Struvite image analysis and its application for product purity prediction

Abstract

Phosphorus is an essential element for agricultural food production. While the global phosphate rock reserves are predicted to be depleted within the next 100 years, its current utilization efficiency is still below 20%. Struvite crystallization has been investigated intensively for cost-effective phosphorus recovery. However, there is little focus on the product quality measurement, not to say an industrialized online purity measurement strategy. Most current quality measurements are off-line and are more suitable for laboratory scales with expensive analyzers such as XRD. Therefore, this work aimed to develop a fast and simple soft sensor that can be used for real-time measurements to predict struvite purity using image analysis. Quantitative morphological information was extracted from images of struvite precipitates and analyzed using principal component analysis (PCA). Soft sensors using linear and nonlinear principal component regressions (PCRs) were developed. The results indicate that the proposed soft sensor could predict the product purity accurately, quickly and robustly. This finding suggests that image analysis could be effective in quantifying struvite purity online.