Softsensor for the Characterisation of the Process Fluid

1.5 System & Preliminary Design

The main function of fans is supplying the required volumetric flow of process gas. Among others for or from combustion processes, for chemical processes or for energy transport. The knowledge of the gas composition, i.e. its characteristic properties, is an important information for process control as uncertainty in gas composition is often accompanied by an uncertainty in the process mass flow, which may even lead to critical operation points and early failure. For diagnosis applications, e.g. for leakage detection, the knowledge of the gas composition is as well essential.
The paper introduces a method to substitute expensive volume flow and gas composition sensors using process knowledge and available data from cheap and often already existing sensors. It is shown that for a two-phase fluid, the characteristic map of the fan and the mixture laws of ideal gases contain enough information to calculate the volume flow rate and the gas composition if the inlet gas temperature, the pressure rise over the fan and the fan converter data, i.e. rotational speed and electrical power input, are known.
The concept is validated experimentally. The used test rig consists of a temperature sensor, the radial fan, a pressure sensor, a throttle valve and a volume flow sensor. Validation was carried out by comparing the data of the latter as well as the used gas composition with the results of the softsensor.