AT the Impact for Process Systems Engineering

Transcription

1 AT the Impact for Process Systems Engineering lian Morris 1, Zeng-ping Chen 2, David Lovett 2 & David Littlejohn Centre for Process Analytics and Control Technologies ewcastle University, 2 Perceptive Engineering, 3 Strathclyde Universit

2 Overview of Presentation What is Process Systems Engineering? Where are we now, where are we going and how important is variability? Look at some novel methodologies for building fit-for-purpose calibration and statistical monitoring models Highlight some process analytical technologies through work aiming to improve understanding of monitoring and controlling batch crystallisation processes using advanced Chemometrics Show how some new closed loop process PAT control ideas are coming to fruition through innovative process systems engineering - Closing the Analytical Control Loop Closure The EU provides 32% of the worlds chemicals manufacturing through some,000 enterprises of which 98% are SMEs which account for 45% of the secto added value, and 46% of all employees are in SMEs

3 What is Process Systems Engineering? Process Systems Engineering (PSE) is a well established scientific area, concerned with the development of methods and computer-based tools for an integrated approach to all aspects of process modelling, simulation, design, operation, control, optimisation and management of complexity in uncertain systems across multiple time and scale lengths. These tools enable Process Systems Engineers to systematically develop products and processes across a wide range of systems involving chemical, biological and physical changes from molecular and genetic phenomena to manufacturing processes and related business processes.

4 Where are we Now and Where are we Going in Process Analytics and Control Technologies?

5 Impurities and Polymorphism (1)

6 Impurities and Polymorphism (2) Impurities effect nucleation and growth processes, and hence can stabilise meta-stable polymorphic forms. As product purity improves during process chemistry work-up, the stable polymorphic form can change! e.g. RITONOVIR aids drug which changed from anhydrous to hydrate crystal after launch: with lower solubility and hence bio-availability. product was withdrawn for a year and reformulated. new FDA approval needed mega cost implication!

7 From this. to this Where Are We Going?. to this T, ph Enablir PC Turbidity FTIR spectrometer FTIR probe Manhole Pharma Process Control N 2 purging tube ph probe Turbidity probe Reactor Pharmaceutical companies typically operate at around 2 to 2½ Sigma compared with 6 Sigma in world class manufacturing

8 Variability (or PAT) by Edwards Deming Cease reliance on mass inspection to achieve quality. Eliminate the need for mass inspection by building quality into the product in the first place. Dr W. Edwards Deming (Circa 1980s) Learning is not compulsory,. Neither is survival

9 ariability and Coping with Different Product Recipes, ifferent Processing Units, Different Production Sites, Different Spectroscopic Probes & Probe Locations,, etc

10 Variability with Different Product Recipes, Processing Units, Production Sites, Spectroscopic Probe Locations, The elimination of differences between different modelling or calibration applications is a major issue in flexible process manufacturing Different spectroscopic probes in different vessels or on different sites Different reactors (unit operations), multiple recipes, etc Normally this would require the construction of separate models for each individual situation. A simple but partial solution is the subtraction of local or global mean levels A better solution is to develop a multi-group (sub-space) model or calibration such that the interest can then focus on within process (or product) variability. Martin, E.B. and A.J. Morris, (2003), Monitoring Performance in Flexible Process Manufacturing, Proceedings of IFAC ADCHEM 2003, Hong Kong Lane S., et al, Performance Monitoring of a Multi-Product Semi-Batch Process." Journal of Process Control, 2001, 11(1), 1-11.

11 Case Study: Between and Within Group Variation Multi product manufacturing with Unilever (Lever Fabergé). Approximately 50 different products and five main production units, complicated by multiple recipes. Monitoring the process using standard MSPC would mean the building and maintenance of approximately 250 statistical monitoring models. Too complicated for quality and operational personnel. High model-maintenance costs

12 Composition of the Multi-Recipe Data Sets To demonstrate the application of multiple group PLS the production of two recipes are considered. Each process mixer is also considered as a separate Recipe and thus the process model contains four distinct groups. Recipe Group Mixer Number of Batches Raw Materials Quality Variables 1 1 ( ) (+) (o) (x) Process Variables Note that Recipe 2 has fewer raw materials and fewer process variables than Recipe 1.

13 Impact of Different Recipes This model contains both within group and between group variation. Subtle process events cannot be detected; the greater the number of distinct groups in the data set, the greater the impact of between group variation. 10 Principal Component Principal Component 1

14 Impact of Spectroscopic Probe Location in a Reaction Vessel Location 1 Location 2 Location 3 Location 4 Location 1 Location 2 Location 3 Agitator PC2 Location 4 PC1 Spectroscopic Probe Location PCA Plot of Measured Spectra

15 6 Multi-group PLS Pooled Covariance Model The pooled variance-covariance matrix is constructed as a weighted sum of the individual elements of the individual variance-covariance matrices. 4 Latent Variable Latent Variable 1

16 Case Study Dosing Valve Problem 5 Principal Component Principal Component 1 The monitoring chart shows the scores jumping outside the action and warning limits as the process malfunction impacts.

17 Fault Diagnostics Variables (62-65) were identified as being related to raw material addition. A faulty dosing-control valve was located as being to source of the problem. 8 6 Contribution Variables

18 Process Issues: Variability Modelling and Calibration Challenges Equipment characteristics; site-to-site process differences, etc Fluctuations in both control and external process variables Cell improvement; cell line changes; media changes Small data sets are an issue but can be enhanced through Bootstrap Aggregation Analytical Issues: Separating absorbance from multiplicative light scattering effects caused by the variations in optical path length Inter probe variability: impact of component variance on PLS calibration can probe differences be accomodated or eliminated? Can calibration models be made generic for different production unit operations / production lines?

19 Variability - Coping with Minimal Experimental Data The need for fit-for-purpose statistical models can be limited due to intrinsic data sparsity resulting from the small number of objects available, e.g. experiments, batches, runs, etc, compared with the large number of process variables wavelengths measured. This has been successfully addressed in PLS calibrations and in building statistical monitoring and predictive models through the addition of Gaussian noise to the original data and combining multiple models using Bootstrap Aggregated Regression. This allows the development of robust calibration or process models and has been shown to lead to a decrease in prediction errors as a consequence of the increase in the data density. Martin, E.B. and A.J. Morris, Enhanced bio-manufacturing through advanced multivariate statistical technologies, Journal of Biotechnology 99, 2002, Conlin, A. et al, Data augmentation: an alternative approach to the analysis of spectroscopic data, Chemometrics and Intelligent Laboratory Systems 44,1998, Zhang, J. et al, Estimation of impurity and fouling in batch polymerisation reactors through the application of neural networks, Computers and Chemical Engineering 23, 1999, 30,

20 Advanced Chemometric Methods

21 Chemicals Behaving Badly (CBBII): Integrated In-Process Analytics and Closed Loop Control FTIR CBB#2 Project: collaboration with Leeds, Heriot-Watt and Newcastle University Video microscopy Shape Reactant rheology Mixing & LDA/PIV scale-up Supersa- Growth turation Nucleation USS kinetics kinetics UVvis Size MSZW Batch process Process monitoring Polymor- & control phic form XRD Process conditions Heat transfer Reaction Calorimetry Partners AEA Technology AstraZeneca Bede Scientific Instruments BNFL Clairet Scientific DTI EPSRC GlaxoSmithKline HEL Malvern Instruments Pfizer Syngenta CFD

22 Smoothed Principal Component Analysis (SPCA) Enhancing signal-to-noise ratio T X Xr ) T i = λi ( I + k Q Q ri i = 1, 2, Λ, m F = r, r, Λ r ] [ r, r, Λ r ] [ 1 2 c 1 2 c + x = Fx = Fx + F s Fx n a SPCA Intensity b b 0.800% 0.800% 0.800% 0.533% 0.533% 0.533% 0.389% 0.389% 0.389% 0.0% 0.000% 0.178% 0.178% 1.000% 1.000% θ θ Raw (a) and Processed (b) XRD profiles (by SPCA) for 6 XRD data sets of mannitol-methanol suspensions with the contents of mannitol equalling to 0.0%, 0.178%, 0.389%, 0.533%, 0.8% and 1.0% g/ml, respectively

23 SPCA in Morphology Monitoring (β - Form) 5000 B1 B2 β -form R= R= Intensity B3 Peak height Intensity θ B1 B2?-form β B3 Peak height θ 2 θ = Concentration (%) R = θ θ θ = = Concentration (%) The raw and pre-processed XRD profiles of the β-form with concentration varying from 0.02% to 8.00%; relationships between concentrations and peak heights at peaks B3 of the raw and processed spectra.

