Chapter 14 - Operations Scheduling and Sequencing
Chapter 14 - Operations Scheduling and Sequencing
Friendly Version
User Name: Trong Nghia Ho
email Id: [email protected]
Book: OM
© 2007 Cengage Learning Inc. All rights reserved. No part of this work may by reproduced or used in any form or by any
means graphic, electronic, or mechanical, or in any other manner without the written permission of the copyright holder.
Chapter 14 : Operations Scheduling and Sequencing
(pp. 3060)
Operations Scheduling and Sequencing: Chapter Objectives
Learning Outcomes
After studying this chapter you should be able to:
141 Explain the concepts of scheduling and sequencing.
142 Describe staff scheduling and appointment system decisions.
143 Explain sequencing performance criteria and rules.
144 Describe how to solve singleand tworesource sequencing problems.
145 Explain the need for monitoring schedules using Gantt charts.
What Do You Think?
As a student, how do you schedule your homework, school projects, and study activities? What criteria do you use?
Jose Girarte/iStockphoto.com
Operations Scheduling and Sequencing: Chapter Overview
Jean Rowecamp, clinical coordinator of nursing services, was faced with a deluge of complaints by her nursing staff about their
work schedules and complaints by floor supervisors about inadequate staffing. The nurses complained they were having too many
shift changes each month. Supervisors said they had too many nurses during the days and not enough at night and on the
weekends. It seemed that nothing she did would satisfy everyone. The nurses were unionized, so she couldn't schedule them for
more than seven consecutive working days and the nurses required at least 16 hours between shift changes. Nurses were
constantly making “special requests” for personal time off, despite the negotiated procedures for bidding for shifts and vacation
times. Jean lamented that she became an administrator and longed for the days before she had these responsibilities.
P. 306
Creating schedules is not easy. The chapteropening nursing example highlights the complexity of scheduling. For example,
union workforce rules and special requests can complicate the scheduling process. Nevertheless, good schedules have to be
developed to provide high levels of patient care and to minimize costs.
Scheduling and sequencing are some of the more common activities that operations managers perform every day in every business.
They are fundamental to all three levels of aggregation and disaggregation planning that we described in the previous chapter.
Good schedules and sequences lead to efficient execution of manufacturing and service plans and better customer service. It's one
thing to promise great customer service and another to actually achieve it. For example, having
It's one thing to promise great customer service and another to actually achieve it.
the right number of employees at a call center at different times of the day and week will ensure that customers do not have to wait
long. Good scheduling of jobs in a factory will ensure that customers receive their orders as promised and increase productivity
and efficiency.
141 Understanding Scheduling and Sequencing
Scheduling refers to the assignment of start and completion times to particular jobs, people, or equipment. For
example, fastfood restaurants, hospitals, and call centers need to schedule employees for
Scheduling is complicated by union regulations and “special requests” for personal time off. Javier Larrea/age
fotostock/Photolibrary
P. 307
work shifts; doctors, dentists, and stockbrokers need to schedule patients and customers; airlines must schedule crews and flight
attendants; sports organizations must schedule teams and officials; court systems must schedule hearings and trials; factory
managers need to schedule jobs on machines and preventive maintenance work; and salespersons need to schedule customer
deliveries and visits to potential customers. Many schedules are repeatable over the long term, such as those for retail store staff
and assemblyline employees. Others might change on a monthly, weekly, or even daily basis, as might be the case with call
center employees, nurses, or salespeople.
A concept related to scheduling is sequencing. Sequencing refers to determining the order in which jobs or tasks are processed.
For example, triage nurses must decide on the order in which emergency patients are treated; housekeepers in hotels must
sequence the order of rooms to clean; operations managers who run an automobile assembly line must determine the sequence by
which different models are produced; and airport managers must sequence outgoing flights on runways. Note that in all these
situations, processing takes place using a common resource with limited capacity. Thus, the sequence will ultimately determine
how well the resource is used to achieve some objective, such as meeting demand or customer due dates. Generally, a sequence
specifies a schedule, and we will see this in various examples later in this chapter.
Scheduling and sequencing in backoffice or lowcontact service processes are similar to that for goodsproducing processes. The
same scheduling and sequencing concepts and methods used in manufacturing are beneficial in lowcontact service processes.
It is not uncommon for a manufacturing facility to have hundreds of work stations or machine centers and to process thousands of
different parts. Managers of such facilities also need daily or even hourly updates on the status of production to meet the
information needs of supply chain managers, sales and marketing personnel, and customers. Similarly, service managers often
manage dozens of parttime workers with varying work availability times (think of a fastfood restaurant manager near a college
campus), or everchanging workloads and demands (think of a hospital nurse administrator). The complexity of these situations
dictates that effective scheduling systems be computerized, not only for generating schedules but also for retrieving information
so that
Telling Umpires Where to Go
One of the authors of this book developed annual schedules for umpires in the American Baseball League for many years before
this activity was merged with the National League. Some of the critical factors in developing these schedules were to ensure that
umpire crews were not assigned to consecutive series with the same team if possible; that the number of times a crew was assigned
to a team was balanced over the course of the season; that travel sequences be rational and realistic; and that a variety of
constraints be met. For instance, it makes more sense to schedule a crew to several consecutive series out on the East Coast or West
Coast and move the crew to nearby cities rather than shuttling back and forth across the country.
Various constraints limited the scheduling possibilities. For example, one could not schedule a crew for a day game in another
city after working a night game on the previous day. In addition, crews need time to rest and travel between game assignments.
All these factors needed to be considered in the context of the game schedule, which was created well in advance.
© Scott Boehm/Getty Images
a salesperson can check the status of a customer's order or project a delivery date. Thus, implementing scheduling systems requires
good information technology support.
Scheduling applies to all aspects of the value chain—planning and releasing orders in a factory, determining work
shifts for employees, and making deliveries to customers.
P. 308
UPS: Sustainable Scheduling
Bob Stoffel has the longest title at UPS: Senior Vice President, Supply Chain, Strategy, Engineering, and Sustainability. The logic
of the job title is that these functions fit together in a global economy where supply chains may be changing radically and UPS's
customers must increasingly demonstrate sustainability to their own customers and investors. With 100,000 trucks, he is
experimenting with virtually every green technology and reducing the company's carbon emissions by millions of tons annually.
