See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/310149252

Development Of A Decision Support System In Determining Optimum Number

Of Server For NNPC Mega Petroleum Stations

Article in International Journal of Scientific & Technology Research · December 2014

CITATIONS READS

0 150

3 authors:

Victor Mmerichukwu Mbachu Pius Chukwukelue Onyechi

Nnamdi Azikiwe University, Awka Nnamdi Azikiwe University, Awka
23 PUBLICATIONS 77 CITATIONS 53 PUBLICATIONS 235 CITATIONS

SEE PROFILE SEE PROFILE

Ogunoh Victor Arinze

Enugu State University of Science and Technology
11 PUBLICATIONS 22 CITATIONS

SEE PROFILE

All content following this page was uploaded by Pius Chukwukelue Onyechi on 17 July 2017.

The user has requested enhancement of the downloaded file.

INTERNATIONAL JOURNAL OF SCIENTIFIC & TECHNOLOGY RESEARCH VOLUME 3, ISSUE 12, December 2014 ISSN 2277-8616

Development Of A Decision Support System In

Determining Optimum Number Of Server For
Nnpc Mega Petroleum Stations
Mbachu, Victor M. , Onyechi, Pius C., Ogunoh, Victor A.

Abstract: Customers queue up at the Nigerian National Petroleum Corporation (NNPC) mega stations to purchase the Premium Motor Spirit (PMS),
Dual Purpose Kerosene (DPK) and Automotive Gas Oil (AGO) popularly called petrol, kerosene and diesel respectively in the country. The PMS usually
has the highest demand due to its various uses, and the queue for the product at NNPC Mega station during period of scarcity (referred to here as peak
periods) has varying average arrival rate at various time of the day. The waiting situation is exasperating to the customers and the management. Thus
an operational plan module for determining basically the optimum number of active servers to salve the queue problem in the Stations, while attending to
vehicles with minimum demand of 10 litres of PMS was developed. The following optimal active servers were obtained and recommended for the
corresponded arrival rate of 1 car per minute to 4 cars per minute, at 72- 80% system utilization rate, a known average service rate of 0.457 cars per
minute and average waiting time of 2.6954 minutes to 0.9737 minutes. Simulation of the system was done generating a model which suggests optimum
number of server given arrival rates and average service rate.
————————————————————

1. Introduction The quest to address the supply needs of petroleum

A queue, which could deteriorate to congestion if effective products such as the Premium Motor Spirit (PMS), Dual
queue management decisions are not implemented, is a Purpose Kerosene (DPK) and Automotive Gas Oil (AGO)
waiting line for service that forms either due to inability to popularly called petrol, kerosene and diesel respectively
meet up with arriving demands as a result of insufficient necessitated the introduction of NNPC Mega Petroleum
service capacity or due to stochastic nature of customer Stations across Nigeria. Whenever these petroleum
arrival and demand. All the components of production products are scarce in the country, the NNPC Mega
system are unhappy with lengthy and longer queue length Petroleum stations serves as a good and reliable source
and waiting time respectively. Unfortunately queuing where consumers purchase these products due to quality
inevitably occurs in the present world with explosive and being the fact that the products are being sold at
population demanding regular services like personal, public, government approved prizes. A mega station is
transport, telecommunication, maintenance, computer, characterized by six pump dispensing devices with two
health and administrative services. It is therefore nozzles on each pump, totaling twelve servers per station
encountered almost every day in our daily lives in one form for PMS refill. This value seems arbitrarily fixed, and affects
or another e.g. on our way to work, waiting for services in their queue performance. At maximum demand periods,
hospitals, phone booths, filling stations etc (Hiller and service systems are being over utilized and customers
Lieberman, 2005). However, since waiting line is part of our queue up for longer time than necessary before being
daily life, all we should hope to achieve is to minimize its served while at minimum demand period, server idleness
inconvenience to some acceptable levels. The basic becomes high and increases running cost of operation. The
strategy of combating queue build-up is by increasing the need of however, developing a decision support system
service capacity of a service facility. Some of the direct that gives a good tradeoff between system utilization and
methods of increasing service capacity are by increasing waiting time became imperative. This will assist the
the number of servers, increasing the speed of the servers, operations manager in deciding the minimum number of
replacing the existing servers with faster ones etc. servers for each mega petroleum station that can serve
both maximum and minimum demand periods.

