Vous êtes sur la page 1sur 38

Arena IŞIKIE

Worldview
process interaction
Entity, Attribute, Process, Activity
Modeling Components in Arena
Resources & Schedules
Animation
Multiple Runs

Sim.w.Arena.Chp7: Steady-State Statistical Analysis


warm-up, run-length, truncated replications, …
Steady State Simulations
transient state
steady state

Sim.w.Arena.Chp6: Terminating Simulations

Chp8 entity transfer


Chp9 further modeling issues

Optimization (parameter experiments, OptQuest, Process Analyzer)


IE303 System Simulation – Fall 2010 Arena 1
Arena IŞIKIE

Example: one operator call center


model+graphs+statistics
extensions: discrete interarrival and service times
multiple operators
multiple customer types
Example: Able-Baker call center
Example: Baker’s Bagel problem (Hwk#1)
Example: Newsvendor problem
Example: Inventory management with/without backlogs
Example: Input Analysis + Output Analysis

Example: Call Center (Chp5)

IE303 System Simulation – Fall 2010 Arena 2


Arena IŞIKIE

Lab#1:
Arena worldview
Parts of Arena window (attaching/detaching panels, RunSetup)
Entity, attribute, process, activity
Modeling Components in Arena
Modules (create, process, dispose)
Resources
Expressions
Running a simulation
Plots (Process.WIP, NQ(Queue), MR(Resource))
Run Setup (terminating endtime/conditional)
Reports
Little’s Law
Example: Checkout Counter (Single Channel Queue)

IE303 System Simulation – Fall 2010 Arena 3


Arena IŞIKIE

Lab#2:
Resources
Modules (assign, decide)
Entity Types
Attributes
Example: simple bank
Example: dump trucks

Lab#3:
Exercise: Microscope Exhibit

Lab#4:
Exercise: (s,S) Inventory

IE303 System Simulation – Fall 2010 Arena 4


Single Channel Queue IŞIKIE

IE303 System Simulation – Fall 2010 Arena 5


Single Channel Queue IŞIKIE

IE303 System Simulation – Fall 2010 Arena 6


Single Channel Queue IŞIKIE

IE303 System Simulation – Fall 2010 Arena 7


Single Channel Queue IŞIKIE

IE303 System Simulation – Fall 2010 Arena 8


Single Channel Queue IŞIKIE

IE303 System Simulation – Fall 2010 Arena 9


Single Channel Queue IŞIKIE

IE303 System Simulation – Fall 2010 Arena 10


Single Channel Queue IŞIKIE

IE303 System Simulation – Fall 2010 Arena 11


Simple Bank – 2 Tellers IŞIKIE

IE303 System Simulation – Fall 2010 Arena 12


Simple Bank – 2 Tellers IŞIKIE

IE303 System Simulation – Fall 2010 Arena 13


Simple Bank – 2 Tellers IŞIKIE

IE303 System Simulation – Fall 2010 Arena 14


Simple Bank – 2 Tellers IŞIKIE

Routing Policy 1: join the shortest queue

Routing Policy 2: join the designated queue

IE303 System Simulation – Fall 2010 Arena 15


Simple Bank – 2 Tellers IŞIKIE

Routing Policy 1: join the shortest queue

Routing Policy 2: join the designated queue

IE303 System Simulation – Fall 2010 Arena 16


Dump Trucks IŞIKIE

IE303 System Simulation – Fall 2010 Arena 17


Dump Trucks IŞIKIE

IE303 System Simulation – Fall 2010 Arena 18


Dump Trucks IŞIKIE

IE303 System Simulation – Fall 2010 Arena 19


Dump Trucks IŞIKIE

IE303 System Simulation – Fall 2010 Arena 20


Dump Trucks IŞIKIE

IE303 System Simulation – Fall 2010 Arena 21


Microscope Exhibit IŞIKIE

Exercise: Recall the microscope exhibit example discussed before. The viewing
times have a triangular distribution with a =1, b =3, andc =5. The arrival rate of the
visitors is 10 per hour with exponentially distributed interarrival times. The waiting
line discipline is FIFO and the ticket purchases are not modeled.
a) Build an Arena model when the view-again probability is 30%
and is independent of how many times the visitor has viewed the
exhibit already.
b) Build an Arena model when the number of views of a visitor has
the following distribution: p(1) = 60%, p(2) = 30%, p(3) = 10%.
c) Build the Arena model for the previous part but this time model
the ticket purchases where the visitors with tickets have priority
over visitors without tickets. Assume that visitors decide to buy
a ticket with 50% chance at the the of arrival only when there is at
least one visitor at the waiting line.

