Download presentation
Presentation is loading. Please wait.
Published byCalvin Jeffries Modified over 9 years ago
1
University Politehnica of Bucharest - Doctor Honoris Causa Professor Stratos Pistikopoulos FREng
2
Outline A brief introduction Chemical Engineering Process Systems Engineering On-going research areas & projects Multi-parametric programming & control
3
Stratos Pistikopoulos Diploma (Chem Eng) AUTh, 1984 PhD (Chem Eng) CMU, 1988 1991 – Imperial College London; since 1999 Professor of Chemical Engineering 2002 - 2009 Director, Centre for Process Systems Engineering (CPSE), Imperial 2009 - 2013 Director of Research, Chem Eng, Imperial 2009 - 2013 Member, Faculty of Engineering Research Committee, Imperial
4
Stratos Pistikopoulos Process systems engineering Modelling, optimization & control Process networks, energy & sustainable systems, bioprocesses, biomedical systems 250+ major journal publications, 8 books, 2 patents h-index 40; ~5000 citations
5
Stratos Pistikopoulos FREng, FIChemE (Co-) Editor, Comp & Chem Eng Co-Editor, Book Series (Elsevier & Wiley) Editorial Boards – I&ECR, JOGO, CMS Founder/Co-founder & Director – PSE Ltd, ParOS 2007 – co-recipient Mac Robert Award, RAEng 2008 – Advanced Investigator Award, ERC 2009 – Bayer Lecture, CMU 2012 – Computing in Chemical Engineering Award, CAST, AIChE 2014 – 21 st Professor Roger Sargent Lecture, Imperial
6
Chemical Engineering
7
Emerging Chemical Engineering Relatively young[er] profession (societies founded in early part of 19 th century, Manchester, UCL, Imperial - 1880s; MIT 1888) (Most likely the) most versatile engineering profession (strong societies & academic programmes, highly-paid in manufacturing, business, banking, consulting) Central discipline towards addressing societal grand challenges (energy & the environment/sustainability, health & the bio-(mics) ‘revolution’, Nano-engineering, Info-’revolution’, central to almost all Top 10 emerging technologies for 2012 World Economic Forum!) Multi-scale & multi-discipline chemical engineering
8
Evolution of Chemical Engineering Recognition of length and time scales
9
Evolution of Chemical Engineering Length-scale Time-scale Factors Energy (algae, energy-based metabolic engineering & optimisation) Product (quality, formulation, quantity) Control (model-based Information pathways) Transport (Molecular Design of Nanoparticles) Only Chemical Engineering integrates TIME, LENGTH, FACTORS (input/output)
10
Chemical Engineering - research Research.. – strong core chemical engineering, new opportunities in nano-driven chemical engineering, biochemical and biomedical-driven chemical engineering, energy/sustainability-driven chemical engineering, info-driven chemical engineering Interactions/interfaces with chemistry, materials, medicine, biology, computing/applied math & beyond – molecular level, nano-materials, nano/micro-reaction, ‘micro-human’, carbon dioxide conversion, bio-energy, resource efficiency & novel manufacturing, from ‘mind to factory’, systems of systems,...
