Introduction to Manufacturing
Systems Engineering: TIE5215
T R CHIKOWORE
Synopsis
Queuing Models and Theory
Systems Configuration and Descriptions
Analysis and Mathematical Modelling
Simulation and Modelling
Hand Simulation
Arena Simulation
Intelligent Manufacturing Systems
Modern Manufacturing Methods
Queuing Models and Theory
Waiting lines are non-value added occurrences
Waiting in lines does not add enjoyment
Waiting in lines does not generate revenue
Queuing theory: Mathematical approach to the
analysis of waiting lines.
Goal of queuing analysis is to minimize the sum of
two costs:
Customer waiting costs
Service capacity costs
Implications of Waiting Lines
Cost to provide waiting space
Loss of business
Customers leaving
Customers refusing to wait
Loss of goodwill
Reduction in customer satisfaction
Congestion may disrupt other business operations
Cost/Capacity Trade off Model
Examples of Real World Queuing Systems
Commercial Queuing Systems
Commercial organizations serving external customers
e.g. Dentist, bank, ATM, gas stations, plumber, garage
Transportation service systems
Vehicles are customers or servers
e.g. Vehicles waiting at toll stations and traffic lights, trucks or ships
waiting to be loaded, taxi cabs, fire engines, elevators, buses
Business-internal service systems
Customers receiving service are internal to the organization providing
the service
e.g. Inspection stations, conveyor belts, computer support
Social service systems
e.g. Judicial process, the ER at a hospital, waiting lists for organ
transplants or student dorm rooms
Components of a Basic Queuing Process
Components of a Basic Queuing Process
Components of a Basic Queuing Process
The calling population
The population from which customers/jobs originate
The size can be finite or infinite (the latter is most common)
Can be homogeneous (only one type of customers/ jobs) or
heterogeneous (several different kinds of customers/jobs)
The Arrival Process
Determines how, when and where customer/jobs arrive to the
system
Important characteristic is the customers’/jobs’ inter-arrival
times
To correctly specify the arrival process requires data collection of
inter-arrival times and statistical analysis.
Components of a Basic Queuing Process
The queue characteristics
Queue discipline, FCFS, FIFO, Priority rules
Their effect on customer behaviour (Balking, Reneging,
Jockeying)
Queue length
Their maximum size (# of jobs the queue can hold)
Patient: Balking: Reneging: Jockeying
Patient: The customer enters the waiting line and
remain until served
Balking: The customer decides not to enter the
waiting line.
Reneging: The customer enters the waiting line but
decides to leave before being served.
Jockeying: When the customer enters one waiting
line and then switches to a different one in an effort
to reduce the waiting time.
Components of a Basic Queuing Process
The System configuration
Channels and phases (Single or multiple)
Number of servers
Number of phases (stations)
Their location
Distinction between infinite and finite capacity
System Configurations
Multiple channel
Multiple phase
Channel: A server in a service system
Phase: A stage in the service system
System Configurations
Exercise 1
Draw and describe the following configurations:
Single Channel, Single Phase System
Multiple Channel, Single Phase System
Single Channel, Multiple Phase System
Multiple Channel, Multiple Phase System
Queuing Theory Basics: Conditions
Steady State condition
Enough time has passed for the system state to be independent of the
initial state as well as the elapsed time
The probability distribution of the state of the system remains the
same over time (is stationary).
Transient condition
Prevalent when a queuing system has recently begun operations
The state of the system is greatly affected by the initial state and by
the time elapsed since operations started
The probability distribution of the state of the system changes with
time
With few exceptions Queuing Theory has focused on
analyzing steady state behavior
Transient and Steady State Conditions
Queuing Theory Basics: Notation
Service times as well as interarrival times are assumed independent and identically
distributed if not otherwise specified
Commonly used notation principle: A/B/C
A = The interarrival time distribution
B = The service time distribution
C = The number of parallel servers
Commonly used distributions
M = Markovian (exponential) – Stochastic memoryless model describing a
sequence of possible events
D = Deterministic distribution
G = General distribution
Example: M/M/c
Queuing system with exponentially distributed service and inter-arrival times
and c servers
Queuing Theory: (M/M/1 System)
Queuing Theory: (M/M/1 System)
The M/M/1 queue assumes that arrivals are a
Poisson process and the service time is exponentially
distributed.
