AIIE Transactions

ISSN: 0569-5554 (Print) (Online) Journal homepage: www.tandfonline.com/journals/uiie19

Reliable Preventive Maintenance Scheduling

Mazhar Ali Khan Malik

To cite this article: Mazhar Ali Khan Malik (1979) Reliable Preventive Maintenance Scheduling,
AIIE Transactions, 11:3, 221-228, DOI: 10.1080/05695557908974463
To link to this article: https://doi.org/10.1080/05695557908974463

Published online: 06 Jul 2007.

Submit your article to this journal

Article views: 721

View related articles

Citing articles: 50 View citing articles

Full Terms & Conditions of access and use can be found at

https://www.tandfonline.com/action/journalInformation?journalCode=uiie21
 Reliable
Preventive Maintenance
Scheduling

MAZHAR ALI KHAN MALIK

SENIORMEMBER, AIIE
Arab Development Institute
PO Box 8004, Tripoli, Libya

Abstract: Maintenance of goods producing systems is undertaken on the principle of minimum cost
whereas the maintenance of service producing systems is done on the principle of operational reliability.
High pressure boilers, elevator ropes, aeroplane engines, subway tunnels, suspension bridges, air condition-
ing and transportation networks are examples of such systems. In this paper, reliability models are built
for a service producing system which works intermittently, is subject t o wear, and can be improved
through maintenance actions like cleaning, lubrication, realignment, etc.-short of replacement. Finally
an application is shown and maintenance scheduling is developed through an example.

System Improvement through preventive maintenance is acceptable levels ot reliability-irrespective ot cost. 'l'he
as old as the proverb, "A stitch in time saves nine." Yet service producing systems are further sub-divided into
there is more isolation between the practitioners of mainte- standby systems which are subject to chance failures only,
nance and the researchers than in any other professional and working systems which are subject to wear-out failures.
activity. Practitioners use 'common sense' in developing The maintenance of standby systems has been demonstrated
maintenance scheduling and researchers imply assumptions previously [13]. This paper deals with working systems and
that do not exist in real-world situations. Often, the word complements the earlier work for goods producing systems
maintenance is used for replacement and 'system' is confused and standby service producing systems. Constantly working
with components. Wide gaps exist in treating the realities of is, of course, only a special case of intermittently working.
equipment wear, intermittent working, chance failures, and Table 1 summarizes different maintenance criteria for
practical effects of a maintenance-action upon equipment different functions.
reliability. It is hoped that this paper will contribute towards
bridging some of these gaps.
Table 1: MaintenanceCriteria.
A system may produce goods or services. Maintenance of
Systems Goods Service
goods producing systems is undertaken on the principle of Producing Producing
minimizing total cost, as shown by Malik [ l 3 ] , where the
total cost includes the repair cost and expected production Maintenance Minimum Cost Acceptable level of
Criteria (1) Reliability
losses. However when systems produce services,the "revenuen
of the services produced cannot be realistically calculated. TYpe Standby Working
Further difficulties arise if such failures endanger human life.
Failure Modes Chance Wear-out
Here, systems should be maintained at predetermined and
(2) (3)

(1): Methodology in 191

(2): Methodology in [ I 3 1
Received June 1978; revised April 1979. Paper was handled by (3) This paper.
Applied Probability and Statistics Department.

September 1979, AIIE T RANSACTIONS 0569-5554/79/0900-0221/$02.00/0 @ 1979 AIIE 22 1

 good as new.' Under wear-out failures, a m'aintenance is
not likely to restore a system to its original condition.