11
Chemical Engineering – a model Core Multi-scale Understanding & Modelling
12
Chemical Engineering – a model Core Multi-scale Understanding & Modelling Simulation/ Optimization Measurements/ Visualization/ Analytics Design/ Products & Processes Properties/ Transport/ Reaction/ Separation Experiments/ Validation
13
Chemical Engineering – a model Bio & Medical driven Chemical Engineering Energy/ Sustainability Chemical Engineering Nano-Chemical Engineering Molecular & Materials/Product Chemical Engineering Core Multi-scale Understanding & Modelling Simulation/ Optimization Measurements/ Visualization/ Analytics Design/ Products & Processes Properties/ Transport/ Reaction/ Separation Experiments/ Validation
14
Chemical Engineering – a model Bio & Med driven Chemical Engineering Energy/ Sustainability Chemical Engineering Nano- & Multi-scale Chemical Engineering Molecular/Materials Chemical Engineering Core Multi-scale Understanding & Modelling Materials Analytical Sciences Systems Transport & Separation Reaction & Catalysis
15
Outline A brief introduction Chemical Engineering Process Systems Engineering On-going research areas & projects Multi-parametric programming & control
16
Process Systems Engineering
17
Scientific discipline which focuses on the ‘study & development of theoretical approaches, computational techniques and computer-aided tools for modelling, analysis, design, optimization and control of complex engineering & natural systems – with the aim to systematically generate and develop products and processes across a wide range of systems involving chemical and physical change; from molecular and genetic information and phenomena, to manufacturing processes, to energy systems and their enterprise-wide supply chain networks’
18
PSE – brief historical overview Relatively ‘new’ area in chemical engineering – started in the sixties/early seventies [Roger Sargent, Dale Rudd, Richard Hughes, and others & their academic trees] Chemical Engineering – around 1890+ [MIT, UCL, Imperial] AIChE - 1908; IChemE - 1922
19
PSE – brief historical overview Relatively ‘new’ area in chemical engineering – started in the sixties/early seventies [Roger Sargent, Dale Rudd, Richard Hughes, and others & their academic trees] Key historical dates – 1961 the term introduced [special volume of AIChE Symposium Series]; 1964 first paper on SPEEDUP [simulation programme for the economic evaluation and design of unsteady-state processes]; 1968 first textbook ‘Strategy of Process Engineering’ by Rudd & Watson (Wiley); 1970 CACHE Corporation; 1977 CAST division of AIChE; 1977 Computers & Chemical Engineering Journal
20
PSE – brief historical overview 1980s – FOCAPD 1980; PSE 1982; CPC, FOCAPO Early 90s – ESCAPE series Significant growth Centres of excellence & critical mass – CMU, Purdue, UMIST, Imperial, DTU, MIT, others around the world (US, Europe, Asia – Japan, Singapore, Korea, China, Malaysia )
21
PSE – Current Status Well recognized field within chemical engineering PSE academics in many [most?] chemical engineering departments Undergraduate level – standard courses [& textbooks] on process analysis, process design, process control, optimization, etc Research level – major activity & strong research programmes [US & Canada, Europe, Asia, Latin America, Australia]
22
PSE – Current Status Well established global international events & conferences Highly respected journals, books & publications Strong relevance to & acceptance by industry- across wide range of sectors [from oil & gas to chemicals, fine chemicals & consumer goods,..] PSE software tools – essential in industry & beyond [simulation, MPC, optimization, heat integration, etc – PSE linked companies]
23
PSE – impact Training & education Significant research advances process design process control process operations numerical methods & optimization [software & other] tools Beyond chemical engineering.. [?]
24
‘Traditional’ PSE PSE Core PSE Core Mathematical Modelling Process Synthesis Product & Process Design Process Operations Process Control Numerical Methods & Optimization
25
PSE Core Recognition of length and time scales nano-scale From nano-scale (molecular) micro-scale to micro-scale (particles, crystals) meso-scale to meso-scale (materials, equipment, products) mega-scale to mega-scale (supply chain networks, environment) PSE evolution..
26
PSE Core Recognition of length and time scales nano-scale From nano-scale (molecular) micro-scale to micro-scale (particles, crystals) meso-scale to meso-scale (materials, equipment, products) mega-scale to mega-scale (supply chain networks, environment) Multi-scale Modelling PSE evolution..
