Our Belle Plaine facility uses a technique called solution mining to recover potash ore from deep underground. Salt and potash brine is pumped from underground for processing through a series of evaporator and crystallizer vessels where potash crystals are formed.  

Use of water in this process is guided by analysis of several variables, and machine learning models developed by our employees are now being used in one of Belle Plaine’s crystallizers to predict exactly how much water flow is needed.

Crystallizer vessels are used in solution mining as part of the process to form potash crystals. 
Water addition to the crystallizers impacts production volume and product quality.    

The machine learning models use algorithms and statistics to analyze these variables and provide a recommended water addition to the crystallizer. Our operators are then able to adjust water flow to the recommended set point, ensuring we use the least about of water possible while consistently producing the quality products our customers depend on. Through teamwork and use of next generation technology – fresh water usage is down, and production rates are up. 

“Implementing technology and innovative solutions provides Belle Plaine new opportunities to reduce freshwater use within our process,” said Sarah Huyghebaert, Process Engineer and Belle Plaine’s Sustainability Lead. “Executing and seeing results from this machine learning pilot and other projects help us make progress on reaching our sustainability target to reduce freshwater use by 20 percent per tonne by 2025.”