Unit 2

Management of Conversion System

Chapter 10: Inventory Management
Lesson 32 - Spare Parts Inventory

Learning Objectives

After reading this lesson you will be able to understand

What is spare parts inventory management
Different types of spares
Spare parts inventory management
Bar coding

Good Morning students, today we are going to introduce the concept of what
is known as Spare Parts Inventory Management. We would learn to
appreciate the importance of the concept as applicable in Production and
operations systems.
Let’s get started then.

Spare Parts Inventory Management

Spare parts management needs special treatment, somewhat different from
the inventory management of regular items. This is because the purpose of
keeping a stock of these I different – to serve as a replacement to the worn-
out parts in the machinery.
The principle that A class items need to be stocked lower than B and C class
items, will provide important guidelines to spare parts inventory control.
The inventory models we have discussed can also be applied to spare parts
control. Just as the behaviour of the consumption of spares is different from
the consumption of the regular items in inventory, so also the supply of
spares is different from the regular items. This being so, some modifications
are necessary to the conventional inventory and safety-stock models.

Spare parts can be classified for stocking policy analysis:-

1. Maintenance or breakdown spares
These are the spares which are required in large quantities at more or less
frequent periodic intervals as and when the breakdowns occur. These
resemble, somewhat, the regular inventory items in their consumption
patterns. To some extent, the analysis for stocking policies of the spares
could be similar to that of the regular items in inventory.

2. Insurance spares
The purpose of these spares is to provide an insurance against the
relatively remotely possible breakdown or failure of an
equipment/component. The probability that such a component /
equipment will survive the life-time of the machinery or plant is quite
high. The reliability of such spares has been observed to be as high as 95
to 99% over the life-span of the machinery. These spares are sparingly
needed.
But they are needed all the same because they may hold up production
resulting in considerable losses for want of them. Many of these spares
are, also, high value items. These spares are, by and large, procured along
with the capital equipments.
At the time of the purchase of the capital equipment itself a decision
regarding the purchase of the insurance spare is also made. Generally, the
decision with regard to insurance spares may be to buy either no spare or
to buy a spare.
 3. Capital spares
These are also high-reliability spares, but not as high as the insurance
spares. The reliability is not as low as that of the maintenance spares as
well. Moreover, these spares have relatively higher purchase cost than the
breakdown spares. The decision regarding these spares is usually made at
the time of purchase of the capital equipment itself. But the decision may
be to buy anywhere from 0 to say 6 or 7 spares.
4. Ratable spares
These are the reusable spare parts, which after their breakdown can be
reconditioned and re-used. Typical examples are the stepany in the car,
jet-engine in aircraft, tyre tubes in cycles, electrical motors, etc. Since
these have more than one life, the cycle of their various lives needs to be
taken into consideration in the analysis of their inventory policy.
After Spare Parts Inventory Management, we now turn our attention
to:-

Maintenance or breakdown spares

The rate of consumption or usage of spares can be derived from historical
data regarding failure of the different components in the machinery. Failure
statistics is an important basic information for this analysis. If the failure
times show a negative exponential distribution, the failure rates are
distributed by means of Poisson distribution.
If the failure times show a normal distribution due to aging or wear, then the
failure rates will also show a normal distribution. From the failure statistics
one can know the mean consumption rate of these spares and also find the
level of consumption expected with a corresponding probability of its
occurrence.
Based on service level, the inventory level can be easily arrived at. The
service level is given by the formula:

Service level = Ku / (Ku + Ko)

Where
Ku = opportunity cost of understock of one unit
Ko = opportunity cost of overstock of one unit

The following example will illustrate the calculations of the level of

maintenance spares.

Let’s consider an example now.

Example
Compute the average level of inventory given the following information:
(a) Fixed order quantity system
(b) Average consumption rate: 16 units per quarter,
(Poisson distribution appears to hold good for the consumption)
(c) Service level: 95%
(d) Lead time for procurement: 1 quarter
(e) Cost of ordering: Rs500 per order
(f) Cost of carrying: Rs18 per unit per year

You know our methodology- don’t you?

As usual, you are supposed to first work on the problem and then refer
to the solution given here.

Solution
Mean consumption rate, f... = 16 units per quarter
Standard deviation = √λ = 4

Referring to Poisson distribution tables furnished in the appendix to this

chapter, we have:
x = 23 approximately
Since the lead time is 1 quarter,
Buffer stock = 23 - 16 = 7 units

.
Now the EOQ = √ (2 (500) (64) / 18 ) = 60 approximately

Since the EOQ is very close to the annual requirement, the item may be
ordered only once in a year.

