EPCOR’s Gold Bar Wastewater Treatment Plant in Edmonton, Canada operates several H2S (Hydrogen Sulfide) scrubbers to clean the exhaust air from the plant. One of these scrubbers was operating below its target level of efficiency.
THE CHALLENGE
Improve H2S scrubber efficiency while conserving chemical costs. EPCOR’s goal was to improve this efficiency to a 95% threshold in order to improve the air quality of the residential area near the plant and reduce resident complaints. As a secondary goal, they were also interested in minimising the cost of chemicals being used while maintaining the target efficiency.
THE SOLUTION
To address this challenge, RLCore deployed their RLTune software, an advanced reinforcement learning (RL) agent, to automate and optimise the scrubber process. This AI-powered solution intelligently adjusted key setpoints, such as pH, oxidation-reduction potential (ORP), and softened water flow rate, in real-time.
These adjustments were made based on a combination of:
- Historical data provided by EPCOR
- Data collected live as RLTune adjusted setpoints in real-time
This system was integrated with the plant’s HMI (Human-Machine Interface) through a secure, OPC-based interface, and included operator override and monitoring capabilities to ensure transparency and trust.
RESULTS
- Scrubber efficiency was increased to meet the 95% target (up from <80%)
- Chemical cost per unit of H2S removed was nearly halved