Improvement through Maintenance

We shall now study the effect of "improvement through

preventive maintenance, overhaul or repair" under the
assumption that maintenance will definitely enhance the
equipment condition but it will not restore it to the original
Fig. 1 Extreme Assumptions. state short of replacement. Here are a few practical examples
where the wear effect can in fact be lessened through main-
tenance :
Role of Maintenance
Present Practices a) When a pump shaft and packing are worn and fluid in
the pump leaks, the maintenance is to readjust or
Present assumptions in maintenance practices fall under four tighten the packing glands; it stops the leak.
extreme assumptions, as given in Fig. 1.
The first assumption, A, is that replacement is the only b) The excessive clearance in crankshaft bearings is reduced
maintenance. Even if any other maintenance is done, it does through maintenance; here maintenance is the scraping
not enhance the survival probability. This is the assumption of the bearing surface to make it smooth and then remove
followed by Flehinger [S, 61, Malik [ l 1; pp. 27-46], and one of the shims (which act as spacers in between the
Pierushka [16, pp. 142, 149, 153, 1581. two split pieces of the bearing). The bearing becomes
The second, B, is the assumption taken by arather large "like" new.
number of authors, among them Barlow and Proschan [3],
Derman [4], McCall [15], and Wagner [17], that a repair c) Worn-out shafts can be metalized to make them "like"
can reinstate a machine to its new condition; it will only new.
cost more.
The third, C, is the assumption of the so-called 'bad as d) Cleaning of Diesel fuel injectors improves performance.
old' model of Ascher [ I ] that a repair just restores the Adjustment through a spring restores them to their
system to work. The reliability has not been improved by "near" original state.
maintenance; the maintenance has only restored the system
to its pre-breakdown reliability.
The fourth, D, is the assumption that a system becomes System, Reliability, Improvements and Assumptions
new after every maintenance. This is adopted by Barlow Proposed System
and Hunter [2], Gavett [7, p. 3401, Malik [13], and
Morse [15, pp. 157-741. A system is defined as a logical gathering o f many sub-
The difference between Assumptions B and D is that systems or units where the unit itself is a logical combina-
under B, a system can be improved up to its complete new tion of many components. Figure 2 and Fig. 3 show these
condition depending upon the amount of money spent on combinations schematically and mechanically respectively.
improvement whereas under D, every maintenance enhances Fig. 3 shows a system to pump water which consists of
a system to its new condition. four subsystems or units. When the water level rises, the
According to Ascher [I], out of the two assumptions float triggers the electrical motor which in turn drives the
'good as new' (Assumption D) and 'bad as old' (Assumption pump. The pipeline carries the water to some destination.
C), 'bad as old' is a better one. This is true but Assumption This pump house exists at an international airport whose
A is still an improvement over C, that is, the reliability of runways are below sea level, thus necessitating the use of
the sub-systems not replaced is as 'bad as old' but those sub- the pump to discharge the storm as well as regular drainage
systems which have been replaced have higher reliability. water. If the system failed during a rain storm and the air-
The plausible assumption may be somewhere in the port was flooded, the airplane landing would be unsafe. The
middle of the above four assumptions. For example, a diversion to other airports resulted in a loss of revenue, bad
maintenance must enhance the equipment reliability some- publicity, and was equally dangerous in storm situations. It
what; otherwise what would be the purpose of the mainte- is therefore required that this system be maintained at an
nance. Therefore, Assumption A, even though an improve- acceptable level of reliability, irrespective of cost.
ment over 'C,' is not a correct representation. Similarly, As far as the improvement in reliability is concerned, the
complete renewal is possible through replacement only. system reliability can be improved only by improving the
Thus 'B' is not correct. Assumption D would seem all right unit reliability. The unit reliability, in turn, can be improved
if a system were subject to memoryless failures such as under by dismantling the unit, cleaning all the components, lubri-
the exponential distribution when any unfailed system is 'as cating them, and perhaps by replacing some components. In

222 AIIE TRANSACTIONS, Volume 1 1, No. 3

 Where r(t) = Pr. (Unit Operating at time t ) , h(t) has the usual
meaning [8,pp. 135-361. The development of (3) and (4)
is shown in the appendix.
In the reliability Equation (3), we are saying that the
'system7 wears only when it works but it is subject to fail
by chance all the time. Overall the systems would fail as a
components result of both chance and wear-out. Consider a car tire, it
can fail by wear-out. Also, a nail can pierce through the tire
which is a chance failure. But, overall a nail has less likeli-
hood to pierce through a new tire and more likelihood to
pierce through an old one. Similarly a young man may
Fig. 2. Concept of a system. sustain an accident whereas an old man might 'fail.' Thus
the overall "Failure" is a result of "compoun8' action of
this paper, we shall show the improvement in unit reliability chance and wear-out. Now, if the system does not work and
only-because, as shown later, the system reliability is then A is equal to zero, the system returns to simple chance
only a product of unit reliabilities. Researchers have gener- failure-as given by exponential distribution, and r(t) is
ally confused components with system [S, 16, 171. Obvi- given by
ously, a heap of components does not constitute a system. exp (-P t ) .

