Professional Documents
Culture Documents
Scheduling Operations
Chapter 13: Scheduling
Dear students, this third lecture focuses solely on OPT, the related software and
other such practical issues.
There is another approach to the planning and scheduling so far presented is Optimized
Production Technology. This is a computer based system for planning production,
material needs, and recourse utilization. It was first introduced in the USA in 1979 by
Creative Output Inc, a Consulting Form in Milferd, Connecticut.
The key feature of OPT is its emphasis on bottleneck center works- people or machines..
The OPT philosophy is that managing bottlenecks is the key to successful performance:
total system output can be maximized and the in-process inventories reduced.
1. BUILDNET
2. SERVE
3. SPLIT
4. OPT
The above mentioned modules creates a model of the shop according to the data provided
by the user, how each product is made, its build up sequence, materials, and routing
through the shop), the products time requirements ( setup, run time, schedule delay ), the
capacity availability at each resource ( work center, machine , worker), and the order
quantities.
The initial purpose of SERVE is to create a tentative schedule for the jobs waiting in the
shop. Later it creates a more refined schedule. The crucial information obtained in the
SERVE is an estimate of the percentage utilization of the various shop resources.
The SPLIT module distinguishes critical from non critical resources based on their
percent utilizations calculated by SERVE. Resources that are near or above 100 percent
utilization are the bottleneck operations. These bottlenecks, and the operations that follow
them, are the "critical" operations; all others (those with lower percentage utilizations) are
"non critical."
The OPT module reschedules the critical part of the network using forward
scheduling. Then the program cycles back to SERVE to reschedule the non critical
resources.
The OPT package consists not only of software but of consulting services and
training for implementation as well. The specific details of the procedure, especially of
SERVE and OPT (the detailed scheduling modules), are proprietary (not published and
available to the general public). Consequently, detailed comparative evaluations of its
performance with that of other systems are not available.
The Behavioral elements in intermittent systems are of immense significance. You can’t simply
ignore it.
Please pay attention.
Individual Characteristics You may remember that intermittent systems, compared with
continuous flow systems, comprise a variety of tasks. Different types of employee skills
are necessary. Generally, jobs in intermittent systems are already "enlarged": tasks vary,
and a higher degree of employee responsibility is emphasized in executing the tasks. In
hiring, managers seek employees who are highly skilled and who can work independently
without a great deal of supervision. Through monetary rewards, facilitating group
relationships, and allocating work methodically, management can create a working
environment that helps employees feel secure and fulfills their social needs. Doing so
increases motivation in job performance.
Group Characteristics In discussing facility layout, we pointed out that intermittent
systems comprise work centers sharing common processes. A facility might have several
work centers, each using workers skilled in machining, painting, and photography, for
example. Group affiliatioJ1s are often established among commonly skilled workers,
machinists, for example. There are three reasons for these group affiliations: command
structure, physical proximity, and shared craft interest. As a formal basis for group
affiliation among machinists, for example, the organizational structure may specify that
all machinists report to a machining foreman. Second, the physical proximity of
machinists in the facility, since they usually work near one another, tends to facilitate
interaction and communication, both work-related and personal. Since this is likely to
occur on a regular basis, strong group bonds may form. Finally, an important shared
interest-the craft or skill-is a basis for interaction. Unions facilitate this last affiliation; it
is likely that in a unionized facility of any size, more than one union will represent
differing groups of employees.
A work group significantly affects the operations of the system. Although a group
usually adopts a set of group norms and strives to satisfy member needs, the group norms
mayor may not be consistent with management goals. Group norms can strongly
influence its members' productivity, especially in highly cohesive groups. When the
norms of these cohesive groups are consistent with management goals, the groups tend to
produce at higher levels.
SCHEDULING IN SERVICES
So far we have seen the scheduling systems in manufacturing. Now let us see how the
scheduling system works in service area.
One important distinction between manufacturing and services that affects scheduling is
that service operations cannot create inventories to buffer demand uncertainties. A second
distinction is- that in service operations demand often is less predictable. Customers may
decide on the spur of the moment that they need a hamburger, a haircut, or a plumbing
repair. Thus capacity, often in the form of employees, is crucial for service providers. In
this section, we discuss various ways in which scheduling systems can facilitate the
capacity management of service providers.
