Systems Laboratory Case Studies
Advanced Signal Analysis to Enable Better Operation of Chemical Processes
When you are ill a doctor takes several measurements from properties of your body and analyses them to find out if something is wrong; and if so, where and why.
An industrial chemical process is in this sense just like our body: it can suffer from disturbances, which are reflected in the measurements of its properties, and these can be analysed to detect, diagnose and help to correct such disturbances.
A process deviating from normal operation can negatively impact industry as well as our everyday life; this was evident recently when the reserves of natural gas in the UK came close to exhaustion due to maintenance in overseas gas plants that supply the UK.
The original work of the Process Automation group investigates methods to analyse collectively measurements taken directly from the process as well as its electrical and mechanical equipment, in order to detect and diagnose disturbances.
Early detection and quick diagnoses of disturbances could not only save industry precious resources and money, but potentially also reduce the risk of serious injury or loss of life.
One of the methods developed1 looks for sudden and brief disturbances to indicate which properties were affected and when, giving an indication of the severity of the disruption and an idea of where it started. This method involves a technique known as anomaly detection, which is also used to detect fraud in credit card transactions, combined with a statistical technique known as principal component analysis, which finds the anomalies which are most representative across the plant.
Another method developed by the group2 tracks down the propagation of a disturbance along the plant. It does so because it is sensitive to delays and smoothing of the disturbance, which happens when it propagates along chemical, electrical or mechanical systems.
This method is currently being extended to measurements sampled at different rates, a complex problem involving combining slow-sampled process measurements with fast-sampled electrical and mechanical measurements.
[1] I. M. Cecílio, and N. F. Thornhill (2013), “Nearest Neighbors methods for detecting and removing interfering disturbances from oscillating signals.” Manuscript submitted for publication in IFAC DYCOPS 10, Mumbai, Dec. 2013.
[2] M. Bauer, J.W. Cox, M.H. Caveness, J.J. Downs, N.F. Thornhill, “Nearest Neighbors Methods for Root Cause Analysis of Plantwide Disturbances”, Ind. Eng. Chem. Res. 46 (2007) 5977-5984.
Synthesis of Biorefinery Treatment Plants for Energy and Nutrient Recovery
Processing organic wastes and wastewater together in a single biorefinery treatment plant offers many benefits and might also have a profound impact on the structure and the approach of conventional depollution strategies.
Apart from the effective treatment of waste/wastewater, a highly valuable product can be produced such as methane, biofuels, phosphorus, nitrogen and heavy metals.
The use of recycled waste streams also presents the additional benefit of reducing the raw material requirement, again reducing costs. Clearly, the ultimate goal is a closed-cycle process, where all waste streams are recycled and the only output is saleable/valuable product.
This project aims to develop and apply a systematic, model- based methodology for the synthesis of biorefinery treatment plants that are both economically attractive and sustainable.
The main objective is to assess the merits and viability of such integrated concepts using this methodology. A superstructure modelling approach is considered, which can account for a large number of promising treatment and separation technologies (units) along with all feasible interconnections between them.
The optimization objective is to determine the process configuration (unit types, interconnections and flows) that maximizes the net present value over the project lifetime (based on capital and operating costs and revenues from energy/product sales). Constraints are also imposed for the treated effluent to comply with local or federal regulations.
The results of a case study are shown in the figure below. Here, a biorefinery treatment plant fed with 100 m3/h of tequila vinasse (100 gCOD/L, 8 gTSS/L, 120 mgN/L and 700 mgP/L) is synthesized based on a small superstructure that consists of 3 biological treatment units (UASB, SAMBR, activated sludge), 1 filtration unit (sand filter), and 2 nutrient recovery units (struvite crystallizer, zeolite adsorber).
Maximum allowable limits are enforced for these concentrations, in agreement with current Mexican regulations. The model recommends the installation of submerged anaerobic membrane bioreactors (SAMBRs) that enable large COD concentration abatement and biogas production. It then recommends to further process part of the SAMBR outlet in struvite crystallizers (36%) and sand filters (37%), the remaining fraction (26%) being discharged without further treatment.