If on the other hand, the system works full time and h is

equal to one, r(t) is given by

exp [-fit - 8 t q ]

which is termed as compound failure by Malik [ 1 2 ] . Thus

standby systems and full time working systems are only
special cases of intermittenth working systems.
The r(t) of (3) can also be written as

Fig. 3. Simplified pump house

rj(ti)= exp [ -,ujti-6i(hjti)vj] (5)

where rj(ti)is the reliability of the jth unit at the end of the
ith maintenance point; j = 1 to 4; i = 1 to n .
Now suppose we require that
System Reliability

Let the hazard function of an intermittently working unit

be expressed as where a2 is a predetermined constant. Then we will under-
h ( t ) = a + bt take maintenance actions on Unit j whenever its reliability
falls to a2 .
where 'a' is a constant factor, perhaps due to environments, Again, let the System reliability be equal to R(ti) where
and 'b' is a wear-out factor [ 1 2 ] .To bring the non-linearity
effect, we change h(t) to

where c is more than one. Now, if we suppose that equip- Then R(ti) will also have to be more than a predetermined
ment wears only when it works, we have level of reliability, say a l , for the whole system. Symbols
are reclarified in the Appendix.
h ( t )= a + b(ht)c (2)

where h is the fraction of time, a unit works. This then I mprovement Factor
gives [ l o , pp. 38-39]
Now suppose that when rj(t)= a2,preventive maintenance
r(t) = exp [-pt - 8 (X t)2 ] (3) is done on this unit which could be just an overhaul, lubri-
where cating and greasing or even the replacement of those compo-
nents which were thought to break down soon. We repeat
a = p, c+ 1 = q, b/(c+l)~=d-,and r (t) = f (t)/h(t) (4) that many components connected in logical fashion make

September 1979, AIIE TRANSACTIONS

what we call a unit (or a subsystem) and many units nance; maintenance time is assumed to be negligible. Also,
connected logically form a system, as shown in Fig. 2. let Oji be the improvement factor in unit j at the end of the
When maintenance has improved the unit, its survival ith interval. Then
probability is enhanced. One way of interpreting this irn-
provement is to think that the t years old item is no longer
that old and its post-maintenance age is reduced from t to
t/O, where P varies between one and infinity, and premainte-
nance reliability rj(t) has become rj(t/P) after the mainte-
nance. Again, during the second interval, t l < t < tz, the
If 0 = 1, maintenance has not improved equipment and reliability of unit j drops to a2 at t = t2. Then the reliability
it is like Flehinger's, Malik's and Pierushka's Assumption A immediately before maintenance:
in Section 1. When 0 = .q t/O -t 0; this is exactly equivalent
to Assumption D and is also equivalent to Assumption B
when the additional assumption, that a maintenance can
renew an item, is adopted.
Thus, just by the addition of one parameter P, it is
possible to have complete generality in maintenance im- Now when maintenance is done, the age t2-tl during the
provement, from none to full 'renewal.' second interval is reduced to (t2-tl)lPj,Z. The improvement
by maintenance does not affect tl/Oj, l , which is the portion
of age permanently consumed as a result of wear and envir-
Assumptions onmental effects during the first interval. Then the reliability
immediately after maintenance at the end of the second
Here we shall assume that a maintenance expert is able to interval:
estimate how much improvement has been done on Unit j as
a result of the maintenance done at the end of ith interval. rj(f2,~)=eXp{-&{tl/&,~+(t2-t~YOj,~}- (10)
Here reliability of a unit is assumed to be enhanced without
a replacement. In our approach, P is necessarily more than
one and less than infinity. Think of a Diesel engine fuel
injector. Suppose it is dismantled and cleaned after 300
hours of operation. It is tested for atomization pressure; if
the pressure is low, it is adjusted. This injector is no longer
like the uncleaned one which was additionally working at Substitution of (10) in (6) will give the system reliability
sub-atomization pressure. At the same time, it cannot be immediately after the end of the second interval. Continu-
like a new one. In 300 hours of working, for a four stroke ing the same way, we obtain
Diesel engine working at 600 R.P.M., the injector's needle
has reciprocated 5.4 million times; this wear cannot be ~ ~ ( t 3 , 0 ) = e ~ p [ - ~ g 3 , o - e j ( ~ j g 3 , o ) ~ j I(11)
taken away. After cleaning and readjustment, this 300 hour
old injector may be like another uncleaned and unmaintained where
injector of 10 hours, 50 hours or 100 hours, etc. age. If P = 5,
the maintenance has reduced the age from 300 hours before + t3-t2
g3,0 = tl/Pj,l +(f~-tl)lP~,2
maintenance to 60 hours post maintenance.
and