This is a Single Channel, Single Phase System
Interarrival times of a Poisson process are IID
(Independent and Identically Distributed)
exponential random variables with parameter
Service time is exponentially distributed with
parameter
Single Channel, Single Phase System
Arrivals
The Average Arrival Rate or Average Arrival
Intensity, , is the number of arrivals per given unit
time
Examples are: number of customers per hr, number
of orders per week, etc - independent from each other!
- each interarrival i follows
an exponential distribution
1 2
t
Service Process
The Average Service Rate or Average Service
Intensity, , is the number of arrivals serviced per
unit time
The assumption is that arrivals are served on a
FCFS basis and that arrivals are patient
Queuing Theory: Results
Pn = The probability that there are exactly n
customers/jobs in the system (in steady state, i.e.,
when t)
= Server utilisation
L = Expected number of customers in the system (in
steady state)
Lq = Expected number of customers in the queue (in
steady state)
W = Expected time a job spends in the system
(Throughput time)
Wq = Expected time a job spends in the queue
Queuing Theory: Results
W
Queuing Theory: M/M/1 and M/M/C Results
APPENDIX A – Table A: M/M/1 and M/M/C Queuing Results
M/M/1 M/M/C
L L s
Lq
2 (c ) c p(o)
Lq Lq ( ) c!(1 ) 2
Wq (c ) c p(o)
W q ( )
c!c (1 ) 2
W 1 Wq 1
ws
P(0) (c ) c c 1
(c ) n
1
Po 1
c! (1 ) n o n!
Example 1
Suppose a manufacturing facility operates as a flow
shop. Interarrival times are exponentially distributed
with an average arrival rate of 10 orders per week.
Orders are processed on a FCFS basis with a
constant production schedule from week to week.
Orders vary in complexity; but the system is capable
of processing 12 orders per week.
Find the average time from order arrival to
completion.
Solution to Example 1
This is an M/M/1 system, therefore:
Throughput time is 0.5 weeks
Find also the average waiting time, average number of
orders in queue, average number of orders in the
system, the probability of having no orders in the
system
Open Networks
A network has more than 1 workstations with jobs
mowing between workstations according to their
processing plans.
Material handling systems are responsible for
moving jobs from one workstation to another.
In an open network an external arrive process
generates jobs that arrive at one or more
workstations and enter the network.
Important Facts for Open Networks
FACT 1: Poisson Reproductive Property. The sum of
independent Poisson random variables is Poisson
FACT 2: If the number of arrivals per time is
Poisson distributed, then the time between arrivals
has an exponential distribution
FACT 3: The inter departure time from an (M/M/c)
system with infinite queue capacity is exponential
FACT 1: Illustration
= 1+ 2 + 3+ 4
Example 2
Six machines stamp part batches and send them to
one inspector. The inspector then sorts batches
sending 97 percent to packing and scrapping 3
percent. Parts are stamped in batches. Each machine
has its own queue and stamps about 10 batches per
day Find the arrival processes at inspection, packing,
and rework.
Solution to Example 2
The six machines merge their output into one stream for inspection.
Assuming that each machine has exponential interarrival and service
times, their output streams are Poisson by Fact 3. The Reproductive
Property then indicates that input to inspection is a Poisson arrival
process with rate = 60 batches per day. If inspection time is
exponential, then the output stream is Poisson. Since this output is
randomly divided into two streams, the input to Packing is Poisson
with rate p = (0.97)60 = 58.2 batches per day. Input to scrap is
Poisson with rate s = (0.03)60 = 1.8 batches per day.
Note that this "network" of workstations conserves; at each
workstation and for the network as a whole, all jobs that
enter eventually leave.
Solution to Example 2: Illustrated
Packing
1 =10
1 P
2 =10
2
3 =10 p =0.97(60)
3
4 =10 =60
4 Inspection
5 =10
5 s =0.03(60)
6 =10
6
S
SERVICE RATE
Scrap
Example 3
Parts are withdrawn from a warehouse and kitted into packets for filling
production orders. The kit may have enough parts for one finished item or
a batch of these items. The kit is sent first to a workstation for assembly.
