CeniCristal: a tool to enhance sucrose recovery through reliable monitoring of crystal size and growth

CeniCristal was developed within the Industry 4.0 framework to accurately monitor crystal growth during evaporating crystallization, utilizing advanced digital image processing and artificial intelligence. Trained and validated with sugar crystal images, the system achieved an error margin of less than 3% compared to the standardized sieving method. A subsequent transition to a web-based platform improved accessibility, allowing crystal segmentation with response times of under 30 s. The implementation of a new slurry preparation mill at a sugar factory, aimed at increasing capacity, was anticipated to affect the crystal size coefficient of variation (CV). CeniCristal confirmed a 19% reduction in slurry crystal CV following the slurry mill’s installation, leading to enhanced uniformity across subsequent processes. Specifically, CV reductions of 5% to 12% were recorded in A and B massecuite crystallization, significantly improving final product quality. Moreover, CeniCristal identified unwanted crystals in B run-off syrup due to centrifuge screen wear, increasing final molasses purity by 4% and resulting in a loss of 13.7 t of sugar, equivalent to USD3805, in just 24 h. Early detection enabled timely maintenance, minimizing sucrose losses and restoring process efficiency. Currently deployed in eight Colombian sugar factories, CeniCristal has become a key tool for enhancing product quality by reducing crystal variability and optimizing operational performance. Its early detection capabilities, such as identifying crystals in B run-off syrup, have proven instrumental in preventing losses and maximizing sugar yields. Plans to incorporate CeniCristal into broader monitoring and control systems within the Industry 4.0 framework aim to achieve full process automation, including predictive maintenance and crystallization optimization, thereby further strengthening the competitiveness of the sugarcane agroindustry sector.