lecture 5 : systems models and diagrams

1

• lecture 5 systems models and diagrams
• this lecture is about modelling
• modelling
• influence diagrams
• flow charts, precedence charts, spray diagrams
  etc.

2

• a model is
• a representation of reality intended for some
  purpose
• a representation of reality intended to be of use
  to someone charged with managing or understanding
  that reality
• a representation of reality intended to be of use
  to someone in understanding, changing, managing
  and controlling that reality
• a representation of part of reality as seen by
  the people who wish to use it to understand, to
  change, to manage and to control that part of
  reality
• an external and explicit representation of part
  of reality as seen by the people who wish to use
  it to understand, to change, to manage and to
  control that part of reality
• a model is a tool for thinking

3

• a system model specifies
• the transformation processes or activities of the
  system
• the boundary, ie. the narrow and the wider
  systems of interest
• subsystems of the narrow system involved in
  transformation the dynamic relationships ie.
  processes stable relationships ie. structure
• uncontrollable inputs control inputs decisions
  and decision rules
• outputs that are desired, undesired, planned,
  unplanned
• outputs serving as performance measures

4
question is aspect affected by system variables or inputs? is aspect affected by system variables or inputs? is aspect affected by system variables or inputs?
does aspect affect system? answer YES NO
does aspect affect system? YES component or relationship input
does aspect affect system? NO output irrelevant
5

• models should be
• simple but complete, not trivial or irrelevant
• simple models are easier to understand, to
  evaluate, to work with and to communicate
• they can be explained to problem users and
  owners this helps to convince them to use models
  as tools for thinking (for example spreadsheets
  are fine)
• even if users cannot comprehend its inner working
  properly, a model should be easy to manipulate
  and it should be tested by owners and users
• simplicity in doing something is parsimony
  (Occams razor the fewest possible assumptions
  should be made in explaining a thing ) model
  should observe parsimony

6

• start small, then add and refine
• start with a small model even if it is not
  sufficiently realistic
• make sure you immediately test it numerically if
  data is not yet ready, use approximate data and
  guesstimates this will show if the model makes
  sense see if it helps you to understand the
  problem situation
• as you discover the shortcomings of your model,
  refine it
• either by enriching, ie. adding new features,
• or by reformulating, ie. replacing with a
  different, new model

7

• a good model should be
• adaptive and robust so that it can be modified
  when data or the problem situation changes
• appropriate to the problem situation, eg.
  practicable to run within the timeframe needed
• relevant and approprite for decision making eg.
  without extensive need to process model outputs
• aided by diagrams such as influence diagrams
• used to go back to rich pictures and mind maps
  for revision of both these and the model etc.

8

• influence diagrams

output
control i/p, decision rules, etc
parameters, constraints, etc
influence
A
B
system variables rates, levels etc
rate flow (a process element) level stock (a
structural element)
9
stocks and flows or levels and rates
10

• flow charts, spray diagrams etc.
• flow charts are used to show logical or temporal
  flows such as
• material flows
• information flows
• decision flows etc.
• spray diagrams are useful to map detailed
  cause-and-effect relationships
• precedence charts are useful in project management

11
(No Transcript)
12
(No Transcript)
13
(No Transcript)