CONSTRAINTS. The technical constraints imposed on the workforce schedule are the
resources provided by the staffing plan and the requirements placed on the operating
system. However, other constraints, including legal and behavioral considerations, also
can be imposed. For example, Air New Zealand is required to have at least a minimum
number of flight attendants on duty at all times. Similarly, a minimum number of fire and
safety personnel must be on duty at a fire station at all times. Such constraints limit
management’s flexibility in developing workforce schedules.
The constraints imposed by the psychological needs of workers complicate scheduling
even more. Some of these constraints are written into labor agreements. For example, an
employer may agree to give employees a certain. number of consecutive days off per
week or to limit employees' consecutive workdays to a certain maximum. Other
provisions might govern the allocation of vacation, days off for holidays, or rotating shift
assignments. In addition, preferences of the employees themselves need to be considered.
One way that managers deal with certain undesirable aspects of scheduling is to use a
rotating schedule, which rotates employees through a series of workdays or hours. Thus,
over a period of time, each person has the same opportunity to have weekends and
holidays off and to work days, as well as evenings and nights. A rotating schedule gives
each employee the next employee's schedule the following week. In contrast, a fixed
schedule calls for each employee to work the same days and hours each week.
Students, this can not be effectively accomplished without:-
Step 1. From the schedule of net requirements for the week, find all the pairs of con-
secutive days that exclude the maximum daily requirements. Select the unique pair that
has the lowest total requirements for the two days. In some unusual situations, all
pairs may contain a day with the maximum requirements. If so, select the pair with the
lowest total requirements. Suppose that the numbers of employees required are
Monday: 8 Friday: 7
Tuesday: 9 Saturday: 4
Wednesday
2 Sunday: 2
:
Thursday: 12
The maximum capacity requirement is 12 employees, on Thursday. The pair having the
lowest total requirements is Saturday-Sunday, with 4 + 2 = 6.
Step 2. If a tie occurs, choose one of the tied pairs, consistent with provisions written into
the labor agreement, if any. Alternatively, the tie could be broken by asking the employee
being scheduled to make the choice. As a last resort, the tie could be broken arbitrarily.
For example, preference could be given to Saturday-Sunday pairs.
Step 3. Assign the employee the selected pair of days off. Subtract the requirements
satisfied by the employee from the net requirements for each day the employee is to
work. In this case, the employee is assigned Saturday and Sunday off. After requirements
are subtracted, Monday's requirement is 7, Tuesday's is 8, Wednesday's is 1, Thursday's
is 11, and Friday's is 6. Saturday's and Sunday's requirements do not change because no
employee is yet scheduled to work those days.
Step 4. Repeat steps 1-3 until all requirements have been satisfied or a certain number of
employees have been scheduled.
This method reduces the amount of slack capacity assigned to days having low
requirements and forces the days having high requirements to be scheduled first. It also
recognizes some of the behavioral and contractual aspects of workforce scheduling in the
tie-breaking rules. However, the schedules produced might not minimize total slack
capacity. Different rules for finding the days-off pair and breaking ties are needed to
ensure minimal total slack capacity.
The Amalgamated Parcel Service is open seven days a week. The schedule of
requirements is
Day M T W Th F S Su
No of Employees 6 4 8 9 10* 3 2
The manager needs a workforce schedule that provides two consecutive days off and
minimizes the amount of total slack capacity. To break ties in the selection of off days,
the scheduler gives preference to Saturday-Sunday if it is one of the tied pairs. If not, she
selects one of the tied pairs arbitrarily.
Well my dears, the same rules apply.
Solve first.
Tally later.
SOLUTION
Friday contains the maximum requirements (designated by an *), and the pair S-Su has
the lowest total requirements. Therefore, employee 1 is scheduled to work Monday-
Friday. The revised set of requirements, after scheduling employee 1, is
Day M T W Th F S Su
Number of employees 5 3 7 8 9* 3 2
TABLE 13.2 Scheduling Days Off
EMPLOY
M T W Th F S Su COMMENTS
EE
4 2 6 7 8* 3 2 3 S-Su has the lowest total requirements. Reduce the
requirements to reflect a M-F schedule for employee 3.
