January 21, 2016

A New Approach to Online Data Validation

By Jeff Gunnell

How would you like an online analyzer that, after commissioning, starts up already in perfect calibration? That during its life, checks that it is in calibration without needing test samples? That works for very long periods without maintenance, but when eventually intervention is required, the analyzer raises a flag, gives a description of why maintenance is needed, and explains exactly what to do? Would this be desirable? Of course! But is it feasible?

Cascade Technologies analyzers from Emerson open up a powerful new approach to real-time validation. They are based on Quantum Cascade Lasers (QCLs): semiconductor devices which produce radiation in the mid infrared range. For spectroscopy, this is the fingerprint region where most molecules exhibit strong, sharp absorptions.


An Information Rich Spectral Region Showing the Absorptions of Many Components


It might seem that congested spectra would make reliable measurements difficult to achieve, but in fact the opposite is the case. Using the complete set of information in a spectrum can reveal sources of error and hence remove the need for validation tests.

In order to develop a measurement application, first the absorption spectra of all components that are active in the measurement region must be obtained. This is done in the development lab and the information created is called a spectral database: it includes wavelengths, absorptions, and peak shapes as well as the effects that temperature and pressure have on spectra.

To make a measurement on a process stream, a spectrum is obtained. A mathematical process is then used to work out the concentrations of the components in the stream. An initial estimate of the stream composition is combined with the spectral databases to calculate a spectrum. This is compared to the measured spectrum and the result is used to generate a better estimate. As this process is repeated, a path is followed which eventually leads to the calculated spectrum almost exactly equalling the measured spectrum, and at that point, the estimated concentrations of the components almost exactly equal the real concentrations.


The Measured and Calculated Spectra of CO Plus Ethylene with the Residuals


This iterative methodology gives a novel way of validating measurement integrity. The process automatically takes into account all of the factors which could affect a spectrum. It starts with fundamental, unchangeable properties of the components in a sample, such as the absorptivity at each wavelength across a spectrum and the effects of temperature and pressure on peak shapes. It combines these properties in a way which matches the measured spectrum and looks at how close the calculated spectrum is to the measured one. A poor match is evidence of a failure somewhere in the system. For example, if an unexpected new component is present, that will be detected by failure to model peak shapes correctly and the details of why the match failed will indicate that it was due to something new in the stream. Similarly, such effects as laser drift can also be readily identified.

In addition to raising alarms when a failure is present, there is another powerful advantage of this approach. If the quality of fit is high, that is strong, positive evidence that the analyzer is operating perfectly; a good fit could never be obtained unless every part of the system, including its calibration, was working correctly.

It is the case, backed by field experience that the iterative approach is extremely robust and leads to very high reliability. The capability to have very long periods between maintenance visits, positive evidence of good operation, and smart identification of problems – together with a highly stable calibration – makes this technology especially valuable when analyzers are installed remotely from skilled maintenance resources, such as on natural gas pipelines. These are also important characteristics for legislative reporting, such as CEMS measurements. In a production environment robust, reliable measurements minimize losses so QCL technology is advantageous for online monitoring in manufacturing industries as diverse as petrochemicals and pharmaceuticals.

Emerson’s Cascade Technologies QCLs are innovative spectroscopic analyzers which provide multicomponent measurements at high sensitivity and with extremely high reliability. You can find out more about the technology in this blog: Cascade Technologies QCL Analyzers. And if you have a potential application that might be solved by this approach, please contact the Emerson team to discuss your needs.