40
Measurement and Testing
Figure 3: Repeat injections over 3 years for a reference oil standard: original fi les on the left and the same data chemometrics-corrected on the right. The use of correlation optimized warping to correct for retention time variability, here processed by a commercial package called LineUp from Infometrix.
Multivariate calibrations of instrument systems are performed when the instrument is fi rst set up, but often suffer as either the instrument quality drifts, or the process conditions change.
We did this work better 20 years ago than we do today. Now, the computed result is monitored but there is little or no considerations of any of the quality diagnostics inherent in the processing. This is exacerbated by the lack of training: apparently an expense and a time commitment to be avoided.
There have been advances in chemometric processing and with some simple concepts integrated into the communication loop, the calibration procedures can be streamlined and automated. It is possible to integrate spectroscopic measurements with nearly the same simplicity as the effort with which we put simple temperature sensors in place.
Conclusions
The oft-quoted statement “you can’t control what you don’t measure” leads easily to “the quality of your measurement dictates the quality of your control”. We can further break
down measurement quality into two sub-parts: 1) the precision and specifi city of the measurement; and 2) the system we employ to extract the information content from these data. In the hydrocarbon processing industry, we have three primary sources of data to utilize, with the collection of univariate sensors (typically temperature, pressure, fl ow, and level), spectrometers (mostly optical), and chromatographs (mostly GCs).
Each of these data source categories require a calibration step to enter the world of multivariate control.
• For a collection of process variables, we would expect the models to last a long time; most of the upset warnings will be tied to failure in individual sensors.
• In chromatography, much of the variability is solved using correlation-based chemometric alignment, giving these analyzers vastly-lower calibration requirements and giving us the opportunity for instrument interchangeability and a common interpretive base instrument to instrument or even plant to plant.
• Even though spectral analysis has long been tied to multivariate analysis, few companies are taking advantage
of tools that can simplify calibration, ease the burdens of maintenance, and improve model quality to be less dependent on different operator experience levels.
The discussion in this paper addressed the information extraction process and how leveraging chemometric analysis can evaluate the data more quickly and automatically. The output of this analysis is objective, can be validated, and is a universal approach. The benefi t is a simplifi ed procedure, freeing up analyst time for other tasks. We need to invest in new technologies if we hope to make this all work; we cannot do things like we used to. Think of how many analyzers we deploy that have identical function; we need to manage them in a global way. Chemometric tools are off-the-shelf that facilitate this management task and will work with most computer- sentient legacy equipment. That really does make the transition to multivariate essentially free and takes us a major step beyond the current analyzer maintenance “IIoT” focus on simply digitizing paper control charts. It also enables a streamlining and simplifi cation of the calibration process and provides an objective, automatable evaluation system.
Figure 4: Dashboards (in this case using Tableau On-Line) can provide visualization techniques to track all data and all process models across the organization in one place. Combined with automated and optimized creation of calibration models, here using Infometrix’ Ai-Metrix system, true quality control of the quality control process can be achieved.
Author Contact Details
Brian Rohrback - President, Infometrix, Inc. • 11807 North Creek Parkway S, Suite B-11, Bothell, WA 98011, USA • Tel +1 425 402-1450 x118 • Email:
brian_rohrback@infometrix.com • Web:
www.infometrix.com
PIN Annual Buyers’ Guide 2023
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44 |
Page 45 |
Page 46 |
Page 47 |
Page 48 |
Page 49 |
Page 50 |
Page 51 |
Page 52 |
Page 53 |
Page 54 |
Page 55 |
Page 56 |
Page 57 |
Page 58 |
Page 59 |
Page 60 |
Page 61 |
Page 62 |
Page 63 |
Page 64 |
Page 65 |
Page 66 |
Page 67 |
Page 68 |
Page 69 |
Page 70 |
Page 71 |
Page 72 |
Page 73 |
Page 74 |
Page 75 |
Page 76
