The scale of cheese making has changed considerably since the time, legend has it, that the foodstuff was made in small quantities in animal-skin bags. Today cheese is made in vats thousands of gallons in size, with much better process control.Ideally, the monitoring of cheese making would be performed in real time and online, enabling processors to spot and correct problems quickly. Now scientists from University College Dublin in Ireland, from the University of Kentucky in Lexington and from Moorepark Food Research Centre in Cork, Ireland, have developed an optical sensor that promises to do just that. The system measures backscattered light as milk is coagulated and cut to produce liquid whey and the curds that eventually become cheese.A light source sends visible and near-IR light into a vat where cheese is being made. A spectrometer measures the backscattered light, and the resulting readings reveal characteristics of the mixture of milk, whey and curds. A system like this could be used in commercial cheese making for process control. Reprinted with permission of the American Chemical Society. Team member Colette C. Fagan of University College Dublin noted that some issues still must be addressed but that the new technique is promising. “Such a sensor should have the potential for comprehensive process control during cheese manufacture.”Today cheese is made commercially in facilities that often have 10 to 15 vats of 20,000-liter capacity running in synchronous operation. Within the vats are curd-cutting blades, stirring impellers, pumping systems and other gear. The vats are enclosed and programmed to cut the coagulated milk and to stir the whey and curd mixture. This helps expel the whey from the curd in a process of syneresis, reducing the curd’s moisture content. The curd then is pressed, salted and matured to make cheese.“Online monitoring of coagulation and syneresis enables variations to be identified and compensated for instantly, with the possibility of producing consistent curd moisture and texture in the vat. The benefit would be more consistent cheese grading,” Fagan said.Monitoring syneresis, however, has proved difficult. The methods tried have had problems because they have disturbed the cheese-making process, have involved the addition of trace markers prohibited by food safety regulations, or because they could monitor only a small part of the inhomogeneous mixture of curd and whey.The researchers, therefore, developed a system based on an optical sensor from the University of Kentucky that has a large field of view relative to curd particle size. As reported in the Oct. 31 issue of Journal of Agricultural and Food Chemistry, they used a tungsten halogen source to transmit visible and near-IR light into a cheese vat through a large-diameter optical fiber. They collected the backscattered light through a glass window and transmitted the light to a fiber optic spectrometer made by Ocean Optics Inc. of Dunedin, Fla. The device used a Sony 2048-pixel linear array CCD as a detector. They collected and averaged spectra over the 360- to 1100-nm range.Fagan noted that the system builds upon previously developed fiber optic technology that has been able to monitor coagulation for cutting-time prediction but that has a larger field of view than existing systems. As a result, the prototype provides a reasonable balance between optical penetration into the sample and the collected scattered light that yields a measure of curd formation.The researchers varied temperature, cutting time and amount of calcium chloride to achieve a wide range of coagulation and curd syneresis rates. They found that the response of their sensor was related to aggregation of the milk protein casein and to firming of the curd. Because the sensor reacted to the latter in a manner similar to that of previous optical systems, it likewise could be used for cutting-time predictions. Significantly, the sensor also responded to syneresis with a sharp decrease in the light backscatter ratio profile, a response that the researchers attributed to changes in content in the curd moisture and in the whey fat.The scientists devised models of the response that allowed them to predict cheese characteristics. When curd moisture was calculated during syneresis, the agreement was very good, with >95 percent correlation. The agreement was not as good for other parameters, however.The group has continued modifying the sensor — to reduce noise, for example. Fagan noted, however, that some of the noise recorded during syneresis actually results from curd particles passing the sensor at random and so perhaps should not be eliminated. “There is useful information within these random signals, and we have developed some algorithms for extracting meaningful trends from these signals.”As for commercial use, the system still requires validation. The sensitivity of the prediction algorithm to error must be addressed, the technique scaled up and a control strategy designed. The researchers believe that this may be challenging but that the technology’s future is both exciting and promising.Contact: Colette C. Fagan, University College Dublin; e-mail: firstname.lastname@example.org.