#2: Fire Assay gap analysis April 2006

Proposed by

lemoene

Proposal type

Gap Analysis

Assigned to release

• Bika Fire Assay 1

State

ready-for-merge

Definitions

See Glossary in the Documentation section

Gap analysis
The process of determining, documenting, and approving the variance between business requirements and system capabilities in terms of packaged application features and technical architecture. www.georgetown.edu

Fire assay
www.newmont.com

Fire assay involves adding a ground portion of the solid (or carbon) to a crucible containing a flux. Flux is used to lower the melting point of the samples to ensure the entire sample becomes liquid during firing

The flux contains borax, soda ash, silica flour, litharge (lead oxide) and silver nitrate in various quantities. In the firing stage everything turns liquid and the lead oxide forms into small globules of metallic lead

The globules fall through the liquid and form an amalgam with any other metals it comes into contact with, including gold and silver. This all collects in the bottom of the crucible

After about an hour at 1,000 degrees Celsius the contents of the crucible are poured into a conical mould to cool. Again the lead, being more dense than other material, quickly settles to the tip of the conical mould and solidifies along with the glass-like rock remains

When cooled, the lead is separated from the glass and then cupelled. Cupellation involves placing the lead ‘button’ into a pre-heated cupel at 1,000 degrees C. A cupel is a small cup made of magnesium sulphate which is a material that is capable of absorbing the lead

At this temperature the metals liquify and lead is absorbed into the cupel. Gold and silver having a higher surface tension than lead, are not absorbed and remain as a small ‘prill’ in the bottom of the cupel

The cupel and prill are removed from the furnace and allowed to cool. The prill is then dissolved in aqua regia. The resulting solution is aspirated into an AA Spectrophotometer to determine the gold content

Control Samples
In the context of this analysis, a control sample indicates all or any of Blanks, Duplicates or Standard samples
Blanks contain none of the mineral assayed for, typically quartz for gold ore
Duplicates indicate duplicate samples in the same tray
Standard samples indicate 'Grade Control' samples of which the analyses are known

Motivation

Requested by sponsor

Assumptions

• The system is built in Plone on an MSSQL database shared with other applications
• All attemps will be made to retain ZODB functionality for subsequent roll-outs - SQL will slot in easily
• The client maintains the MSSQL server and ODBC drivers
  
• Some samples are analised more than once. Up to 5% of fire assay analyses are routinely rejected in QC. Standard samples are analysed over and over. Therefor Samples and Analysis Requests (ARs) are maintained in a 1 to many relationship in the system. Technically in the texts here further-on 'Samples' mostly indicate current Bika ARs
  
• 100 samples per day
• For the purpose of this analysis, a gold mine laboratory

Proposal

List and detail the gap between current Bika LIMS 1.1 and Bika Fire Assay 1.0

Implementation

A design based on the gap analysis will be built in html wireframe including a running Bika LIMS demo customised via the schema editor for geology samples

Deliverables

Analysis workflow gap

Sample Preparation Stage

The main gap here is in a considerable sample preparation stage - standard Bika workflow catches up at data capturing again and remains the same there-after

Much of the preparation stages are currently exclusively done on printed works sheets - Bika job cards - with sign-offs after every stage. In the analysis here, the new system does not capture and track theses statuses

This must be addressed should a sponsor require - either as data captured from the printed sheet at the end of the preparation stage or where conditions and computer litteracy allows, in 'realtime'. The latter will be good for status tracking and performance analysis

Though most work flow proceed per work sheet, standard AR workflow is maintained

Gap summary

Bika Fire Assay Actions Current Bika LIMS Actions Status Change User LIMS interaction 1 Register Gap Sample instruction by other department. Picked up by the LIMS in the database 2 Sample Gap The sample arrives from the field but not at lab yet. 3 Record Create AR Sample due Lab clerks Clients on the shared database 4 Wet weight Gap Technician 5 Dry none 6 Catch weight Gap Technician 7 Flux none 8 Fuse Gap Weigh none 8 Hammer none 10 Cupelate none 11 Digest none 12 Results capture AA spectrometer Instrument Interface Work sheet download Gap to be verified Technician 13 Validate Verify Gap: Auto rejection on duplicates, blank or standards verified Manager 14 Report Publish publish Clerk

Sampling

This will be separated into 2 stages in Fire Assay 2.0 - often the sample container is created and barcoded first with instructions to the 'sampler' as to where the sample must be collected. The sample itself is collected often days later

No gap. This data is currently captured by other departments in the database from where it can be picked up by the LIMS if necessary, all in status Sample due

Some samples never arrive - such records are 'deleted' in the database by their creators. A neater option would be to give these another status, 'expired', and don't surface them except on a special view/report. Would help towards accreditation?

Receipt

Samples arrive barcoded by the 'sampler'

Bar-code scanning triggers the 'receive' action and 'next field' navigation

The scan has to pass validation before the sample goes to 'in the lab'. The bar-code is validated for i) correct format and ii) record exists

Rejections are logged and there is a standard report available to list them with

Standard functionality to create sample records here remains - some times records are not created earlier. It is possible to create samples without barcodes and edit sample title if previously created

Weighing workflow starts, the bag weight is captured. Sample receipt reports are available  at this point - insufficient sample material, sample received/not received, etc.