2. Review of Past Work

In recent times, queuing theory and the diverse areas of its
applications has grown tremendously. Takagi (1991)
_________________________ considered queuing phenomena with regard to its
applications and performance evaluation in computer and
communication systems. Obamiro (2003) Applied Queuing
 Mbachu, Victor M. & Onyechi, Pius C. are Model in Determining the Optimum number of Service
lecturers in Department of Industrial and Facility in Nigerian Hospitals. He however achieved this by
Production Engineering Nnamdi Azikiwe determining some queuing parameters which enabled him
University, Awka. Nigeria to improve the performance of the system. Mgbemena
[email protected] (2010) was able to model the queuing system of some
banks in Nigeria using regression analysis. In her work, she
 Ogunoh, Victor A. is currently a postgraduate created queuing management software in MATLAB that
student of Department of Industrial and shows at a glance, the behavior of the queuing system and
Production Engineering Nnamdi Azikiwe the unit that needs attention at any time. The essence was
University, Awka. Nigeria to improve the customer service system in Nigerian banks.
Chinwuko and Nwosu (2014) adopted the single line multi-

207
IJSTR©2014
www.ijstr.org
INTERNATIONAL JOURNAL OF SCIENTIFIC & TECHNOLOGY RESEARCH VOLUME 3, ISSUE 12, December 2014 ISSN 2277-8616

server queuing existing model to analyze the queuing line. An example is a petroleum service station. As
system of First Bank Nigeria PLC. In their work, they illustrated by figure 3.
suggested a need to increase the number of servers in
order to serve customers better in the case study d. Multiple servers, Multiple-phases System:
organization. Ohaneme et al, (2011) proposed the single According to Singh (2007), this type of system has
line multi-server queuing system which they simulated numerous queues and a complex network of multiple
using c-programming to be adopted at the Nigerian National phases of services involved as can be seen in figure 4. This
Petroleum Corporation (NNPC) Mega petroleum station in type of service is typically seen in a hospital setting, multi-
Awka, Anambra State in order to avoid congestion and specialty outpatient clinics, patient first form the queue for
delay of customers. Here, the queue performance at NNPC registration, and then he/she is triage for assessment, then
mega petroleum station Enugu and Owerri at peak demand for diagnostics, review, treatment, intervention or
period were evaluated, establishing a decision support prescription and finally exits from the system or triage to
system for operation management in the Mega petroleum different provider.
stations.

3. Characteristics of Queuing System

The mechanism of the queuing process is very simple.
Customers (not necessarily human customers) are arriving Arrivals Queue service facility Departures
for service, waiting for service if it is not immediate, and
leaving the system as soon as they are served. There are Figure 1.: Single-server, Single phase System.
six basic characteristics of queuing processes which Source: Adopted from Obamiro (2003)
provide an adequate description of a queuing system: (1)
arrival pattern of customers, (2) service pattern of servers,
(3) number of service channels, (4) system capacity and (5)
queue discipline. In usual queuing systems the arrival
pattern of customers is stochastic. Queue discipline refers
to the manner in which customers are selected for service Figure 2: Single-server, multiple phases System
when a queue has formed. The most common discipline is Source: Adopted from Obamiro (2003)
first come, first served (FCFS), but there are many others
like last come, first served (LCFS) which is applicable in
many inventory systems as it is easier to reach the nearest
item; randomly selecting for service (RSS) independent of
the arrival time of the customer; and a variety of priority
schemes, the customers with higher priority being served
ahead of the lower priority customers regardless of the
order in which they arrived to the system.