3 1 5 6 7* 3 2 4 M- T has the lowest total requirements. Assign
employee 4 to a W-Su schedule and update the
requirements.
3 1 4 5 6* 2 1 5 S-Su has the lowest total requirements. Assign
employee 5 to a M-F schedule and update the
requirements.
2 0 3 4 S* 2 1 6 M-T has the lowest total requirements. Assign
employee 6 to a W-Su schedule and update the
requirements.
2 0 2 3 4* 1 0 7 S-Su has the lowest total requirements. Assign
employee 7 to a M-F schedule and update the
requirements.
9 off X X X X X off
10 X X X X X off off
Capacity, C 7 7 10 10 10 3 2 50
Requirement
6 4 8 9 10 3 2 42
s, R
Slack C - R 1 3 2 1 0 0 1 8
Decision Point With its substantial amount of slack capacity, the schedule is not unique.
Employee 9, for example, could have Su-M, M-T, or T-W off without causing a capacity
shortage. Indeed, the company might be able to get by with one fewer employee because
of the total of eight slack days of capacity. However, all 10 employees are needed on
Fridays. If the manager were willing to get by with only 9 employees on Fridays or if
someone could work one day of overtime on a rotating basis, he would not need
employee 10. As indicated in the table, the net requirement left for employee 10 to satisfy
amounts to only one day, Friday. Thus, employee 10 can be used to fill in for vacationing
or sick employees.
SYNCHRONOUS MANUFACTURING
Optimized Production Technology (OPT) was evolved through software. This software
was developed by Creative Output Inc USA and the person responsible for it is Dr. Eli
Goldratt. Here the scheduling logic is based on the separation of “bottleneck” and non-
bottleneck operations.
Further Dr Goldratt developed the “Theory of Constraints” (TOC), which has become
popular as a problem solving approach that can be applied to many business areas. So
let’s go through TOC briefly;
1. Identify the system constraints (No improvement is possible unless the weak link
or constraint is found out)
2. Decide how to exploit the system constraints (Make the constraints as effective
as possible)
3. Subordinate everything to that decision (align every other part of the system to
support the constraints even if this reduces the efficiency of non constraint
resources).
4. Elevate the system constraints (If output is still inadequate acquire more of this
resource so that it is no longer a constraint)
5. If, in the previous steps, the constraint is broken back, go back to step 1, but do
not let inertia become the system constraint (After the constraint problem is
solved, go back to the beginning and start all over again. This is a continuous
process of improvements
Therefore by removing the constraints and moving forward we get into a situation where
the entire operations or production process work in harmony to achieve the ultimate goal
of an organization i.e. Profit. From the point of view of operation or production
management in order to achieve profit, the goal would be
UNBALANCED CAPACITY
Historically (and still typically in most firms) manufacturers have tried to balance
capacity across a sequence of processes in an attempt to match capacity with market
demand. However, this is the wrong thing to do-unbalanced capacity is better.
Consider a simple process line with several stations, for example. Once the output rate of
the line has been established, production people try to make the capacities of all stations
the same. This is done by adjusting machines or equipment used, workloads, skill and
type of labor assigned, tools used, overtime budgeting and so on.
In synchronous manufacturing thinking, however, making all the capacities is viewed
as a bad decision. Such a balance would be possible only if the output time of stations
were constant or had a very narrow distribution. A normal variation in output causes
downstream stations to have idle time when upstream stations take longer to process.
Conversely, when upstream stations process in a shorter time, inventor builds up between
the stations. The effect of the statistical variation is cumulative. The only way that these
stations can be smoothed is by increasing work-in-process to absorb the variation (a bad
idea because we should be trying to reduce work-in-process) or increasing capacities.
Stream to be able to make up for the longer upstream times. The rule here is that capacity
within the process sequence should not be balanced to the same levels. Rather, attempt
should be made to balance the flow of product through the system. When flow is balance
capacities are unbalanced.
