Transcript Discrete Event Process Models and Museum Curation

Discrete Event
Process Models
and
Museum Curation
Louis G. Zachos
Ann Molineux
Non-vertebrate Paleontology Laboratory
Texas Natural Science Center
The University of Texas at Austin
Discrete Event Simulation
• What is DES?
• Many processes can be represented as a
series of discrete events or activities.
Discrete Event Simulation
• Events occur at an instant in time, persist for
some period of time, and mark a change of
state in the process – they are the individual –
discrete - steps in the staircase of a process.
• DES is a computational (i.e., computer) model
of a system of real-life processes modeled as
multiple series of discrete events
Functionality of DES
Modeling Environment
• In practical terms, a DES is comprised of a
model and the environment in which it is
executed
• It is possible to design a DES as a single
computer program – but there is software to
create a modeling environment for a DES
DES Modeling Environment
Components
(House-Keeping Functions)
• Clock
• Random Number Generators for a Variety of
Probability Density Functions
• Statistics Collation and Graphing Capability
• Events, Resources, Stores Lists Handling
• Conditions and System State Handling
SimPy
Simulation in Python
• An Open Source object-oriented discreteevent simulation language based on
http://simpy.sourceforge.net/
•
“Many users claim that SimPy is one of the cleanest, easiest to use discrete event
simulation packages!” (from http://simpy.sourceforge.net/)
Process Object Model
• DES in SimPy is based on the definition of
Object Classes
• There are 3 classes:
• Process class – the object that “does
something”
• Resource class – objects required to “do
something”
• Monitor class – an object to record
information
Model Design
• A system can be decomposed in a top-down,
hierarchical manner
• Start with the most general
Model Design
• Break each process into sub-processes
Resources
• Resources are things like people, cameras,
computer workstations, etc. – required to
perform processing.
Stores
• The entities being processed – museum
specimens – are represented as stores
• Stores act like queuing bins -
NPL Model
• Photography of type specimens
• Scan labels
• Prepare and scan
• Photograph specimens
• Prepare and photograph
• Convert raw imagery
• Process multi-focus imagery with Helicon
• Cleanup and standardize imagery in
Photoshop
NPL Model
• Resources
• People
• Cameras
• Computer workstations
• Stores – fossil specimens and labels
• Simplest case – individual resources are alike
• Variability is modeled stochastically
Modeling Results
Can capture various
aspects of a process and
realistically model
throughput and variability
Modeling Results
Bottlenecks in the process become readily
apparent – in this example the process waits
on human resources – just adding another
camera would not improve throughput
Validation
• Model results must be
validated against
actual system
throughput
• Actual process is timed
and variability
modeled
Extrapolation
• Once a working model has been validated:
• Bottlenecks can be quantified
• The effects of varying resources or
changing order of processes can be
evaluated
• Reliable estimates of time to completion
for entire projects can be made
Conclusion
• Discrete event simulations can be a useful tool
for evaluating long-term projects in the
museum environment
• The methodology makes the results easier to
justify for budget or grant applications
• The development of a model aids in
understanding the underlying processes