In addition to more efficient vehicles, better scheduling using GPS technology and telematics resulted in a reduction of 53,000
miles each day while delivering 350,000 more packages each day. Telematics involves sensors that monitor 200 different vehicle
activities, such as braking, speeds, and even whether GPS coordinates match package addresses so that routes can be better
planned and scheduled. UPS also developed software to provide routes that rely on right turns whenever possible to reduce fuel
consumption as drivers wait to make a left turn.1
code6d/iStockphoto.com
142 Scheduling Applications and Approaches
Scheduling applies to all aspects of the value chain—planning and releasing orders in a factory, determining work shifts
for employees, and making deliveries to customers. Many problems, such as staff scheduling, are similar across different
organizations. Quite often, however (as with the baseball umpiring situation or scheduling classrooms and teachers at a
university), unique situational factors require a unique solution approach.
Many organizations use spreadsheets, desktop software packages, or webbased tools for scheduling. Customized scheduling
spreadsheets use readily available software such as Microsoft Excel, but may be rather expensive to develop. Commercial
spreadsheet templates are simple to use and are generally inexpensive. One example is Simple Scheduler, which can be purchased
from schedule.com for an annual fee of $39. Shiftschedules.com provides more sophisticated capabilities in its Template
Scheduler, which starts at $79 per year.
Desktop and webbased software packages are more powerful and have optimization capabilities. Typical features include the
ability to schedule days on and days off for both fulland parttime employees, create alternative schedules to compare costs, and
handling multiple shifts. One product is SNAP Schedule, a $450 package available from www.bmscentral.com . Webbased
scheduling systems have the advantage of being accessible anywhere but often cost more. Two popular systems are When To
Work (www.whentowork.com ) and Kronos (www.kronos.com ).
Many software packages have been developed for specific industries, such as call centers, law enforcement, and health care. Each
industry has unique problem structures and decision variables. Concerro (formerly known as BidShift), for example, is a web
based software package specifically designed for scheduling nurses. Over 170 hospitals have signed up to use this system; they
estimate savings of $1 to $4 million per year, along with improved employee morale and quality of patient care.2 In this section
we present two common applications of scheduling that are prevalent in operations management.
142a Staff Scheduling
Staff scheduling problems are prevalent in service organizations because of high variability in customer demand. Examples
include scheduling call center representatives, hotel housekeepers, tollbooth operators,
P. 309
nurses, airline reservation clerks, police officers, fastfood restaurant employees, and many others.
Staff scheduling attempts to match available personnel with the needs of the organization by
1. accurately forecasting demand and translating it into the quantity and timing of work to be done;
2. determining the staffing required to perform the work by time period;
3. determining the personnel available and the full and parttime mix; and
4. matching capacity to demand requirements, and developing a work schedule that maximizes service and minimizes costs.
The first step requires converting demand to a capacity measure—that is, the number of staff required. For instance, we might
determine that for every $400 of sales forecast, we need one additional fulltime employee. The second step determines the
quantity and timing of the work to be done in detail, usually by hour of the day, and sometimes in 5to 10minute time intervals.
Determining the staffing required must take into account worker productivity factors, personal allowances, sickness, vacations,
noshows, and so on.
Step 4 focuses on the matching of capacity to demand requirements; this is the essence of scheduling. Different approaches are
required for different situations because of the nature of constraints. If service demands are relatively level over time, as in the case
of hotel housekeepers, it is usually easy to schedule personnel on standard weekly work shifts. If the workload varies greatly
within a shift, as is the case for telephone customer service representatives, the problem becomes one of scheduling shifts to meet
the varying demand. Let us examine a relatively simple problem of scheduling personnel with consecutive days off in the face of
fluctuating requirements.3
T.R. Accounting Service is developing a workforce schedule for three weeks from now and has forecasted demand and translated
it into the following minimum personnel requirements for the week:
The staff requirements are for fulltime accountants who do accounting work such as endofmonth financial statements, tax record
organization, and federal, state, and local tax payments. T.R., the owner of the accounting service, wants to schedule the
employees so that each employee has two consecutive days off and all demand requirements are met.
The staffing procedure is as follows. First, we locate the set of at least two consecutive days with the smallest requirements. That
is, we find the day with the smallest staff requirements, the nextsmallest, and so on, until there are at least two consecutive days.
Sunday and Saturday, for example, have requirements of 3 and 5, respectively, while all others are greater than 5. We then circle
the requirements for those two consecutive days. Thus we have the following, for employee 1:
We assign accountant 1 to work on all days that are not circled, that is, Monday through Friday. Then we subtract 1 from the
requirement for each day that accountant will work. This gives us the following requirements that remain:
The procedure is repeated with this new set of requirements for accountant 2.
When there are several alternatives, as in this case, we do one of two things. First, we try to choose a pair of days with the lowest
total requirement. If there are still ties, we are to choose the first available pair that makes the most sense to the scheduler. Hence,
we again use Saturday and Sunday as days off for accountant 2, since this pair has the smallest total requirement of 8. We subtract
1 from each working day's requirement, yielding the following:
Circling the smallest requirements until we obtain at least two consecutive days again yields the following for employee 3:
Notice that Sunday is not adjacent to Tuesday, Wednesday, or Thursday, so we cannot use Sunday in the schedule. Remember we
are looking for consecutive pairs of days. Let's choose the TuesdayWednesday pair. The remaining requirements are:
P. 310
Continuing with this procedure, we obtain the sequence of requirements shown in Exhibit 14.1 (with circled numbers
Continuing with this procedure, we obtain the sequence of requirements shown in Exhibit 14.1 (with circled numbers
representing the lowestrequirement pair selected). The final accountant schedule is shown in Exhibit 14.2. Even though some
requirements are exceeded, such as Thursday with a demand for six accountants yet we schedule eight, the solution minimizes the
number of employees required. A more difficult problem that we do not address is that of determining a schedule of rotating shifts
so that employees do not always have the same two days off. Over a predetermined longer cycle such as a quarter, all employees
rotate through all possible days off. This makes for a fair and more equitable staff schedule, but it is complicated and beyond the
scope of this book.
Many software packages are available to help with staff scheduling. However, scheduling is so integrated with the practices and
culture of the organization that these standardized software packages normally need to be modified to work well in specific
operating environments. Accurate input data and the user's understanding of how the software techniques develop the schedules
are other challenges when adopting offtheshelf scheduling software.
Exhibit 14.1 Scheduling Procedure for T.R. Accounting Service © Cengage Learning 2013
Exhibit 14.2 Final Accountant Schedule
Employee Number Mon. Tue. Wed. Thur. Fri. Sat. Sun.