We have earlier said about constraints and bottlenecks, now let us see its definition and
further understanding of the topic.
A bottleneck is defined as any resource whose capacity is less than the demand placed
upon it. A bottleneck is a constraint within the system that limits throughput. It is that
point in the manufacturing process where flow thins to a narrow stream. A bottleneck
may be a machine, scarce or highly skilled labor, or a specialized tool. Observations in
industry have shown that most plants have very few bottleneck operations.
If there is no bottleneck, then excess capacity exists and the system should be changed
to create bottleneck (such as more setups or reduced capacity), which we will discuss
later.
Capacity is defined as the available time for production. This excludes maintenance
and other downtime. A non bottleneck is any resource whose capacity is greater than the
demand placed on it. A non bottleneck, therefore, should not be working constantly
because it can produce more than is needed. A non bottleneck contains idle time.
A capacity-constrained resource (CCR) is one whose utilization is close to capacity
and could be a bottleneck if it is not scheduled carefully. For example, a CCR may be
receiving work in a job-shop environment from several sources. If these source:; schedule
their flow in a way that causes occasional idle time for the CCR in excess of its unused
capacity time, the CCR becomes a bottleneck when the surge of work arrives at a later
time. This can happen if batch sizes are changed or if one of the upstream operations is
not working for some reason and does not feed enough work to the CCR.
Friends, the question that comes to mind is how do we remove these constraints?
Well, the answer is:-
We usually can not.
Therefore, we must have adequate
The figure below shows how bottleneck and non bottleneck resources should be
managed.
Resource X and Resource Yare work centers that can produce a variety of products. Each
of these work centers has 200 hours available per month. For simplicity, assume that we
are, dealing with only one product and we will alter the conditions and makeup for four
different situations. Each unit of X takes one hour of production time and the market
demand is 200 units per month. Each unit of Y takes 45 minutes of production time and
the market demand is also 200 units per month.
The figure shows a bottleneck feeding a non bottleneck. Product flows from Work
Center X to Work Center Y. X is the bottleneck because it has a capacity of 200 units
(200 hours/l hour per unit) and Y has a capacity of 267 units (200 hours/45 minutes per
unit). Because Y has to wait for X and Y has a higher capacity than X, no extra product
accumulates in the system. It all flows through to the market.
B part in the figure is the reverse of A, with Y feeding X. This is a non bottleneck
feeding a bottleneck. Because Y has a capacity of 267 units and X has a capacity of only
200 units, we should produce only 200 units of Y (75 percent of capacity) or else work-in-
process will accumulate in front of X.
C part in the figure shows that the products produced by X and Y are assembled and
then sold to the market. Because one unit from X and one unit from Y form an assembly,
X is the bottleneck with 200 unit_ of capacity and, therefore, Y should not work more
than 75 percent or else extra parts will accumulate.
D part in the figure equal quantities of product from X and Y are demanded by the
market. In this case we can call these products "finished goods" because they face
independent demands. Here Y has access to material independent of X and, with a higher
capacity than needed to satisfy the market, it can produce more products than the market
will take. However, this could create an inventory of unneeded finished goods.
A X Y B Y X
Market Market
FG
X Y
X Y (Spare parts)
Y can be used only 75% of the time X can be used only 75% of
the
or spare parts will accumulate time or Finished Goods
inventory will build up
The four situations just discussed demonstrate bottleneck and non bottleneck resources
and their relationships to production and market demand. They show that the industry
practice of using resource utilization as a measure of performance can encourage the
overuse of non bottlenecks and result in excess inventories.
Dear friends, now Let us look at the components of production cycle time
T I M E C OM P 0 N E N T S
1. Setup time- the time that a part spends waiting for a resource to be set up to work
on this same part.
3. Queue time –the time that a part waits for a resource while the resource is busy
with something else.
4. Wait time-the the time that part waits not for a resource but for another part so
that they can be assembled together.
5. Idle time-the unused time; that is, the cycle time_ less the sum of the setup time
processing time and the wait time.