27
Product Value Chain (Marquardt; Grossmann et al) Recognition of length and time scales
28
PSE evolution... Multi-scale Modelling
29
PSE evolution... Multiscale Modelling simulation control optimization Product/process design synthesis
30
Recognition of length and time scales nano-scale From nano-scale (molecular) micro-scale to micro-scale (particles, crystals) meso-scale to meso-scale (materials, equipment, products) mega-scale to mega-scale (supply chain networks, environment) Core, generic enabling technology provider to other domains molecular genomic biological materials energy automation plants oilfields global supply chains Multi-scale process systems engineering PSE evolution
31
Multi-scale Process Systems Engineering Biological & Biomedical Systems Engineering Energy/Sustainability Systems Engineering Supply Chain Systems Engineering Multi-scale Modelling Molecular Systems Engineering simulation control optimization Product/process design synthesis
32
Multi-scale PSE PSE Core PSE Core Domain-driven PSE Domain-driven PSE Problem-centric PSE Problem-centric PSE
33
PSE Core Multi-scale Modelling Multi-scale Optimization Product & Process Design Process Operations Control & Automation
34
Domain-driven PSE Molecular Systems Engineering Materials Systems Engineering Biological Systems Engineering Energy Systems Engineering
35
Problem-centric PSE Environmental systems engineering Safety systems engineering Manufacturing supply chains
36
Multi-scale Process Systems Engineering Biological & Biomedical Systems Engineering Energy/Sustainability Systems Engineering Supply Chain Systems Engineering Multi-scale Modelling Molecular Systems Engineering simulation control optimization design synthesis
37
Multi-scale Process Systems Engineering leads to.. Biological & Biomedical Systems Engineering Energy/Sustainability Systems Engineering Supply Chain Systems Engineering Multi-scale Modelling Molecular Systems Engineering simulation control optimization design synthesis
38
CONCEPT OPERATION DESIGN Detailed design of complex equipment Process flowsheeting Optimization of plant and operating procedures Process developmen t Operational optimization TC A Plant Troubleshooting/ Safety Model- based automation Model Based Innovation across the Process Lifecycle
39
Process Systems Engineering.. provides the ‘scientific glue’ within chemical engineering (Perkins, 2008) Bio-driven Chemical Engineering Energy -driven Chemical Engineering Multi-scale Chemical Engineering Process Systems Engineering Molecular Driven Chemical Engineering Materials Analytics/ Experimental Properties Reaction engineering Transport Phenomena
40
Process Systems Engineering ‘systems thinking & practice’ – essential to address societal grand challenges Health Energy Sustainable Manufacturing Systems Engineering Nano - materials simulation control optimization design synthesis
41
Outline A brief introduction Chemical Engineering Process Systems Engineering On-going research areas & projects Multi-parametric programming & control
42
Research Group - research areas & current projects
43
Acknowledgements Funding EPSRC - GR/T02560/01, EP/E047017, EP/E054285/1 EU - MOBILE, OPTICO, PRISM, PROMATCH, DIAMANTE, HY2SEPS, IRSES CPSE Industrial Consortium, KAUST Air Products People J. Acevedo, V. Dua, V. Sakizlis, P. Dua, N. Bozinis, P. Liu, N. Faisca, K. Kouramas, C. Panos, L. Dominguez, A. Voelker, H. Khajuria, M. Wittmann- Hohlbein, H. Chang P. Rivotti, A. Krieger, R. Lambert, E. Pefani, M. Zavitsanou, E. Velliou, G. Kopanos, A. Manthanwar, I. Nascu, M. Papathanasiou, N. Diangelakis, M. Sun, R. Oberdieck John Perkins, Manfred Morari, Frank Doyle, Berc Rustem, Michael Georgiadis Imperial & ParOS R&D Teams, Tsinghua BP Energy Centre
