Analysis and Management

of Production System
Lesson 4: Variability of processing time

Prof. Giulia Bruno

Department of Management and

Production Engineering

[email protected]
Introduction

• Approximation of the cycle time in a queue system

consists of four parameters
• coefficient of variation of the inter-arrival time process
(Ca2)
• squared coefficient of variation of the service time
process (Cs2)
• machine utilization (u)
• mean service time (E[Ts])
Introduction

● Reducing one of the variability components, Ca2 or Cs2,

will reduce the cycle time in the queue

● Example: consider a single machine system with the

following values Ca2 = 1, Cs2, = 1, u = 0.8, E[Ts] = 2 hr
● Thus the cycle time in the queue CTq is
Introduction

● If Cs2 is reduced by 10% to 0.9, the resulting cycle time is 7.6

hours (a reduction of 5%)
● If C is not changed, to accomplish the same reduction in cycle time
would take a reduction in machine utilization from 0.8 to 0.7917

● A 50% reduction in the service time variability would reduce the

cycle time measure to 6 hours
● The equivalent machine utilization factor for 6 hours given the
original system parameters is 0.75
● Either of these changes would result in a cycle time in the queue of
6 hours which is a 25% reduction from the original 8 hours
Introduction

● Reducing variability equivalent to reducing the machine

utilization
 a reduction of utilization with a constant arrival rate implies a
reduction of mean processing time (i.e., an increase in the mean
processing rate)
● Very important to concentrate on reducing variability for the
inter-arrival and service time processes since these reductions
are like finding “free” machine capacity
Main contributors to processing time variability

● Natural processing time variability - the variability evident in the

time it takes to actually process a specific job type
● Random breakdowns and repairs during processing - the
variability of the time between breakdowns and the variability of
the time to repair a broken machine
● Operator unavailability - can induce random delays in the time
a job spends “in control of” a machine (machine and job are
available, but the operator is busy serving another machine/job)
● Job class setup and take-down times - the time caused by a
job-type change on a machine (change-over time generally
occurs at the end of processing of one job type and the starting
of a different job class)
Natural process time variability

● Consider a job with processing time random variable, T, with

known mean and variance parameters E[T] and V[T], respectively
● If the processing time is made up of several separate tasks, it is
possible to reduce the total processing time variability by reducing
the variability of the individual tasks
 sub-tasks that can be studied separately or possibly assigned to
different workers for manual task operations
Natural process time variability
Example:
Consider that the natural processing time random variable T is made
up of three separate (independent) sub-tasks

Additionally consider that these three sub-processes times are

independent and identically distributed random variables so that
Natural process time variability

The individual processing time random variables Ti, for i = 1, 2, 3,

have distributional parameters

the squared coefficient of variation of the individual tasks are

If the total processing time is made up of three identical sub-tasks,

then the squared coefficient of variation of the individual tasks is
actually three times that of the total time squared coefficient of
variation
Example

● Consider a natural processing time that is exponentially distributed

with a mean time of 3 hours
 the squared coefficient of variationC2[T] is equal to 1
● Assume that this job consists of three distinct but identically
distributed sub-tasks
 these sub-tasks have processing times random variables Ti that have
distributional parameters E[Ti] = 1 and V[Ti] = 3, for each I
● After further study of the three sub-tasks, it is found that the
variability of each task can be reduced and the resulting times are
i.i.d. exponentially distributed times each with a mean of one hour
(it is assumed that these variabilities can be reduced while the
mean processing times remain unchanged)
 C2[Ti] = 1, for each sub-task i
Example

● The total processing time random variable now has parameters

● Total processing time variability reduced to 1/3 of its original value

● This reduction will in turn reduce the associated workstation cycle
time in the queue by 1/6
 extra processing capability has been found
 this new system is equivalent in cycle time response to a system with a faster
processing time
Exercise

● Consider a processing time, T, with measured

parameters E[T]= 6 and C2[T] =2, that has 4 i.i.d. sub-
tasks
 Determine E[Ti] and C2[Ti] for the sub-tasks

● Assume that the variability of each sub-task can be

reduced (identically) so that C2[Ti] = 2
 Determine the squared coefficient of variation of the total
processing time and the percentage improvement over the
“old” processing time variability
Solution
Machine breakdown

● Major source of processing time variability is due to the breakdown

of an operating machine and the subsequent delay while the
machine is being repaired
 job might not be recoverable (i.e., lost)
 job might require additional processing before resumption of “normal”
processing
 job might not be effected by the breakdown and normal processing
can resume immediately after the repair is complete (as if the
breakdown never occurred)
● Only the latter case is considered
 although for the second case, the additional processing time needed to
resume service can be included in the machine repair time so that the second
and third situations become equivalent
Effective processing time

