Engineering Complex Systems Julio M. Ottino R.R. McCormick School of Engineering
by user
Comments
Transcript
Engineering Complex Systems Julio M. Ottino R.R. McCormick School of Engineering
McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science Engineering Complex Systems Julio M. Ottino R.R. McCormick School of Engineering and Applied Sciences Northwestern University JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science Collaborators • • • • • • • • Ashley Smart Ben Severson Nick Pohlman Pengfei Chen Stan Fiedor Gabe Juarez Steve Cisar Steve Meier • Nitin Jain • Jim Gilchrist • Joe McCarthy • • • • • Luis Amaral Rich Lueptow D.V. Khakhar Paul Umbanhowar Randy Snurr JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science What does Complex Mean? “Complex”, signifying “composed of parts”, comes from the French, ca 1650 the adjective meaning of “not easily analyzed” is first recorded in 1715 JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science Complex Systems: Numerous Examples in…. Biological Systems Social Systems & Organizations Physical & Chemical systems www.northwestern.edu/nico/ JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science Sociology “From Micromotives to Macrobehavior” Thomas Schelling, 1978 Nobel Prize Economics 2005 JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science From back cover of Schelling’s book • Through familiar and readily grasped examples, [the writer] demonstrates what happens when behavior in the aggregate is more than a simple summation of individual behaviors, how members of a society tend to be blind to the collective consequences of their separate decisions, and why attempts to infer individual intentions from group phenomena are tricky at best and often downright impossible. JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science Uri Wilensky, [email protected] JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science Granular Materials and Suspensions “dry” (DGS) air “wet” (LGS) liquid Examples in nature and technology JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science Segregation heap formation S2 systems In, uniform mixture Magnified view Makse et al., Nature (1997) JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science Shallow cylinders and long cylinders JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science Mixing by avalanching (1) Geometry wedges ⇒ wedges (2) Dynamics Random mixing within wedges Model Model Experiment Experiment JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science γ = rate of mixing, χ=γ x total area being mixed Metcalfe, Ottino, et al. Nature 1995 JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science Energy in: Vibration Small brass spheres Amplitude A frequency f Umbanhowar, Swinney JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science How particles interact, bottom-up understanding Contact Forces Normal Forces Tangential Forces α Spring and Slider Fn = k nα 3 / 2 − k d vn α Hertz repulsion with dissipation Ft = min ( kt s , µFn ) Linear tangential spring with maximum JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science top view perspective view side view Vibrated granular matter Umbanhowar, Swinney et al. 1996 JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science S+D system 1mm steel 2mm glass # steel 3455 #glass particles 432 (volume ratio ~ 1:1) 5cm diameter tumbler axial length of 6.4mm (3.2 times diameter of large particles) flat endwall condition angular velocity 0.21rad/s Pengfei Chen (2006) JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science Segregation, Unmixing f=4 f=6 f=8 Fiedor & Ottino J. Fluid Mech. (2005). DGS f=4 f=6 f=8 LGS n=3 n=4 n=5 Poincaré Sections JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science Poincaré Model Exp (S-DGS) Interpenetrating Continua Model S.E. Cisar, P.B. Umbanhowar, and J.M. Ottino, 2006 JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science Long cylinder, band dynamics (LGS) JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science Effect of rotation rate - DGS 5 RPM 10 RPM 15 RPM 20 RPM 25 RPM JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science Evolution granular matter time Fiedor &. Ottino, Phys. Rev. Lett. (2003). Fiedor, Umbanhowar & Ottino, Phys. Rev. E, (2006). JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science DGS LGS Laws? Logarithmic decay JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science Complex system… recognize by… (1) What is does: Display organization without any organizing principle being applied, i.e. behavior emerges (2) How can be analyzed: Decomposing the system and analyzing a part does not give a clue as to the behavior of the whole. Rich behavior with simple parts JMOttino 9/06 McCormick simple R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science complicated complex JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science GO CHESS JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science “More is different” Philip Anderson Science, 1972 John von Neumann 1903-1957 JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science Avalanches… Per Bak “How Nature Works: The Science of Self Organized Criticality” JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science Osborne Reynolds (1842-1912) JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science • Osborne Reynolds “On the dilatancy of media composed of rigid particles in contact. With experimental illustrations”. Philosophical Magazine, December 1885. • Rode Lecture in 1902 (“On an inversion of ideas as to the structure of the