Assembly combines parts from the kit to produce a batch of finished parts.
Part batches are then transported to an inspection and packing station.
There are always orders waiting to be kitted, in fact a dispatching system is
designed to keep one hour's worth of work in the input queue for kitters at
all times: Kitting turns out an average of 10 kits per hour. Service rates at
assembly (A) and-inspection/pack (I) are 12 and 15 per hour, respectively.
Kitting, assembly, and inspection/packing times are exponentially
distributed. This system is a serial production system with random
processing times.
Find the expected number of jobs in the kit, assembly, and
inspection/pack system along with the expected throughput time.
Each station has a single server.
Solution to Example 3
Step 1: Draw the configuration of the system
and label appropriately
WHSE
K A I/P
Kitting Assembly Inspection/Pack
= 10 = 12 ; c = 1 = 15 ; c = 1
Solution to Example 3
Step 2: Analyse each station independently
For the Assembly station = 10 ; = 12 ; c = 1
Solution to Example 3
For the Inspection station = 10 ; = 15 ; c = 1
Solution to Example 3
Step 3: Combine results across stations
Total number of jobs in the system:
Throughput Time:
Example 4
Suppose in the previous Example that each time jobs
are inspected they fail with probability = 0.1 and are
returned to assembly. A feedback loop now exists
from Inspection to Assembly. Rework averages the
same 5 minutes as the initial assembly process.
How is throughput time affected?
Solution to Example 4
Step 1: Draw the configuration of the system
and label appropriately
Rework at a probability of = 0.1
WHSE
K A I
Kitting Assembly Inspection/Pack
= 12 = 12 ; c = 1 = 15 ; c = 1
Solution to Example 4
Step 2: Determine effective Arrival rates for the
Assembly and Inspection station
For the Assembly station: Assembly receives arrivals
from both kitting and rework from inspection, therefore:
For the Inspection station: What arrives at Assembly,
arrives at Inspection station, therefore:
Solution to Example 4
By solving the two equations simultaneously, using
= 0.1 and = 10, we get:
Step 3: Analyse each station independently
Solution to Example 4
For the Assembly station = 11.1111 ; = 12 ; c = 1
Solution to Example 4
For the Inspection station = 11.1111 ; = 15 ; c = 1
Solution to Example 4
Step 3: Combine results across stations; Determine the
number of visits per station
W
Managerial Implications of Waiting Lines
Managers have a number of very good reasons to be
concerned with waiting lines. Chief among those
reasons are the following:
The cost to provide waiting space.
A possible loss of business should customers leave
the line before being served or refuse to wait at all.
A possible loss of goodwill.
A possible reduction in customer satisfaction.
The resulting congestion that may disrupt other
business operations and/or customers.
Mitigating Effects of Long Queues
1. Concealing the queue from arriving customers
Restaurants divert people to the bar, amusement parks require people
to buy tickets outside the park
2. Use the customer as a resource
Patient filling out medical history form while waiting for physician
3. Making the customer’s wait comfortable and distracting
their attention
Complementary drinks at restaurants, TVs, computer games, internet
stations, food courts, shops
4. Explain reason for the wait
5. Provide pessimistic estimates of the remaining wait time
Wait seems shorter if a time estimate is given.
6. Be fair and open about the queuing disciplines used
Maximum Queue Length
The approximate line length that will satisfy a
specified percentage can be determined by solving
the following equation for n:
Where
Maximum Queue Length
The resulting value of n will not usually be an
integer.
Generally, round up to the next integer and treat the
value as n. However, as a practical matter, if the
computed value of n is less than 0.10 above the next
lower integer, round down.
Thus, 15.2 would be rounded to 16, but 15.06 would
be rounded to 15
Example 5
Determine the maximum length of a waiting line for
specified probabilities of 95 % and 98 %, for a system
in which number of servers = 2, arrival rate = 8 per
hour, and service rate = 5 per hour
Example 5: Solution