IE303 System Simulation – Fall 2010 Arena 22


Microscope Exhibit IŞIKIE

a)

IE303 System Simulation – Fall 2010 Arena 23


Microscope Exhibit IŞIKIE

a) (cont)

IE303 System Simulation – Fall 2010 Arena 24


Microscope Exhibit IŞIKIE

b)

IE303 System Simulation – Fall 2010 Arena 25


Microscope Exhibit IŞIKIE

b) (cont)

IE303 System Simulation – Fall 2010 Arena 26


Microscope Exhibit IŞIKIE

b) (cont)
There are two equivalent ways to
specify the decide module, as
depicted on the left top and
bottom:

Alternative 1: The condition can be


based on an attribute, NoV, as
depicted on the top, selecting the
Alternative 1 relationship (=,≥,≤,etc) and the
value (“1” in this case).

Alternative 2: The condition can


be also specified as an expression
depicted on the bottom
(“NoV==1”).

Alternative 2
IE303 System Simulation – Fall 2010 Arena 27
Microscope Exhibit IŞIKIE

c)

1 2

IE303 System Simulation – Fall 2010 2 Arena 28


Microscope Exhibit IŞIKIE

c) (cont)

IE303 System Simulation – Fall 2010 Arena 29


Microscope Exhibit IŞIKIE

c) (cont)

IE303 System Simulation – Fall 2010 Arena 30


Microscope Exhibit IŞIKIE

c) (cont)

IE303 System Simulation – Fall 2010 Arena 31


(s,S) Inventory System IŞIKIE

Exercise: Model 5-4 (s,S) Inventory Simulation (KSS p244)


Widgets By Bucky Co. carries inventory of only one kind of item, called a widget.
Initially there are 60 widgets on hand. Customer orders arrive with interarrival
times distributed as exponential with mean 0.1 day. Customers demand 1, 2, 3, or 4
widgets with respective probabilities 0.2, 0.3, 0.3, and 0.2. If a customer’s demand
can be met out of on-hand inventory, the customer gets the full demand and goes
away happy. But if the on-hand inventory is less than the customer’s demand, the
customer gets whatever is on-hand (which might be nothing), and the rest of the
demand is backlogged and the customer gets it later when inventory will have been
sufficiently replenished; this is kept track by allowing the inventory level to go
negative. Customers with backlogged items are infinitely patient and never cancel
their orders. If the inventory level is already negative and more customers arrive
with demands, it just goes more negative.
Ordering policy is (s = 20, S = 40) with a . Lead Time is uniformly distributed
between 1 and 4 days.
An alternative to continuous review is the periodic review (e.g. review the
inventory level once a day and decide whether to order or not).

IE303 System Simulation – Fall 2010 Arena 32


(s,S) Inventory System IŞIKIE

AvgCustomerDemand=2.5widget/customer
CustomerArrivalRate=10customer/day
Reordering Point = s
Order Upto Level = S

We define:
InvOnHand : widgets physically present in the warehouse
InvBackordered : widgets promised to customers but not delivered yet

InvLevel = InvOnHand – InvBackordered

InvOnOrder : widgets ordered with the supplier but not received yet

InvPosition = InvLevel + InvOnOrder

IE303 System Simulation – Fall 2010 Arena 33


(s,S) Inventory System IŞIKIE

Simulate the (s,S) system for 300 days and estimate the Total Cost per day. Total
Cost is the sum of three components: fixed ordering cost, holding cost, and
backordering cost. Fixed ordering cost is incurred every time Widgets By Bucky
places an order to its supplier. Fixed ordering cost (K=$32/order) does not depend
on how many widgets are ordered. Holding cost (h=$1/widget/day) is incurred per
widget onhand held per day. Backordering cost (b=$5/widget/day) is incurred,
similarly, per widget backordered per day. In order to find calculate Total Cost per
day we need to calculate the Total Cost over time [0,T] where T is the time
simulation ends.