44
Current Research Focus Overview Multi-parametric programming & Model Predictive Control [MPC] Energy & Sustainability (driven) Systems Engineering Biomedical Systems Engineering
45
Energy and Sustainability (driven) Systems Synthesis and Design Design of micro-CHP systems for residential applications Design of poly-generation systems Long-term design and planning of general energy systems under uncertainty Operations and control Scheduling under uncertainty of micro-CHP systems for residential applications Supply chain optimization of energy systems Integration of design and control for energy systems – fuel cells, CHPs Integration of scheduling and control of energy systems under uncertainty
46
Biomedical Systems Engineering Leukaemia – Development of optimal protocols for chemotherapy drug delivery for: Acute Myeloid Leukaemia (AML) Chronic Lymphocytic Leukaemia (CLL) Experimental, modelling and optimization activity Anaesthesia & Diabetes Emphasis on modelling and control in volatile anaesthesia the artificial pancreas Collaboration with Prof. Mantalaris and Dr. Panoskaltsis Collaboration with Prof Frank Doyle, UC Santa-Barbara
47
Multi-Parametric Programming & Explicit MPC a progress report Professor Stratos Pistikopoulos FREng
48
Outline Key concepts & historical overview Recent developments in multi-parametric programming and mp-MPC MPC-on-a-chip applications
49
What is On-line Optimization? MODEL/OPTIMIZER SYSTEM Data - Measurements Control Actions
50
What is Multi-parametric Programming? Given: a performance criterion to minimize/maximize a vector of constraints a vector of parameters
51
What is Multi-parametric Programming? Given: a performance criterion to minimize/maximize a vector of constraints a vector of parameters Obtain: the performance criterion and the optimization variables as a function of the parameters the regions in the space of parameters where these functions remain valid
52
Multi-parametric programming (2) Critical Regions (1) Optimal look-up function Obtain optimal solution as a function of the parameters Obtain optimal solution u(x) as a function of the parameters x
53
Multi-parametric programming Problem Formulation
54
Multi-parametric programming Critical Regions x2x2 x1x1
55
Multi-parametric programming Multi-parametric Solution
56
Multi-parametric programming Only 4 optimization problems solved!
57
On-line Optimization via off-line Optimization System State Control Actions OPTIMIZER SYSTEM POP PARAMETRIC PROFILE SYSTEM System State Control Actions Function Evaluation!
58
Multi-parametric/Explicit Model Predictive Control Compute the optimal sequence of manipulated inputs which minimizes On-line re-planning: Receding Horizon Control tracking error = output – reference subject to constraints on inputs and outputs tracking error = output – reference subject to constraints on inputs and outputs
59
Compute the optimal sequence of manipulated inputs which minimizes On-line re-planning: Receding Horizon Control Multi-parametric/Explicit Model Predictive Control Solve a QP at each time interval
60
Multi-parametric Programming Approach State variables Parameters Control variables Optimization variables MPC Multi-Parametric Programming problem Control variables F(State variables) Multi-parametric Quadratic Program
61
Explicit Control Law
62
Multi-parametric Controllers SYSTEM Parametric Controller Optimization Model (2) Critical Regions (1) Optimal look-up function Measurements Control Action Input Disturbances System Outputs Explicit Control Law Eliminate expensive, on-line computations Valuable insights ! MPC-on-a-chip!
