However, the quality of raw materials in the food industry fluctuates significantly. There are seasonal, climatic and geographic effects. This is clearly seen in the dairy industry where milk is collected over large areas which includes hills and plains, dry and humid conditions. The composition and quality of the milk changes daily and it is simply not possible to obtain an entirely consistent quality of product with a manufacturing process using set control parameters. The difficulty of tracing the cause of sudden changes in product quality remains a key problem for the food industry. Often, quality control in the New Zealand food industry consists in setting rigorous specifications for the final product and rejecting or recycling the failed product. Inherently this control mechanism is more wasteful than the alternative method: the prevention of products out of specification.

This second control method can be applied if we can identify a variable that characterises the material being processed and relates to the operation being controlled. Cascading control based on a floating set point strategy can only be applied once cause-effect relationships are understood. In the past, production research has centred on determining the effect of particular process operations on the properties of the final product. The only way to know whether you cooked your egg right is to eat it.

In order to apply a floating set point control one needs to:

• Identify a measurable property which can characterise the material being processed

• Investigate the effect of the process operation on this characteristic property of the intermediary product at the operation outcome

• Analyse the relationship between the property of the intermediary material processed in the particular unit operation being controlled and the quality of the final product.

At the 7^th World Congress on Engineering and Food held in the UK in 1997, Professors Lillford and Fryer gave an illustration of this more challenging approach. The product in question was ice cream, the quality smoothness. In order to understand which operation and control parameter was crucial in determining smoothness, structural analysis was performed. Researchers at Unilever were able to identify from SEM (scanning electron microscope) micrographs that water can crystallise in different sizes depending on its position in the food matrix. Having identified the target structure, the task was then for the engineer to develop a freezing process which led to crystallisation in the desired mode. That done, the freezing operation could then be controlled to a predetermined schedule to achieve smoothness.

For floating set point control some simple physical measurement must still be devised to measure the quality of the ice cream mix exiting the freezer. In many cases, the development of in-line measuring devices is difficult. For example, it has been argued for the past twenty years that the viscosity of milk concentrates entering the spray drier has a direct impact on the functional properties of the powder. This viscosity can be manipulated by heating the concentrate immediately before feeding into the drier. Thus the operation of the heater should be controlled not by a fixed temperature set point but by a measure of the viscosity which in turn decides the temperature set point. This cascading control is much more efficient in the manufacture of product of consistent functional quality. Unfortunately, the development of in-line viscometers for milk powder plants has not yet been successful and control at the present is still by fixed temperature set point and final product quality screening.

Glass box

There is, in my view, a definite need for a strategic shift in food production research, from black box, cook book experiments to more fundamental analysis of mechanisms and structures and how they link to specific physical properties and functional/sensory properties of the food. Examples of this approach do exist, as discussed above, but they are still few and far between. More effort is also required in the development of in-line sensors for the physical properties identified. The job of researchers is to convince industry that this is a better way to secure our competitive edge in the next century and introduce true food process control rather than just automatic control in the food industry. In the medium to long run a strategic programme based on unit operations and process analysis will prove more economically attractive.