Total Fixed Order Holding Backordering


Cost = Cost + Cost + Cost
over [0, T] over [0, T] over [0, T] over [0, T]

Total K Number of h Area under b Area under


Cost = ⋅ Orders Placed + ⋅ InvOnHand + ⋅ InvBackordered
per day T over [0, T] T over [0, T] T over [0, T]

Total Number of Avg Avg


Cost = K ⋅ Orders Placed + h ⋅ InvOnHand + b ⋅ InvBackordered
per day per day over [0, T] over [0, T]
IE303 System Simulation – Fall 2010 Arena 34
(s,S) Inventory System IŞIKIE

IE303 System Simulation – Fall 2010 Arena 35


(s,S) Inventory System IŞIKIE

CREATE “Customers”
Entity Type: customer
Time Between Arivals Type: Random (Expo) Value: 0.1 Units: Days
Entities per Arrival: 1 Max Arrivals: Infinite First Creation: 0.0
ASSIGN “Widgets Ordered”
Assignment1: Attribute, ordersize, DISC(0.2, 1, 0.5, 2, 0.8, 3, 1.0, 4)
ASSIGN “Update Inventory Level”
Assignment1: Variable, InvLevel, InvLevel – ordersize
ASSIGN “Order Widgets from Supplier”
Assignment1: Entity Type, widgetorder
Assignment2: Attribute, widgetno,
OrderUptoLevel – (InvLevel + InvOnOrder)
Assignment3: Variable, InvOnOrder, InvOnOrder + widgetno
Assignment4: Attribute, leadtime, UNIF(1,4)
SEPARATE “Separate 1”
Type: Duplicate Original Percent Cost: 100 #Duplicates: 1
DECIDE “Order Decision”
Type: 2-way by Condition If: Expression
Value: InvLevel + InvOnOrder <= ReorderPoint
IE303 System Simulation – Fall 2010 Arena 36
(s,S) Inventory System IŞIKIE

PROCESS “Lead Time Delay”


Type: Standard Logic Action: Delay
Delay Type: Expression Units: Days Allocation: Other
Expression: leadtime

ASSIGN “Update Inventory Level 2”


Assignment1: Variable, InvLevel, InvLevel + widgetno
Assignment2: Variable, InvOnOrder, InvOnOrder – widgetno
Variables
InvLevel (initialvalue= 60) ReportStatistics: yes
OrderUptoLevel (initialvalue=140)
ReorderPoint (initialvalue=40)
InvOnOrder (initialvalue=0) ReportStatistics: yes
Plots
Plot1:Expressions: InvLevel, ReorderPoint, OrderUptoLevel, 0
TimeRange: 50 InitMin: –100 InitMax: 200
Plot2: Expressions: Lead Time Delay.WIP, 0
TimeRange: 50 InitMin: 1 InitMax: 3
Plot3: Expressions: DAVG(InvLevel Value) #HistPoints: 1000
TimeRange: 50 InitMin: 0 InitMax: 150

IE303 System Simulation – Fall 2010 Arena 37


(s,S) Inventory System IŞIKIE

Statistics
Name Type Expression ReportLabel
InvOnHand TimePers.
MAX(InvLevel,0) InvOnHand
InvBackord TimePers.
MAX(–InvLevel,0) InvBackord
OrdersPlaced Output EntitiesIn(widgetorder)/TFIN OrdersPlaced
TotalCost Output 1*DAVG(InvOnHand)+ TotalCost
5*DAVG(InvBackord)+
32*EntitiesIn(widgetorder)/TFIN
Reports – UserSpecified – TimePersistent
Reports – UserSpecified – Output

Run Setup
NumberofReplications: 1
Warmup Period: 0.0 Time Units: Days
Replication Length: 300 Time Units: Days
HoursPerDay: 24 BaseTimeUnit: Days
TerminatingCondition: “”

IE303 System Simulation – Fall 2010 Arena 38

Vous aimerez peut-être aussi