universe”) “I have in my hand the first experimental model universe, a soft India rubber bag filled with small shot”. • “The Sub-Mechanics of the Universe” (Reynolds 1903). “By this research it is shown that there is one, and only one, conceivable purely mechanical system capable of accounting for all the physical evidence, as we know it, in the Universe.” JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science • Mechanical theory of the ether. Universe filled with rigid grains size: 5.534x10-18 cm, mean free path 8.612x10-28 cm. • Planck length (‘quantum of length’) smallest measurement of length with any meaning: 1.6 x 10-37 cm or about 10-20 times the size of a proton. • ….but Reynolds was a teacher of J.J. Thomson (discoverer of the electron). …so much for the modernity of far-reaching analogies… JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science Unmixing, self-organization f=4 f=6 f=8 f=4 f=6 f=8 JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science Questions • Can we guide systems that want to design themselves ? • Is it possible to design elements – built in interactions – so that systems design themselves in an “intelligent” manner? • Given a final picture, can we uncover the organizing rules? JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science Complexity in materials self-assembly Structures emerge from basic thermodynamic principles and forces among building blocks (e.g. atoms, molecules, nanoparticles, and colloids). Designer building blocks Reference S.C. Glotzer, Science, 306, 419 (2004). JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science First Steps… Packing rigid shapes on a square lattice An agent-based approach for modeling molecular self-organization Troisi, Wong, and Ratner Proc. Nat. Acad. 2005 25 rigid shapes, “molecules” consisting of four “atoms” neutral (red), positive (blue), negative (black). JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science Initial definition of agents and moves Pick one agent A shape for the elementary object is selected. This is rigid and cannot be + broken. Examples − + − Randomly or one at the time The agent moves Given an agent to be moved, the energy Em of each allowed move (m) is computed. The move is selected following a Metropolis MC move. Does the agent need to be merged? If there are no moves that lower the energy (a fact established in the previous step) the agent is merged with one other agent. Merge with another agent The current agent is merged with a second agent selected to have the largest interaction (negative) energy with the first one. JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science An Example MC shape: AB Monte Carlo Agent-based JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science James Clerk Maxwell 1831-1897 Insight: From individual trajectories to distributions Statistical Mechanics Adolphe Quetelet 1796-1874 Mécanique Sociale Ref. Critical Mass, Philip Ball, FSG, 2004 JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science Granular force network g P(f) Force Distribution, P(f) Ashley Smart 2006 f /<f> JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science Networks: Nodes •Grains Nodes •Genes •Proteins •Metabolites •Neurons •Persons •Species •Organizations •Cities JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science Networks: Nodes & Links •Grains Links Nodes •Genes Forces Cross regulation •Proteins Complex participation •Metabolites Chemical reaction •Neurons Signaling •Persons Relationship •Species Trophic relation •Organizations Supply chains •Cities Travel connections JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science Predictions… Northwest Atlantic fishery http://www.imma.org/orlando.pdf JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science Harp seals Food chain Cod Crustaceans Harp seals Cod Crustaceans Sand eels Haddock Food tree JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science Harp seals Cod Crustaceans Harp seals The real web Cod Crustaceans Sand eels Haddock Luis Amaral et al. Northwestern, 2005 JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science Structure, Scaling Food Webs Amaral et al. Phys Rev E 56, 030901(R) (2002) Phys Rev Lett 88, 228102 (2002) JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science Power Grid Transportation… Critical Infrastructure JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science Snaptshot of the Internet K. C. Claffy JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science Cell Metabolism Two Prevailing Points of View: Pathways Network reactions metabolites or e.g. Edwards and Palsson, (2000) e.g. Ma and Zeng (2003) Biological “assembly line” JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science A “network interpretation” of the cell metabolism Connect a pair of nodes (metabolites) with a directed edge if they participate in the same reaction A+B+C → D+E A D B E C JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science An alternative interpretation of the cell metabolism Bipartite graph (S-graph) consists of metabolite and reaction nodes A+B+C → D+E A D 1 B C E (1 = Enzyme) JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science How Robust is this Network to Node Failure? Algorithm for cascading failure A D 4 1 I E 5 B 2 F J H C 3 K G JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science Algorithm for Cascading Failure Steady State: consumption must be balanced by production The concentration of a metabolite with outgoing edges and no incoming edges goes to zero A D 4 1 I E 5 B 2 F H C 3 Remove Reaction 2 J K G JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science Algorithm for Cascading Failure (Reverse Direction) Steady State: consumption must be balanced by production The concentration of a metabolite with incoming edges but no outgoing edges goes to infinity A D 4 1 I E 5 B 2 F H C 3 Remove Reaction 4 J K G JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science Algorithm for Cascading Failure (Reverse Direction) Steady State: consumption must be balanced by production The concentration of a metabolite with incoming edges but no outgoing edges goes to infinity A D 4 1 I E 5 B 2 F H C 3 Remove Reaction 4 J K G JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science Special Cases • Feed Metabolites (allowed to deplete) • Product Metabolites (allowed to accumulate) • Equilibrium Reactions (coupled “half reaction” nodes) A+B↔C+D A B C 1 D ⇒ A B 1F 1R C D JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science Simple “Complex” “Complex” Simple JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science Engineering, Engineer JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science • “… if a situation is complex and poorly understood, and if the solution is constrained by limited resources (knowledge?)…. then you are in the presence of an engineering problem. “To be human is to be an engineer” *Discussion of the Method, Billy Vaughn Koen JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science • “… if a situation is complex and poorly understood, and if the solution is constrained by limited resources (knowledge?)…. then you are in the presence of an engineering problem. “To be human is to be an engineer” *Discussion of the Method, Billy Vaughn Koen JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science Complexity and Engineering Conflict • The hallmarks of complex systems are adaptation, self-organization and emergence. • Engineers engineer. Engineering is about making things happen, about consistency of design, about convergence. • Engineering has typically dealt with complicated systems as opposed to complex JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science Barriers to Inter-Disciplinary Work • Language, Mindset/Culture, Academic Silos • Cultural differences, expectations, practices, community norms • Epistemological Roots: – Engineering: engineering method, heuristics (since dawn of civilization) – Mathematics: True for all time, deductive (Euclid, Gödel, Russell …) – Natural Science: Observation, reductionist, inductive vs falsifiable (Galileo, Newton, Popper, …) • Understanding of what constitutes “truth”, proof and belief for different scientific approaches From Shail Patel 2006 JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science Complexity and Engineering • Benefit by adoption of complex systems tools to standard toolkit. Ecological systems, supply chains, and materials selfassembly; new disciplines (e.g. security engineering). • Expanding modelling and prediction • Expanding understanding. • If modelling is defined as “all happens as if...”, we define understanding as “not being surprised or making sense of outcomes”. • What does “solution” mean? • Choice for engineering is not between designing everything at the outset and letting systems designing themselves – it is attacking the most pressing problems facing us today. JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science Security Engineering: Protecting Critical Infrastructure • • • • • • Natural disasters Weather Congestion of facilities Business failures Labor disruptions Terrorist actions JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science Evolution of Disciplines Engineering in no danger of becoming obsolete (same not true of branches) JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science Large Fight to keep in Crisis of Control Size of the Discipline Crisis of leadership Growth through coordination Fight to get in Small Growth through Creativity Young Mature Age of the Discipline JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science Core and periphery; Nature of knowledge expansions State of the art of domain at time t “Break-with” Breakthrough JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science Core-periphery balance Core Core Tools unify picture Pieces only loosely connected by tools JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science How could the Future be like? The Central Question: • Anything that should be in the (new) core that it is missing? and if so….What types of “new” problems could be engineering problems? JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science The Tools of Complexity • • • • Nonlinear dynamics Broad-based statistical mechanics Agent-based models Network theory Ref. J.M. Ottino, Complex Systems, AIChEJ, 49, 292-299 (2003). JMOttino 9/06 McCormick R.R. McCormick School of Engineering and Applied Science Robert R. McCormick School of Engineering and Applied Science Further Reading (about this seminar) • J.M. Ottino, Complex Systems, AIChEJ, 49, 292-299 (2003). • Amaral, L.A.N. and J.M. Ottino, Complex Networks: Completing the framework for the study of complex systems, European Physics Journal, 38, 47-162 (2004). • J.M. Ottino, Engineering Complex Systems, Nature, 427, 399 (2004). • Philip Ball, Critical Mass: How one Things leads to Another, Farrar, Straus and Giroux, 2004 • A.-L. Barabási, E. Bonabeau, Scale-free Networks, Scientific American 288, 60-69 (2003). • Alessandro Troisi, Vance Wong and Mark A. Ratner, An agentbased approach for modeling molecular self-organization, PNAS, 2004, January 11, 2005 | JMOttino 9/06