IE604 Causal Loop Diagram Jayendran Venkateswaran IEOR @ IIT - PowerPoint PPT Presentation

IE604 Causal Loop Diagram Jayendran Venkateswaran IEOR @ IIT Bombay

Causal Loop Diagram (CLD) p Visual representation of the cause-effect relationships between the various elements of the system, forming feedback loops n Conceptualize the problem n Capture hypothesis about causes of dynamics n Communication with others IEOR, IIT Bombay IE604: System Dynamics Modeling and Analysis Jayendran Venkateswaran

CLDs p CLDs consists of variables connected by causal links or arrows p Variable at tail of arrow X Y n Causal or independent variable p Variable at head of arrow n Affected or dependent variable p Arrows show the direction of causation p Causal link is associated with a link polarity n + or S n — or O IEOR, IIT Bombay IE604: System Dynamics Modeling and Analysis Jayendran Venkateswaran

Causal Link Polarity p + or S n If all other things being equal, a change in causal variable generates a change in the same direction in affected variable relative to its prior value n If X increases, then Y increases above what it would have otherwise been n If X decreases, then Y decreases below what it would have otherwise been IEOR, IIT Bombay IE604: System Dynamics Modeling and Analysis Jayendran Venkateswaran

Causal Link Polarity p — or O n If all other things being equal, a change in causal variable generates a change in the opposite direction in affected variable relative to its prior value n If X increases, then Y decreases below what it would have otherwise been n If X decreases, then Y increases above what it would have otherwise been IEOR, IIT Bombay IE604: System Dynamics Modeling and Analysis Jayendran Venkateswaran

CLD Example 1 (Goodman, M. R., Study Notes in System Dynamics) p Hypotheses of dynamics in urban region n Job availability attracts migrants to the city n New arrivals to city expand the labor population n Population absorbs available jobs, decreasing job availability n In long run, as labor also creates demand for additional services & facilities, a further increase in the number of jobs in the city comes about n More jobs increase job availability p The above description is reference mode! IEOR, IIT Bombay IE604: System Dynamics Modeling and Analysis Jayendran Venkateswaran

Loop Polarity p When feedback loop response opposes the original perturbation, the loop is negative or goal seeking p When feedback loop response reinforces the original perturbation, the loop is positive or reinforcing p Fast way: n If the number of negative links in a loop is even à loop is _______________ n If the number of negative links in a loop is odd à loop is _______________ IEOR, IIT Bombay IE604: System Dynamics Modeling and Analysis Jayendran Venkateswaran

Core Concepts p Cause and Effect p Feedback loops n Makes system complicated n Key causal links are those inside feedback loops p Positive Feedback loop n Even number of “-” signs; Amplifying, reinforcing, growth (rapid decline), unstable p Negative Feedback loop n Odd number of “-” signs; Balancing, Stabilizing, Stable equilibrium; often good. p Delays in Feedback loops n E.g., carrying cup of tea; taking a shower; etc. time units important IEOR, IIT Bombay IE604: System Dynamics Modeling and Analysis Jayendran Venkateswaran

Guidelines for CLD (1) p All links should have unambiguous polarities n Ambiguous polarity indicates presence of other pathways p Proper Variable names n Variable names should be nouns or noun phrases. Action (verbs) are captured by the causal link n Choose variable names whose normal sense of direction is positive p Normal Sense of direction is positive p Make intermediate links explicit IEOR, IIT Bombay IE604: System Dynamics Modeling and Analysis Jayendran Venkateswaran

Guidelines for CLD (2) p Capture Causation and NOT correlation n Correlation do NOT represent structure of system. Correlation among variables reflect the past behavior à may change if circumstances change n Make sure the relationships are causal, no matter how strong the correlation is! p Make goals of negative loops explicit p Distinguish between actual and perceived conditions p Indicate important delays in causal links IEOR, IIT Bombay IE604: System Dynamics Modeling and Analysis Jayendran Venkateswaran

Justification of Causal Links p Conservation consideration n Laws of conservation p Accepted Theory p Instructions p Direct observation p Hypothesis or assumptions p Statistical evidence IEOR, IIT Bombay IE604: System Dynamics Modeling and Analysis Jayendran Venkateswaran