● The effective processing time, Te, refers to the time that a job has
control of the processor until the time at which the job releases the
processor so that it is available to begin work on another job
● The job can complete processing without a breakdown interruption,
the machine could breakdown once during service, the machine
could breakdown twice during service, etc.
● So the effective processing time is also a random variable given by

where T is the normal (uninterrupted) processing time random

variable, the Ri’s are the (i.i.d.) repair time random variables, and N is
the random number of failures during the service time T
Availability

● The availability, a, of a processor that is subject to failures is the

long-run average fraction of time that the processor is available for
processing jobs

where F1,F2,… and R1,R2,… are random variables representing

successive failure times and successive repair times, respectively, for
the processor
Mean and variance of effective time

Hopp and Spearman developed an expression for the mean and

variance of the effective service time for processors that are less
than 100% reliable under the assumption that failures are
exponentially distributed:
Example

Consider a single workstation with jobs arriving according to a

Poisson process (i.e., exponential inter-arrival times) with an
average time between arrivals of 75 minutes
Initially we ignore the fact that the machine at the workstation is not
100% reliable and observe that the normal processing time is
described by an Erlang type-3 distribution with mean of 58 minutes

Thus:
E[Ta] = 75 min and Ca² = 1
E[Ts] = 58 min and Cs² = 1/3
u = 58/75 = 0.7733
CTq = 132 min
Example

Now let’s consider that the time between machine breakdowns is

exponentially distributed with a mean time of 3 hours, and the repair
time is distributed according to a lognormal distribution with a mean
time of 30 min and a standard deviation of 15 min
Considerations

● It can be noticed that the inclusion of machine failure in the

model results in over a threefold increase in the mean waiting
time
 ignoring failures can create significant errors in performance measures
● This increase is due to two factors
1) machine failures cause an increase in the effective utilization factor
2) machine failures cause an increase in the service variability
● As the utilization factor approaches one, small changes in the
factor will have major changes in waiting times, and in this case,
the majority of the increase in waiting times is due to the
utilization factor increase
 only about 5%–6% of the increase is due to the increase in service variability
Example 2

Let’s consider a line consisting of 2 machines (M1, M2) with mean

process time E[Ts] = 15 min and natural standard deviation equal to
𝜎𝑠 =3,35 min
Both machines undergoes failures which limit their availability (but in
different ways according to the following table)
Which machine is to be preferred?

Machine Failure type E[F] E[R] C[R]

Long and
M1 744 min 248 min 1
infrequent

Short and
M2 114 min 38 min 1
frequent
Example 2

• Both the machines have the same availability, and thus the same
effective processing time:
a= 0.75 E[Te] = E[Ts]/a = 15/0.75 = 20 min

• Effective capacity of the machines:

μe= 1/E[Te]= 3 item/hr

• The two machines seem to be equivalent by using the mean

values of effective processing time, but we also need to consider
the variability:

M1:
M2:
Exercise 1

● Consider a job with processing time distribution

parameters E[Ts] = 3 hours and Cs2 = 2
● The machine breakdown and repair time
characteristic parameters are:
 E[F] = 7 hr and C2[F] = 1
 E[R] = 1 hr and C2[R] = 1
● Find the parameters of the effective processing
time: E[Te], V[Te], and C2[Te]
Solution
Exercise 2

● Consider an M/M/1/3 system with a server that has

exponential time between breakdowns ( rate) and
exponential repair times ( rate)
 Develop the state diagram of the system
● Consider an arrival rate of 5 jobs per hour, a service
rate of 4 jobs per hour, a breakdown rate of once per
hour, and a mean repair time of 10 minutes
 Determine the steady-state probabilities for the system states
 Compute the system performance measures of WIPs, CTs, and
ths
 Compute the proportion of the time that the machine is idle,
down (i.e., under repair), and processing
Solution

P0 0.14
P1P 0.17
P1B 0.02
P2P 0.23
P2B 0.03
P3P 0.33
P3B 0.08

● Ths = 5*(1-(0.33+0.08)) = 2.95 j/h

● WIPs= (0.17+0.02)+2*(0.23+0.03)+3*(0.33+0.08) = 1.94
● CTs = WIP/Th = 0.66 h

● Idle: P0 = 0.14
● Processing: P1P+P2P+P3P = 0.17+0.23+0.33 = 0.73
● Down: P1B+P2B+P3B = 0.02+0.03+0.08 = 0.13
Exercise 3

● Consider M/M/1/∞ system with a server that has

exponential time between breakdowns and exponential
repair times
● Given an arrival rate of 5 jobs per hour, a service rate of
7 jobs per hour, a breakdown rate of once per hour and
a mean repair time of 15 minutes, compute the system
performance measures of WIPs, CTs, and ths
Solution

= 1/(1+0.25) = 0.8

0.143/0.8=0.18 h

1 + [(1+1)*0.8*(1-0.8)*0.25]/0.143 =1.56

5*0.18 = 0.9

[(1+1.56)/2] * [0.9/(1-0.9)] * 0.18 = 2.07

CT = CTq + E[Te] = 2.07 + 0.18 = 2.25 h

WIP = TH*CT = 5 * 2.25 = 11.25