Sweet and low moisture measurement
15 Jan 2000
If sugar is allowed to become too moist, it tends to set into lumps, causing severe problems for storage and handling. In the worst case, an entire storage silo or tanker of sugar can form into a single lump. It is essential to the manufacturer that moisture content is controlled throughout the process and that final moisture is checked before shipping.
Recently, long term trials with a major UK manufacturer have resulted in successful, stable and accurate on-line measurement of very low final moisture levels in the range 0.04-0.12%. Replacing the traditional Karl-Fisher laboratory analysis with a continuous, on-line measurement, the manufacturer has for the first time been able to remove all possibility of damp sugar being shipped or stored.
In developing this application, Infrared Engineering's engineers had to consider the complex nature of moisture in sugar, and the normal variations in the crystals. Moisture occurs in sugar in free, bound and surface forms, and it is important that the NIR gauge measures the total moisture content. A good quality sugar contains tiny particles of sugar dust or fines which give it its distinctive 'sparkle'. A sugar of lesser quality may contain particles of varying sizes and colours. The NIR measurement must be able to cope with both.
A clear sugar crystal has a long optical path. This yields a high resolution measurement owing to the outstanding signal to noise ratio and low gain requirement of the NIR instrument.
Experiments were initiated on samples of sugar received from the manufacturer. Filters and reference filters were selected and an initial calculation algorithm chosen, based on experience with other NIR measurements. Samples of sugar of various different qualities and grades (castor, granulated, industrial etc) were then divided in two, and half presented to the NIR gauge, and half analysed using the traditional Karl-Fisher technique.
The initial results showed the measurement to be sensitive to the varying scatter characteristics of different sized sugar particles, and those which had suffered surface scratching during silo conditioning and storage. To correct this, engineers altered the gauge to include a continuous measurement Gp40 of the actual optical path length through the crystals, obtained by measuring the NIR absorption of the sugar. This figure was incorporated into a new algorithm, which was further modified using mathematical analysis, to achieve a good linear correlation with the Karl-Fisher results.
Long term trials, over a 12 month period at the manufacturer's site, demonstrated an on-line accuracy of 0.006% and showed the measurement to be highly repeatable indeed better than the manual Karl-Fisher technique. Further work was carried out on instrument calibrations for the individual grades. Although it was established that the highest accuracy was always obtained when individual sugar grade calibrations were used, it was feasible to use a single calibration of all the white castor and granulated sugars. Industrial grades needed separate calibrations.
As a result of the trials, the manufacture has invested in five MM55 NIR sensors from Infrared Engineering for on-line monitoring of final moisture content in refined white sugar.
