Moisture content directly controls stoichiometry in chemical dosing skids, where a 0.5% error in active concentration shifts the mass of active ingredient delivered to process water by an equivalent amount. Selection therefore starts with three engineering questions: what fluid is being dosed, what dose rate window must the analyzer track, and where in the skid the analyzer will physically mount.
The dominant sensing families in service as of mid-2026 are loss-on-drying halogen or infrared balances, Karl Fischer titration, and near-infrared (NIR) optical analyzers — each with a different answer to those three questions. The sections below line those families up against chemical compatibility, response time, and skid integration.
Why moisture matters at the skid
On a chemical feed skid the chemical itself — sodium hypochlorite, polymer emulsion, coagulant, or acid — is the variable being metered, and the dose rate is computed from a target concentration [S6]. If the active concentration in the day tank drifts because of water pickup, evaporation, or reaction, the flow meter downstream still delivers the right volume but the wrong mass of active ingredient. Moisture measurement closes that loop by letting the PLC adjust stroke length or pump speed to a measured concentration rather than a nominal one.
Core and Main's municipal water operator guide treats chemical characteristics as the primary driver of pump technology: diaphragm pumps for corrosive liquids because contact between chemical and mechanism is minimized, peristaltic pumps for sodium hypochlorite specifically because they are not susceptible to out-gassing air lock [S6]. The same logic applies to moisture analyzer selection — the fluid dictates wetted materials, sensor head geometry, and whether inline or grab-sample measurement is even feasible.
Three sensing families and what each does well
Loss-on-drying analyzers — halogen or infrared heated chambers plus a precision balance — determine moisture by measuring weight loss over time under controlled heat, with results repeatable only when sample weight, drying temperature, drying mode, and an automatic switch-off criterion are held constant [S3]. They are robust, work on powders, slurries, and pastes, and are the workhorse in food, plastics, and pharmaceutical QC where government mandates such as FDA moisture limits apply [S1]. Their weakness is cycle time: typical runs run from several minutes upward, which is too slow for closed-loop control on a fast dosing skid.
Karl Fischer titration determines water by reacting it with a reagent containing iodine, sulfur dioxide, a base, and a solvent, then measuring the quantity of reagent consumed [S2]. It is the method of choice for products containing volatiles, amines, or aldehydes where loss-on-drying would over-report moisture [S10]. Inline Karl Fischer probes exist, but most skid-side deployments are sidestream or grab-sample because the titration cell needs reagent replenishment and waste handling.
Near-infrared (NIR) analyzers are optical, non-contact, and read in seconds, which is why they appear on production lines where quick readings during processing are needed [S8]. The catch is calibration: NIR is an indirect method that must be calibrated against a reference method (typically loss-on-drying or Karl Fischer), and instrument readings typically maintain variance around 1.5 or fewer standard deviations of that reference [S8]. On a dosing skid, NIR suits a continuous sidestream cell where the optical head can see a flowing film or a quartz flow cell piped off the industrial valve manifold.
Selection criteria that survive the skid environment
Mettler Toledo frames the four parameters that govern any loss-on-drying measurement — sample weight, drying temperature, drying mode, and switch-off criterion — as the levers that decide whether two readings from the same analyzer will agree [S3]. DSC Balances adds a fifth: the type of moisture analyzer (halogen versus traditional drying oven) and its principle and sensitivity, because the two technologies do not interchange for the same sample [S4]. LabPro's four-question checklist reduces to sample type, accuracy, speed, budget [S5], and Lab Manager extends the sample-type axis to cover volatiles, which require special handling such as lower drying temperatures or sealed crucibles, while also flagging that multiple operators with varying experience affect result consistency [S10].
Translated to a dosing skid, those criteria narrow to six concrete questions before any vendor is named: (1) Is the sample a liquid, slurry, or powder? (2) Does the chemistry contain volatiles, aldehydes, or amines that would bias loss-on-drying? (3) What response time does the dose-control loop in the PLC require? (4) What is the installation space — inline spool, sidestream cell, or bench grab sample? (5) Does the analyzer need to be cleaned in place, and what wetted materials survive the chemical? (6) How many operators will run the device, and what is their training depth [S10]?
Comparing the three options against dosing-skid criteria
Three options dominate dosing-skid installations — loss-on-drying, Karl Fischer titration, and NIR — and the deciding factors line up against cost, response time, chemistry compatibility, and skid-integration geometry [S3][S4][S8][S10].
Loss-on-drying (halogen or IR balance): cost low to mid, response time slow (minutes range), chemistry compatibility wide for non-volatiles, skid integration grab-sample or sidestream batch. Karl Fischer titration: cost mid to high, response time minutes, chemistry compatibility best for water-specific measurement, skid integration sidestream or bench. NIR optical: cost mid, response time seconds, chemistry compatibility requires empirical calibration per fluid, skid integration inline optical cell on the flow meter return line. The pick depends on whether the skid's pressure transmitter loop is fast enough to use a 2-minute Karl Fischer reading as feedback, or only fast enough for a 5-second NIR pass.
Limitations, failure modes, and what to watch
Loss-on-drying over-reports moisture when the sample contains volatiles other than water — solvents, plasticizers, low-molecular-weight alcohols — because those species also evaporate at the chosen drying temperature [S10]. Karl Fischer drift comes from reagent age, humidity ingress into the titration cell, and side reactions with strong acids or ketones; reagent conditioning and routine standardization are mandatory, not optional. NIR drift comes from the calibration itself: window fouling, source aging, and changes in the sample matrix push the reading outside the band where the original calibration model holds, and instruments must be re-referenced to a primary method at a defined interval [S8].
Halogen and traditional drying oven analyzers are not interchangeable, and the calibration philosophy for one does not transfer to the other because they operate on different principles and sensitivities [S4]. For a skid that ships product against an FDA-style or trade-association moisture specification [S1], the method stated in that specification is the method that has to be used in QC — even if it is slower or more operator-intensive than what the PLC feedback loop would prefer.
Standards, sourcing, and what to put on the datasheet
Where moisture is a regulated release parameter, the published specification in the relevant pharmacopeia, food standard, or industry code drives the analyzer choice — not the other way around. FDA is named in [S1] as a government agency that mandates moisture content as a key quality control criterion in food and pharmaceutical applications, and similar industry-specific rules apply to fuels, lubricants, and reagents. The analyzer datasheet needs to confirm detection limit against that published number, supported by repeatability and reproducibility figures from the vendor.
Calibration discipline — the calibration frequency, the reference method, the temperature and humidity environment, and the SOP that ties the two together — is the deciding factor between two analyzers that otherwise look identical on paper [S4]. On a dosing skid, that SOP also has to cover the industrial valve sequence that routes sample to the analyzer and back, so a sample taken at the wrong point in a dose cycle produces a number the flow meter cannot use.