Cheese production was approximately 21.86 million metric tons worldwide, while the USA manufactured over 6 million tons, in 2021 [1
]. Cheese is a dairy product that has played an important role in human nutrition for centuries. Since then, the main objective has been and is today to transform a highly perishable product, such as milk, into another that has a long life and preserves its nutrients. In essence, cheese is made by removing water or whey, allowing the solids or curds to be controlled. Cheese is a healthy and diverse dairy product regularly consumed as part of a dish, as a snack, or as a premade meal. Cheese-making is a process in which the fat and casein of milk are concentrated in five steps: acidification, coagulation, whey separation, molding, and salting [2
]. Several parameters impact the transformation of milk into cheese, and the knowledge of such factors could represent a competitive advantage for the cheesemaker. The chemical composition of the milk plays a crucial role in cheese manufacture due to the complex biochemistry and microbiology involved [4
]. Cheesemaking technology permits the production of various flavors, textures, and consistencies [5
], which involves the meticulous monitorization of the curd cutting time.
Cheesemaking can be considered a sensitive process. Once the gelation point is reached, the curd’s hardness increases with time. Therefore, selecting the optimal hardness moment for the cut is necessary, which induces the drainage process [6
]. The cut should be performed when the curd is sufficiently cohesive but has not reached excessive hardening [7
]. The changes in the optical properties of milk during coagulation have allowed, primarily using optical fibers, the development of a series of instruments based on determinations of reflection, absorbance, dispersion, and refraction of light [8
]. These are continuous and non-destructive methods. Among the properties mentioned, light scattering is one of the properties that yields the best results for the study of gel formation. In a non-absorbing medium, it is possible to directly measure the portion of incident light that is dispersed at a certain angle or indirectly measure the portion of the light that has not interacted with the medium, that is, the transmitted light [9
]. In milk, near-infrared light scattering is directly related to the speed of micellar aggregation and gel hardening, so this optical sensor technology has been proven as a valuable tool to control milk coagulation [10
]. In particular, the fiber optic sensor near-infrared (NIR) scattering, CoAguLite, is an inline sensor that has been well-documented for monitoring milk coagulation and predicting gelation and cutting times [11
] and is currently commercially available for inline control in cheesemaking, mainly in the United States. Several previous studies have shown that such a sensor is a promising instrument for inline monitoring of optical properties during milk coagulation, providing indirect information on relevant technological parameters in cheese production. Arango et al. (2018b) [12
] obtained and validated curd cutting time prediction models using optical time parameters determined by calculating derivatives from the dispersion ratio profile as a function of time. Similarly, Salvador et al. (2019) [13
] found that the optical light scattering sensor predicts well the time the gel reaches values of G
’ = 30 Pa.
On the other hand, Arango and Castillo (2018a) [10
] developed and patented a method for real-time monitoring of gel firmness, which consists of a mathematical model that relates the NIR scattering signal obtained through the sensor optical fiber with the elastic modulus of the gel. This method does not involve time parameters and, therefore, allows the collected light scattering ratio data to be introduced directly into the model to calculate the value of G
’ in real time [14
The curd cutting time is a critical aspect that impacts the quality characteristics of cheese [15
]. Cutting time can significantly affect cheese yield, and the curd can change fat, whey, and moisture contents. A high-moisture curd results from cutting the curd too late, and a poor yield could be caused by cutting the curd too early. The curd cutting time is typically monitored by instruments based on mechanical approaches, vibrational techniques, ultrasonic methods, electrical conductivity, hot wiring, and optical systems [16
]. These apparatuses are used to examine the rheological characteristics of the curd. The curd cutting time could demand a more effective, inline, and non-destructive method.
The light backscatter sensor could be a possible option, and these optical systems have been suggested by various studies [17
] since these systems are non-destructive, rapid, and effective. Optical technology developed by Payne et al. (1999) [18
] was designed for cow milk cheese production, and Castillo (2000) [2
] and Nicolau et al. (2015) [19
] evaluated and adapted NIR (near-infrared) light backscatter technology, respectively, for goat and sheep milk cheese manufacture. Thus, developing an optical sensor is required in spray-dried milk cheese. When predicting the cutting time, the NIR practicality of cow’s, goat’s or sheep’s milk cheese could vary from spray-dried milk cheese. Therefore, the objective of this study was to consider if an inline fiber optic sensor with NIR could be applied to monitor the cutting time of curd hardness milk cheese produced by spray-dried milk. The study represents an innovative approach for the dairy industry to monitor milk coagulation in cheese production, as it uses multifiber probes and specific wavelengths to predict curd elasticity during milk coagulation, optimizing the process economically. The study employed randomized block designs and varying protein levels and wavelengths, with 880 nm showing superior sensitivity and stability in response.
In cheese, choosing the appropriate moment for cutting the curd is important due to its effects on the product’s yield, maturation, and useful shelf-life. The optical time parameters (tmax) and response parameters (Rmax, R1min and R2min) generated from the light scatter curve (R), unlike the initial intensity (I0), do not present significant statistical differences as a function of wavelength. This indicates that I0 is inversely proportional to the wavelength. This was likely because any of the three selected wavelengths corresponded to the maximum light backscatter peak (i.e., the peak of maximum light scatter response was likely located at a wavelength slightly shorter that the wavelengths at which measurements were taken). However, this inversely proportional relationship did not translate clearly into a significant effect on the parameters of the coagulation process. This circumstance is preferable when designing a coagulation sensor, as the three wavelengths generate similar coagulation parameters but with a more intense response at a shorter wavelength. However, other parameters concerning the suitability of the signal should be considered in future studies to identify the wavelength that contains the highest level of information about coagulation and the highest degree of response.
In the coagulation monitoring process as a function of protein concentration, no significant differences are seen in tmax with this new probe, but they are seen in t2min, which is close to the beginning of the hardening reaction. This indicates that hydrolysis and the beginning of aggregation did not change significantly with increasing protein concentration, likely due to the relatively high concentration of enzyme added, but the beginning of hardening did advance with increasing protein concentration, as a result of increasing rate of casein micelle aggregation with increasing protein concentration. The I0 response, increases linearly with protein concentration. This could allow the G’ estimation probe to be used to estimate the protein concentration in milk.
The multifiber probe evaluated with different protein levels represents an improvement in the inline monitoring process of milk coagulation, as it presents a lower cost and provides a coagulation profile like those of previously evaluated probes, which allows the generation of coagulation parameters. Those parameters respond as expected to a coagulation factor relevant to gel hardness as protein concentration. The light backscatter ratio obtained by the multifiber probe at the wavelengths used indicates that it can successfully adapt to the real-time prediction models of the elastic modulus of the gel, G’, previously established by other studies.