Choose your Adventure: Taking NIR from the bench to production

This editorial on NIR calibration transfer from benchtop to production was submitted by Mark Sullivan, a Senior NIR Applications Specialist for BUCHI North America.

Not long after many new users become accustomed to the speed and convenience of laboratory NIR over conventional techniques, it dawns on them that this technology is inherently suited for use directly in production. In fact, NIR is a common technology for enabling the cultural change of bringing the analyzer to the sample rather than transporting the sample to the analyzer. The burning question then is what can you be doing now with NIR application development to facilitate a future transition to NIR online?

Do you remember the old Choose Your Own Adventure books from grade school? This is the NIR blog version. Here’s what you’ll do: (1) Read through the primer on calibration transfer, (2) Choose between the two example plots provided, then (3) follow the prompts to discover a possible path forward from the lab to the process for the adventure you selected!

Foreword: A Primer on Calibration Transfer

The motivation to transfer the calibration to more benchtop or online analyzers grows alongside the confidence of a robust calibration. If you need help navigating the calibration development process, read through BUCHI Application Note No. 394/2020 “Guideline for development of NIR calibration models” linked here. Once you’ve established that the benchtop calibration performance is suitable for the intended application, it can be put into use. Just don’t forget… calibrations are not static; tweaks in formulations and changes in raw material suppliers or even lot-to-lot variability may require calibration update. Similarly, calibration transfer requires a calibration update.

Transferring calibrations between multiple units of the same (or similar) model is like reading the abridged story: short and sweet. Standardization between these instruments is straightforward. Things get more interesting as the number of variations increase (think: different hardware, different wavelength ranges, different sample orientation, different sampling conditions). Although it’s not uncommon for someone using an NIR from Vendor A to transfer to an NIR from Vendor B, a larger calibration transfer sample set or more sophisticated standardization may be required to get satisfactory results. Similarly, method transfer from an at-line to an on-line system will require special attention due to differences in sample presentation and other variables that change because of the online data collection scenario. The interface between the sample and the light source is almost always unique in the at-line case vs. the in-line case, even when both NIR products have the same specifications (e.g., same wavelengths, same detector).  In general, we would expect at least a slight loss of method accuracy in the transfer (relative to the original method on the at-line system, only) for this reason.  

Compensating for differences in the sample interface for a successful at-line to on-line calibration transfer will require additional sample spectra with known property values collected directly on the on-line system. This effort will usually require less time than the original calibration, in part because the calibration statistics are already known. After somewhere in the vicinity of 50-100 additional calibration samples have been collected with the online analyzer it will become more apparent as to how much of the transferred data should be retained.

Plot 1: The Great Escape from 3rd Party Testing

A small to mid-sized feed company that wants to reduce the cost of third-party testing of protein, moisture, protein, fat, and ash in finished products. Rapid in-house testing would also allow them to increase testing of incoming raw materials and to expand their capabilities to in-process intermediate testing, helping ensure that blended products meet their target assays. Their budget is limited. Of course, they want to make sure that a laboratory NIR will deliver the expected ROI before considering NIR online, but they anticipate rapid growth and can envision automated online testing in the future. As such, they want assurance that the vendor can provide that solution, as well.

Choose this adventure to blot out 3rd party testing.

Plot 2: The Case of the (Less) Mysteriously Replicating NIR Sites

A large polymer company that wants to install NIR online analyzers in multiple geographic locations to measure monomer composition in feedstocks, perform end group and intrinsic viscosity analysis to monitor average molecular weight in polymer beads and flake, and measure the color of the final product in real-time. The NIR calibration statistics have been previously estimated at a principal site. The challenge: NIR online analyzer installations will be performed in two stages 1) at line initially until the sample interfaces can be installed in the process lines during a shutdown period and 2) in line where the analyzers will be bolted onto the process

Choose this adventure to learn the secret to cloning NIR methods around the globe (or across your production facility).

Continue with Plot 1: The Great Escape from 3rd Party Testing

Here is a summary of some key elements from the backstory:

  1. Goal: Reduce third-party testing of finished products
  2. Samples: Finished petfood products, incoming raw materials, blends
  3. Special Considerations: Limited budget for online, but expecting rapid growth and envision automated online testing in the future

In terms of hardware, spoiler alert: we’ve read this story before! Photodiode array (PDA)-based analyzers have a proven track record for food and feed applications, plus they are more affordable relative to high-end lab FT systems. (Detour to this blog to get the full story on FT vs DA.) In terms of preparing for a bench to online calibration transfer, you can also benefit from the fact that there are two sampling orientations that can mimic online data collection scenarios, the so-called “up-view” and “down-view.”