1 X X X X X
2 X X X X X
3 X X X X X
4 X X X X X
5 X X X X X
6 X X X X X
7 X X X X X
8 X X X X X
9 X X X X X
10 X X X X X
Total 8 6 6 8 10 6 6
Exhibit 14.2 Final Accountant Schedule © Cengage Learning 2013
P. 311
Using Spreadsheet Models to Schedule Medical Residents
The chief radiology resident from the University of Vermont's College of Medicine and a group of business students who were
enrolled in a semesterlong Master of Science course developed a spreadsheet model for creating a oneyear schedule for a group
of 15 medical residents in the radiology department. Demand for radiology services could occur at any time of the day; thus, they
must ensure that adequate personnel are always available. The majority of demand occurs during the weekday working hours;
approximately 20 to 25 certified radiologists and the residents serve this demand. Certified radiologists do not work on
weeknights; however, a radiology resident is on call and is responsible for all radiology services at the hospital. The resident
works the overnight shift alone and must be physically at the hospital, but is off during the next day shift. The radiology chief
resident has the task of assigning each resident to these oncall and emergency room (ER) shifts. This scheduling task—to assign
one resident to each oncall and ER assignment every day of the year—is incredibly complex because of many requirements and
restrictions, such as having one and only one resident assigned to the ER schedule per fourweek rotation block. The student team
used Microsoft Excel to create a spreadsheet that would allow them to measure the key metrics of the residents’ assignments, such
as the number of days worked in each category (e.g., Thursdays, Fridays, etc.). Then, they added an optimization model to find a
feasible solution. The model was simple to use and increased the quality of the resulting resident assignments, and decreased the
time required to obtain the assignments.4
John Wood Photography/Stock Image/Getty Images
142b Appointment Systems
Appointments can be viewed as a reservation of service time and capacity. Using appointments provides a means to maximize the
use of timedependent service capacity and reduce the risk of noshows. Appointment systems are used in many businesses, such
as consulting, tax preparation, music instruction, and medical, dental, and veterinarian practices. Indirectly, appointments reduce
the cost of providing the service because the service provider is idle less each workday. An appointment system must try to
accommodate customers and forecast their behavior, such as the noshow rate or a difficult customer who demands more
processing time.
Four decisions to make regarding designing an appointment system are the following:
1. Determine the appointment time interval, such as 1 hour or 15 minutes. Some professional services such as dentists and
physicians use smaller appointment intervals and then take multiples of it, depending on the type of procedure thought to
be required by the patient.
2. Based on an analysis of each day's customer mix, determine the length of each workday and the time off duty. Once the on
and offduty days for the year (annual capacity) are determined and assuming a certain customer mix and overbooking rate
(see step 3), the service provider can forecast expected total revenues for the year.
3. Decide how to handle overbooking for each day of the week. Often, customers do not show up as scheduled. If the no
show percentage is low, say 2 percent, then there may be no need to overbook. However, once the noshow percentage
reaches 10 percent or more, overbooking is usually necessary to maximize revenue and make effective use of perishable
and expensive time.
4. Develop customer appointment rules that maximize customer satisfaction. For example, some service providers leave one
appointment interval open at the end of each workday. Others schedule a 60minute lunch interval but can squeeze in a
customer during lunch if necessary. Telephone and electronic appointment reminders are another way to help maximize
serviceprovider utilization.
143 Sequencing
Sequencing is necessary when several activities (manufacturing goods, servicing customers, delivering packages, and so
on) use a common resource. The resource
P. 312
might be a machine, a customer service representative, or a delivery truck. Sequencing can be planned, in which case it creates a
schedule. For example, if a student plans to begin homework at 7:00 p.m. and estimates that it will take 60 minutes to complete an
OM assignment, 45 minutes to read a psychology chapter, and 40 minutes to do statistics homework, then sequencing the work
from most favorite to least favorite—OM, psychology, and statistics—creates the schedule:
143a Sequencing Performance Criteria
In selecting a specific scheduling or sequencing rule, a manager must first consider the criteria on which to evaluate schedules.
These criteria are often classified into three categories:
1. processfocused performance criteria,
1. processfocused performance criteria,
2. customerfocused due date criteria, and
3. costbased criteria.
The applicability of the various criteria depends on the availability of data. Later we will show how these performance measures
are applied to various sequencing rules.
Processfocused performance criteria pertain only to information about the start and end times of jobs and focus on shop
performance such as equipment utilization and workinprocess (WIP) inventory. Two common measures are flow time and
makespan. Flow time is the amount of time a job spent in the shop or factory. Low flow times reduce WIP inventory. Flow time
is computed using Equation 14.1.
Fi = Σpij + Σwij= Ci −Ri [14.1]
where
Fi = flow time of job i
Σpij = sum of all processing times of job i at work station or area j (run + setup times)
Software to Schedule Anywhere
One provider of small business software offers an online employee scheduling system called ScheduleAnywhere
(ScheduleAnywhere.com ). This service allows managers to schedule employees from any computer with Internet access, whether
at work, at home, or on the road. “With over 70,000 users, we get a lot of feedback on what people really need in an employee
scheduling system,” said Jon Forknell, vice president and general manager of Atlas Business Solutions. “Many of our customers
told us they needed an online solution that was affordable and easy to use.” Small firms can use the online scheduling service for
as little as $20 per month, while large organizations such as AT&T, Amazon.com , Bank of America, and the American Red Cross
sign longterm contracts. ScheduleAnywhere gives users the power to do the following:
schedule employees from any computer with Internet access
create schedules by position, department, location, and so on
view schedule information in a 1day, 7day, 14day, or 28day format
enter staffing requirements and view shift coverage
see who's scheduled and who's available
automatically rotate or copy employee schedules
preschedule timeoff requests
avoid scheduling conflicts
give employees read/write or readonly access to schedules5
Atlas Business Solutions, Inc.
P. 313
Σwij = sum of all waiting times of job i at work station or area j
C = completion time of job i
Ci = completion time of job i
Ri= ready time for job i where all materials, specifications, and so on are available
Makespan is the time needed to process a given set of jobs. A short makespan aims to achieve high equipment utilization and
resources by getting all jobs out of the shop quickly. Makespan is computed using Equation 14.2.