Models and Scheduling ri(t3,1) = exP [-pjg3, - 8j ( ~g3,

j )qi] (1 2)

Improvement Models where

According to the second section, when rj(t reaches a*, we g3, I = tl lPj, 1 + (f2-f 1)lOj,2 + (t3-t2)/fij,3 .
maintain this unit. The maintenance reduces the calendar
age tl to effective age tl I&,where Pj is the improvement We can now find rj(ti,o) and rj(ti, ,), the reliabilities of
unit j at ti immediately before and after maintenance. Thus
factor for unit j. Note that for a full time working device,
= 5 reduces "tl = 300" to "tl = 60." Similarly for the
intermittently working device, 150 hours of net working -
rj(ti, o ) = exp [ lyg,, o - ej(Afii, o ))"'I (1 3)

(Aj = 0.5) will be reduced to 30 hours. Thus Aj and Pj cover where

the entire space of various possibilities.
Now let rj(ti,o) be the reliability of the jth unit at the gi,O = tl/Pj,l +(t?-tl)/@,2 +(t3-t2)/Pj,3 + " '

end of the ith interval immediately before the maintenance,

and rj(ti, l ) be the reliability immediately after the mainte-
.... + ti-ti-l
AIIE TRANSACTIONS,
Volume 11, No. 3
 = ti-1 /Pi + ti - ti- 1 . (20)
Similarly and
rj(ti,1 ) = exp [-pjgi,1 - ej(Ajgi,1 )"I (14) I
tl + t2-tl + - . . ti- 1 -ti-2 ti - ti-
+- --
1
gi, I =
where Pi pi (21)
gi,l =t~fOj,l + (t2-tl )/Pj,2 + ' ' ' (ti-ti- 1 )l&,i

= C [(ti - ti- 114,il .

1
i =l Example

Again (13) and (14) Lan be substituted in the system We shall now apply the theory on just one unit of the system
reliability model, (6), to find expressions for system reli- given in Fig. 3. Let us take Unit 3, the pump itself. Table 2
ability immediately before maintenance (which is required gives the values of the parameters of (18).We require that its
to be more than al for all i's) and immediately after mainte- reliability should be equal to or more than 0.75. After the
nance. unit starts working, the reliability falls from 1.00 at to = 0 ,
to 0.75 at tl = 4.17 years. Maintenance is performed at this
I
point. As a result, the reliability is enhanced from r3(tl ,,),
Maintenance Scheduling given by (18 ) to r3(tl ,l ), given by (19). The unit is allowed
to work again till at t2 = 6.67 years, the reliability falls to
The equation rj(ti,o )= a 2 ,where rj(ti,,) is given by (13), can 0.75. Here, it is again overhauled. Reliability at t2 can be
be solved for i = 1 to n by successive numerical approxima- calculated by putting i = 2 in (18) and (19) for immediately
tion. Then before and immediately after maintenance, respectively. We
can thus continue calculating all the maintenance points.
Table 3 gives the maintenance scheduling till tlo which was
Asveloped on a digital computer.
where tfi)is the first interval for the jth unit.
When tfi)is known, we can find t y ) from (6-4), that is:
Table 2: Values of parameters

M e h rl P
1/15 0.067 0.5 2 2.5
which means t2 is inverse of r(o12) when tl is known.
Similarly we can find t3, t4,. . . , tn for the jth unit
successively where Figure 4 shows the respective reliability deteriorations
and improvements as a result of working as well as of
maintenance performed according to the schedule of Table
3.

The Case of Equal Improvement in All Internals

Suppose that the improvement in maintenance is the same

for the jth unit for all the intervals, that is,

p.,J l =p.,J z = . - . .p.;=

J I
..a.
Pj, n = P j , say -
Then

o - &(Ajgi,0)'
rj(ti,0 ) = exp [-~jg;, I (18 )
and
3'
-
rj(ti,1) - ~ X [-pjg;,
P i - ej(Ajg;,1 )qj] (19)
Fig. 4. Reliable Preventive Maintenance Scheduling Deteriorations
where and Improvements in Reliability.

September 1979, AlIE TRANSACTIONS

 Table 3: Reliable preventive maintenance
scheduling for a, = 0.75.
which give the reliability expressions in the case of com-
bined failures.
Maintenance Schedule from Maintenance Schedule from Finally, let Aj = 1, and pj = 0 in (18) and (19). We will
points Zero point points Zero point then have
i ti i ti

1 4.1 7 years 6 9.92 years

7
and ti
2 6.67 years 10.12 years
3 8.17 years 8 10.24 years
rj(ti, ) = exp [-ej (-1 "I
4 9.1 7 years 9 10.32 years
0j
5 9.59 years 10 10.34 years
I which give the reliability expressions in the case of Weibull
failures.
From the above, it also follows that Compound and
Similar maintenance scheduling can be done for the Weibull failures-which depict constantly working systems-
whole system of Fig. 3 by substituting (18) in (6). are special cases of Compositive and Combined failures,
respectively, when A is equal to one. Similarly, for A equal
to zero, exponential distribution is a special case of Composi-
Application to Other Failures and Conclusions tive distribution. Researchers have often assumed constant
working [5, p. 4; 16, pp. 149-165; 17, p. 731 even though
Other Failures constantly working systems rarely exist in industry.
Thus the theory of reliability improvement through main-
We have demonstrated reliability improvement on the tenance can be applied with respect to all failure modes,
Compositive failure model [12] . We now show that (13), and the maintenance scheduling can be developed by sub-
(14), (15) and (16) are readily adaptable to other failure stituting any of the (22) - (23), (24) - (25), or (26) - (27) in
modes, namely, the so-called Compound [12], Combined place of (1 8) - (19).
[lo] , and Weibull failures. Table 4 explains the inter-
relationship of these failure modes. Respective density
functions are summarized in the Appendix. Conclusions

i) There is no need to adopt unrealistic assumptions and

even minor maintenance adjustments done by field
mechanics can be arrested within mathematical model-
ing.

ii) Distinctly different failure rates should be adopted for

working and non-working periods when systems work
intermittently. These different rates should then be
"compounded" to arrive at the overall failure rates.

iii) The final point is highlighted by Fig. 4. When systems

wear, we shall have to resort to successively decreasing
maintenance intervals if we wish to maintain our
Let hj = 1, then from (18) and (19), we obtain systems above given levels of operational reliability.
The constant interval maintenance yields continuously
deteriorating levels o f operational reliability.

where g,!,o is given in (20). Similarly Acknowledgment

The major part of this material is from author's thesis

"Optimum Capital Planning of Reliable Systems" written
Equations (22) and (23) give the reliability expressions under the guidance of Professor Leon Herbach, at Poly-
in the case of compound failures. technic Institute of New York, for a PhD degree from New
Again, let pj = 0,'then from (18) and (19), we obtain York University. Field data from Port Authority of New
York & New Jersey, and assistance from Mr. Iqbal Ahmad
(in the preparation of this paper) are acknowledged. The
and author is thankful to the department editor and referees
ti ~j
rj(ti,1 ) = exp [-6j(hj -) I for suggestions that resulted in the improvement of this
Pi paper.

226 AIIE TRANSACTIONS, Volume 11, No. 3

Appendix
Density Functions under Various Failure Phenomena

.
Title Density Function f (t) Title Density Function f(t)
Exponential General = [ql~1h~l-1t~l-1+02(1-h)q~-1~2tq2-1]
dist. = P exp(-~t) dist.
(1) - exp [-01~~~-1t~~-02(l-h)q~-'tq2]
Weibull Compound
dist. (1) vOtq-' exp(-0 tq) dist. (2) (p + qOtq-l) exp(-pt - d rq)
--
Combined [p(l-A) + 170 hq-' tq-'1 Cornpositive ,-,,
t,-e 't q(p*8
hq-
dist. (1) dist. (2) A"-' tq-' 1
exp [- (1 -h)pt - 0hq-' tq]

(1): Refer to [lo] for density function derivation.

(2): Refer to [12] for density function derivation.

