Grade of Duplicates (g/T)
Mean Vs. Difference PlotData20%-20%Fail
44Analytical QA/QC - Duplicates
Relative difference chart Acceptable chart
80% <0.1 90% <0.2
Chart works only if enough data is collected
45Analytical QA/QC - Duplicates
Field Duplicate Lab Duplicate
46Analytical QA/QC - Duplicates
Two things to notice on this chart
47Analytical QA/QC - Duplicates
Sample swap
48Analytical QA/QC - Duplicates
Accuracy and precision of FA/AA above 3 g/t
ANALYTICAL QA/QC: STANDARD REFERENCE MATERIAL (SRM)
50Analytical QA/QC - SRM
Certified Reference Material (CRM) vs. Standard Reference Material (SRM)
Commercial purchased or created internally
Selection of SRM
Minimum of 3 SRM in a program
Charting SRM
51Analytical QA/QC - SRM
CRM vs. SRM
CRM are certified
Not all standards are certified
Standards are not required to be certified
Best not to use the word “Certified” unless you are sure
52Analytical QA/QC - SRM
List of commercial SRM retailers located globally Brammer Standards Canmet Minerals Exploration & Environmental Geochemical Shea Clark Smith Nevada Bureau of Mines West Coast Minerals CDN Resource Laboratories LQSI Geostats Rocklabs
Ore Research & Exploration
53Analytical QA/QC - SRM
Most common is the small sealed packet with a fixed weight 50 g or 100 g
Larger containers require you to repackage the material
Commercial SRM
54Analytical QA/QC - SRM
Understand the SRM Certificate
Header information
Viewed the most and misunderstood
Method of analysis for this SRM
Who created the SRM
Expected value of SRM
Some certificates only indicate 1 Std.
Dev.
The precision of the SRM
55Analytical QA/QC - SRM
Understand the SRM Certificate Information most people miss or ignore
ORIGIN OF REFERENCE MATERIAL: Standard as prepared using ore supplied by Barrick Gold Inc.
from their Bald Mountain Mine in Nevada, USA. It is Carlin Style Mineralization in the prolific Battle Mountain-Eureka Trend in Northern Nevada, USA. The material is from breccias near the contact between the Mississippian Pilot Shale and the underlying Devonian Guilmette formation. Near the center of the system, micron-sized native Au occurs with base metal sulfides and sulfosalts. In peripheral deposits and in later stages of mineralization, Au is typically submicron in size and resides in pyrite or arsenopyrite.
Would you use this SRM for a shear-hosted gold deposit in Newfoundland?
56Analytical QA/QC - SRM
Understand the SRM certificate Information most people miss or ignore
METHOD OF PREPARATION: Reject ore material was dried, crushed, pulverized and then
passed through a 270 mesh screen. The +270 material was discarded. The -270 material was mixed for 5 days in a double-cone blender. Splits were taken and sent to 14 commercial laboratories for round robin assaying.
Some certificates will show the results of the round robin test
57Analytical QA/QC - SRM
Certificates usually contain a mean value, a standard deviation, and a 95% Confidence Interval (95% CI)
The mean value is the target grade for the material
The standard deviation is the variation expected in the results;68% of results should be within 1 standard deviation of the mean, 95.5% should be within 2, and 99.7% should be within 3
The 95% confidence interval (95% CI) is an indication of the quality of the material and the certification test work; issuers are 95% confident that the true mean of the material lies within this interval of the mean This is not a control limit to be used
58Analytical QA/QC - SRM
Custom SRM
Less expensive in the long run than commercial SRM
SRM material matches the deposit characteristics
Specifically determines the appropriate grades to test
Several suppliers will generate a custom SRM from the material provided There are several papers available on how to generate a proper
custom SRM
Most laboratory companies with provide guidance as well
59Analytical QA/QC - SRM
Round Robin results
10 analyses completed at each of the 14 labs submitted
Means range from 0.198 to 0.232.
60Analytical QA/QC - SRM
Selection of SRM material
Material has a matrix identical to the samples being assayed
Has extremely low heterogeneity
Has a value within the range used for categorizing
A reputation for being reliable and accurate
61Analytical QA/QC - SRM
Selection of SRM Material A suite of standards brackets every category boundary, combined with
every important difference in sample matrix and change in laboratory procedure effecting other samples
Important procedural steps in the laboratory, such as dilution steps or changes in instrument set-up and calibration standards
Important sample matrix differences almost always include oxide and sulfide samples, because fire assay procedures utilize flux adjustments and/or sample weight changes when processing sulfide-rich samples, and copper projects often have assays to help distinguish copper oxide from other ore types
Other potential matrix differences might include high and low levels of copper, silver, platinum group elements, bismuth, telluride, etc., presence or absence of hydrated minerals, high and low amounts of total sulphur in non-oxide samples, varying levels of carbon
Minimum of three SRM per program.
62Analytical QA/QC - SRM
Charting SRM results
Place in sequential time order
Determine accepted value and standard deviation from the SRM certificate
The standard deviation on the certificate can be misleading
63Analytical QA/QC - SRM
Mean of all SRM Assays
Mean + 2 St D
Accepted Value of
SRM
SRM Assay Results
Mean - 2 Std D
64Analytical QA/QC - SRM
UCL
One time failure usually not an
issue
LCL
3 times the median of the relative difference Concern of failure
is due to 3 failure in the moving range
65Analytical QA/QC - SRM
Lab Drift
66Analytical QA/QC - SRM
Insertion rates – no set standard
QUESTIONS?