As a process analyser company operating in minerals processing environments, the reference used to check and calibrate our instruments accuracy is the field or process laboratory. There is no question that the instruments used in the laboratory are more accurate than ours given the same sample. What often lets everyone down and reduces the correlation between the field instrument measurements and the lab results is the sampling method.

Figure 1 shows a spent liquor application discharging past the field instrument and sample point from a one million litre tank.  The lab measurement is in blue and the field instrument measurement is in pink. In this process, the concentration physically cannot change by 2 gpl in 1 hourThe field instrument shows the normal steady rise and fall in concentration while there is noise in the lab measurement (changes in concentration of over 2gpl in 1 hr) caused by sampling errors.

Figure 1. Poor correlation lab vs field instrument concentration measurements

Figure 1. Poor correlation lab vs field instrument concentration measurements

In figure 2 far better results can be seen.  The lab and the field instrument are trending together.  There is no sawtoothing in the lab measurement and the field measurement is following with near perfect precision.

Figure 2. Example of a good correlation lab vs field instrument measurements

Figure 2. Example of a good correlation lab vs field instrument measurements

Below are the top 5 ways to improve the field sampling process when taking reference samples:

  1. Use a clean dry jar. We are pushing accuracies of 1 gpl with field instruments these days so if there is residual water or previous sample in the sample jar, it does not take much in a 400ml sample container to affect the accuracy we are trying to get. If the application is critical, surely it can justify a fresh container every sample (if proper wash and dry facility is not available).
  2. Drain the sample line for at least 15 seconds. Cross contamination of precipitated product in a sample drain line is devastating to a fresh sample. This product has been sitting for hours at ambient temperature and has evaporated and changed chemically between samples. Give the line a really good long high volume flush then take the sample over 4 or 5 stabs at the flowing sample. This will provide a small average of the liquor coming out of the line and it will be more representative.
  3. Cap the sample or use a thermos flask. Evaporation of hot samples will start immediately. In addition, precipitation in the vessel can start to be seen if the sample is hot and the vessel cold. Capping the sample will help. Using a flask will further aid the sample keeping as close to process temperature for as long as possible. If you introduce a safety risk by capping a sample (danger of the vessel building up pressure) open it in a fume cabinet just prior to handing it over to the lab.
  4. Don’t delay. Once you have taken the sample, don’t head off for morning break or stop and answer your phone. Head straight for the lab to keep the chemical properties of the sample as close to process conditions as possible. If you are the person taking a Bayer digestion sample that will be used as a reference for charge control, at that moment and until you deliver the sample, your job escalates to the most important job at the refinery.
  5. Repeat what you always do. If you take more than one sample, make sure that you take the samples in the same order every time. If your first sample is normally the sump pump discharge and your last is the spent liquor after the evaporators, make that route the same every time. Regression maths that is used to calculate the differences between field measurement and lab samples can compensate for fixed offsets, but it can’t compensate for variations that are not predictable or repeatable. So doing the same thing every time in all aspects of sample taking is crucial.

The above is my two cents worth from the field perspective on improving the sample taking process to ensure samples are representative and a closer correlation field instrument measurements vs lab results can be achieved.  I would be interested in hearing any comments on the above and on how the sample is best handled once it makes its way to the lab.

For more information contact our technical team at sales@plapl.com.au or call us on +61 3 9786 1711.

Category: SmartDiver