We’re scaling functional, low-sugar pectin gummies and I’m eyeing an inline NIR moisture/solids sensor right after the continuous vacuum cooker, targeting 78–80 Brix and about 19% moisture at about 200 kg/h to shave hours off drying without cooking out heat-sensitive actives. If you’ve run NIR in-line vs just a refractometer with acids/colors injected at 88–90 C post-cook, did you tighten texture and reduce QC pulls, or did calibration drift/fouling make it not worth the hassle?
We put an inline NIR right after the vacuum cooker on a 200 kg/h pectin line and the single biggest win was adding a short static mixer in a bypass loop before the probe — knocked down microbubbles and gave rock-steady reads at “78–80 Brix”. That let us push to about 19% moisture and shave a couple hours off drying, but if your acids/colors hit at 88–90 C, keep the probe upstream or include them in the calibration since dye shifts wrecked our model.
Building on @simonB78, the bigger unlock for us was treating the NIR like a temperature instrument: put it pre‑acid/color, jacket the sapphire window to about 90 °C and feed the analyzer a temp tag, then calibrate to Karl Fischer at 3–4 moisture points so you can feed‑forward cooker vacuum and reliably hit about 78–80 °Bx/≈19% and cut dryer hours… Small caveat: when we tried post‑add, low pH and anthocyanins drifted the baseline, so we added a 5–10 s moving average to tame pump ripple and stuck with the pre‑add model — keep it warm and it behaves.
The unsexy fix that made our inline NIR usable on a 200 kg/h pectin line was a 2–3 s hot DI water micro‑rinse over the sapphire every 30–45 min, triggered when scatter spiked from film build. That kept the model tight enough to hold 78–80 Brix/≈19% without drift, and we still kept a cheap inline refractometer for Brix because dyes at 88–90 °C would occasionally nudge the NIR baseline. If you don’t want a wet rinse, a low dry‑air purge works, but only if the window’s jacketed near process temp or it fakes a lower moisture read. @simonB78 is right on pre‑acid placement; cleanliness was the bigger win for us.
Microbubbles were our silent killer right after the vacuum cooker; we added a small in-line deaerator and held about 0.3–0.5 bar backpressure ahead of the probe, and the noise dropped from ±0.8 to ±0.1 Brix at your post‑cook acid/color temps. @simonB78 nailed the mixing angle, but without de‑aeration the NIR kept lying whenever the cooker hiccupped — . If you won’t add a deaerator, at least mount the window in a pressurized spool so it doesn’t flash around 90 °C.