Upon successful scanning, samples are divided with the first half going to the lab and the other, for possible re-testing, to Short term retention shelve position, storage at an address made up of: Storage_Area, Hut, Rack, Shelve, Box, NumberinBox

Drying

This process entails the removal of any water residue/moisture from the sample. The samples go into an oven at around  105 C, and barcode strips are added to the drying containers for tracking

Some labs might require the recording of the wet sample weight for calculating sample % moisture

Catch Weight

Weighing out exactly 50 g of a sample for analysis is a time consuming process and as a productivity measure the 'catch weight' principle is applied. As soon as the sample weight is within acceptable limits, the weight is captured, 'next field' navigation triggered

Any difference with the desired weight is factored into calculations further down stream as a 'concentration factor'

Weighing takes place on interfaced balances that fires a value down its port as soon as it has stabilised on a valid weight

• valid catch weight limits are set in the set-up
• out of range values are reported immediately
  
• balances are interfaced and as soon as a reading stabilises, puts a reading out
  

Fluxing

A process used in the heating of metals which may be intended to reduce or eliminate oxidation, confine the products of oxidation, reduce their melting point, and improve fluidity of surface metal layers - www.wmrc.uiuc.edu

No LIMS interaction at this stag

Fusion

Again heated up in an oven to remove non-gold particles. At the end of the fusion process the gold-glass remainder is weighed and captured

Hammering

Physical removal of glass around gold residue. In phase 1 no LIMS interaction

Cupelation

Furnace phase, no interaction with LIMS. In this process the metals are being separated, those oxidised and the non-oxidised gold. Involves adding lead to the sample, the easily oxidisable lead takes away most of the base impurities. www.sciencemadness.org

The gold is separated as Prills,'small pieces of pure precious metal extracted from assay samples'

Digestion

Extraction of the gold particles through dissolution - acid is added to the non-oxidised metals part. For Fire Assay 1 no LIMS interaction or gap

Data capturing - AA Analysis

Sample finally gets analised by atomic absorption spectrometry on an instrument such as the Analytik Jena AA

These instruments typically has a bi-directional interface - previously prepared worksheets are uploaded from the LIMS onto the AA and the results returned from there, all in .csv format

Values returned via the interface to works sheets may be manually overwritten - this is already the case in Bika LIMS, no gap

On receipt of the results, the LIMS puts the data thru auto validation, checking the values for duplicates, blanks and control samples against valid ranges set-up earlier. Any values out of range, would point to contamination or incorrect sample positioning and the work sheet is automatically rejected and the manager alerted. See below

Sample Catch weights are factored into final values here

After inspection the results are submitted for verification

There could be more than one AA and worksheets get assigned to a specific machine. The instrument ID is kept on the WS for QC purposes

Calibrations/ drift corrections are recommended from control sample data

All file movements are logged

Gaps
    Worksheet download
    Auto validation
    File up/download log
    Some extra fields on worksheets

Validation or Retraction, Reporting

In this phase the worksheets are verified. QC, Duplicate and Control Samples' results are compared. Anomalies here prevents the works sheet from being validated

If no anomalies, all values are on spec, the work sheet is automatically validated and released/published

On average 5% of work sheets fail at this stage, fire assaying is a messy business prone to contamination

In the event of anomalies -

• the statuses of the samples & worksheet go all the way back to catch-weight
  

• the worksheet maybe edited, ie the duplicate was in the wrong tray position - by a labmanager and manually validated
  

• Out of range values are clearly indicated on the worksheet with different shading
  

        below spec - yellow
        above spec - red
        in spec - green

Gap: re-testing workflow. To be detailed

New Roles

To be detailed, soon ...
    Sampler
    Hut manager
    Supervisor (semi literate)

New Modules

Samples

Some sample records will be created in the SQL db by other departments

Fields gap

• batch number -samples are collected in batches
• barcode
• Sample ID prefix includes a code indicating where it originated from
• Storage address made up of: Storage_Area, Hut, Rack, Shelve, Box, NumberinBox
• oven tray position

Views gap

• by batch number - batches look-up and view pages
• results per batch
  

Control Samples

Results data per instrument/analyst are available in tables and graphs for blanks, duplicates and standard samples

Work sheet templates

A work sheet is created by a supervisor, by putting the samples on the work sheet in chronological order. A works sheet generally gets restricted to samples from one client/department only

Some departments will compile their own worksheets and submit them as spreadsheets, typically .xls. The system imports these as .csv

The work sheet layout - and NB, its ID is determined by the size of the trays used in the ovens used in the work flow. Tray sizes can be configured in the system set-up, typically 4 columns by 6 rows, and this 24 sample tray is used here

• tray position is captured for each sample. The positions of control samples are indicated on worksheets - results are returned against the control samples' expected values
• the 24 samples will always include control samples:

1. at least 1 'blank' - containing none of the mineral assayed for, typically quartz for gold ore
2. 2 QC/Grade Control samples
3. 2 duplicates - duplicate sample IDs on the same work sheet are catered for

• If fewer than the designated number of samples, 19 + 5, values are returned from the AA, the work sheet is rejected
  

Dynamic work sheet sorting allows for sorting on Control type so Duplicates and QC samples can be displayed next to one another

Worksheets are printed for sign-offs during preparation stages - physically they represent the trays of samples traveling through the lab

The worksheet ID is determined by the position of coppered samples in the tray - they show up blue and provide easy tray recognition

<Graphic to display this principle goes here>

Worksheet is exported in CSV format for upload by Analytical Instrument, typically an atomic absorption spectrometer

New Set-up objects

• Laboratory Instruments
• Control samples
•  Valid ranges

          duplicates, blanks, control samples
          per client - criteria differ per department
            

• Batches
• Samples storage facilities. Controlled by other departments - they should have access to them in the system. Hut managers
• Oven tray sizes
  

Reports

Several reports were mentioned in other paragraphs up to here, they'll be searched for and listed here again

1. Sample receipt reports - insufficient sample material, sample received/not received, etc.
2. QC reports and graphs - control samples, per analyst and instrument
3. Instrument performance (QC)
4. Barcode rejections - sample list
   
5. Turn-around times
6. Statistical Report (to be specced)

Optional

Instrument Maintenance
Consumables Inventory - with alarms for low stock levels