Background PROCESS OPTIMISATION AND CONTROL Air Enhanced Process Understanding Data Analysis and Principal Component Analysis The focus of the two assignments for this module is based on data collected from

the fluidised bed reactor 'A' at BASF, Seal Sands, Teesside. The ultimate objective of the analysis was to attain reliable on-line predictions of 7 quality variables, in particular AN (Acrylonitrile) yield, to be able to provide an estimate of AN Yield to the process operators at the same frequency as the process variable measurements. Yield information is normally provided by the on-line Gas Chromatograph with a 40 minute cycle time. The reactor is cooled by a number of internal cooling coils in the bed of the reactor. The coil coolant flow rates are fixed. As the coils foul, their thermal efficiency is reduced and the flow of raw materials into the reactor change through the action of the multi-loop controllers. During reactor operation, the cooling coils are switched in and out as part of a de-fouling recycling procedure to maintain optimum bed cooling and good production. The operational staff select the required cooling coil patterns based upon experience, observation of changing operating conditions and increases in the bed temperatures as the cooling coils become fouled. The cooling coils are periodically taken off-line to allow fouling deposits to be removed by the high temperatures in the reactor bed. A schematic of the process is provided below. COOLER REACTOR CAPE3321 To boiler Catalyst in/out Gas chromatograph QUENCH Propylene Ammonia To atmosphere Oxygen meter Refrigerated de-mineralised water ABSORBER Product rich water Data Information The process data was five minute averaged data. The initial data set comprised a total of forty-eight variables and 14344 samples. A sub-set of the data will be used for the assignment (391 samples and 25 process and calculated variables). Not all bed temperatures are provided in this data set as detailed in the overview. Please note the focus is on the reactor and not the other elements of the process. The variable names are provided below. The process data is available in Minerva in CAPE3321 Assessment and Feedback - CAPE3321 - Data Analysis and PCA Data Set 2024 Assignment.xlsx Assignment Task Explore the data using appropriate pre-screening techniques and then develop a Principal Component Analysis (PCA) representation. The objective of the analysis is to obtain an understanding of the underlying behaviour of the process from the analysis of the data. The outputs from the graphical representations and the principal component analysis and any other information/representations must be interpreted in the context of the industrial application. Provide a Report and ensure you follow the directions below in terms of the structure. Submission - Maximum 6 pages - does not include Title Page but includes references. 1 page - Executive Summary (definition provided below) 5 pages analysis including interpretation of results based on appropriate figures and tables Font Calibri 11pt, Single Line Spacing, Page Margins (2cm top, bottom, LHS and RHS) An Executive Summary summarises a longer report in such a way that the reader can rapidly become acquainted with a large body of material without having to read it all. The overview is a brief introduction / summary - a paragraph. Marks Breakdown Executive Summary: 20% Overview: 10% Analysis: 30% Interpretation of Results: 30% Conclusions: 10% Summary of Variables 1 2 3 4 5 6 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 Propylene flow rate Ammonia flow rate Air flow rate T.M.F. (total mass flow) rate Catalyst inventory Pressure Propylene partial pressure Ammonia partial pressure Oxygen partial pressure Nitrogen partial pressure Ammonia/Propylene ratio Air/Propylene ratio Velocity W.W.H. (Weight of propylene feed/ Weight of catalyst/hr) Contact time Bed Temperature 1 Bed Temperature 2 Bed Temperature 3 Bed Temperature 4 Bed Temperature 5 Bed Temperature 7 Bed Temperature 8 Bed Temperature 11 Bed Temperature 16 Bed Temperature 21 Process Process Process Calculated Process Process Calculated Calculated Calculated Calculated Calculated Calculated Calculated Calculated Calculated Process Process Process Process Process Process Process Process Process Process