What is a PEMS?

Software-based Predictive Emission Monitoring Systems (PEMS) gained popularity over the past decade as a viable alternative to CEMS by eliminating the need for specialized hardware through the utilization of predictive software capabilities. This in turn reduces the capital and operating expenses of running a CEMS. PEMS fall into two categories: Theoretical First Principal, and Empirical/Data-driven. Theoretical First Principal PEMS derives its model from scientific principles, for example, the conservation of mass and energy. Empirical PEMS uses historical operational data to create predictive models; these include regression models, statistical hybrid models, and neural networks.

Diagram 2: Parts of a PEMS

As mentioned above, PEMS requires minimal hardware which helps reduce not just capital expenses of specialized hardware but also maintenance costs, manpower requirements and space as well. Although the inner mechanism of PEMS are different from each other, they have similar components as depicted in the diagram above. Using an identical diagram for CEMS, we can see that the PEMS software can pretty much replace both the Sampling System and the Emissions Analyzer of a typical CEMS.

That’s right, a PEMS does not require a probe stuck into the smoke stack nor does it require a physical analyzer to measure the concentration of the pollutants in the air 24/7.

How does it work?

Most PEMS utilizes the existing sensor data inputs from the facility to either calculate using first principles(formula-based) or run it in a predictive model(depending on the type of PEMS) where it can produce a prediction/output of the desired pollutant.

PEMS that use predictive models are commonly derived from statistical hybrid or neural network models.