Unformatted Attachment Preview
Letter to the Board
•
The letter to the board should be the first page.
•
The letter should be on white 8 ½” x 11” paper.
•
Use a twelve-point font (Times New Roman, Helvetica, Tahoma, Universe, or
comparable).
•
Do not use different fonts, italics, underscores, or bold print anywhere in the letter.
•
The letter should be in portrait style.
•
The letter should be short, no more than a few paragraphs, on one page.
•
The letter to the board should contain the answers to the question you were asked to
address.
•
The letter should not contain details on methodology or use any statistical language
(e.g. stable, control chart, Ishikawa diagram), tables, or charts.
•
Use direct language and active tense.
•
Check your letter for spelling and grammar.
•
Decide as a group what process you will study.
•
Decide on the location to which you will go. Pick a location that is easy for everyone
to get to.
Every observation must come from the same location, as each location is a
different process, even if it looks similar.
•
Decide on one measurable characteristic to study (e.g. time to buy a Metro ticket, time
to be served, time waiting in line).
•
Operationally define the beginning and end points of the process you will study.
•
Set up a plan for how you will position yourselves as you collect observations, who
will record data, who will take measurements, …
•
•
•
Definition of the process (5 points)
•
Plan for observations (5 points)
•
This is a service process. Our group has five people .We decided to go to the wendy’s to
start our process study from 3:15 pm -4:30 pm .We decide to measure how long it will
takes form they ordered at cashier desk until they get the foods.We all sit around the desk
and observe each person come into the store .When they goes to make an order we start
to measure the time ,each one choose one person as study model ,record and measure
their different models 12 times. Then we combinate all the datas to draw a flow chart of
the process. In order to keep our service study ‘s data
accurately ,during
our process ,we
only choose the person come alone , if someone come together ,we won’t take them as
our study model.
•
Although we choose one person guest as our model but still have many potential factors
make our data inaccurate .Like people can self-order on machine or by their mobile. That will increase
waiting time while other people ordered at cashier desk. Also, during we observe time , passenger flow is
also different .When we just arrive the store almost 20 people stand there and waiting for their food .30
minutes later only have 8-10 people are waiting for their food . Less work piled up in the kitchen ,less time
people waiting
Service Processes and
Variability
Service Process Design
Services cannot be stored in inventory
In services, capacity becomes the dominant issue
Too
much capacity leads to excessive costs
Insufficient capacity leads to lost customers
Analytical models and computer simulation provide
powerful tools for analyzing many common service
situations.
How
many tellers in a bank?
How many EZ pass lanes at Whitestone bridge?
How many checkout lanes at BJ’s?
7-2
Service-System Design Matrix
7-3
Service Blueprinting
The standard tool for service process design is the
flowchart
A unique feature of the service blueprint is the
distinction made between the high customer contact
aspects of the service and those activities that the
customer does not see
Called a service blueprint
Made with a “line of visibility” on the flowchart
Fail-safing involves using the service blueprint to
identify opportunities for failure and then establishing
procedures to prevent mistakes from becoming defects
(poka-yokes)
7-4
Poka-Yokes
Poka-yokes - procedures that block the inevitable mistake
from becoming a service defect (“avoid mistakes”)
Many applications of poka-yokes to services
Poka-yokes are common in factories
Warning methods (e.g. steps that lead to mistakes trigger a
reminder)
Physical or visual contact methods (e.g. parts can only fit together
in the correct way)
Examples of service poka-yokes:
Height bars at amusement parks
Take-a-number systems in DMV
“It is just a reminder”….
7-5
Variability
Understand the impacts of variability on system
performance
Game: simulate a service system
The Game
RM
1
WIP
2
WIP
Infinite Supply of RM Inv.
3
WIP
4
WIP
5
WIP
6
FG
Initial WIP = 4 Pennies
➢
Each workstations processes pennies.
➢
The processing capability of each workstation is determined by the
roll of a six sided die.
➢
The actual number processed by each workstation depends on its
processing capability and the number of pennies available for
processing
7- 7
Place Your Bet
______________________________________
40 45 50 55 60 65 70 75 80 85 90 95 100
(# of pennies processed in 20 rolls)
7- 8
Observations
1. Number of pennies processed after:
1st roll
10th roll
20th roll
2. Work Station with most WIP inventory after:
1st roll
10th roll
20th roll
3. Total amount of WIP in system after:
0th roll
1st roll
10th roll
20th roll
7- 9
Discussion # 1
Why does this system systematically not
achieve what the law of averages predicted?
Starving :A work station (resource) is not able to
take advantage of a high roll (available capacity)
because of lack of WIP.
7-10
Discussion # 2
Why is inventory accumulating in the system?
