Monitoring grain and flour quality

28 October 202211 min reading

Over the past 30 years, I have been involved in developing instrumentation that measures grain, flour, semolina, rice, corn, soybeans and many other products. This article describes several instruments that can provide millers with tools that can measure protein, oil, moisture, starch, milling quality and physical characteristics in the incoming raw materials, the process streams and the final product.

Phillip Clancy
Next Instruments
Sydney, Australia


There are different opinions on the amount of measurement that is required to produce an excellent product. The famous statistician, W. Edwards Deming, stated “Quality comes not from inspection, but from improvement of the production process.” He implies that increasing the number of tests of final product will only result in finding more defects. On the other hand, Peter Druker’s famous statement, “You cannot manage what you cannot measure”, implies that more measurements are better. These two quotes are not contradictory but complimentary. You need to make measurements of the process stream and the final product so that you can use the data to improve the process and thereby the final product. If you only test the output and do nothing to change the process, then you cannot expect to change the quality of your product.

Over the past 30 years, I have been involved in developing instrumentation that measures grain, flour, semolina, rice, corn, soybeans and many other products. This article describes several instruments that can provide millers with tools that can measure protein, oil, moisture, starch, milling quality and physical characteristics in the incoming raw materials, the process streams and the final product.


Near Infrared Spectroscopy and Image Analysis are two tools that are available for millers to make rapid and continuous measurements within a milling process. Near Infrared Spectroscopy, NIR, measures protein, oil, moisture and starch in grains as well as the final products, i.e., flour, semolina, rice flour, soy flour, soymeal, corn flour, corn meal, pasta, noodles, bread etc. NIR can also provide a measurement for ash, starch damage and water absorption in flour and semolina. Image analysis provides millers with a qualitative assessment of physical parameters such as, blacktip in wheat, black specks in semolina and flour, rice kernel size and colour, crowning and horneous endosperm in corn and others.


Starting at the grain intake platform in a mill, NIR analysers have been used for decades to measure the protein, gluten and moisture in wheat, oats and barley, protein, amylose and moisture in rice, protein, oil and moisture in corn, soybeans and oil seeds. It is most common that Near Infrared Transmission (NIT) analyser be used to make measurements of whole grains and seeds. Near Infrared Reflectance (NIR) analysers can also be used however the seeds are first ground in a mill and presented to the analyser as a powder. Figure 1 shows a diagram of how NIT and NIR analysers operate. Spectra of whole and ground wheat, rice, corn and soybeans are shown in figure 2. Note that the wavelength scales differ for NIT and NIR analysers.

Protein, moisture, oil and starch absorb infrared energy within the spectral region from 720 to 2500nm. The amount of energy absorbed at the resonant frequencies for Nitrogen-Hydrogen bonds (Protein), Oxygen-Hydrogen bonds (Water and Sugars) and the Carbon-Hydrogen bonds (Oil/Fat), are proportional to the concentration of each component. Only N-H, C-H and O-H bonds show absorption bands in the NIR region. Salts and minerals do not exhibit absorption bands in the NIR region and as such cannot be measured directly using NIR. However other chemical compounds in grains and oil seeds may be measured indirectly.

In Europe, W and Zeleny are common measurements associated with wheat. W is a measure of the hardness of grains and Zeleny is a measure of the quality of the grain based on a sedimentation rate of flour suspended in water. There are no single NIR absorption bands for W nor Zeleny, however both W and Zeleny correlate well with grain moisture content, protein strength and grain conditioning. As such, it is reasonable to accept that NIR can be used to measure W and Zeleny since NIR measures protein, starch and moisture.

Likewise, water absorption and starch damage in flour cannot be measured directly based on specific absorption bands but can be measured based on the overall interaction between NIR energy and protein, starch and moisture.

Falling Number is a very common parameter used by millers and grain processors to evaluate the enzymatic breakdown (Amylase Activity) in the grains. Millers require wheat that has a certain Falling Number value as this parameter indicates if the seeds have begun to germinate. High amylase activity reduces the available starch in the seeds and a reduction in the yield and quality of the flour. Falling Number Apparatus consists of a constant temperature bath in which a glass tube is placed. The tube is filled with a dilute solution of water and ground grain. A ball is dropped into the tube and sensors at the bottom detect how long the ball took to drop a specific length. The longer the time period the higher the Falling Number value and the less Amylase Activity indicated.

The Falling Number Apparatus is difficult to master and the poor precision for most tests results in a broad range for determining low or high Amylase Activity. The question is whether NIR can be used to measure or estimate the Falling Number Apparatus. There are no NIR absorption bands that correlate directly with Falling Number, however starch, protein and water contents will influence the viscosity of a water flour suspension and thereby should correlate to the Falling Number test. Many NIR instrument manufacturers supply calibrations which include Falling Number, however the error of these NIR calibrations or methods can be quite large and should only be used a rough guide. Typically, a NIR calibration for Falling Number has a Standard Error of Prediction between 30 and 50 seconds. When the useable Falling Number value for most wheat flour operation is greater than 250 seconds, the NIR calibration only provides a 95% confidence level between 190 and 310 seconds.