M = C − S [14.2]
where
M = makespan of a group of jobs
C = completion time of last job in the group
S = start time of first job in the group
Duedate criteria pertain to customers’ required due dates or internally determined shipping dates. Common performance measures
are lateness and tardiness, or the number of jobs tardy or late. Lateness is the difference between the completion time and the due
date (either positive or negative). Tardiness is the amount of time by which the completion time exceeds the due date. (Tardiness
is defined as zero if the job is completed before the due date, and therefore no credit is given for completing a job early.) In
contrast to processfocused performance criteria, these measures focus externally on customer satisfaction and service. They are
calculated using Equations 14.3 and 14.4.
Li = Ci − Di [14.3]
Ti = Max(0, Li) [14.4]
where
Li = lateness of job i
Di = due date of job i
Ti = tardiness of job i
A third type of performance criteria is costbased. Typical cost includes inventory, changeover or setup, processing or run, and
material handling costs. This costbased category might seem to be the most obvious criteria, but it is often difficult to identify the
relevant cost categories, obtain accurate estimates of their values, and allocate costs to manufactured parts or services correctly. In
most cases, costs are considered implicitly in process performance and duedate criteria.
143b Sequencing Rules
Two of the most popular sequencing rules for prioritizing jobs are
shortest processing time (SPT), and
earliest due date (EDD).
In using one of these rules, a manager would compute the measure for all competing jobs and select them in the sequence
according to the criterion. For example, suppose that the student we discussed earlier sequenced the homework according to SPT.
The sequence would be statistics, psychology, and OM. These rules are often applied when a fixed set of jobs needs to be
sequenced at one point in time.
In other situations, new jobs arrive in an intermittent fashion, resulting in a constantly changing mix of jobs needing to be
sequenced. In this case, we assign priorities to whatever jobs are available at a specific time and then update the priorities when
new jobs arrive. Some examples of these priority rules are
first come, first served (FCFS),
fewest number of operations remaining (FNO),
least work remaining (LWR)—sum of all processing times for operations not yet performed, and
least amount of work at the next process queue (LWNQ)—amount of work awaiting the next process in a job's sequence.
The SPT and EDD rules generally work well in the short term, but in most situations new orders and jobs arrive intermittently and
the schedule must accommodate them. If SPT were used in a dynamic environment, a job with a large processing time might never
get processed. In this case, some timebased exception rule (such as “if a job waits more than 40 hours, schedule it next”) must be
used to avoid this problem.
Different sequencing rules lead to very different results and performance. The SPT rule tends to minimize average flow time and
WIP inventory and maximize resource utilization. The EDD rule minimizes the maximum of jobs past due but doesn't perform
well on average flow time, WIP inventory, or resource utilization. The FCFS rule is used in many servicedelivery systems and
does not consider any job or customer criterion. FCFS focuses only on the time of arrival for the customer or job. The FNO rule
does not consider the
P. 314
length of time for each operation; for example, a job may have many small operations and be scheduled last. Generally, this is not
a very good rule. The LWNQ rule tries to keep downstream work stations and associated resources busy.
144 Applications of Sequencing Rules
Sequencing in a job shop, in which several different goods or services are processed, each of which may have a unique
routing among process stages, is generally very complex, but some special cases lend themselves to simple solutions. These
special cases provide understanding and insight into more complicated scheduling problems. One that we illustrate is scheduling
on a single work station or processor.
144a SingleResource Sequencing Problem
The simplest sequencing problem is that of processing a set of jobs on a single processor. This situation occurs in many firms. For
example, in a serial manufacturing process, a bottleneck work station controls the output of the entire process. Thus, it is critical to
schedule the bottleneck equipment efficiently. In other cases, such as in a chemical plant, the entire plant may be viewed as a
single processor. Single processors for service situations include processing patients through an Xray or CATscanning machine,
trucks through a loading/unloading dock, or financial transactions through a control work station. For the singleprocessor
sequencing problem, a very simple rule—shortest processing time—finds a minimal average flowtime sequence. An example of
its use follows.
Consider a work station that has one maintenance mechanic to repair failed machines. We can think of the mechanic as the
processor (scarce resource) and the machines awaiting repair as the jobs. Let us assume that six machines are down, with estimated
repair times given here, and that no new jobs arrive.
Job (fix machine #) 1 2 3 4 5 6
Processing time (hours) 10 3 7 2 9 6
No matter which sequence is chosen, the makespan is the same, because the time to process all the jobs
Monkey Business
Images/Shutterstock.com
is the sum of the processing times, or in this example, 37 hours. Therefore, we use average flow time as the criterion to minimize
the average time a job spends in the work station. The idea here is to get the most jobs done as soon as possible. Applying the SPT
rule, we use the job sequence 426351. We assume that all jobs are ready for processing at time zero (that is, Ri = 0 for all jobs
i). Then the flow times (Fi) for the jobs are computed as follows:
Job Sequence Flow Time
4 2 hours
2 2 + 3 = 5 hours
6 5 + 6 = 11 hours
3 11 + 7 = 18 hours
5 18 + 9 = 27 hours
1 27 + 10 = 37 hours
The average flow time for these six jobs is (2 + 5 + 11 + 18 + 27 + 37)/6 = 100/6 = 16.67 hours. This means that the average time a
machine will be out of service is 16.7 hours. The SPT sequencing rule maximizes work station utilization and minimizes average
job flow time and WIP inventory. For example, if you switch jobs 4 and 6 so the job sequence is 624351, note that the average
flow time increases to 18 hours. We encourage you to work through the calculations to show this. As long as no additional jobs
enter the mix, all will eventually be processed. Of course, the job with the longest processing time will wait the longest (and this
customer might not be very happy), but on average, SPT will reduce the average flow time.
When processing times are relatively equal, then most operating systems default to the firstcome, firstserved (FCFS) sequencing
rule. There are, of course, exceptions to this rule. For example, a job for a firm's most important customer might be pushed to the
front
P. 315
of the sequence, or the maître d’ at a restaurant might seat a celebrity or VIP before other patrons.
In many situations, jobs have due dates that have been promised to customers. Although SPT provides the smallest average flow
time and smallest average lateness of all scheduling rules that might be chosen, in a dynamic environment, jobs with long
processing times are continually pushed back and may remain in the shop a long time. Thus it is advantageous to consider
sequencing rules that take into account the due dates of jobs.
