Virtual Sensor

Key measurements like quality, efficiency, composition are very often not available to  operators in real time. Although these measurements are often the most important KPI’s of a plant, they are only made available a few times per hour or much less frequently. Additionally, laboratory sampling and analyses are also subject to human error.

This leads many to opt for an online analyzer, which are often very expensive and require a lot of maintenance efforts like frequent re-calibration. Reliable online analyzers are not always available for every measurement, such as multi-phase flows in the upstream industry, color of chemical components, etc. Analyzers like Gas Chromatographs have often sampling frequencies of 20 minutes or even slower.

Soft sensing, also called virtual metering or inferentials, combines de data from reliable standard sensors with the help of a model into a desired reading. With the use of soft sensors based on good models and real time data, robust “virtual sensors” can be built in a very cost effective way.