Therefore, the average inventory = (64 / 2)+ 7 = 39 units

After this, we march ahead to :-

CAPITAL SPARES
As mentioned earlier, the decision here is to buy spares ranging anywhere
from 0 to say 7 spares. These spares are bought along with the capital
equipment. The reliabilities of such spares are much higher than those of the
maintenance spares. Let us have the following notation:
C = Cost of spare parts,

Cs = Cost of shortage per unit,

S = Salvage value of the spare parts when they are salvaged,

Pi = Probability that the demand for the capital spare parts is 'i' in
number.

N = Optimal number of spares required, and

TC = Total cost for stocking N items.

The demand may be either more than the optimal number N, or less than
or equal to optimal number N. These are the two situations that are
considered which have their own associated costs. The policy of buying N
spare parts should be such that the total costs (i.e. summation for both the
aforesaid situations) are minimum.
 Situation No.1
Demand is less than or equal to N.

In this case, there is no shortage cost. There is some revenue to be expected

from salvaging the excess spares, which is:

N
S∑ Pi . (N - i)
i=O

Situation No.2
Demand is more than N.

Naturally, we shall have the shortage cost which will be:

Cs ∑ Pi.(i-N)
i=N+i

Therefore, Total cost for the policy of buying N spares would be:

= NC + Cs∑Pi( (i - N) i=N+!

N
- S∑ Pi (N - 1)
i=0

Because N is the optimal policy, we have:

TC (N* - 1) - TC (N*) >= 0

And

TC (N* + 1) - TC (N*) >= 0

The above two equations are obvious. From the former of these two, it
follows that: Since
TC (N* - 1) - TC (N*) >= 0

Therefore,

(N*-1).C - N*.C + Cs∑. Pi .(i-N*+I)

i=N*

- Cs ∑. P(i - N*) - S∑ Pi (N* - I - i)

i=N*+1 i=0

N*
- S∑. Pi (N* - i) >= 0

i=0

That is,

- C + Cs [ I - i_ Fj + S _ Ii "? 0

That is,

N*-I
- C + Cs + I. Pi[S - Cs] "? 0
i=O

That is,

N*-l I. Pi, [S - Cs] "? C - Cs

i=O

Therefore,

NLI C-C
I. p. < s
i=O I - Cs – S

Similarly, we have from TC (N* + 1) - TC (N*)"? 0:

N*
I. Pi "? Cs - C
i=O

Cs-S

Therefore, the optimal number of capital spares to be stocked shall be such

that:

NLI C-C N*
I,P< s <I,P
i=O I - Cs - S - i=O I

Let’s take another example.

The probability of the demand of spares in a textile unit is observed to be:

No. of Spares Required 0 1 2 3 4 5 6

Probability 0.05 0.15 0.30 0.25 0.15 0.05 0.05

The cost of stock-out for lack of a spare, is computed to be Rs.5,000. The

price of spares is Rs1,00 each. The salvage value of the spare part is
estimated to be almost nil. What is the optimal number of spares to be
stocked?

Solution

The optimal N* should be such that it should satisfy:

 N*-l 5000 - 1000 N*
L P; 5, 5, L P;
;=0 5000 - 0 ;=0

At N* = 4, ∑ Pi = 0.90

And
N*-l
∑Pi=0.75.
i=0

This satisfies the required condition.

Therefore, N* = 4 or 4 spares' should be stocked.

Let us see if we could have arrived at the same solution by another method.

Table 10.3 lists the cost of different stocking policies.

Expected total cost is the lowest for the stocking policy of 4 spares. This is
the same as the result v got earlier, which is to be expected since the
previous formula was a condensed version of the procedure given in Table
10.3.

Only a few small concepts are now to be considered.

Let’s do it, NOW.

INSURANCE SPARES

Since the question here is whether to procure no spare at all or one spare
along with the equipment, the analysis is fairly simple. A spare is to be
procured only if the expected value of the down-time costs exceeds (or
equals) the purchase value of the spare.

That is,

If cost of down-time (in the event of failure) x Probability of failure during

the life-time of the
equipment
 >= Cost of the spare

then, purchase a spare along with the equipment.

The probability of failure = (1 - Reliability).