A popular and effective rule for scheduling on a single processor (resource) is the earliestduedate (EDD) rule, which dictates
sequencing jobs in order of earliest due date first. This rule minimizes the maximum job tardiness and job lateness. It does not
minimize the average flow time or average lateness, as SPT does, however. An example of how the EDD rule is used follows.
Suppose an insurance underwriting work area (that is, the single processor) has five commercial insurance jobs to quote that have
these processing times and due dates:
If the jobs are sequenced “by the numbers” in the order 12345, then the flow time, tardiness, and lateness for each job are
calculated using Equations 14.1, 14.3, and 14.4, as shown in the table below.
Using Equation 14.2, the makespan is Mt = CtSt = 22 − 0 = 22. If we use the SPT rule to schedule the jobs, we obtain the
sequence 35142. The flow time, tardiness, and lateness are then given as follows:
5 2 + 3 = 5 9 −4 0
1 5 + 4 = 9 15 −6 0
4 9 + 6 = 15 21 −6 0
2 15 + 7 = 22 16 6 6
Average 10.6 −3.2 1.2
Note that the makespan is 22 and that the maximum tardiness and the maximum lateness are both 6. Using the EDD rule, we
obtain the sequence 35124. The flow time, tardiness, and lateness for this sequence are given in the following table:
P. 316
Exhibit 14.3 Comparison of Three Ways to Sequence the Five Jobs
Performance Criteria Sequence 12345 Sequence 35142 (SPT) Sequence 35124 (EDD)
Average Flow Time 13.8 10.6 10.8
Average Lateness 0 −3.2 −3.0
Maximum Lateness 13 6 1
Average Tardiness 3.6 1.2 0.2
Maximum Tardiness 13 6 1
Exhibit 14.3 Comparison of Three Ways to Sequence the Five Jobs © Cengage Learning 2013
The results of applying three different sequencing rules to the five jobs are shown in Exhibit 14.3. Note that the SPT rule
minimizes the average flow time and number of jobs in the system. The EDD rule minimizes the maximum lateness and tardiness.
As previously noted, the SPT rule is internally focused, whereas the EDD rule is focused on external customers. Using a bythe
numbers sequencing rule, as in 12345, results in very poor relative performance. This result helps illustrate that random or
commonsense sequencing rules seldom give better results than the SPT or EDD rules for sequencing jobs over a single processor.
Solved Problem
Five tax analysis jobs are waiting to be processed by Martha at T.R. Accounting Service. Use the shortestprocessingtime (SPT)
and earliestduedate (EDD) sequencing rules to sequence the jobs. Compute the flow time, tardiness, and lateness for each job,
and the average flow time, average tardiness, and average lateness for all jobs. Which rule do you recommend? Why?
2 3 10
3 5 8
4 2 5
5 6 17
Solution
The SPT sequence is 42351.
The EDD sequence is 43215.
P. 317
Exhibit 14.4 Excel Sequencing Template for Calculating SPT Sequence Performance Measures
Given the nature of the data, this is not an easy decision. The SPT rule minimizes average flow time and average lateness but job 1
is extremely late by 12 days. The EDD rule minimizes the maximum job tardiness and lateness. Jobs 1 and 5 are tardy by six days.
If job 5 is a big client with significant revenue potential, then the EDD rule is probably best.
Exhibit 14.4 shows the Excel Sequencing template, which can be used to evaluate the performance of different sequences.
144b TwoResource Sequencing Problem
In this section, we consider a flow shop with only two resources or work stations. We assume that each job must be processed first
on Resource #1 and then on Resource #2. Processing times for each job on each resource are known. In contrast to sequencing jobs
on a single resource, the makespan can vary for each different sequence. Therefore, for the tworesource sequencing problem, it
makes sense to try to find a sequence with the smallest makespan.
S.M. Johnson developed the following algorithm in 1954 for finding a minimum makespan schedule.6 The following algorithm
(procedure) defines Johnson's sequencing rule for the tworesource problem structure.
1. List the jobs and their processing times on Resources #1 and #2.
2. Find the job with the shortest processing time (on either resource).
3. If this time corresponds to Resource #1, sequence the job first; if it corresponds to Resource #2, sequence the job last.
4. Repeat steps 2 and 3, using the nextshortest processing time and working inward from both ends of the sequence until all
jobs have been scheduled.
Consider the tworesource sequencing problem posed by Hirsch Products. It manufactures certain custom parts that first require a
shearing operation (Resource #1) and then a punchpress operation (Resource #2). Hirsch currently has orders for five jobs, which
shearing operation (Resource #1) and then a punchpress operation (Resource #2). Hirsch currently has orders for five jobs, which
have processing times (days) estimated as follows:
5 2 3
P. 318
Roger Tully/Stone/Getty Images
The jobs can be sequenced in any order but they must be sheared first. Therefore, we have a flow shop situation where each job
must first be sequenced on the shear operation and then on the punch operation.
Suppose the jobs are sequenced “by the numbers” in the order 12345. This schedule can be represented by a simple Gantt chart
showing the schedule of each job on each machine along a horizontal time axis (see Exhibit 14.5). This shows, for instance, that
job 1 is scheduled on the shear for the first four days, job 2 for the next four days, and so on. We construct a Gantt chart for a given
sequence by scheduling the first job as early as possible on the first machine (shear). Then, as soon as the job is completed, it can
be scheduled on the punch press, provided that no other job is currently in progress. First, note that all jobs follow each other on
the shearing machine. Because of variations in processing times, however, the punch press, the second operation, is often idle
while awaiting the next job. The makespan is 37 days, and the flow times in days for the jobs are as follows:
Applying Johnson's rule, we find that the shortest processing time is for job 2 on the punch press.
Because the minimum time on either machine is on the second machine, job 2, with a oneday processing time, it is scheduled last.
Next, we find the secondshortest processing time. It is two days, for job 5 on machine 1. Therefore, job 5 is scheduled first.
In the next step, we have a tie of four days between job 1 on the shear and job 3 on the punch press. When a tie occurs, either job
can be chosen. If we pick job 1, we have the following sequence:
Exhibit 14.5
Gantt Job Sequence Chart for Hirsch Product Sequence 12345 © Cengage Learning 2013
P. 319
Solved Problem
A manufacturing process involving machined components consists of two operations done on two different machines. The status
of the queue at the beginning of a particular week is as follows:
The processing on machine 2 must follow processing on machine 1. Schedule these jobs to minimize the makespan. Illustrate the
schedule you arrive at with a bar chart.
