0
1.9kviews
What is the purpose of model verification? What are the different ways available to verify a model?
1
7views

Model Verification

Verification is the process of building the model correctly and implementing it with good input and structure.

Figure above illustrates Model Building, Verification and Validation

After the phase of model building, the purpose is to ensure that the conceptual model is represented accurately in computerized representation or computer simulation.

Areas involved:

I) Program Testing

1. Verify initial condition

2. Independent tester for testing a model

3. Walk through or trace program

5. Debugging

6. Use graphical representation

7. Create internal model partitions

II) Simulation knowledge

• When to monitor and display information
• Attention to a particular block or entity
• Values of which components are required
• When to suspend or pause simulation
• Close examination of model o/p

Different ways to verify are:

A) Examination of model output for Reasonableness

• Test simulation model for reasonableness under extreme value conditions
• Test simulation model under variation in parameter
• Does response follow direction of the change?
• Examine inputs and outputs of subsystems
• compare test conditions to analytical results
• Check current contents and total counts

B) Current contents and total counts

• Current contents refer to no of transactions in each of the system at a given time.
• Total count refers to number of transactions that have entered each part of system at a given time.

If current contents in some part of s/m is high $\rightarrow$ Large no of transactions are delayed

If current contents increases in linear fashion $\rightarrow$ queue is unstable

Total count of subsystem is zero $\rightarrow$ No transactions entered the subsystem

Current count = total count - 1 $\rightarrow$ Entity captured resource but never freed it

C) Performance measure

The aim of simulation is to estimate long run measure of performance.

If a simulation is predicting one performance measure correctly, then there is an increase in confidence in the model's prediction ability for related performance measures.

D) Documentation

Model builder should

• Comment for operational model
• Definition for all variables and parameters
• Description of each major section of the operational model

It clarifies logic of a model.

E) Trace:

It is detailed computer printout of the stale of the simulation model over a period of time. It is design for use in simulation program to get value of selected variables each time the simulation clock is advanced selective trace includes.

• Trace set for specific location
• Trace set for particular entity
• Trace set for occurrence of a particular condition.