Table 10.3 Costs of different stocking policies

Expected Cost of Expected Total
Stocki Probable
Stock-out Cost of Spares Costs
ng Require-
Policy ment (Rs.) (Rs.) (Rs.)
0 0 -
1 5,000 x 0.15 750
+
2 + 10,000 x 0.30
3,000
+
3 + 15,000 x 0.25 = = 13,250 Nil 13,250
3,750
+
4 + 20,000 x 0.15
3,000
+1
5 + 25,000 x 0.05
,250
+1
6 + 30,000 x 0.05
,500
0
1
2 5,000 x 0.30 1,500
+
3 + 10,000 x 0.25
2,500
+ 15,000 x 0.15 = +
4 = 8,500 1,000 9,500
2,250
+
5 + 20,000 x 0.05
1,000
+
6 + 25,000 x 0.05
1,250
2 cr---- ----- - .
1
I Y;
2 -
 1'25
3 5,000 x 0.25
0
150
4 + 10,000 x 0.15 = + 2,000 6,750
0 = 4,750
5 + 15,000 x 0.05 + 750
125
6 + 25,000 x 0.05 +
0
3 0
1
2
3
4 5,000 x 0.15 750
5 + 10,000 x 0.05 = + 500 = 2,000 3,000 5,000
6 + 15,000 x 0.05 + 750
4 0
1
2
3
4
5,000 x 0.05
5
250
6 + 10,000 x 0.05 = + 500 = 750 4,000 4,750
5 0
1
2
3
4
5
6 5,000 x 0.05 = 250 5,000 5,250
6 0
1
2
3 Nil 6,000 6,000
4
5
6

The reliability figures for such spares are generally provided by the
manufacturer.
Typical examples of insurance spares are a stand-by generator, or special
purpose gear, or a turbine rotor.

Figure 32.1 Rotable spares’ flow

ROT ABLE SPARES

Items such as a jet-engine or an electric motor can be reconditioned after

failure and put back in operation. These are called the rotable spares. (see
Figure. 10.6).
The situation can be viewed in the Queueing Theory format.

where,

Arrivals (customers) == failed equipment/components

Servers == the spares inventory

Queue discipline == first come first served

Service times == given by the distribution of times to recondition

a spare
Arrival times == given by the distribution of the times (intervals) at
which the failed equipments come for replacement

Please note that replacement takes very little time (theoretically almost
nil). But while counting the service times one must consider the time
required to recondition the failed spares as well. Reconditioning and
replacement are one system.

The down-time significantly appears when no reconditioned spare is

available and the equipment is waiting for a replacement. This means all the
'servers' (spares) are busy in this case. Assuming different values of the
number of spares to be kept, we can compute the probability of all servers
being busy in the Multi-server Queueing model. This can be used to
calculate the expected value of the down-time cost. The cost of carrying
spares should balance the expected down-time cost, for the optimal numbers
of spares.

Probability that all servers are busy in the Cost of Cost

of carrying
n-servers Queueing model x down-time = n number of spares

Here, the cost of down-time is the cost incurred whenever an equipment is

stranded for spares replacement.

If the cost of non-operation of the failed equipment is related to the length of

down-time, then we may compute the average waiting time for the ll-servers'
model and use it thus:

Average waiting time x Down-time cost per unit of time = Cost of

carrying n spares.

All of us have seen bar coding on various commodities and have a basic
idea.
Let’s build up on it further.

Bar coding
The data in a bar code is just a reference number which the computer uses to
look up associated computer disk record(s) which contain descriptive data
and other pertinent information.
For example, the bar codes found on food items at grocery stores don't
contain the price or description of the food item; instead the bar code has a
"product number" (12 digits) in it. When read by a bar code reader and
transmitted to the computer, the computer finds the disk file item record(s)
associated with that item number.

In the disk file is the price, vendor name, quantity on-hand, description, etc.
The computer does a "price lookup" by reading the bar code, and then it
creates a register of the items and adds the price to the subtotal of the
groceries purchased.

(It also subtracts the quantity from the "on-hand" total.)

Another example of bar code data might be in a quality reporting

application, the bar code may have only a single digit in it, but it may be
titled "Failed Vibration Test". The computer associates the single digit with

the test result.

So, bar codes typically have only ID data in them; the ID data is used by the
computer to look up all the pertinent detailed data associated with the ID
data.

A bar code is a series of varying width vertical lines (called bars) and spaces.
Bars and spaces together are named "elements". There are different
combinations of the bars and spaces which represent different characters.

When a bar code scanner is passed over the bar code, the light source from
the scanner is absorbed by the dark bars and not reflected, but it is reflected
by the light spaces. A photocell detector in the scanner receives the reflected
light and converts the light into an electrical signal.
As the wand is passed over the bar code, the scanner creates a low electrical
signal for the spaces (reflected light) and a high electrical signal for the bars
(nothing is reflected); the duration of the electrical signal determines wide
vs. narrow elements. This signal can be "decoded" by the bar code reader's
decoder into the characters that the bar code represents. The decoded data is
then passed to the computer in a traditional data format.

There are lots of different bar codes. Some bar codes are numeric only, (i.e.
UPC, EAN, Interleaved 2 of 5). Some bar codes are fixed length, (i.e. UPC-
A is 12 digits, UPC-E is 6 digits, EAN-13 is 13 digits, and EAN-8 is 8
digits). Some bar codes can have numbers and alphabetic characters, (i.e.
Code 93, Code 128, and Code 39). One bar code allows you to encode all
128 characters, (Code 128) and other bar codes allow you to encode a lot of
data into a small space (PDF417 and MaxiCode).