Drawbacks of CLD p Richardson, G. P., 1986, Problems with causal-loop diagrams, System Dynamics Review, 2(2), 158-170 n http://sysdyn.clexchange.org/sdep/Roadmaps/R M4/D-3312-2.pdf IEOR, IIT Bombay IE604: System Dynamics Modeling and Analysis Jayendran Venkateswaran

Example: Population, Economy & Land Use (1) p As employment opportunities increase in a city, people are attracted into the urban area. However, in-migration do not immediately react to opportunities. Since migrants react to perceived opportunities, there may be 5-10 year lag in response. p Population growth from the influx of migrants tends to encourage business expansion in the growing urban area. This business expansion creates demand for additional labor which increases employment opportunities. IEOR, IIT Bombay IE604: System Dynamics Modeling and Analysis Jayendran Venkateswaran

Example: Population, Economy & Land Use (2) p Population growth also tends to drive housing construction at a greater pace to match the population. Assuming only a fixed land is available for business and housing use, increasing housing stock makes less land available for business expansion. p As the unavailability of more land begin to suppress business expansion in the area, the demand for labor decreases. Consequently local employment opportunities decline. Once potential migrants perceive the lack of opportunities, declining in-migration generates a reduction in the population growth of the area. IEOR, IIT Bombay IE604: System Dynamics Modeling and Analysis Jayendran Venkateswaran

Example of CLD: Factory p In a factory, the customer order rates are usually accumulated into order backlog. p As backlog increases, the production orders to the shopfloor is increased. After some production delay these orders are converted into finished products and stored in inventory. With more inventory, orders will be filled at a higher rate and correspondingly backlogs will reduce. p With early filling of orders, the delivery delay will be reduced which improves customer satisfaction. Higher customer satisfaction leads to more orders from customers. p An increase in order backlog worsens the delivery schedule and has an adverse impact on customer satisfaction IEOR, IIT Bombay IE604: System Dynamics Modeling and Analysis Jayendran Venkateswaran

Case Study: Problem of Traffic Congestion p Trend in road traffic n Freight transport by road has risen from 6 Billion Tonne Km (BTK) in 1951 to 1100 BTK in 2000 and passenger traffic has risen from 23 Billion Passenger Km (BPK) to 2875 BPK during the same period. n The annual growth of road traffic is expected to be 10 to 11%. Current boom in the automobile sector may even increase the future growth rate of road traffic. p Growth in road network n NHs carry nearly 40% of road traffic n In NH & SH, only 2% of their length is four-lane, 34% two-lane and 64% single-lane Source: http://siadipp.nic.in/publicat/books/roads.pdf IEOR, IIT Bombay IE604: System Dynamics Modeling and Analysis Jayendran Venkateswaran

Model for Traffic Congestion p Open Loop View p Closed loop view n Travel Time n Pressure to build roads IEOR, IIT Bombay IE604: System Dynamics Modeling and Analysis Jayendran Venkateswaran

Capacity Expansion: Road Traffic Model Source: Sterman, John D. Business Dynamics (Fig 5-33) IEOR, IIT Bombay IE604: System Dynamics Modeling and Analysis Jayendran Venkateswaran

Traffic Volume dynamics: Road Traffic Model Source: Sterman, John D. Business Dynamics (Fig 5-34) IEOR, IIT Bombay IE604: System Dynamics Modeling and Analysis Jayendran Venkateswaran

Suburban Development: Road Traffic Model Source: Sterman, John D. Business Dynamics (Fig 5-35) IEOR, IIT Bombay IE604: System Dynamics Modeling and Analysis Jayendran Venkateswaran

Mass Transit Death Spiral: Road Traffic Model Source: Sterman, John D. Business Dynamics (Fig 5-36) IEOR, IIT Bombay IE604: System Dynamics Modeling and Analysis Jayendran Venkateswaran

You can ’ t get there on bus! Road Traffic Model Source: Sterman, John D. Business Dynamics (Fig 5-37) IEOR, IIT Bombay IE604: System Dynamics Modeling and Analysis Jayendran Venkateswaran

Model for Traffic Congestion p Open Loop View p Closed loop view n Travel Time n Pressure to build roads p Dynamics of Traffic volume p Growth of suburban regions p Mass Transit Death Spiral n You can ’ t get there on the bus! IEOR, IIT Bombay IE604: System Dynamics Modeling and Analysis Jayendran Venkateswaran