Solution
Because this is a twomachine flow shop problem, Johnson's rule is applicable. Total time in minutes on each machine is the
product of the number of components and the unit times, as shown here.
The sequence specified by Johnson's rule is 201213184101185176. The schedules are shown in the following two different
versions of Gantt charts.
©
Cengage Learning 2013
Continuing with Johnson's rule, the last two steps yield the complete sequence.
The Gantt chart for this sequence is shown in Exhibit 14.6. The makespan is reduced from 37 to 27 days, and the average flow
time is also improved from 22.4 to 18.2 days. As noted, the total idle time on the punch press is now only four days, resulting in
Different sequencing rules lead to very different results and performance.
P. 320
Exhibit 14.6
Gantt Job Sequence Chart for Hirsch Product Sequence 51432 © Cengage Learning 2013
a punchpress resource utilization of 23/27, or 85.2 percent, and we gain 10 days to schedule other jobs. If the sequencing problem
structure fits the assumptions of Johnson's rule, it is a powerful algorithm. Again, commonsense scheduling is seldom as good as
Johnson's rule.
145 Schedule Monitoring and Control
Murphy's Law states that if something can go wrong it will, and this is especially true with schedules. Thus, it is
important that progress be monitored on a continuing basis. For example, in manufacturing, the master scheduler must know the
status of orders that are ahead of schedule or behind schedule due to shortages of material, work stations that are backlogged,
changes in inventory, labor turnover, and sales commitments. Schedules must be changed when these things occur. Therefore,
reschedules are a normal part of scheduling.
reschedules are a normal part of scheduling.
Shortterm capacity fluctuations also necessitate changes in schedules. Factors affecting shortterm
Murphy's Law states that if something can go wrong it will, and this is especially true with schedules.
Mobil oil: ComputerBased Scheduling
Mobil Oil Corporation runs a nationwide system for dispatching and processing customer orders for gasoline and distillates. It is
an integrated operating system that controls the flow of billions in annual sales from initial order entry to final delivery,
confirmation, and billing. Although the entire dispatching process is overseen by a handful of people in a small office, it operates
more efficiently than the old manual system in all respects: It provides better customer service; greatly improved credit, inventory,
and operating cost control; and significantly reduced distribution costs. Central to this new system is computerassisted dispatch
(called CAD at Mobil), designed to assist schedulers in real time as they determine the means by which ordered products will be
safely and efficiently delivered to customers.
Scheduling decisions include (1) assigning orders to terminals; (2) assigning orders to delivery trucks; (3) adjusting order
quantities to fit truck compartments; (4) loading trucks to their maximum legal weight; and (5) routing trucks and sequencing
deliveries. Annual net cost savings is in the millions of dollars.7
Fred Prouser/Reuters/Landov
P. 321
Exhibit
14.7 Gantt Chart Example for Monitoring Schedule Progress © Cengage Learning 2013
capacity include absenteeism, labor performance, equipment failures, tooling problems, labor turnover, and material shortages.
They are inevitable and unavoidable. Some alternatives available to operations managers for coping with capacity shortages are
overtime, shortterm subcontracting, alternate process routing, and reallocations of the workforce, as described in the previous
chapter.
Gantt (bar) charts are useful tools for monitoring schedules, and an example is shown in Exhibit 14.7. The dark shaded areas
indicate completed work. This chart shows, for example, that job 4 has not yet started on machine 2, job 1 is currently behind
schedule on machine 3, and jobs 2 and 5 are ahead of schedule. Perhaps needed material has not yet been delivered for job 4, or
perhaps machine 3 has had a breakdown. In any event, it is up to productioncontrol personnel to revise the schedule or to
expedite jobs that are behind schedule. Many other types of graphical aids are useful and commercially available.
Operations Scheduling and Sequencing: Discussion Questions
1. Discuss how you decide to schedule your school assignments. Do your informal scheduling rules correspond to any of the
rules described in this chapter?
2. Explain why appointments are necessary for many professional services. (Hint: How do services differ from goods, as
described in Chapter 1?) List and explain some key issues and decisions that must be addressed in designing appointment
systems.
3. Scheduling and sequencing are typically viewed from a technical perspective; that is, they are focused on minimizing
quantitative measures such as lateness or cost. However, schedules also have intangible effects on customers, employees,
and the perception of service quality. Discuss what some of these intangible effects might be and how managers should
consider them when constructing schedules.
4. Why is staff scheduling in a service environment a difficult task? What can managers do to ensure that staff schedules are
effective and efficient?
5. Explain the advantages and disadvantages of SPT and EDD sequencing rules. Under what circumstances might you prefer
one over the other?
Operations Scheduling and Sequencing: Problem and Activities
Note: an asterisk denotes problems for which an Excel spreadsheet template on the CourseMate Web site may be used.
1. Write a short onepage paper listing the advantages and disadvantages of using parttime employees to help meet demand.
2. Research a desktop or webbased scheduling software system and write a short onepage paper describing its features and
capabilities, and any
P. 322
benefits obtained by realworld organizations that adopted the software.
3. Research and write a short paper (two pages maximum) on how efficient vehicle scheduling and routing can be accomplished. If
possible, provide some examples of commercial software programs that are used for vehicle scheduling and routing and describe
their capabilities.
4. Interview an operations manager at a nearby manufacturing or service company to find out about scheduling problems the
company faces and how they are addressed.
5. A hospital emergency room needs the following numbers of nurses:
Day MTWTFSS
Min. number 4 3 2 4 7 8 5
Each nurse should have two consecutive days off. How many fulltime nurses are required and what is a good nurse schedule?
6. A supermarket has the following minimum personnel requirements during the week. Each employee is required to have two
6. A supermarket has the following minimum personnel requirements during the week. Each employee is required to have two
consecutive days off. How many regular employees are required, and what is a good schedule?
7. These six jobs are to be scheduled on a single machine:
Job 1 2 3 4 5 6
a. Suppose the jobs are processed in numerical order. Compute the average flowtime after each job is completed.
b. In what order would the jobs be processed using the SPT rule? Compute the average flowtime after each job is completed.
Compare this answer with your answer to part a.
8.* An insurance claims work area has five claims waiting for processing as follows:
Compute the average flow time, tardiness, and lateness for the following sequences: SPT sequence, EDD sequence, and the
sequence 21534.What sequencing rule do you recommend and why?
9.* Mike Reynolds has four assignments due in class tomorrow, and his class times are as follows:
Class Time
Marketing 304 8 a.m.
OM 385 10 a.m.
Finance 216 1 p.m.
Psychology 200 3:30 p.m.
Each class lasts one hour, and Mike has no other classes. It is now midnight, and Mike estimates that the finance, OM, marketing,
and psychology assignments will take him five, three, six, and two hours, respectively. How should he schedule the work? Can he
complete all of it?
10.* Eight jobs have arrived in the following order:
5 6 17
6 9 16
7 14 38
8 4 12
Find and compare the average flowtime, lateness, and tardiness for the following sequencing rules:
a. Process in the order they have arrived
b. Shortest processing time
c. Earliest due date
11.* In this chapter we noted that the EDD rule minimizes the maximum job tardiness and maximum lateness, whereas the SPT
rule minimizes the average flowtime. However, neither of these rules minimizes the average lateness or average tardiness. Use the
data in problem 10 to do the following:
a. Try to find a sequence that minimizes the average lateness.
b. Try to find a sequence that minimizes the average tardiness.
c. Can you generalize your logic into a rule or procedure that will accomplish these objectives most of the time?
P. 323
12. Monday morning Baxter Industries has the following jobs waiting for processing in two departments, milling and drilling, in
that order:
Time Required (hours)
Job Mill Drill
216 7 4
327 6 10
462 10 3
519 5 6
258 3 9
617 7 2
Develop a minimum makespan schedule using Johnson's rule.
13. Graph the minimum makespan schedule in problem 12 on a Gantt chart.
14. Dan's Auto Detailing business performs two major activities: exterior cleanup and interior detailing. Based on the size of the
car and its condition, time estimates for six cars on Monday morning are as follows:
Car Number
1 2 3 4 5 6
Exterior 30 35 90 65 45 80
Interior 10 40 20 45 25 55
Sequence the cars so that all exterior detailing is done first and total completion time is minimized.
15. Draw a Gantt chart of your solution to problem 14. Evaluate the idle time, if any, for these two resources—exterior and interior
cleaning capability.
Operations Scheduling and Sequencing: Balloons Aloha Case Study
The company fills the balloons on the day of
the party, so the workload is hectic. Michael Newman/PhotoEdit
Susie Davis owns Balloons Aloha and must fill balloons with helium and assemble them into certain configurations today for six
major parties. Her six customer jobs all need to use the same helium tank (that is, the single processor), and she was wondering
what might be the best way to sequence these jobs. Client (job) number 5 is Balloons Aloha's top customer. Her assistant store
manager, Lee Sailboat, wants to process them in sequential order (i.e., 1, 2, 3, 4, 5, and 6). Because the balloons lose air quickly,
the company waits until the day of the parties to fill them and then the workload is hectic. Business is booming and growing
about 15 percent per year in their new store location. The job processing time estimates are as follows:
Job 1 2 3 4 5 6
Processing time (min.) 240 130 210 90 170 165
Due dates (6 a.m. to midnight in minutes from opening) 240 360 480 240 720 780
Case Questions for Discussion
1. Compute the average flow time, lateness, and tardiness for this group of jobs using Mr. Sailboat's sequential order of 1
(first), 2, 3, 4, 5, and 6 (last).
2. In what order would the jobs be processed using the SPT rule? Compute the average flow time, lateness, and tardiness for
this group of jobs.
3. Compare the answers in questions 1 and 2.
4. What are your shortterm recommendations for this set of six jobs? Justify and explain.
5. What are your longterm recommendations with respect to sequencing jobs at Balloons Aloha? Justify and explain.
P. 324
Operations Scheduling and Sequencing: Reviewcard/OM4 Chapter 14 Operations Scheduling and
Sequencing: Learning Outcomes
141 Explain the concepts of scheduling and sequencing.
Scheduling and sequencing are some of the more common activities that operations managers perform every day in every business.
They are fundamental to all three levels of aggregation and disaggregation planning that we described in the previous chapter.
Good schedules and sequences lead to efficient execution of manufacturing and service plans. Some examples of scheduling: fast
food restaurants, hospitals, and call centers need to schedule employees for work shifts; doctors, dentists, and stockbrokers need to
schedule patients and customers; airlines must schedule crews and flight attendants; sports organizations must schedule teams and
officials; court systems must schedule hearings and trials; factory managers need to schedule jobs on machines and preventive
maintenance work; and salespersons need to schedule customer deliveries and visits to potential customers. Some examples of
sequencing: triage nurses must decide on the order in which emergency patients are treated; housekeepers in hotels must sequence
the order of rooms to clean; operations managers who run an automobile assembly line must determine the sequence by which
different models are produced; and airport managers must sequence outgoing flights on runways.
142 Describe staff scheduling and appointment system decisions.
Staff scheduling problems are also prevalent in service organizations because of high variability in customer demand. Examples
include scheduling call center representatives, hotel housekeepers, tollbooth operators, nurses, airline reservation clerks, police
officers, fastfood restaurant employees, and many others.
Staff scheduling attempts to match available personnel with the needs of the organization by
1. accurately forecasting demand and translating it into the quantity and timing of work to be done;
2. determining the staffing required to perform the work by time period;
3. determining the personnel available and the fulland parttime mix; and
4. matching capacity to demand requirements, and developing a work schedule that maximizes service and minimizes costs.
The first step requires converting demand to a capacity measure, that is, the number of staff required. The second step determines
the quantity and timing of the work to be done in detail, usually by hour of the day, and sometimes in 5to 10minute time
intervals. Determining the staffing required must take into account worker productivity factors, personal allowances, sickness,
vacations, noshows, and so on. Step 4 focuses on the matching of capacity to demand requirements; this is the essence of
scheduling.
A simple problem of scheduling personnel with consecutive days off in the face of fluctuating requirements can be solved as
follows. First, locate the set of at least two consecutive days with the smallest requirements. Circle the requirements for those two
consecutive days. Assign staff to work on all days not circled and update the remaining requirements. When there are several
alternatives, do one of two things. First, try to choose a pair of days with the lowest total requirement. If there are still ties, choose
the first available pair that makes the most sense to the scheduler. Continue circling the smallest requirements until you obtain at
least two consecutive days and repeat the procedure until all requirements are scheduled.
Appointments can be viewed as a reservation of service time and capacity. Using appointments provides a means to maximize the
use of timedependent service capacity and reduce the risk of no shows. Appointment systems are used in many businesses, such
as consulting, tax preparation, music instruction, and medical, dental, and veterinarian practices. Indirectly, appointments reduce
the cost of providing the service because the service provider is idle less each workday. An appointment system must try to
accommodate customers and forecast their behavior, such as the noshow rate or a difficult customer who demands more
processing time.
Four decisions to make regarding designing an appointment system are the following:
1. Determine the appointment time interval, such as 1 hour or 15 minutes.
2. Based on an analysis of each day's customer mix, determine the length of each workday and the time off duty.
3. Decide how to handle overbooking for each day of the week.
4. Develop customer appointment rules that maximize customer satisfaction.
P. RC_29
143 Explain sequencing performance criteria and rules.
Sequencing criteria are often classified into three categories:
1. processfocused performance criteria,
2. customerfocused due date criteria, and
3. costbased criteria.
Processfocused performance criteria pertain only to information about the start and end times of jobs and focus on shop
performance such as equipment utilization and WIP inventory. Two common measures are flow time and makespan.
Due date criteria pertain to customers’ required due dates or internally determined shipping dates. Common performance measures
are lateness and tardiness, or the number of jobs tardy or late.
Costbased performance criteria include inventory, changeover or setup, processing or run, and materialhandling costs. In most
cases, costs are considered implicitly in process performance and duedate criteria.
Two of the most popular sequencing rules for prioritizing jobs are the following:
Shortest processing time (SPT)
Earliest due date (EDD)
In other situations, new jobs arrive in an intermittent fashion, resulting in a constantly changing mix of jobs needing to be
sequenced. In this case, we assign priorities to whatever jobs are available at a specific time and then update the priorities when
new jobs arrive. Some examples of these priority rules are the following:
First come, first served (FCFS)
Fewest number of operations remaining (FNO)
Least work remaining (LWR)—sum of all processing times for operations not yet performed
Least amount of work at the next process queue (LWNQ)—amount of work awaiting the next process in a job's sequence
The SPT rule tends to minimize average flow time and WIP inventory and maximize resource utilization. The EDD rule minimizes
the maximum of jobs past due but doesn't perform well on average flow time, WIP inventory, or resource utilization. The FCFS
rule is used in many servicedelivery systems and does not consider any job or customer criterion. FCFS focuses only on the time
of arrival for the customer or job. The FNO rule does not consider the length of time for each operation; for example, a job may
have many small operations and be scheduled last. Generally, this is not a very good rule. The LWNQ rule tries to keep
downstream work stations and associated resources busy.
144 Describe how to solve singleand tworesource sequencing problems.
The simplest sequencing problem is that of processing a set of jobs on a single processor. For the singleprocessor sequencing
problem, a very simple rule—shortest processing time—finds a minimal average flow time sequence. The EDD, which dictates
sequencing jobs in order of earliest due date first, minimizes the maximum job tardiness and job lateness. It does not minimize the
average flow time or average lateness, as SPT does, however.
S.M. Johnson developed an algorithm for finding a minimum makespan schedule for a tworesource sequencing problem.
1. List the jobs and their processing times on Resources #1 and #2.
2. Find the job with the shortest processing time (on either resource).
3. If this time corresponds to Resource #1, sequence the job first; if it corresponds to Resource #2, sequence the job last.
3. If this time corresponds to Resource #1, sequence the job first; if it corresponds to Resource #2, sequence the job last.
4. Repeat steps 2 and 3, using the nextshortest processing time and working inward from both ends of the sequence until all
jobs have been scheduled.
Gantt charts are often used to display the results, such as those shown in Exhibit 14.5.
145 Explain the need for monitoring schedules using Gantt charts.
Murphy's Law states that if something can go wrong it will, and this is especially true with schedules. Thus, it is important that
progress be monitored on a continuing basis. Reschedules are a normal part of scheduling.
Shortterm capacity fluctuations also necessitate changes in schedules. Factors affecting shortterm capacity include absenteeism,
labor performance, equipment failures, tooling problems, labor turnover, and material shortages. They are inevitable and
unavoidable. Gantt (bar) charts are useful tools for monitoring schedules (see Exhibit 14.7).
Exhibit 14.5 Gantt Job Sequence Chart for Hirsch
Product Sequence 12345 © Cengage Learning 2013
Exhibit 14.7 Gantt Chart Example for
Monitoring Schedule Progress © Cengage Learning 2013
P. RC_30
Operations Scheduling and Sequencing: Reviewcard/OM4 Chapter 14 Operations Scheduling and
Sequencing: Key Terms
141
Scheduling refers to the assignment of start and completion times to particular jobs, people, or equipment.
Sequencing refers to determining the order in which jobs or tasks are processed.
143
Flow time is the amount of time a job spent in the shop or factory.
Makespan is the time needed to process a given set of jobs.
Lateness is the difference between the completion time and the due date (either positive or negative).
Tardiness is the amount of time by which the completion time exceeds the due date.
Footnotes
1. “The UPS Green Dream,” Fortune, December 27, 2010, p. 44–51; Tricia Bisoux, “The Connected Capitalists, BizEd, May/June
2010, pp. 24–28 and pp. 44–51.
2. G. M. Campbell, “Overview of Workforce Scheduling Software,” Production and Inventory Management, 45, 2, pp. 7–22.
3. This approach is suggested in R. Tibrewala, D. Phillippe, and J. Browne, “Optimal Scheduling of Two Consecutive Idle
Periods,” Management Science, 19, 1, September 1972, pp. 71–75.
4. Anton Ovchinnikov and Joseph Milner, “Spreadsheet Model Helps to Assign Medical Residents at the University of Vermont's
College of Medicine,” Interfaces, 38, 4, July–August 2008, pp. 311–323.
College of Medicine,” Interfaces, 38, 4, July–August 2008, pp. 311–323.
5. http://www.absusa.com.
6. S. M. Johnson, “Optimal Twoand ThreeStage Production Schedules with Setup Times Included,” Naval Research Logistics
Quarterly, 1, 1, March 1954, pp. 61–68.
7. Gerald G. Brown, Carol J. Ellis, Glenn W. Graves, and David Ronen, “RealTime, WideArea Dispatch of Mobil Tank Trucks,”
Interfaces, 17, 1, 1987, pp. 107–120.
P. RC_29
© 2007 Cengage Learning Inc. All rights reserved. No part of this work may by reproduced or used in any form or by any
means graphic, electronic, or mechanical, or in any other manner without the written permission of the copyright holder.