Figure 1: Example Down View (left) and Up View (right) configurations of BUCHI ProxiMate NIR System. Note that in the Down View configuration, the light source is located above the sample and irradiates the top surface of the sample. In the Up View configuration, the light source is located under the sample and irradiates the bottom of the sample container.

When samples are collected in a plastic or glass cup and the NIR beam irradiates from the bottom of the cup (i.e. the light is “looking up at the sample” in up view.  Here, the sample plane is fixed and sample presentation is highly reproducible. This collection orientation is comparable to an online NIR mounted via a flange to the bottom of a chain conveyor. When used in conjunction with a transflectance adapter, liquids may also be measured using the up view option. This orientation is recommended for the best measurement repeatability or for researchers requiring the flexibility to measure both solids and liquids.

Illumination of a sample from above (“down view”) offers non-contact, remote measurements. The large beam diameter in this configuration is advantageous for very coarse samples like whole grains or heterogeneous mixtures and is the preferred geometry for an NIR online analyzer operating over a belt conveyor. In food processing plants or other workplaces where glass is not allowed and plastic containers will be used, the down view orientation has the advantage of avoiding spectral interference from the polymer composition of the sampling cup.

Because the initial feasibility was performed on a laboratory analyzer with the same spectroscopic engine as the NIR online analyzer, we have a high degree of assurance that the application can be performed online. Cliffhanger alert: It remains to be determined if the laboratory calibration data can be effectively transferred to the online analyzer because of inherent differences in sampling conditions (e.g., temperature, flow vs static) and sample presentation.

Continue with Plot 2: The Case of the (Less) Mysteriously Replicating NIR Sites

Here were some key elements in this back story:

  1. Goal: Install NIR online analyzers in multiple geographic locations
  2. Samples: feedstocks, polymer beads and flake
  3. Special Considerations: Two-stage NIR-Online implementation plan (at-line to in-line)

Since color is a major property of interest in this adventure, the best-fit hardware configuration is a process hardened photodiode array (PDA) analyzer with NIR and VIS spectral ranges. In this configuration, the visible range (350 – 920 nm) is measured simultaneously with NIR.

The best-fit in-line sampling interface would be a weld-in flange attached to a conveyor which would allow for direct contact analysis of the flowing stream of polymer beads. However, an online simulation cell (OSC) would be a good stand-in during the at-line trial.

Figure 2: NIR-Online X-One system in up-view configuration (left) and Online Simulation Cell (OSC, right).

An OSC is fixed to the NIR online analyzer supported by an up-view stand. The OSC can be filled with polymer beads (similar to a traditional sample cup) and is placed over the window of the analyzer to make a measurement. In this case the sample interface simulates the optical path of the in-line installation, although the sample temperature in the OSC is not controlled.

Because the spectrometer model is the same (and/or it is the identical analyzer) and the OSC closely matches the process sample interface the initial calibration data collected at line have a high probability for creating a successful online application. Cliffhanger: It remains to be determined if the calibration obtained at line can be applied directly when the analyzer is bolted onto the process or if corrections will be required to compensate for sample temperature differences.

How does your adventure story begin and end?

Irrespective of the adventure your applications and goals lead you to, your plot will have a guide. Our NIR Application specialists support everything from NIR sensor selection to sample preparation to calibration development. You may ask, “What about the earlier-mentioned cliffhangers?” BUCHI will work with you to navigate the efficient way to transfer and use calibrations between NIR sensors at the bench, on-line and in-line. And no, you do not require technical expertise in chemometrics and NIR spectroscopy. Our custom-made and you-centered training will equip you with all the necessary knowledge and tools to get the adventure going. Our dedicated, continuous support (and smartly designed tools like the auto-calibration feature AutoCal) will ensure that the journey is as smooth as possible. 

What would be an immediate next step? Explore our hardware solutions to achieve your goals for rapid at-line analysis with ProxiMate NIR, or process control with one of our NIR-Online sensors. When you’re ready to dig deeper, contact us for a tailored conversation aimed at meeting your specific needs.   

It’s time to start your adventure!

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

Website Powered by WordPress.com.

Up ↑

%d bloggers like this: