A Decision-Theoretic Perspective of Maintenance Optimization of a

Complex System
Fahad Izhar Khan, University of Arizona
Allan T. Mense, University of Arizona

Key Words: Maintenance Optimization, Decision Theory, Risk Analysis

SUMMARY safety of people managing these systems. The purpose of this

paper is to resolve conceptual deficiencies in preventive
Maintenance optimization has been a major topic among
maintenance models to achieve enhanced and better interval
research communities belonging to physical asset management
2025 Annual Reliability and Maintainability Symposium (RAMS) | 979-8-3503-6774-4/25/$31.00 ©2025 IEEE | DOI: 10.1109/RAMS48127.2025.10935204

selection approach. For that purpose, two different methods i.e.,

where the sole purpose of research is to define a modeling
Utility Decision Trees and PRA have been integrated to
framework for selection of a maintenance interval. All these
formulate a preventive maintenance optimization model for a
formulations are based on optimization principle of balancing
complex system. The paper is composed of different portions
yearly unplanned failure costs against yearly planned
starting from deficiencies of existing models in Section 2 to the
maintenance costs under different objectives. Any preventive
proposed model formulation in Section 3. Section 4 provides a
maintenance task is applicable only if it is feasible and effective
case study example to show the feasibility of implementing the
means the task should reduce the chances of unplanned failure
proposed framework. The paper ends with discussion of
and cost less than not doing anything. This paper proposes a
limitations of proposed model and pathway for future research
unique formulation for preventive maintenance optimization of
in Section 5.
a complex system using Decision Theory framework. The
probabilistic risk analysis (PRA) approach has been used to 2 LITERATURE REVIEW
quantify the uncertainty associated with multi-dimensional
The researchers used different modeling approaches in past
consequences. A decision-theoretic framework augmented with
such as mathematical programming, monte-carlo simulation
a probabilistic risk analysis approach has been applied for risk
and stochastic modeling to determine the optimal preventive
management in nuclear and space missions but its use for
maintenance intervals under given constraints. Apart from the
maintenance optimization of a complex system is novel.
related algorithms and solvers involved, these models vary due
1 INTRODUCTION to haphazard use of factors such as level of analysis, time vs.
age-based approach, repair effectiveness, and time to repair.
The current technological age is composed of complex
The asset hierarchy as per ISO-14224, consisting of plant,
physical and socio-technological systems like transportation,
subsystem, equipment, parts and components, provides
manufacturing, and telecommunication that provide unique
opportunity to conduct analysis at any hierarchal level using
capabilities to end users under extreme operating conditions.
temporal data of failure modes and assets downtime. The
Failures of physical assets for such systems, being uncertain,
incorporation of maintenance errors, repair quality, and spare
are highly consequential due to production loss, uncontrolled
conditions prevents the next operation cycle from starting as a
damage, safety issues and environmental impacts. Preventive
new one. Time-based approaches use fixed time of the year
maintenance has a great role in controlling physical assets,
while age-based approaches use running hours of the asset for
either single or redundant, failures by restoring the functional
maintenance interval. The consideration of time to repair adds
state of assets at specified intervals. The research related to
more complications in model due to repair distributions.
preventive maintenance optimization is focused on selecting an
All above-mentioned approaches have their own modeling
appropriate interval using formulations and modeling approach
realities and limitations which are described as follows:
of mathematical programming, monte-carlo simulation and
a) Non-Repairable Data Models: The integration-based
stochastic modeling. Though existing maintenance models are
models optimize preventive maintenance interval for failure
backed up by promising concepts and body of knowledge of
modes [1][2] under cost objectives and can be extended to an
related fields, these models lack consideration of non-monetary
asset level only by including failure modes of all parts. These
factors like safety, environment, and reputation in optimization
models can be configured for age or time-based intervals with
problem. There is also a serious conceptual deficiency in model
or without consideration of repair duration. The major
formulation for a system’s risk assessment and comprehension
assumption in this model is perfect replacement i.e. each part is
of human psychology involved with huge costs. This not only
restored to its initial condition after each replacement. There is
results in sub-optimal outcomes but also compromises the
no consideration of issues such as common cause, human

979-8-3503-6774-4/25/$31.00 ©2025 IEEE

Authorized licensed use limited to: INDIAN INSTITUTE OF MANAGEMENT LUCKNOW. Downloaded on April 27,2025 at 14:38:54 UTC from IEEE Xplore. Restrictions apply.
 reliability, and repair effectiveness to express dependency e) Integer Programming with NHPP Model: This model
between different parts. They also lack the consideration of builds over the integer programming-based model with further
nonmonetary values such as human life, environmental inclusion of NHPP parameters [6],[7] and employs the failure
damages, and organizational repute in model objectives. behavior of assets to consider impact of reactive maintenance
b) Repairable data models: These models are based upon in the decision framework. Again, there is no consideration of
repairable data analysis techniques such as Non-Homogenous the non-monetary issues in the objective function of the
Poisson process (NHPP) and Homogenous Poisson process [3] optimization model.
of renewal theory which add the impact of repair effectiveness
3 PROPOSED DECISION FORMULATION
in the model behavior. Though these models are created directly
at the asset level, they cannot be expanded to the system level. This section elaborates a new formulation of preventive
Like non-repairable data models, these optimization models maintenance model using decision theory and PRA framework
also do not consider non-monetary parameters including human that provides better optimization results. The superlative
life, environmental damages, or organizational repute in the interval for preventive maintenance is achieved due to factors
model objectives. like monetary and non-monetary consequences, comprehensive
c) Simulations: Reliability (RAMS) modeling incorporates uncertainty management, psychological impact of spending
repairable and non-repairable models to simulate system-level huge amount of money and standard measurement scale for all
conditions using discrete events and monte carlo simulation consequences. The proposed decision-making approach for a
techniques. These complicated models are based on operational complex system preventive maintenance can be visualized as
logic (load, standby/redundant assets), maintenance strategies, an influence diagram in Figure 1 which helps to bring together
and sparing policies. This simulation can be conducted at any multiple aspects of the problem such as alternatives, objectives,
asset hierarchy ranging from failure modes to whole plant. The and uncertainties.
output of these models also fails to consider other non-monetary To warrant the conceptual claims associated with proposed
risks [4]. formulation, the influence diagram has been segregated and
d) Integer Programming: This model uses an integer reviewed separately for better reasoning as follows:
programming-based models to select a suitable preventive • Decision Objectives
maintenance schedule with or without time to repair [5],[8] at • Decision Alternatives
system level but fails to consider inherent damage mechanisms • Failure Probabilities
and associated failure modes for each asset in the system. There • Decision Making Measures
is no mention of reactive maintenance due to failures, and the • Resolution of Decision Problem
selection criteria is composed of monetary values ignoring the
non-monetary attributes in the objective function.

Figure 1– Maintenance Decision Influence Diagram

and system-wise risk acceptability. Similarly, decomposition of
3.1 Decision Objectives
same frame in risk assessment helps to focus on the deeply
The objective for preventive maintenance optimization is buried failure modes for each asset included in the complex
minimization of Net Risk which itself is composed of two system. Using the decision frame, an objective tree for
competing sub-objectives i.e. planned maintenance cost and maintenance problem, including fundamental and mean
failure consequences. To consolidate failure consequences, a objectives, is defined as shown in Figure 2. The leaves of
decision-making frame is used to comprehend full decision objective tree become categories of consequences during risk
context and evaluate conflicting expectations of the related assessment to evaluate the impact of failure. The fundamental
stakeholders. The frame boundary is expanded to consider all objectives are those which exist on their own and define the real
emerging issues that are not considered in the amalgamation of needs of the user such as safety issues, environmental damages,
individual assets such as long-term stoppage of plant, inventory and financial loss, while mean objectives provide a surrogate
management costs, market deliveries, contractor’s availability, for achieving fundamental maintenance objectives when the

Authorized licensed use limited to: INDIAN INSTITUTE OF MANAGEMENT LUCKNOW. Downloaded on April 27,2025 at 14:38:54 UTC from IEEE Xplore. Restrictions apply.
 fundamental objective cannot be measured directly. Thus, this previous maintenance optimization models by considering all
modeling formulation addresses the most important gap of kinds of consequences.

Figure 2– Maintenance Decision Objectives Tree

3.2 Decision Alternatives 3.3 Failure Probabilities
The alternatives for this decision include different time The PRA technique as mentioned in Figure 4 is normally
intervals at which preventive maintenance of the whole system used to quantify risk triplets [10]. It is a highly systematic and
can be conducted as shown in Figure 3. Increasing the interval logical approach to comprehend probabilities of consequences
would reduce planned maintenance cost per year but would lead in objective tree shown in Figure 2. PRA starts with a master
to increased chances of failure consequences as shown in risk logic diagram which includes different logical functions of the
exceedance curves in Figure 1 while decreasing the interval system. The end nodes of master logic diagram represent
would reduce the chances of failure but increase the planned functional failure of the system. Each functional failure node is
maintenance cost. further decomposed into its specific failure modes using the
conditional gates of the fault tree. The major benefit of using
the fault tree is the inclusion of common cause events in the
model that leads to multiple assets failure even with redundancy
protection. Each identified functional failure becomes the
initiating event leading to the separate event sequence diagram
where the layer of protection comes to the rescue of the system.

Figure 4– Probabilistic Risk Analysis [9]

Since the protection system itself is prone to failure risk
Figure 3– Decision Tree
due to hidden and age-based failure modes, a single protection

Authorized licensed use limited to: INDIAN INSTITUTE OF MANAGEMENT LUCKNOW. Downloaded on April 27,2025 at 14:38:54 UTC from IEEE Xplore. Restrictions apply.
 can become unavailable and directs the event progression to the particular consequence measure corresponding value is read
next pivotal event where another protection acts as a barrier to from value curve. Different value curves are then combined
the propagation of the initiating event. So different layers of using the additive value function such that requirement of
protection are used to provide a safe shutdown of the system mutually preferential independence is fulfilled. The additive
otherwise system failure would become out of control. value function is defined as:
By combining the fault trees from the master logic diagram 𝑉𝑉𝑌𝑌 (𝑦𝑦) = 𝑤𝑤1 𝑉𝑉𝑥𝑥1 (𝑥𝑥1 ) + 𝑤𝑤2 𝑉𝑉𝑥𝑥2 (𝑥𝑥2 ) … 𝑤𝑤𝑛𝑛 𝑉𝑉𝑥𝑥𝑛𝑛 (𝑥𝑥𝑛𝑛 ) (2)
with the individual event sequence diagram and related fault
tree of the protection system, the probabilities of different
failure end states are obtained for an initiating failure. This step
is repeated for all the initiating events along with connected
event trees and summed up to get the final probabilities for each
level of consequence. Normally the results of risk assessment
are represented in the form of risk exceedance curves, as shown
in Figure 1, such that probability decrement occur with
increasing level of each consequence. Thus, PRA provides
great robust approach to include uncertainty associated with
failure modes and complicated operational logic of the system.
However, for decision making continuous risk curves are Figure 5– Value Function [12]
converted to standard discrete distribution as per Extended
Swanson-Megill (ES-M) or Extended Pearson-Tukey (EP-T) The weights are derived using swing weight method of
approximations methods as shown in Table 1. For this purpose, Mult-Attribute Utility Theory (MAUT) [13] as shown in Table
each specific fractile of original distribution is assigned a preset 2. The major assumption required for the swing weight method
probability [11]. is given by equation 3 as follows:
∑𝑛𝑛1 𝑤𝑤𝑖𝑖 = 1 (3)
Table 1– Conversion Of Continuous To ES-M Discrete
Probabilities [11] Table 2– Estimation of Weights Using Swing Weight Method

Number Continuous Discrete Number Measure Rank Score Weight

Probabilities Fractile Probabilities 1 Planned Cost 4 100 0.40
1 .10 .167 2 Safety 1 50 0.20
2 .50 .66 3 Environment 2 50 0.20
3 .90 .167 4 Unplanned Cost 3 50 0.20

3.4 Decision Making Measures b) Utility Curve: The utility curve is developed based on
Rather than combining values of consequences when each decision maker risk acceptability and risk attitude. To simplify
having different scale ranges, a standardization step is done to the elicitation of the utility curve, the assumption is made that
bring all consequence types on common scale and then the decision maker is a delta person who does not consider
combined into a single digit appropriately. The values are initial wealth in the utility curve or can be assumed to possess
mapped on developed utility curve to make explicit mention of same selling and buying prices for all kinds of ownership cycles
psychology associated with huge resource consumptions and [14]. The utility curve of the delta person can be defined as a
risk acceptability in decision evaluation. Different steps of the linear or exponential curve and can be elicited using the risk
process are as follows: odd or risk tolerance evaluation. The exponential value function
• Multi-Attribute Values is given as follows [12]:
• Utility Curve 1−𝑒𝑒 −�𝑣𝑣−𝑣𝑣𝑚𝑚𝑚𝑚𝑚𝑚� /𝜌𝜌
• Risk Attitude −�𝑣𝑣𝑚𝑚𝑚𝑚𝑚𝑚 −𝑣𝑣𝑚𝑚𝑚𝑚𝑚𝑚 � /𝜌𝜌 𝑖𝑖𝑖𝑖 𝜌𝜌 ≠ ∞
𝑈𝑈(𝑥𝑥) = �1−𝑒𝑒 (4)
a) Multi-Attribute Values: A value curve is developed (𝑣𝑣−𝑣𝑣𝑚𝑚𝑚𝑚𝑚𝑚 ) 𝑖𝑖𝑖𝑖 𝜌𝜌 = ∞
using an expert elicitation process to bring all consequences on 𝑣𝑣𝑚𝑚𝑚𝑚𝑚𝑚 −𝑣𝑣𝑚𝑚𝑚𝑚𝑚𝑚
the same scale of 0 to 1. Apart from the linear, the value The value of ρ, which is called risk tolerance, is evaluated from
function can be represented using the exponential value certainty equivalent involving different bets such that decision
function given as follows [12]: maker is indifferent to two options as per assigned probabilities
1−𝑒𝑒 −�𝑥𝑥−𝑥𝑥𝑚𝑚𝑚𝑚𝑚𝑚� /𝜌𝜌 to each option. Reference [14] provides further details for the
−�𝑥𝑥𝑚𝑚𝑚𝑚𝑚𝑚 −𝑥𝑥𝑚𝑚𝑚𝑚𝑚𝑚 � /𝜌𝜌 𝑖𝑖𝑖𝑖 𝜌𝜌 ≠ ∞ evaluation of the exponential utility curve. The single values
𝑉𝑉𝑥𝑥 (𝑥𝑥) = �1−𝑒𝑒 (1)
(𝑥𝑥−𝑥𝑥𝑚𝑚𝑚𝑚𝑚𝑚 ) 𝑖𝑖𝑖𝑖 𝜌𝜌 = ∞ obtained from equation 2 are mapped on developed utility curve
𝑥𝑥𝑚𝑚𝑚𝑚𝑚𝑚 −𝑥𝑥𝑚𝑚𝑚𝑚𝑚𝑚 to convert individual values into respective utility figures.
The value of ρ is evaluated from the mid-value assessment as
shown in Figure 5. Each consequence category defined in the
risk curve of PRA is mapped on value curves such that for a

Authorized licensed use limited to: INDIAN INSTITUTE OF MANAGEMENT LUCKNOW. Downloaded on April 27,2025 at 14:38:54 UTC from IEEE Xplore. Restrictions apply.
 was developed in Precision Tree software based on the
influence diagram defined in Figure 1. From the evaluation of
the tree, it was derived that the system should be overhauled
every 6 years as shown in Table 3. With integration of decision
framework and risk assessment, selected preventive
maintenance interval showed the right balance between
acceptable level of risk against resources available for planned
and unplanned failures.
Figure 6– Utility Function [12] Table 3– Optimal Interval Evaluation for Case Study
c) Risk Attitude: Depending upon the behavior of the Year Discrete Values Utility Final
utility curve, the decision maker is considered as risk averse, Probability
risk neutral, and risk taker. A person is risk averse if his certain 5 0.167 0.26 0.81
equivalent for any monetary deal is always less than the mean 0.551
0.66 0.54 0.54
of the monetary measure. A risk-neutral person has his certainty 0.167 0.72 0.35
equivalent equal to the mean of the monetary measure while a 0.167 0.24 0.82
risk taker would always overestimate the certainty equivalent 6 0.66 0.53 0.55 0.565
against the mean of the monetary measure. This type of 0.167 0.70 0.37
behavior is the same in both the increasing and decreasing
utility curves as shown in Figure 6. 5 SUMMARY & CONCLUSION

3.5 Resolution of Decision Tree The purpose of this paper is to provide better formulation
for maintenance optimization by considering all applicable non-
The consequences obtained from the PRA after being monetary variables along with costs, uncertainty management
mapped on value and utility curves are entered in the decision of failure modes, logical redundancy and protections in the
tree form. The decision tree is folded backwards using the tree system and most importantly decision-maker preferences and
resolution method and the branch with maximum end utility is risk attitudes as huge costs are involved in the maintenance of
selected as right interval for preventive maintenance decision. the whole system together. These considerations are important
4 CASE STUDY to manage the safety of operating plants under the given
budgetary constraints. Though there are some assumptions as
The proposed decision-making approach was implemented well as complications with the use of this method like failure of
for a fluidized bed reactor system of a petrochemical plant. In aggregation of multiple decision makers as per Arrow’s
case of emergency polymerization reaction must be stopped impossibility theorem, but elegant justification of risk-averse
using the kill system with immediate effect. The last line of decision makers cannot be explained in such a simple manner
defense in case of failure of kill system includes fixed fire by any other approach. As decision theory distinguish between
protection systems (e.g., sprinklers, water sprays, deluge the decision-making process and outcomes of the decision,
systems, monitor guns, etc.), dikes and drainage systems. The decision-making should be judged by its conformity to
assets included in the reactor process were used to conduct the principles of rationality, the validity of reasoning, and the
PRA study and consequence were mapped on plant specific clarity of the process, which are also the hallmark of the process
value and utility curves as shown in Figure 7. mentioned.
However, there are still some serious challenges for
decision theoretic framework which include consideration of
expected value, valuation of life and dummy bet which no
decision maker experiences in real life. The most important
issue associated with decision-making process in practice is
biases [15]. The decision-making process to manage the risks
can become faulty when the decision-maker does not consider
readily available options, fails to consider relevant constraints,
or does not project consequences of the various courses of
action. The decision makers are sometimes also overwhelmed
by multiple objectives as company owners want maximization
of profit and minimization of organizational liabilities while
regulatory agencies demand that public safety should be the
topmost priority. Market competition, company image, and
sales requirements build another contextual dimension for
Figure 7– Case Study Workflow decision-makers to complicate things. The communication
language among stakeholders and decision-makers is money
A decision tree for maintenance optimization of the system