24 Loading Space Standardisation (LSS) Correcting non-linear shift and broadening in spectral bands caused by temperature fluctuations PCA ' X( t1) TP ( t Μ Μ Μ X( t m PCA ) TP K x i ( ti ) = ci, ks k ( ti ) + e ' 1 ( t ) m ) k = 1 i ( t i ) LSS P ( t ) ~ t LSS x t ) x( t ) i i ( test ref LSS Absorbance (AU) Wavelength (nm) Wavelength (nm)

25 Optical Path-Length Estimation and Correction (OPLEC) Separating absorbance from multiplicative light scattering effects caused by the variations in optical path length i J = bi j =1 x c s + ε b i - i, j j i multiplicative parameter bi = OPLEC( X, c j ), X = [ x, x2, Λ, xm], c j = [ c1 j, c2 j, Λ, c 1 mj ] Wavelength (nm) OPLEC Abs. (AU) Wavelength (nm) OPLEC has recently been applied for Raman Scattering applications in complex crystallisation processes.

26 Bede MONITOR TM In-process XRD Crystal Polymorph Monitoring & Control Using SPCA β peaks t = t = 600 t = 100 t = 140 t = 180 t = 220 Temperature Temperature monitors readers System provides capability to monitor polymorphic form in-process unaffected by product separation prior to analysis. Typically circa 1 wt % detectable via in-process XRD, much lower with advanced chemometric analysis (Smoothed PCA) Intensity Theta [degree] t = 260 t = 300 t = 340 t = 380 t = 420 t = 440 t = 460 α peaks

27 250l Pilot Plant Batch Agitated Vessel α-sizer Outlet (8mm) port Inlet (12mm) po

28 Supersaturation Control System Upgrade to PI Capability Crystallisation Vessel FTIR Spectrometer ENABLIR WINISO SOFTWARE HEL PC FTIR PC C(t) Spectra PLS Cal. model C(t) Data Processing Block Macro Solubility Model SOFTWARE 4 20 ma INPUT Block User Define Model S(t) Control Block IMC Based PI Controller PI Controller User Define S set point

29 Supersaturation Control of L-Glutamic Acid 250 litre Plant Crystalliser Temperature, Concentration, Solubility & Turbidity S = 1.1 Smax = Started Supersaturation Smin = Control % seeds added Temperature ( C) Concentration (g/500ml) Solubility (g/500ml) Turbidity (%) Supersaturation Slimit Slimit Time (min)

30 Model Based Closed Loop Process Control

31 Challenges for Real Time Closed Loop Control At least 15 years worth of academic control theory, simulations and occasionally pilot studies have reported on the benefits and challenges of improving operational performance through improved control, more recently in the Pharmaceutical Industry. Since then the FDA has opened the way to a technology-led quality assurance regime there has been hectic activity directed towards defining an implementation methodology for measurement, modelling and data analysis. Many of the larger automation companies have focussed on the overall infrastructure to ensure that the integration of data from many sources is recorded securely and a traceable batch record is generated. Hopefully one of the key outcomes is that the integration of the data will also integrate the users of the data such as the chemists, biologist, engineers, to provide a process systems engineering approach to design, build, operate, monitor and improve production facilities.

32 The Impact of Process Dynamics (Auto-correlated Data) In the presence of auto-correlation, the action and warning limits calculated for a process performance monitoring scheme based on the usual assumptions that the observations are Independent and Identically Distributed (IID), are inappropriate making the monitoring charts unreliable, if not useless. Developing an MSPC representation for the monitoring of processes, exhibiting time varying characteristics, will result in an increase in the number of false alarms and operational changes not being detected. So, how do we deal with auto-correlated data?

