Analysis of complex interdependencies between soil-specific properties and cleaning behavior in the food industry by using a decision tree

Abstract Plate heat exchangers are commonly used for food processing, but tend to foul rapidly. Frequent cleaning guarantees product safety and quality, but accounts for high consumption of water, chemicals, energy, and time. Targeted soil characterization is essential for cleaning cost reduction. The effects of soil properties on cleaning are difficult to estimate since the interdependencies are complex. Principal component analysis and a decision tree were used to comprehend this complexity. A variety of carbohydrates formed the basis of the statistical exploration (eight native and modified starches, guar gum, xanthan, pudding, and whey protein isolate). Method adjustments allowed the characterization of soils during contact with different cleaning fluids. Solubility, fluid binding capacity, swelling-induced thickness increase, rheological properties, and pull-off forces were characterized. The cleaning time was obtained from planar channel flow experiments. The decision tree revealed an ordered relevance of the soil properties on cleaning. The effects are largely determined by the active cleaning mechanisms, which were quantified to help explain the results.