63
A framework for multi-parametric programming & MPC (Pistikopoulos 2008, 2009) ‘High-Fidelity’ Dynamic Model Model Reduction Techniques System Identification Modelling/ Simulation Identification/ Approximation Model-Based Control & Validation Closed-Loop Control System Validation Extraction of Parametric Controllers u = u ( x ( θ ) ) ‘Approximate Model’ Multi-Parametric Programming (POP)
64
‘High-Fidelity’ Dynamic Model Model Reduction Techniques System Identification Modelling/ Simulation Identification/ Approximation Model-Based Control & Validation Closed-Loop Control System Validation Extraction of Parametric Controllers u = u ( x ( θ ) ) ‘Approximate Model’ Multi-Parametric Programming (POP) REAL SYSTEM EMBEDDED CONTROLLER On-line Embedded Control: Off-line Robust Explicit Control Design: A framework for multi-parametric programming and MPC (Pistikopoulos 2010)
65
Key milestones-Historical Overview Number of publications 2002 Automatica paper - citations [Sep 2014]: 900+ WoS; 1200+ Scopus; 1650+ Google Scholar Multi-parametric programming – until 1992 mostly analysis & linear models Multi-parametric/explicit MPC – post-2002 much wider attention Multi-Parametric Programming Multi-Parametric MPC & applications Pre-1999>100 0 Post-1999~70250+ AIChE J.,Perspective (2009)
66
Multi-parametric Programming Theory mp-LP Gass & Saaty [1954], Gal & Nedoma [1972], Propoi [1975], Adler and Monterio [1992], Gal [1995], Acevedo and Pistikopoulos[1997], Dua et al [2002], Pistikopoulos et al [2007] mp-QP Townsley [1972], Propoi [1978], Best [1995], Dua et al [2002], Pistikopoulos et al [2002,2007] mp-NLP Fiacco [1976],Kojima [1979], Bank et al [1983], Fiacco [1983], Fiacco & Kyoarisis [1986], Acevedo & Pistikopoulos [1996], Dua and Pistikopoulos [1998], Pistikopoulos et al [2007] mp-DO Sakizlis et al.[2002], Bansal [2003], Sakizlis et al [2005], Pistikopoulos et al [2007] mp-GO Fiacco [1990], Dua et al [1999,2004], Pistikopoulos et al [2007] mp-MILP Marsten & Morin [1975], Geoffrion & Nauss [1977], Joseph [1995], Acevedo & Pistikopoulos [1997,1999], Dua & Pistikopoulos[ 2000] mp-MINLP McBride & Yorkmark [1980], Chern [1991], Dua & Pistikopoulos [1999], Hene et al [2002], Dua et al [2002]
67
Multi-parametric/Explicit Model Predictive Control Theory mp-MPC Pistikopoulos [1997, 2000], Bemporad, Morari, Dua & Pistikopoulos [2000], Sakizlis & Pistikopoulos [ 2001], Tondel et al [2001], Pistikopoulos et al [2002], Bemporad et al [2002], Johansen and Grancharova [2003], Sakizlis et al [2003], Pistikopoulos et al [2007] mp-Continuous MPC Sakizlis et al [2002], Kojima & Morari[ 2004], Sakizlis et al [2005], Pistikopoulos et al [2007] Hybrid mp-MPC Bemporad et al [2000], Sakizlis & Pistikopoulos [2001], Pistikopoulos et al [2007] Robust mp- MPC Kakalis & Pistikopoulos [2001], Bemporad et al [2001], Sakizlis et al [2002], Sakizlis & Pistikopoulos [2002], Sakizlis et al [2004], Olaru et al [2005], Faisca et al [2008] mp-DP Nunoz de la Pena et al [2004],Pistikopoulos et al [2007],Faisca et al [2008] mp-NMPC Johansen [2002], Bemporad [2003], Sakizlis et al [2007], Dobre et al [2007], Narciso & Pistikopoulos [2009]
68
68
69
Patented Technology Improved Process Control European Patent No EP1399784, 2004 Process Control Using Co-ordinate Space United States Patent No US7433743, 2008
70
Multi-parametric programming & Model Predictive Control [MPC] Theory of multi-parametric programming Multi-parametric mixed integer quadratic programming [mp-MIQP] Multi-parametric dynamic optimization [continuous-time, mp-DO] Multi-parametric global optimization Theory of multi-parametric/explicit model predictive control [mp-MPC] Explicit robust MPC of hybrid systems Explicit MPC of continuous time-varying [dynamic] systems Explicit MPC of periodic systems Moving Horizon Estimation & mp-MPC
71