Near Infrared Spectroscopy lends itself to in line or continuous measurement of quality parameters because it is rapid and non-destructive. The challenge is how to capture the NIR spectra from a flowing stream of wheat, rice, corn, soybean, flour and meals.

Figure 3 shows an in-line NIT analyser (CropScan 3000S In Line Analyser) installed in the intake silo of a flour mill in Australia. The analyser was originally developed for measuring whole grains in a combine harvester but has been adapted to the in-line measurement within the flour mill. As grain passes down a shute, it falls into the sample head of the CropScan 3000S where it is trapped for 3-5 seconds. A NIT scan is performed and the grain is released back into the stream of grains. The measurement of the NIR spectra are performed using a Silicon Diode Array Spectrometer and a Fibre Optic Cable to connect the Sample Head and the Spectrometer. Protein and Moisture readings are averaged for the load of grain that has been dumped into the receival silo and the data is displayed on the computer screen located in the mill laboratory. Figure 4 shows the plot of the Protein data from the CropScan 3000S compared with the laboratory’s benchtop NIR analyser (Foss Infratech 1241).

In line analysis of the flour streams is also an established application for NIR. In this case Near Infrared Reflectance analysers are used. Companies including Buhler, Buchii, Foss and Perten offer in line NIR analysers for measuring flour streams. Typically, these measurements are made through a window in a pipe or shute where the flour is moving past the NIR analyser. The major problem with such a setup is that the packing density of the flour varies. To improve the consistency of presentation of the sample of flour to the analyser, various chokes and traps have been implemented. The best device is where the flour is trapped, compressed against the window and then released. Such a device introduces a mechanical element into the process stream however it does provide a more consistent and reliable measurements.

There are several suppliers of grain colour sorting machines, i.e., Satake, Buhler etc. These machines rapidly sort and segregate grains based on colour so that grains which are weather damaged, exhibit blacktip or are fungal stained, are rejected before milling. In line measurement of aleurone and bran specks in flour streams is also available, i.e., Branscan Flouroscan F4000, which uses UV fluorescence to detect these components that indicate the quality and purity of the flour or semolina.

Moisture is very critical component in flour milling. Wheat is conditioned to 15-16% moisture and allowed to stand for many hours before milling. A NIR in line analyser was developed to measure the water content of wheat as it was being conditioned. The objective was to better control the final moisture content. The difficulty in making this measurement lies in that the water is not bound in the grain but lies on the outside of the grain until it is absorbed which takes many hours.


Physical characteristics of seeds are important quality parameter for the milling industry. The presence of blacktip and blackpoint on wheat and barley affect the quality of flour and malt. Fungal stains, black, red and green streaks in rice effect the quality of the milled rice. Broken grains, chipped grains, white spots, chalkiness etc are all parameters that affect the quality of the final milled product. Since these parameters are visual effects rather than chemical defects, then NIR and UV sensors cannot measure them. The human eye and brain can make these measurements, however Image Analysis provides a rapid measurement tool which can be more cost effective and consistent than humans.

Image analysis is based on collecting a visual scan or image of the seeds and then applying computer based algorithms to measure a wide range of physical parameters, including, length, width, thickness, broken seeds, chipped seeds, colour, colour defects, fungal stains, chalkiness, yellow berry, and many other parameters that are calculated from these primary measurements. Figure 5 shows an image collected for wheat placed on a tray using the SeedCount SC6000 Image Analyser. The SeedCount software measures 12 parameters in approximately 45 seconds as shown in the picture. The SeedCount has several application modules including long grain rice, medium and arborio rice, wheat, barley, durum, canola, sorghum, corn, and coffee. Each module requires different sample trays and measures different parameters.

An Italian company, SCE s.r.l, offers an application module for SeedCount called Specktek which detects black specks in fluor and semolina. The flour or semolina is spread out on a SeedCount tray an inserted into the SeedCount scanner. N image is collected and the Specktek software detects black specks and displays the data on the SeedCount screen. Figure 6 shows a typical Specktek analysis.

The Branscan Flouroscan is also an image analyser that is designed for powered materials such as flour and semolina. The Flouroscan measures the quality of the milling process by detecting the number and size of aleurone and bran particles in the flour or semolina.

Satake and Kett Instruments offer specialised rice image analysis systems that measure whiteness index which is used to evaluate the milling process. Other image analysers include the EyeFoss (Foss, Denmark), MARVITECH (Germany), Vibe Q3mi (VIBE, USA).