Authorized licensed use limited to: INDIAN INSTITUTE OF MANAGEMENT LUCKNOW. Downloaded on April 27,2025 at 14:38:54 UTC from IEEE Xplore. Restrictions apply.
 such that losses and profits are projected in monetary value. In 5. S. Perez Canto, “Using 0/1 mixed integer linear
this way, the system context is reduced to the profit programming to solve a reliability-centered problem of
maximization only. This is the point where the decision maker power plant preventive maintenance scheduling,”
underestimates the risk, fails to simulate outcomes associated Optimization and Engineering, vol. 12, pp. 333–347, 2011.
with various courses of action, and leads to the objectives 6. Q. Ahmed, K. S. Moghaddam, S. A. Raza, and F. I. Khan,
related to the profits, dominating in the decision space. The “A multiconstrained maintenance scheduling optimization
perspective vision of the decision maker also plays an important model for a hydrocarbon processing facility,” Proceedings
role in achieving balanced targets. If a decision maker judges a of the Institution of Mechanical Engineers, Part O: Journal
long-term duration decision using short hindsight goals, then an of Risk and Reliability, vol. 229, no. 2, pp. 151–168, 2015.
unrealistic picture would appear in front of the decision maker 7. K. S. Moghaddam, Preventive maintenance and
which results in the removal of highly attractive alternatives. replacement scheduling: models and algorithms.
The issues like sustainability, environment, and safety require University of Louisville, 2010.
long-term goals and cannot be compared against other short- 8. M. Alardhi, R. G. Hannam, and A. W. Labib, “Preventive
term objectives. Motivational biases are also implanted in maintenance scheduling for multi-cogeneration plants with
people’s minds by the kind of treatment given to them. In the production constraints,” Journal of Quality in Maintenance
maintenance world, organizations reward people for heroic acts Engineering, vol. 13, no. 3, pp. 276–292, 2007.
that are related to either an increase in production/revenue or 9. M. Stamatelatos, H. Dezfuli, G. Apostolakis, C. Everline,
reducing the interval of shutdown after a failure. This results in S. Guarro, D. Mathias, A. Mosleh, T. Paulos, D. Riha, C.
a general motivation among all to show monetary results by Smith et al., “Probabilistic risk assessment procedures
cost-cutting and adopting other shortcuts. Similarly, immediate guide for nasa managers and practitioners,” Tech. Rep.,
reinforcement is more powerful than delayed reinforcement. 2011.
Though doing timely planned investment is good, its results 10. S. Kaplan and B. J. Garrick, “On the quantitative definition
appear after a long time and the reward for decision-making is of risk,” Risk analysis, vol. 1, no. 1, pp. 11–27, 1981.
delayed. This biases the approach of the decision maker to use 11. R. T. Clemen and T. Reilly, Making hard decisions with
wrong assumptions during decision making process. What was DecisionTools. Cengage Learning, 2013.
explained for motivation to get reward can be related to fear of 12. P. R. Garvey, Analytical methods for risk management: A
negative criticism i.e., being held responsible for the over- systems engineering perspective. Crc Press, 2008.
budget or bad financial health of the company. A major change 13. K. Chelst and Y. B. Canbolat, Value-added decision
that appears at this stage is that a decision maker who might be making for managers. CRC press, 2011.
risk-averse may turn out to be a risk-taker due to his motivation 14. A. E. Abbas and R. A. Howard, Foundations of decision
and vision dilemma. Any wrong assumption in preferences also analysis. Pearson Higher Ed, 2015.
aggravates the final ranking process as explained in an Allais 15. Daniel, Kahneman. Thinking, fast and slow. 2017.
Paradox, whereby individuals tend to place too much weight on
BIOGRAPHIES
a certain outcome as compared to uncertain outcomes. Impact
of the safety and environment targets are uncertain in long term, Fahad Izhar Khan
and the postponement of investment benefits is obvious, the Department of Systems and Industrial Engineering
decision maker prefers the delay option as its expected utility is University of Arizona
more than other options on the selected preference curve. 1127 James E. Rogers Way,
Though proper awareness and calibration techniques have Tucson, AZ 85721 USA
been developed to guard against the biases in uncertainty
e-mail: [email protected]
elicitation and other motivational biases, there is still a need to
model and quantify such behavioral errors in maintenance Fahad Izhar is a graduate student of systems engineering at the
decision making. This would make the optimization process University of Arizona. He graduated in Reliability Engineering
true representative of natural decision-making scenario and from the University of Maryland USA and Monash University
bring further realistic predictions of preventive maintenance Australia. He holds different certifications in the field of
interval. Reliability Engineering and Systems Engineering from ASQ
USA, INCOSE USA, Object Management Group, and Institute
REFERENCES
of Asset Management UK. His research interests include
1. A. K. Jardine and A. H. Tsang, Maintenance, replacement, Simulation, Bayesian Decision Theory, Systems Engineering,
and reliability: theory and applications. CRC press, 2005. Prognostics, Operation Research, and Probabilistic Risk
2. R. S. Beebe and R. S. Beebe, Predictive maintenance of Analysis.
pumps using condition monitoring. Elsevier, 2004.
Allan T. Mense
3. M. Modarres, M. P. Kaminskiy, and V. Krivtsov,
Department of Systems and Industrial Engineering
Reliability engineering and risk analysis: a practical guide.
University of Arizona
CRC press, 2016.
1127 James E. Rogers Way,
4. Calixto, Eduardo. Gas and oil reliability engineering:
Tucson, AZ 85721 USA
modeling and analysis. Gulf Professional Publishing, 2016.

Authorized licensed use limited to: INDIAN INSTITUTE OF MANAGEMENT LUCKNOW. Downloaded on April 27,2025 at 14:38:54 UTC from IEEE Xplore. Restrictions apply.
 e-mail: [email protected] Professional Engineer in the state of Arizona and a Certified
Reliability Engineer by ASQ. Dr. Mense received a B.S. in
Dr. Mense is a principal engineering fellow with Raytheon
Engineering Physics and M.S. in Nuclear Engineering from the
Missile Systems. He has authored over 70 publications, chaired
University of Arizona, a Ph.D. in Nuclear Engineering/Plasma
IEEE Committees at the national level. He was a member of the
Physics from the University of Wisconsin-Madison, and an
U.S. Army Science Board for 8 years, member of the Nat’1
M.S. in Industrial and Management Science Engineering from
Academy of Engineering, BAST panel. He is a registered
Arizona State University.

Authorized licensed use limited to: INDIAN INSTITUTE OF MANAGEMENT LUCKNOW. Downloaded on April 27,2025 at 14:38:54 UTC from IEEE Xplore. Restrictions apply.