Oil and gas and other chemical process industries are always under pressure to minimise costs and environmental footprint. To achieve these objectives effective control and optimisation of the processes are required.
Two key enabling technologies are needed to achieve effective control and optimisation:
- online measurement systems that provide continuous information on key physical and chemical attributes of the process streams
- enabling technology is for the extraction of critical information from process measurements through the application of big data analytics and process modelling.
These technologies are essential to developing reliable predictive models that effectively utilise a large amount of data that are typically collected in petroleum and chemical plants. Combining this databased approach with first-principles-based knowledge of process systems can potentially provide significantly improved predictive models.
While our focus is mainly on the oil and gas sector, the technologies we are developing have broad applications across the chemical and agricultural sectors.
We welcome enquiries from prospective industry and academic collaborators as well as from researchers interested in pursuing a PhD within the following project areas:
Data-driven process analysis: Big data analytics
In many areas of engineering, the commercial availability of various sensors based on different measurement techniques has increased due to them becoming progressively cheap and thus affordable. This has led to the collection of vast amounts of data. Most of these data are of disparate types. For example, in a chemical process data collected consists of a large number of qualitative and quantitative measurements taken at different time intervals.
We are investigating cutting-edge machine learning techniques for building process models to provide information on the current status of the process regarding critical product attributes and also to predict the endpoint conditions so that timely corrective actions can be taken. The handling of disparate data types is also an active area of investigation in our group.
We are investigating approaches based on Bayesian principles for combining data from different sources. The goal is to develop data and model fusion techniques that can be applied to upstream and downstream oil and gas processes.
Online monitoring of critical attributes of a process stream
We are engaged in the development of a measurement platform that can measure multiple chemical and physical properties of a process stream. Optical spectroscopy has been our primary focus. We are developing a novel fibre-optic probe system that utilises Ultraviolet-Visible-Near Infrared spectroscopy.
We are investigating some measurement configurations based on spatially and angularly resolved measurements that will enable the simultaneous determination of particle/droplet size and composition of process streams that are slurries or emulsions.
The challenge of extracting the required information from these measurements is being investigated through a combination of machine learning techniques and fundamental light propagation models.
Research Coordinator: Prof Suresh Thennadil.