33 Model-based Performance Monitoring Inputs Plant Outputs First Principles Model or Dynamic Empirical Empirical Model Model First Principles Model Dynamic Empirical Model - + Plant-Model Mismatch Dynamic Dynamic Empirical Error-whitening Empirical Error Model Model Multivariate Monitoring Charts Multivariate Monitoring Charts At the heart of a model-based control or monitoring system is a DYNAMIC model

34 Impact of Non-linearity Normal Probability Plots Process Data Plant-Model Residuals Percentage Model Residuals Linear PLS Percentage Model Residuals Dynamic Non-linear PLS Data Data

35 Impact of Autocorreltion Partial Correlation Plots Process Data Residuals Plant-Model Mismatch Model Residuals Dynamic PLS Model Residuals Time Series Model

36 Incorporating PAT Sensors into Real Time Process Control and PAT in Process Systems Engineering

37 Key Challenges in PAT Pharma-Batch Control Availability of Critical to Quality (CtQ) measurements: Are the CtQ parameters continuously measured? Are the CtQ parameters measured only by lab assay at the end of the batch or a-periodically during the batch? Can the CtQ parameters be inferred from PAT sensors with associated calibration model(s)? Process non-linearity: As a function of process operating point (e.g. Arrhenius reaction rate, etc) Time varying dynamics as batch evolves - varying time constants and process gains Changes in sign of process gain can be especially challenging Trajectory tracking: Many control systems are excellent at regulating to a fixed set-point and rejecting disturbances BUT not so good at tracking a time varying set-point profiles.

38 Real Time Quality Control (Batch Control) Imported from Model Development File Critical to Quality Parameters Spectral Data Process Data Discrete Data Data Quality Monitor Pre-Processing Sequence Batch PLS Batch Controller 21 CFR part 11 Records Courtesy Perceptive Engineerin

39 Batch Model Predictive Tracking CtQ parameters controlled indirectly by virtue of controlling a process variable. At the heart of the controller is a dynamic process model that captures dynamic, multivariable process interactions. The model predicts the trajectory of controlled variable(s) over a future horizon (typically much shorter than the batch duration). The controller computes a trajectory of MV moves over this future horizon to minimise the current and future SP error. Only the first move is applied; at the next control interval the whole process is repeated - this is a moving window approach Courtesy Perceptive Engineeri

40 Process Systems Engineering The Future A Process Systems Engineering, holistic, approach should help integrate all aspects - sensor, data, chemometrics, and the process and control system in a modular way that provides a readily validatable system. Clean Data Set Imported from Model Development File Spectral Data Process Data Discrete Data Data Quality Monitor Real Time Pre- Processing Sequence Spectral Data Courtesy Perceptive Engineerin

41 Process Systems Engineering The Future eal time pre-processed data CtQ Parameters Continuously measured OR A-periodically measured OR Real time value Inferred from calibration model OR End-point value inferred from calibration model OR Scores of calibration model are CTQ parameters Spectral Data Process Data Discrete Data PLS/PCA Calibration Model Dynamic PCA Controller Normal Operating Space Design Space 21 CFR part 11 Compliant Records Courtesy Perceptive Engineerin

42 PAT Sensors in Closed Loop Process Control Some Challenges Real-Time Management of Process Data - maintaining a traceable record of Data Quality. Real time pre-processing needs to be consistent with and traceable to the unique calibration model. No control system is going to control a spectrum of several hundred simultaneous values; so what is important? Is there is a fit-for-purpose calibration model to infer specific product properties? Are there particular features / segments of the spectrum of interest? Should the scores of the PCA/PLS calibration model be controlled directly using Latent Variable controllers?

43 Round up Be realistic about what can be achieved at each development stage. Use generic nature of development process to your advantage. Identify which unit operations are the source of greatest product quality variation, eg upstream or downstream processing. Start early and use data derived from existing development strategies. Data infrastructure is critical for acquisition and collation of large data sets such as NIR spectra. Data Robustness is critical to operate in a validated environment. Essential to work with equipment vendors on system specification (both analytical and control systems providers). Ease-of-Use and highly visual software is critical On-line Spectral methods coupled with closed loop process control will be critical and useful tools for process transfer and comparability.

44 Courtesy Prof Al Sacco, Chemical Engineer, Space Scientist So - Where will IFPAC be in 2020?

45 Closure and Thanks In this talk, we have tried to Overview some novel methodologies for building robust calibration and statistical monitoring models using Data Augmentation and Bootstrap Aggregated Regression. Highlight some process analytical technologies through work aiming to improve understanding of batch crystallisation processes and its scale-up, and the provision of enhanced process understanding through advanced chemometrics. And show how some new closed loop process PAT control ideas are coming to fruition through innovative process systems engineering.

46 any thanks to the IFPAC Organisers for their kind invitation and of course you for your kind attention will be happy to attempt to answer your questions