Mitsubishi Electric Corporation has announced the development of aeration control technology to reduce the electric power consumption for supplying air (aeration) to biological reactors which is essential for biological wastewater treatment. By leveraging the company’s Maisart® artificial-intelligence (AI) technologies, the system accurately predicts the quality (ammonia concentration) of the water flowing into the reactor over the following few hours.
The control of aeration levels at individual section of the reactor will achieve approximately 10% reduction in the total amount of aeration, compared to the conventional methods. This will lead to a reduction in the power consumption of biological wastewater treatment plants, which consume about 7 billion kWh of electricity annually, equivalent to about 0.7% of total electric power consumption in Japan.
The company is currently verifying the stability and effectiveness of this technology with the cooperation of domestic wastewater treatment plants, and is aiming to commercialise operation control systems using the new technology within fiscal year ending in March 2021.
Key features include:
AI-based aeration control by highly accurate prediction of quality of the water flowing into the reactor
Even in clear weather where the quality of the water flowing into the reactor is relatively stable, the ammonia concentrations of the water may fluctuate by as much as 50%. In conventional systems, in order to maintain the quality of treated water, an excessive amount of air has to be supplied owing to delays in the aeration control, and thus the ammonia concentration may temporarily decrease by more than needed, resulting in excessive aeration.
In order to improve responsiveness, FF (feed forward) control based on the quality (ammonia concentration) of the water flowing into the reactor, is combined with conventional FB (feedback) control based on the measured value of the quality of the treated water. The company’s new algorithm further improves responsiveness by utilising AI to predict the quality of the inflowing water over the following few hours. It achieves this by analysing the current fluctuation patterns using an accumulated database. By searching for multiple data patterns that resemble the current fluctuations and calculating a predicted value using those patterns, the system can select the optimal data on which to base its predictions. This makes the prediction less susceptible to abnormal data caused by factors such as heavy rain or instrument failure. In addition, the database can be automatically updated to maintain accurate predictions. This technology is particularly effective when the flowrate and/or ammonia concentration of water flowing into the reactor gradually decreases.
Aeration control at individual section of the reactor realises approximately 10% reduction in overall aeration levels
In conventional processing plants, the aeration levels across the all sections of the reactor is uniformly controlled, resulting in uneven treated water quality and excessive aeration. The new algorithms accurately adjust aeration levels by applying weighting to the control parameters for each section. As a result, overall the aeration levels can be reduced by approximately 10% compared to conventional methods, while the quality of the treated water is maintained.