Variability in processing times and uncertainty in
demand have a profound effect on system
performance.
Was the number of pennies processed in the previous
class the same?
7-11
Discussion # 3
How could we increase the number of pennies
processed in 20 rolls?
Increase
WIP
Reduce variability
Reduce the # of workstations
Any act which reduces starving will increase
the number of pennies processed.
7-12
Simulation
•
Introduce Monte Carlo simulation to solve
some common waiting line problems and
estimate server utilization, the length of a
waiting line, and average customer wait time.
Waiting Line Problems (Queues)
A central problem in many service settings is the
management of waiting time
Reducing
waiting time costs money, but raises customer
satisfaction and throughput
When people waiting are employees, it is easy to
value their time
When people waiting are customers, it is more
difficult to value their time
Lost
sales is one value (often a low estimate)
7-14
Simulation
Sim ulation: the act of reproducing the
performance of a system.
Monte Carlo Sim ulation: the act of reproducing
the performance of a system that involves
random events.
Random events: events whose outcome is
unknown before the event occurs.
7-15
Monte Carlo Simulation Step-by-Step
Step 1:
Step 2:
Step 3:
Step 4:
Define the model.
Collect data on the random variables
Construct a Monte Carlo Table
Reproduce the performance of the
system
7-16
Motivating Example
Traffic planners are considering the addition of
either one or two toll booths on Tiffany Lane.
The decision of whether to build one or two
booths depends on the effects on average
customer wait time and the number of cars in
line.
7-17
Monte Carlo Simulation Step-by-Step
Step 1: Define the model.
What is the issue?
What are the random variables?
What are the criteria?
Step 2: Collect data on the random variables
Methods of data collection
1. Monitor current system
2. Find a similar situation
3. Expert Opinion
7-18
# of cars
Service
Time
# of
customers
1
50
1
10
2
32
2
20
3
8
4
10
3
70
Interarrival time
The inter-arrival tim e of a customer is how soon they
arrive after the previous customer.
The service tim e of a customer is how long they take to
be serviced through the toll booth.
Step 3: Construct a Monte Carlo Table
a.
Construct a probability distribution over the
possible outcomes of each random event
b.
Assign numbers that could be generated by the
random number generator to events in the same
proportion as the probability of occurrence of the
event .
Note: Cumulative distribution functions make it
easier to assign numbers to events.
7-20
Interarrival time
# of Cars
1
50
2
32
3
8
4
10
Service
Time
# of Cars
1
10
2
20
3
70
Probability
Cum.
Probability
#s assigned
to event
Probability
Cum.
Probability
#s assigned
to event
Step 4: Analyze the functioning of the system
Setup the chart
A.
a.
b.
In the first column, list what your simulating (jobs, cars,
days, breakdowns).
Then you’ll need two columns for each random variable.
In the 1st column you’ll put the random number which is
generated.
In the 2nd column you’ll put the value of the random
variable which the generated random number is
associated with on the Monte Carlo Table.
7-22
A. continued…….
c.
Add columns to evaluate the performance of the
system in terms of identified criteria.
Note: In simple queuing situations, to determine wait
time, resource idle time and # of items waiting, you’ll
need a column for each of the following:
the time something arrives,
the time service begins,
the time service ends.
7-23
B.
Simulate the system
a. Generate the needed random numbers.
b. Go to the appropriate Monte Table and determine the value
of the random variable.
c. Evaluate the performance of the system in terms of the
relevant criteria.
Note: It is common to have to add column after a first attempt
because you realize that other things have to be kept track of the
answer the question.
7-24
Simulation of One Toll-Booth Option
CAR
1
2
3
4
5
6
7
8
9
10
RN
Interarrival
Time
RN
Service
time
Arrival
Time
Time
Service
Begins
Time
Service
Ends
Wait
Time
Length
of
Line
Simulation of Two Toll Booth Option
Assumptions:
1. If both toll booths are empty, the car will use
toll booth 1
2. If toll booth 1 is occupied the car will go to
toll booth 2.
3. If both toll booths are occupied, the car
will go to the
first available
7-26
Simulation of Two Toll Booths
CAR
1
2
3
4
5
6
7
8
9
10
InterArrival
Time
Service
Time
Arrival
Time
Time
Service
Begins
Toll 1
Time
Service
Ends
Toll 1
Time
Service
Begins
Toll 2
Time
Service
Ends
Toll 2
Wait
Time
Length
Of
Line
Toll 1
Length
Of
Line
Toll 2
Common Queuing Situations
1.
2.
3.
Single-server single-line
Multiple server multiple line
Multiple-server single line
Note: Most processes can be modeled as a
sequence of queues.
7-28