Simulation 


Befriend
Uncertainty 
'Measure what is measurable,
and make measurable what is not so.' Galilei,
Galileo (1564  1642)
'Why
speculate when you can calculate?' 
John Baez
Simulation
is 'modeling the uncertain' in a riskfree environment.
Simulation forms an essential numerical tool
 to solve a complex problem where an analytic solution is
not feasible
 when there are large numbers of interacting variables with
nonlinear relationships
 when choices are limited but decision variables have significant
uncertainty
Simulation
has been used extensively in business for:
 Queuing applications: Number of service counters required
for a desired service level at different times of the day/month for
given customer arrival rates. A few application
areas are Call Centers, ATMs, Banks,
Airline /Rail reservation counters, Retail billing counters.
 Supply chain: Number
and size of warehouses, stocking and ordering policy for given demand
and desired order fill rates; identifying information gaps and bottlenecks
in a supply chain so as to improve
response time.
 Scheduling: Sequencing and allocation of capacity for given
demand and operational constraints. Examples: Shop floor scheduling,
Public transport  Airline / rail / bus scheduling.
 Customer Behavior: Simulation of wordofmouth
diffusion of new product concept in a population to chart the
evolution of its demand.
 Competitor Behavior: Modeling competitor
moves resulting from entering
a new market, using game theory.
 Stock markets: Arriving at circuit filters, given desired
levels of volatility, by simulating trades.
Consider two
business examples where simulation can make a difference in decisionmaking:
1 
Nationwide New Product Launch: 

Managers rely heavily on experience to
make decisions for a new product launch. Success of such a launch
hinges on many variables, which include  perception of
product by target segment, reach of the promotional campaign, price
sensitivity and competitors' moves.
Simulation can model these uncertainties by simulating the entire
market place and help the manager make the decisions by generating
and evaluating alternative scenarios, and deriving the chances of
success. For example, Gametheoretic approaches can
be used to model competitor moves.

2 
Warehouse Capacity Planning: 

A firm has developed alternative sets of demand
forecasts for the next five years, each with a certain probability.
The decision to be taken is whether to increase the present warehousing
capacity in the light of the demand forecast and desired service
level. The alternatives could be physical capacity addition or deploying
an extra shift with different cost impacts.
The above situation of supply chain can be simulated with different
demand scenarios and alternatives for capacity increase to come
up with different service levels. Examining a variety of scenarios
leads to an informed decision on capacity planning. 
Development
of any simulation goes through four phases:
 Identifying goals: Defining the goals of simulation and decisions
to be addressed.
 Model Design: Understanding the problem domain, in terms
of relevant variables, their distributions and their interrelationships.
 Model Execution: Programming the design using statistical
techniques and algorithms.
 Model validation: Using statistical techniques and confidence
intervals to analyze execution.
Key characteristics
of a good simulation:
 HiFidelity
 should represent the actual situation closely
 Whatif analysis
 should generate all possible scenarios effortlessly
 Interactive Visualization 
derive insights about all possible scenarios
 Generation of and fitment to a range of probability
distributions for variables
 Model validation using advanced statistical
techniques
©
200010 DecisionCraft Inc.
