This
Energy efficiency and the reduction of pollutants is important in energy intensive industrial processes. Being able to measure the energy efficiency and pollutant levels in real time is an important first step in achieving this goal.
This application note shows there is a surprising amount of information in the video images of a boiler flame. Most furnace and boiler systems have such a video system already installed to determine if the flame is present; but far more can be done.
Challenges
The initial challenge was to determine the energy content from a steam boiler burning natural gas and a liquid waste stream of variable energy content.
Consecutive images show a highly turbulent flame, yet the underlying process has not changed during that short time. This poses considerable difficulty in extracting stable information.
These 4 consecutive frames show that the flame is turbulent and unstable. The luminous area continually shifts in time.
Multivariate Image Analysis however does extract stable information and only true changes in process conditions are detected. This allows us to predict the heat of combustion from the boiler, a property that varies as the liquid waste stream changes.
Another very interesting feature was our ability to predict NOx and SO2 emissions. Data from the current emissions monitoring system, the video images, and routine process measurements (flow rates and temperatures) were combined to build a soft sensor for emissions with good prediction results.
Consider your process data currently being acquired every minute, every day, from laboratories, from video images, and other sensors. To find out how you can make the most from this data, please contact us.
