Taber is now offering Hg PAC Control application built using the Griffin AI Toolkit™. This application optimizes for tighter control versus rolling average emission limits and for PAC injection savings. Many current implementations have the PAC feed-rate adjustment reacting to either instantaneous Hg stack emissions or period manual stack samples with little or no regard to the rolling average impact or the impact on other combustion variables such as opacity.

The Griffin AI Toolkit is easily tunable to adjust factors for different control strategies and time horizons. The basic concept uses a 30 day rolling average goal for the plant or site and determines the most effective amount of PAC required for maintaining the average within the selected margin. Generally, as more PAC is added to the flue stream each incremental add of PAC will reduce Hg by a lesser amount. Griffin is setup to recognize this and seeks to keep the PAC average rate at the lowest possible level.

Short term instantaneous readings are smoothed out, preventing oversupply of PAC to a temporary spike in Hg, which is often over before the PAC can affect the process. Feedrates are biased or controlled directly on a time cycle more closely matching the expected changes in fuel over time. Adjustments may be made to accommodate high opacity or other combustion situations where it is desirable from an operator’s standpoint to have PAC reduced. When conditions warrant and the PAC injection is deemed to be in an effective control zone, margin may be built up in the rolling average.


  1. In this example, Opacity is reduced and kept out of alarm conditions
  2. The PAC feed-rate is much steadier, helping smooth the overall process
  3. The Hg emissions are noisier short term, but the average is the same or better.