For a part waiting to go through a bottleneck, queue time is the greatest. As we discuss
latter in this chapter, this is because the bottleneck has a fairly large amount of work to do
in front of it (to make sure that it is always working). For a non bottleneck, wait time is
the greatest. The part is just sitting there waiting for the arrival of other parts so that an
assembly can take place.
Schedulers are tempted to save setup times. Suppose that the batch sizes are doubled to
save half the setup times. Then, with a double batch size, all of the other times
(processing time, queue time, and wait time) increase twofold. Because these times are
doubled while saving only half of the setup time, the net -result is that the work-in-
process is approximately doubled, as is the investment in inventory.
A capacity resource profile is obtained by looking at the loads placed on each resource
by the products that are scheduled through them. In running a capacity profile, we assume
that the data are reasonably accurate, although not necessarily perfect. As an example,
consider that products have been routed through Resources Ml through M5. Suppose that
our first computation of the resource loads on each resource caused by these products
shows the following:
For this first analysis, we can disregard any resources at lower percentages because
they are non bottlenecks and should not be a problem. With this list in hand, we should
physically go to the facility and check all five operations. Note that MI, M2, 'and M3 are
overloaded; that is, they are scheduled above their capacities.
We would expect to see large quantities of inventory in front of M I. If this is not the
case, errors must exist somewhere-perhaps in the bill of materials or in the routing sheets.
Let's say that our observations and discussions with shop personnel showed that there
were errors in M1, M2, M3, and M4. We tracked them down, made the appropriate
corrections, and made the capacity profile again:
M1, M2 and M3 are still showing a lack of sufficient capacity, but M2 is the most
serious. If we now have confidence in our numbers, we use Iv'l2 as our bottleneck. If the
data contain too many errors for a reliable data analysis, it may not be worth spending
time (it could take months) nuking all the corrections.
SAVING TIME
Recall that a bottleneck is a resource whose capacity is less than the demand placed on it.
Because we focus on bottlenecks as restricting throughput (defined as sales), a
bottleneck's capacity is less than the market demand. There are a number of ways we can
save time on a bottleneck (better tooling, higher-quality labor, larger batch sizes,
reducing setup times, and so forth), but:-
Because a non bottleneck has more capacity than the system needs for its current
throughput, it already contains idle time. Implementing any measures to save more time
does not increase throughput but only serves to increase its idle time.
Finally, my dear students, we inch towards the end of today’s lecture. But before
we formally close our shops for the day, let’s focus on:-
(i) MRP uses backward scheduling after having been fed a master production schedule.
MRP schedules production through a bill of materials explosion in a backward manner-
working backward in time from the desired completion date. As a secondary procedure;
MRP, through its capacity resource planning module, develops capacity utilization
profiles of work centers. When work centers are overloaded, either the master production
schedule must be adjusted or enough slack capacity must be left unscheduled in the
system so that work can be smoothed at the local level (by work center supervisors or the
work themselves). Trying to smooth capacity using MRP is so difficult and would require
many computer runs that capacity overloads and under loads are best left to local decision
such as at the machine centers. An MRP schedule becomes invalid just days after it was
created.
The synchronous manufacturing approach uses forward scheduling because it focuses
on the critical resources. These are scheduled forward in time, ensuring that loads placed
on them are within capacity.
The no critical (or non bottleneck) resources are then scheduled to support the critical
resources. (This can be done backward to minimize the length-of time that inventories are
held.) This procedure ensures a feasible schedule. To reduce lead time and work-in-
process, in synchronous manufacturing the process batch size and transfer batch size are
varied-a procedure that MRP is not able to do.
3) JIT does not allow very much flexibility in the products produced. (Products must be
similar with a limited number of options.)
4) JIT still requires work-in-process when used with kanban so that there is "some
demand to pull." This means that completed work must be stored on the downstream side
each workstation to be pulled by the next, workstation.
5) Vendors need to be located nearby because the system depends on smaller, more
frequent deliveries.
With that, we have come to the end of today’s discussions. I hope it has been an
enriching and satisfying experience.