Multi-parametric programming & Model Predictive Control [MPC] – cont’d Framework for multi-parametric programming & control Model approximation [from high fidelity models to the design of explicit MPC controllers] Software development, prototype & demonstrations [for teaching & research] Application areas Fuel cell energy system – experimental/laboratory Car system control – prototypes/laboratory Energy systems [CHP and micro-CHP] Bio-processing [continuous production & control of monoclonal antibodies] Pressure Swing Absorption [PSA] and hybrid systems Biomedical Systems
72
MPC-on-a-chip Applications – Recent Developments Process Control Air Separation (Air Products) Hybrid PSA/Membrane Hydrogen Separation (EU/HY2SEPS, KAUST) Automotive Active Valve Train Control (Lotus Engineering) Energy Systems Hydrogen Storage (EU/DIAMANTE) Fuel Cell
73
MPC-on-a-chip Applications – Recent Developments Biomedical Systems (MOBILE - ERC Advanced Grant Award) Drug/Insulin, Anaesthesia and Chemotherapeutic Agents Delivery Systems Imperial Racing Green Fuel cell powered Student Formula Car Aeronautics (EPSRC) (Multiple) Unmanned Air Vehicles – with Cranfield University
74
Small Air Separation Units (Air Products, Mandler et al,2006) Enable advanced MPC for small separation units Optimize performance Minimize operating costs Satisfy product and equipment constraints Parametric MPC ideally suited Supervises existing regulatory control Off-line solution with minimum on-line load Runs on existing PLC Rapid installation compared to traditional MPC Advantages of Parametric MPC 5% increased throughput 5% less energy usage 90% less waste Installation on PLC in 1-day
75
Active Valve Train Control (Lotus Engineering, Kosmidis et al, 2006) Active Valve Trains (AVT): Optimum combustion efficiency, Reduced Emissions, Elimination of butterfly valve, Cylinder deactivation, Controlled auto-ignition (CAI), Quieter operation Basic idea: Control System sends signal to valve This actuates piston attached to engine valve Enables optimal control of valve timing over entire engine rpm range Challenges for the AVT control Nonlinear system dynamics: Saturation, flow non-linearity, variation in fluid properties, non-linear opening of the orifices Robustness to various valve lift profiles Fast dynamics and sampling times (0.1ms)
76
Multi-parametric Control of H 2 Storage in Metal-Hydride Beds (EU-DIAMANTE, Georgiadis et al, 2008) Tracking the optimal temperature profile Ensure economic storage – expressed by the total required storage time Satisfy temperature and pressure constraints Optimal look-up table (Projected on the y t - u t plane)
77
PEM Fuel Cell Unit Collaborative work with Process Systems Design & Implementation Lab (PSDI) at CERTH - Greece
78
Unit Specifications Fuel Cell : 1.2kW Anode Flow : 5..10 lt/min Cathode Flow : 8..16 lt/min Operating Temperature : 65 – 75 °C Ambient Pressure Control Control Strategy Start-up Operation Heat-up Stage : Control of coolant loop Nominal Operation Control Variables : Mass Flow Rate of Hydrogen & Air Humidity via Hydrators temperature Cooling system via pump regulation Known Disturbance : Current Unit Design CERTH Unit Design : Centre For Research & Technology Hellas (CERTH) (2) Critical Regions (1) Optimal look-up function PEM Fuel Cell System m H2 m Air m cool TY Hydrators V fan T st HT st PEM Fuel Cell Unit
79
79
80
80
81
81
82
82
83
Imperial Racing Green Car Student Formula Project Control of Start-up/Shut- down of the FC Traction Motion Control Control & Acquisition System FPGA (MPC-on-a-Chip)
84
Biomedical Systems (MOBILE ERC Advanced Grant) Step 1: The sensor measures the glucose concentration from the patient Step 2: The sensor then inputs the data to the controller which analyses it and implements the algorithm Step 3: After analyzing the data the controller then signals the pump to carry out the required action Step 4: The Insulin Pump delivers the required dose to the patient intravenously Controller Sensor Patient Insulin Pump 12 34
85
University Politehnica of Bucharest - Doctor Honoris Causa Mulumesc!
86
University Politehnica of Bucharest - Doctor Honoris Causa Professor Stratos Pistikopoulos FREng
Similar presentations
© 2024 SlidePlayer.com Inc.
All rights reserved.