Models offer benefits even before they are put on line. Based on years of experience, the authors reveal in New Directions in Bioprocess Modeling and Control that significant improvements can result from the process knowledge and insight that are gained when building experimental and first-principle models for process monitoring and control. Doing modeling in the process development and early commercialization phases is advantageous because it increases process efficiency and provides ongoing opportunities for improving process control. This technology is important for maximizing benefits from analyzers and control tool investments. If you are a process design, quality control, information systems, or automation engineer in the biopharmaceutical, brewing, or bio-fuel industry, this handy resource will help you define, develop, and apply a virtual plant, model predictive control, first-principle models, neural networks, and multivariate statistical process control. The synergistic knowledge discovery on bench top or pilot plant scale can be ported to industrial scale processes. This learning process is consistent with the intent in the Process Analyzer and Process Control Tools sections of the FDA s Guidance for Industry PAT A Framework for Innovative Pharmaceutical Development, Manufacturing and Quality Assurance. It states in the Process Analyzer section of the FDA s guidance: For certain applications, sensor-based measurements can provide a useful process signature that may be related to the underlying process steps or transformations. Based on the level of process understanding these signatures may also be useful for the process monitoring, control, and end point determination when these patterns or signatures relate to product and process quality.
Author(s): Michael A. Boudreau, Gregory K. McMillan
Publisher: ISA
Year: 2006
Language: English
Pages: 307
Tags: Биологические дисциплины;Матметоды и моделирование в биологии;
Index......Page 0
Front Matter......Page 1
Acknowledgments......Page 3
About the Authors......Page 4
Table of Contents......Page 5
Preface......Page 7
1.1 Introduction......Page 9
1.2 Analysis of Variability......Page 12
1.2.1 Process Inputs and Outputs......Page 15
1.2.2 Field and Laboratory Measurements......Page 16
1.2.3 Experimental Models......Page 17
1.2.4 Quality and Quantity of Data......Page 18
1.2.6 Dynamic First-Principle Models......Page 19
1.2.7 Batch, Fed-Batch, and Continuous Modes......Page 20
1.3 Transfer of Variability......Page 22
1.3.1 Discrete Process Actions......Page 24
1.3.2 Feedback Control......Page 25
1.3.3 Selecting MSPC, PLS, and ANN Inputs......Page 27
1.4 Online Indication of Performance......Page 30
1.5 Optimizing Performance......Page 33
1.6.1 FDA Initiative......Page 34
1.6.2 Business Drivers......Page 35
1.6.3 PAT Tools......Page 36
References......Page 37
2.1.1 Learning Objectives......Page 38
2.2.1 Impact of Loop Dead Time......Page 39
2.2.2 Impact of Controller Tuning Settings......Page 41
2.2.3 Sources of Dead Time......Page 45
2.2.4 Limit Cycles......Page 49
2.3 Self-Regulating Processes......Page 50
2.4 Integrating Processes......Page 54
References......Page 57
3.1 Introduction......Page 58
3.1.1 Learning Objectives......Page 60
3.2.1 Proportional Mode Structure and Settings......Page 61
3.2.2 Integral Mode Structure and Settings......Page 63
3.2.3 Summary of Structures......Page 64
3.2.4 Algorithms......Page 65
3.3.1 Focus......Page 72
3.3.2 Temperature Loops......Page 73
3.3.4 Closed and Open Loop Responses......Page 74
3.3.5 Lambda Tuning......Page 75
3.3.6 Contribution of Loop Components......Page 77
3.3.7 Unified Approach......Page 78
3.3.9 Near Integrators......Page 80
3.3.10 Loop Cycling......Page 81
3.3.11 Composition Loops......Page 82
3.3.12 Biomass and Product Profile Control......Page 86
3.4.2 Road Maps and Terrain......Page 88
3.4.4 Watching but Not Waiting......Page 89
3.4.5 Back to the Future......Page 91
3.5.1 Nothing Says Forever Like Tradition......Page 92
3.5.3 Without Dead Time I Would Be Out of a Job......Page 94
References......Page 97
4.1 Introduction......Page 98
4.2 Capabilities and Limitations......Page 99
4.2.1 Situations Where MPC Can Be Beneficial......Page 106
4.3.1 High- and Low-Cost MV......Page 108
4.3.2 Coarse and Fine MV......Page 111
4.4 Optimization......Page 115
References......Page 126
5.1.1 Learning Objectives......Page 127
5.2 Key Features......Page 128
5.3 Spectrum of Uses......Page 134
5.4 Implementation......Page 137
References......Page 143
6.1.1 Learning Objectives......Page 144
6.2 Our Location on the Model Landscape......Page 145
6.3 Mass, Energy, and Component Balances......Page 146
6.4 Heat of Reaction......Page 151
6.5 Charge Balance......Page 152
6.6 Parameters and Their Engineering Units......Page 155
6.7 Kinetics......Page 160
6.7.1 Penicillin Production......Page 161
6.7.2 Human Epidermal Growth Factor Production......Page 166
6.7.3 Hybridoma Growth and Monoclonal Antibody Production......Page 170
6.8.1 Oxygen Interphase Mass Transfer......Page 173
6.8.3 Power Requirement versus Volumetric Oxygen Mass Transfer Coefficient......Page 174
6.8.5 Oxygen and Carbon Dioxide Mass Balances in Penicillin Production......Page 175
6.8.7 Gas Liquid Mass Transfer in a Mammalian Cell Culture......Page 176
6.8.8 Evaporation Loss from a Fed-Batch Bioreactor......Page 177
6.9 Simulated Batch Profiles......Page 178
References......Page 181
7.1 Introduction......Page 183
7.1.1 Learning Objectives......Page 184
7.2 Types of Networks and Uses......Page 188
7.3 Training a Neural Network......Page 190
7.3.1 Other Learning Methods......Page 192
7.4 Timing Is Everything......Page 193
7.5 Network Generalization: More Isn't Always Better......Page 196
7.6 Network Development: Just How Do You Go About Developing a Network?......Page 198
7.7 Neural Network Example One......Page 201
7.7.1 Automatic Control Method......Page 206
7.8 Neural Network Example Two......Page 207
7.8.1 Network Construction......Page 209
7.8.2 Nine-Parameter Model......Page 211
7.8.3 Six-Parameter Model......Page 213
7.8.4 Four-Parameter Model......Page 214
7.8.5 Two-Parameter Model......Page 215
7.9 Designing Neural Network Control Systems......Page 223
7.10 Discussion and Future Direction......Page 225
7.10.1 Neural Networks for PAT Compliance......Page 227
7.10.2 Correcting Neural Network Predictions Automatically......Page 228
7.11 Neural Network Point-Counterpoint......Page 229
7.11.1 Point......Page 230
7.11.2 Counterpoint......Page 231
References......Page 232
8.1 Introduction......Page 234
8.1.1 Learning Objectives......Page 235
8.2.2 Simple Example Applying PCA......Page 236
8.2.3 PCA Algorithm......Page 247
8.3 Multiway PCA......Page 252
8.3.2 PCA of a Batch Process......Page 253
8.3.3 Batch Data Synchronization......Page 257
8.4 Model-Based PCA (MB-PCA)......Page 259
8.4.2 Super Model-Based PCA......Page 260
8.5.2 PCA Application......Page 263
References......Page 270
Appendix A: Definition of Terms......Page 274
Appendix B: Condition Number......Page 284
Appendix C: Unification of Controller Tuning Relationships......Page 286
Appendix D: Modern Myths......Page 295
Appendix E: Enzyme Inactivity Decreased by Controlling the pH with a Family of Bezier Curves......Page 297
