NIR in the Dairy Plant: Fat, Protein and Lactose

Liquid Dairy: From the Tanker Bay to Pasteurized Product

Dairy processing is a continuous standardization problem. The fat content of incoming milk varies by farm, season, and breed — and every downstream product is made from milk split and blended back to a precise fat target. NIR at the tanker intake gives the full picture before unloading. At the standardizer, it verifies the fat target was actually hit.

Powder Production: Evaporator to Finished Dairy Powder

Spray drying amplifies every quality deviation from the liquid stage. Moisture going into the dryer directly affects both the energy cost and the final powder moisture, which must land in a tight window — typically 3.5–4.5% for WPC, under 5% for NFDM. NIR before the dryer lets operators adjust evaporator density in real time. At powder QC, NIR confirms protein grade and moisture before release.

Dairy NIR in Practice — Eight Specific Deployment Points

NIR isn't one technology deployed the same way everywhere in a dairy plant. The measurement mode, calibration requirements, and decision it supports differ at each stage. Here's how it actually works at each point.

Tanker Bay Testing: Know What You're Paying For Before It Unloads

Raw milk varies by farm, season, and breed — sometimes within the same collection route. Fat content in spring milk from Holstein cows can run 3.2–3.6% while late-lactation autumn milk from the same herd reaches 4.2%. If you're standardizing to 3.5% fat and the incoming tanker is at 3.1%, every litre of cream you add to compensate costs margin. NIR at the intake bay gives you the full picture — fat, protein, lactose, and total solids — before a single litre is unloaded. It also flags potential adulteration: added water suppresses solids-not-fat (SNF) in a pattern NIR detects rapidly as a screening tool; confirmation of adulteration uses cryoscopy as the regulatory reference method.

Hitting Fat Targets at the Standardization Line — Not Guessing After Separation

Standardization is where raw milk becomes product. You split it — cream goes one way, skim the other — then blend back to a legal fat target: 3.5% for whole milk in most markets, 1.5% for semi-skim. Getting this wrong means rework or a mislabeled product. NIR inline or at-line on the standardized stream closes the loop on what the separator and blending valve actually delivered, not what the recipe assumed. In plants running continuous standardizers with closed-loop control, NIR is what drives the valve — not the lab.

Cheesemaking: Moisture at the Vat Determines Your Yield

Cheese yield is directly tied to moisture retention in the curd. A 1% swing in final cheese moisture on a 10-tonne batch is hundreds of kilograms of product that either exist or don't. Traditional moisture testing on cheese requires oven drying — an hour minimum, useless for in-process decisions. NIR on curd at the vat and on pressed cheese gives the cheesemaker a result while the press is still loaded. Fat in dry matter (FDM) is the grade specification buyers use, and salt content determines both flavor and regulatory compliance. A single NIR scan returns all three after pressing without destroying the sample.

Butter Production: Moisture Control Before You Hit the Legal Limit

Legal maximum moisture in butter is 16% in most markets — the Codex Alimentarius and EU standard. That ceiling is closer than it sounds — overworked butter or excess water addition during working can push moisture toward 15.5% before operators notice. The traditional check is gravimetric drying, which takes two hours and reports history, not the current churn. NIR during the working phase gives moisture readings in real time, while there's still time to adjust the working speed or water-addition rate. Salt distribution in butter is also measurable by NIR and matters for quality consistency across different production batches.

WPC Grading Before Spray Drying — Not After You've Already Run It Through

WPC 34 and WPC 80 are different products with different customers, different prices, and different label requirements. The distinction is made by the ultrafiltration process before spray drying — but the only way to confirm you're on track for an 80% protein product is to measure the concentrate going into the dryer. NIR on liquid concentrate before spray drying gives protein and lactose content in under a minute. Catching a borderline concentrate before the dryer runs means you can divert to WPC 34 instead of producing an off-spec WPC 80 batch and downgrading the whole lot.

NFDM and Skim Milk Powder: Moisture Release Testing at Scale

Non-fat dry milk is traded globally and blended at multiple points in the supply chain. The specification that matters most at packaging is moisture: CODEX and most buyer specs require maximum 4–5% moisture, and caking in storage begins above that threshold. Traditional oven drying on NFDM takes 2–5 hours and tests history — the batch has already been packed by the time you have the number. NIR on finished powder before packing gives a moisture result in 30 seconds, allowing every bag to be cleared rather than sampling one in fifty. In a plant running two 5-tonne batches per shift, that difference in test frequency is material.

MPC for Infant Formula and Clinical Nutrition: Tolerances That Don't Move

Milk protein concentrate going into infant formula operates under tighter tolerances than most dairy products. MPC 70 means 70% protein on a dry basis — and the buyer's specification window might be ±0.5%. A calibration drift of 0.8% on your NIR means you've been shipping borderline product for a week before the lab check catches it. This is one application where NIR replaces convenience rather than replaces accuracy — the lab Kjeldahl still runs as the reference, but NIR screens every batch so only borderline results go to the full reference analysis. The calibration validation regime here is also more rigorous than a commodity cheese plant.

Lactose Powder: A Narrow Moisture Window With No Margin for Error

Food-grade alpha-lactose monohydrate has a moisture window of roughly 4.5–5.5%. Too dry and it becomes electrostatically charged — dusty, difficult to convey, and prone to segregation in blends. Too wet and it begins to transform to anhydrous lactose, which blocks screens and causes caking. Specialty lactose grades are held to an even tighter window. NIR is the only at-line method fast enough to monitor moisture during the drying and milling stages. Crystallization degree is also predictable from the NIR spectrum — a parameter that matters for downstream handling and tableting performance.

One Plant, Two Different Measurement Setups

Dairy NIR spans two fundamentally different physical setups. Liquid milk and liquid whey are measured in transmission — NIR light passes through a defined pathlength of the liquid, typically 0.1–1 mm for whole milk. Pathlength control is critical: milk scatters light heavily due to fat globules, so the optical path must be consistent across every reading. Most dedicated dairy analyzers and FT-NIR systems with liquid cells operate this way, with the sample homogenized and temperature-equilibrated to 40°C before measurement.

Liquids — Transmission Geometry

• Raw milk, cream, skim milk, liquid whey, WPC concentrate

• Pathlength 0.1–1 mm — fat globules scatter heavily at longer paths

• Sample temperature equilibrated to 40°C before measurement

• Homogenization required — inhomogeneous fat affects readings

• Flow-through cell for inline or at-line continuous monitoring

Powders and Solids — Diffuse Reflectance

• NFDM, WPC powder, MPC, lactose, cheese, butter

• Consistent fill depth and packing pressure in sample cup

• Cool powder to room temperature before scanning — heat causes moisture drift

• Particle size must be consistent — mill samples to standard fineness

• Reference spectra updated when process or supplier changes