Many were invented some time ago and have been superseded by newer bar
codes. Some industries standardized on older bar codes before the better
ones had been invented, thus there is a continuing requirement for their use
in particular industries.

Variable Allowable
Bar Code Industries in use
Length Characters
Older Bar Codes
Code 11 Yes 0-9 AT&T pre 1990
Blood Banks, Cotton,
Codabar Yes 0-9,$+.:/
Transportation
Plessey Yes 0-9,A-F Shelf Labels
MSI Yes 0-9 Shelf Labels
2 of 5 Yes 0-9 UPC Shipping Container
UPC and
No 0-9 Food/Discount Store Items
EAN
Newer Bar Codes
0-9,A-Z./+-%$Spc
LOGMARS, HIBCC,
Code 39 Yes (2 character pairings
AIAG,TCIF
for Full ASCII )
Code 128 Yes Full ASCII UCC-128, EAN-128
0-9,A-Z./+-%$Spc
HIBCC Alternative, Canadian
Code 93 Yes (2 character pairings
Postal Service
for Full ASCII)
This is a “stacked” code, used
mainly by AIAG, LOGMARS
PDF 417 Yes Full ASCII
and identification card
applications.
This is a “bulls-eye” type 2-D
MaxiCode Yes Full ASCII code created and used
primarily by UPS.
Many readers have to comply with their customer's or industry's bar coding
specifications; no choice is possible, just compliance. Look at the following
samples of printed bar codes:

The classic bar code type is Code 39, (also called Code 3 of 9) which has 9
bars and spaces; three are wide, and the other 6 are narrow. In Code 39, 3 of
9 total bars and spaces are wide; hence the name, Code 3 of 9. For example,
look at the following character representations with Code 39:
Notice there are two widths of bars and two widths of spaces. If you wished
to print a bar code of ABCD, you would need to start and end it with a
special Start/Stop code character - the * (asterisk) is used for Code 39. So to
print a bar code of ABCD, it would need to be printed as *ABCD*. There
should be at least 1/4" of white space to the left and right of the code; this
helps the reader pick out where a bar code begins and ends.

Other bar code types are similarly constructed. UPC and EAN bar codes
have four widths of bars and spaces; so does Code 128.

For new bar coding projects that don't have industry or customer standards,
Code 39 is the typical non-food standard, because almost all bar code
equipment reads/prints Code 39. However, Code 39 produces relatively long
bar codes; it is not particularly efficient in bar code density, (the maximum
density is 9.4 characters per inch including 2 start/stop characters). Where
the label width is an issue and there is numeric data or lower case data, Code
128 is the best alternative; Code 128 also has an extra efficient numeric only
packing scheme to produce very dense bar codes, and Code 128 has all 128
ASCII characters. Not all readers read Code 128, so before you settle on it as
a standard, be sure that your reader is 128 capable. Only only one vendor has
promoted code 93; it requires two characters to make Full ASCII; and it
doesn't have a numeric packing option. For these reasons, Code 128 is
preferable over Code 93.

The larger the width of the elements, the more space it takes to print the bar
code; therefore, the lower the bar code density. The thinner the bar and
spaces, the less space is required and the higher the bar code density. Look
at the samples below of different densities:

Lower density bar codes are more reliably printed and more consistently
read than higher density bar codes, because minor variations (due to printing
or damage) are much more serious with high density bar codes - the
percentage of distortion is larger.

There are three basic types of bar code readers: -

Fixed,

Portable batch, and

Portable RF.

Fixed readers remain attached to their host computer and terminal and
transmit one data item at a time as the data is scanned. Portable batch readers
are battery operated and store data into memory for later batch transfer to a
host computer.

Some advanced portable readers can operate in non-portable mode too, often
eliminating the need for a separate fixed reader. Portable RF Readers are
battery operated and transmit data real-time, on-line. More importantly, the
real-time, two-way communication allows the host to instruct the operator
what to do next based on what just happened.

A basic bar code reader consists of a decoder and a scanner, (a cable is also
required to interface the decoder to the computer or terminal). The basic
operation of a scanner is to scan a bar code symbol and provide an electrical
output that corresponds to the bars and spaces of a bar code.

A decoder is usually a separate box which takes the digitized bar space
patterns, decodes them to the correct data, and transmits the data to the
computer over wires or wireless, immediately or on a batch basis. here is a
mystique surrounding bar codes which intimidates many people. Let's
eliminate it quickly.

First the bar code usually doesn't contain descriptive data, (just like your
social security number or car's license plate number doesn't have anything
about your name or where you live).

With that, we have come to the end of today’s discussions. I hope it has
been an enriching and satisfying experience.