Glossary of Symbols Used

Latin Greek
a Constant failure rate a, Required system reliability
b Variable failure rate Require subsystem or unit reliability
a2
c Intensity of wear 0 Maintenance improvement factor
fct) Failure density Shape parameter of Weibull distribution
17
h(t) Hazard function 8 Scale parameter of Weibull distribution
i Subscript for maintenance points h Proportion of time a device works
j Subscript for subsystems or units 6 Transformed scale parameter, 6 = 0/h
n Total number of maintenance points p Exponential distribution parameter
r(t) Reliability of a subsystem or unit II Multiplication symbol.
R(t) System reliability Z Summation symbol.
Development of Reliability Function
References
From Lloyd and Lipow [a, pp. 135-361, we have
[I] E. Harold Ascher, "Evaluation of System Reliability Using
'Bad-as-Old' Concept," ZEEE (Institute of Electrical and
f(t) - f(t>
h (t) = ------ - - Electronic Engineers), Transactionson Reliability, R-17, 103-
1 - F(t) r(t) 110 (1968).
[2] Richard Barlow and Larry Hunter, "Optimum Preventive
then Maintenance Policies," OperationsResearch, 8,90-100 (1960).
[3] Richard Barlow and F. Proschan, "Availability Theory for
Multicomponents Systems," AFOSR Technical Report No. 4,
Department of Statistics, Florida State University, Tallahassee
(1972).
[4] C. Derman, "On Optimal Replacement Rules when Changes
of State are Markovian," kfathematical Optimization Tech-
niques, 201-210, University of California Press, Berkeley,
California (1963).
[S] Betty J. Flehinger, "A General Model for the Reliability
Analysis of Systems under Various Preventive Maintenance
Policies," PhD Thesis, Columbia University, New York(1960).
[6] Betty J. Flehinger, "System Reliability as a Function of
System Age, Effects of Intermittent Component Usage and
Periodic Maintenance," Operations Research, 8,3344 (1960).
by substituting from Eq. (4). [7] William J. Gavett, Production and Operations Management.
(A-5) is then the same as Eq. (3). Harcourt, Brace &World, Inc., New York (1968).

September 1979, AIIE TRANSACTIONS

David K. Lloyd and Myron Lipow, Reliability: Management, [16] Erich Pierushka, Principles of Reliability, Prentice-Hall, Inc.
Methods, and Mathematics, Prentice-Hall, Inc., Englewood Englewood Cliffs, New Jersey, (1963).
Cliffs, New Jersey, (1962). [17] John V. Wagner, Jr., "Maintenance Models for Stochastically
Mazhar Ali Khan Malik, "Optimal Preventive Maintenance Failing Equipment," Technical Report No. 153, Stanford
Scheduling" ASME (American Society o f Mechanical Engi- University Operations Research and Statistics Department,
neers), Paper No. 71-PEM-4 (1971). Stanford, California (1973).
Mazhar Ali Khan Malik, "A Note on the Physical Meaning of
Weibull Distribution," IEEE Transactions on Reliability,
R-24, 1 (1975).
Mazhar Ali Khan Malik, "Optimum Capital Planning of Reli-
able Systems," PhD Thesis, New York University (1975).
Mazhar Ali Khan Malik, "Average Life of Equipment subject Dr. Mazhar Ali Khan Malik is the chairman of Industrial Engineering
toCompr . ~ r d Failures:' ASME Paper No. 76-PEM-A (1976). and Planning at the Arab Development Institute, Tripoli, Libya. He
(Repubhsl.ed ASME Transactions, Journal of Engineering for has a BS from Lahore, Pakistan. He completed his MS and PhD from
Industry, 99, Series B, Number 2 (1977.) New York University. Malik has twenty years experience as engineer,
Mazhar Ali Khan Malik, "Efficient Management of Critically economist and professor in various developing and developed
Important Systems," 1977 Canadian Reliability SRE Sympo- countries of Asia, Africa and Europe which includes seven years
sium, Ottawa, Canada. (Republished Microelectronics and Re- work with New York Port Authority. A Senior Member of AIIE,
liability, 17, 173-178, Oxford, U.K., January 1978.) Malik is also a member of ASME, IEEE, ORSA, TIMS, AEA, British
John J. McCall, "Maintenance Policies for Stochastically Fail- ECO.Society, Australian Industrial Engg. Inst. and Pakistan Soc. for
ing Equipment," Management Science, 11,493-521 (1965). Advancement of Science. In addition to research papers, Malik is a
Philip M. Morse, Queues, Inventories and Maintenance, John frequent contributor on topics of economic and industrial planning
Wiley & Sons, Inc., New York (1958). in developing countries.

AIIE TRANSACTIONS, Volume 11, No. 3