Software and Hardware Solutions for Accurate Data and Profitable Operations. Miguel J. Donald J. Chmielewski Contributor. DuyQuang Nguyen Tanth

Transcription

1 Smart Process Plants Software and Hardware Solutions for Accurate Data and Profitable Operations Miguel J. Bagajewicz, Ph.D. University of Oklahoma Donald J. Chmielewski Contributor DuyQuang Nguyen Tanth Contributor TECHNISCHE INFOR K'l AT! 0 M S,1 ibllothek v UNIVERSlTATSSiP 1fOTH E K HANNOVER, _ / Mc Grauu Hill New York Chicago San Francisco Lisbon London Madrid Mexico City Milan New Delhi San Juan Seoul Singapore Sydney Toronto

2 Contents Preface xiii 1 Smart Plants 1 Our Vision 7 Value of Information 7 Book Focus and Contents 8 References 9 2 Measurement Errors 11 Range and Span 11 Precision 11 Origin of Fluctuations 14 Systematic Error (Bias) 14 Outliers 21 Accuracy 27 Calibration Curves 27 Hysteresis and Dead Band 30 Guide to the Expression of Uncertainty in Measurement 30 References 32 3 Variable Classification 35 Linear Model 35 Observability 36 Redundancy 37 Hardware Redundancy Quantification of Observability and Redundancy Estimability 38 Degree of Observability of Variables 38 Degree of Redundancy of Variables 39 Degree of Estimability of Variables 40 Canonical Representation 41 The General Case 48 References 49 4 Material Balance Data Reconciliation 51 Determination of the Measurement Vector Precision of the Estimates 58 Variance of Observable Quantities 59 Method to Avoid the Classification Step 60 References 62 vii

3 viii Contents 5 Gross Error Detection 65 Gross Error Handling 66 Smearing 66 Tests for Gross Errors 68 Hypothesis 68 Testing Type I and Type II Errors and the Power of a Test 70 Global Test 71 Nodal Test 72 Maximum Power Nodal Test 73 Measurement Test 73 Maximum Power Measurement Test 74 Generalized Likelihood Ratio 74 Principal Component Test 75 Multiple Gross Errors and Gross Error Elimination 75 Serial Elimination Strategy Based on the Global Test 76 Serial Elimination Based on the Measurement Test 77 Failures of Tests 80 References 83 6 Equivalency of Gross Errors 87 Definition 87 Practical Consequences 88 Cardinality of Equivalent Sets 88 Basic Subset of an Equivalent Set 89 Determination of Equivalent Sets 89 Practical Consequence 90 Degeneracy 90 Quasi-Degeneracy 91 Quasi-Equivalency 91 Detection of Leaks 91 Practical Approach to Equivalency 95 References 96 7 Gross Error Size Elimination and Estimation The Compensation Model 97 Serial Identification with Collective Compensation Strategy (SICC) 98 The Unbiased Estimation Model (UBET) 99 Conversion between Equivalent Sets 100 References 103

4 Contents ix 8 Nonlinear Data Reconciliation 105 Component Balances 105 Splitters 106 Number of Components 107 Measurement Pattern 107 Energy Balances 108 Heat Exchangers 109 Full Nonlinear Systems Ill Gross Error Detection in Nonlinear Systems 115 Parameter Estimation 116 References Dynamic Data Reconciliation 121 Filtering 121 Recursive Filters 121 Linear Estimators 122 Discrete Kalman Filter 123 Quasi-Steady State Estimator 123 A Balance-Based Quasi-Steady State Estimator 124 Difference Estimators 125 Integral Approach 127 Nonlinear Case 128 Gross Error Detection 132 References Accuracy of Estimators 139 Accuracy of Measurements 139 Induced Bias 140 Accuracy of Estimators 141 Maximum Undetected Induced Bias 141 Maximum Power Measurement Test-Based Software Accuracy 142 Graphical Representation of Undetected Biases 148 Effect of Equivalency of Errors 152 Stochastic Software Accuracy 156 Monte Carlo Sampling 159 Instantaneous Testing 162 Periodic Testing 162 References Economic Value of Accuracy 177 Value of Precision 177 Value of Accuracy 181 Probabilities 182

5 X Contents Trade-Off between Value and Cost 188 References Data Reconciliation Practical Issues 195 Data Preprocessing 195 Use of Filters 196 Steady-State Recognition and Variance 196 Variance Estimation 196 Steady-State Detection 200 Ratio Test 201 Tanks and Steady-State Data Reconciliation 204 Use of Dynamic Data in Steady-State Reconcilers 205 Random Error Distributions 209 Multiple Measurements of the Same Variable 209 Excessive Number of Gross Errors 210 References Value of Control Strategies 225 Classic Control 225 Model Predictive Control 233 The Hierarchy of the Modern Control Architecture 248 State-Space Process Modeling 251 Disturbance Modeling 253 Expected Dynamic Operating Region Characterization 259 Constrained Minimum Variance Control 265 Connection between CMV Control and MPC 271 Control System Value 273 Impact of Process and Measurement Biases 280 Conclusions 283 References Value of Parametric Fault Identification 287 Introduction 287 Fault Classification 287 Fault Detection and Diagnosis Techniques 288 Fault Observability 290 Single Fault Resolution 297 Multiple Fault Resolution 298 Value of Fault Detection 302 References 309

6 Contents XI 15 Value of Instrumentation Upgrade Monitoring and Faults Perspectives 311 Cost-Optimal Instrumentation Design 313 Cost-Optimal Design 317 Cost-Optimal Design for Precision or... Accuracy 317 Tree Search Procedure for the Cost-Optimal Formulation 320 Branching Criteria 323 Cost-Optimal Design for Parametric Faults 328 Integrated Cost-Optimal Design 332 Value-Optimal Instrumentation Design 334 References Value of Instrumentation Upgrade Control Perspective 341 References Structural Faults and Value of Maintenance 353 Maintenance 354 Maintenance Policies 355 Reliability, Failure Rate, and Mean Time to Failure 358 Failure Density Distributions 362 Exponential Distribution 362 Weibull Distribution 363 Normal Distribution 363 Maintenance Models 364 Renewal Processes 365 Markov Processes 378 Discrete Time Markov Models 385 Monte Carlo Simulation 388 Maintenance Policy and Decision Variables 390 Interfering/Noninterfering Units 391 Input Data 391 Spare Parts Inventory 391 Labor Assignment 392 Imperfect Maintenance 392 Maintenance Rules 392 Monte Carlo Simulation Procedure 393 Advantages and Limitations 401 Renewal Process 401 Markov Process 402 Monte Carlo Simulation 402 References 402

7 xij Contents 18 Maintenance Optimization 405 Components of Maintenance Optimization 406 Renewal Process-Based Models 407 Markov-Based Model 410 Monte Carlo-Based Models Using Genetic Algorithms 411 References Value and Optimization of Instrument Maintenance 423 Financial Loss Evaluation 428 References 435 Index 437