4 Types of Queuing Systems

There are four major types of queuing system. Lapin (1981)
broadly categorized queuing system structures into the Figure 3: Multiple-servers, Single phase System
following. Source: Adopted from Obamiro (2003)

a. Single-server, Single-phase system:

This is a situation in which single queue of customers are to
be served by a single service facility (server) one after the
other. An example is bottles or cans of minerals or beer to
be cocked in a production process. Diagrammatically it is
depicted in figure 1

b. Single-server, Multiple-phases System: Figure 4: Multiple-servers, Multiple-phase System

In this situation, there’s still a single queue but customers Source: Adopted from Obamiro (2003)
receive more than one kind of service before departing the
queuing system as shown in figure 2. For example, in the 5 Performance Evaluation of Queuing
university, students first arrive at the registration desk, get Systems
the registration done and then wait in a queue for their Hillier and Lieberman, (2005) put forth the following
forms to be signed, after signing; they join another queue performance parameters in a queuing system:
for submission. Students have to join queue at each phase a. System Utilization (p): System Utilization is the most
of the system. important measure of a queuing system. It is the ratio
of system capacity used to available capacity. It
c. Multiple-servers, Single-phase System: measures the average time the system is busy. System
This is a queuing system characterized by a situation utilization of zero means that there is nobody in the
whereby there is a more than one service facility (servers) system. On the other hand, a system utilization of one
providing identical service but drawn on a single waiting or more signifies that there is infinite number of people
208
IJSTR©2014
www.ijstr.org
INTERNATIONAL JOURNAL OF SCIENTIFIC & TECHNOLOGY RESEARCH VOLUME 3, ISSUE 12, December 2014 ISSN 2277-8616

on the waiting line. This means that the available v. The average time a customer spends in line waiting for
servers cannot cope with the arriving demand. Thus service:
something has to be done on the service facility
𝐿𝑄
(Egolum, 2001). Based on Egolum (2001), System 𝑊𝑄 = (5)
𝜆
Utilization should be greater than 0 but less than 0.8.
b. Mean Number in the system (Ls): Mean number in the
vi. The average time a customer spends in the system:
system is the average number of system users
(entities) in the system; it includes those in the queue 𝐿𝑠
and those being served by the server(s). 𝑊𝑠 = (6)
𝜆
c. Mean Number in Queue (Lq): Mean number in the
queue is the average or expected number of system vii. The average waiting time of a customer on arrival not
users in the queue (waiting line), waiting for their turn to immediately served:
be served.
1
d. The average waiting time for an arrival not immediately 𝑊𝑎 = (7)
𝑀𝜇 −𝜆
served (Wa)
e. Mean Time in System (Ws): Mean time in the system is
viii. Probability that an arriving customer must wait:
the expected value or average waiting time an entity
will spend in the queuing system. It includes the 𝑊𝑄
average time waiting for service to begin and the 𝑝𝑤 = (8)
𝑊𝑎
average service time.
f. Mean Time in Queue (Wq): Mean time in the queue is (Blanc, 2011; Sztrik, 2011; and Nain, 2004)
the expected value or average time an entity will spend
in the queue, waiting for service to begin.
g. Probability of zero customers in the system (PO )
h. Probability of waiting (Pw): This is the probability that
an arrival will have to wait for its service to begin.

6 Methodology
The single line multi server queuing equations stated below
were adopted and coded in Microsoft Excel which was used
to develop the decision support system that generates the
results of the queue performance at various arrival rates of
customers in both case study establishments. The essence
was to determine the best server utilization that can serve
both maximum and minimum demand periods. Based on
Egolum (2001), System Utilization should be greater than 0
but less than 0.8.Thus, system utilization was optimized
towards the 80% limit as other parameters improves. This Figure 5: Coding Environment of the decision support
will however, assist in decision making as regards number system displaying the results of the queue performance
of servers to be engaged for duty.
when average arrival rate (𝜆) = 1.000, and average service
rate (pr server) (𝜇 ) = 0.45764 cars/minutes with = 2 – 12
Single Line Multi Server Queuing Equations servers in NNPC Mega Petroleum Station Enugu.
i. The average utilization of the system:
𝝀
𝜌= (1)
𝑚 𝜇

ii. The probability that there are no customers in the

system:
𝑛 𝑀 −1
𝝀 𝝀
𝑀−1 𝜇 𝜇
𝑃0 = 𝑛=0 + 𝝀
(2)
𝑛! 𝑀! 1−
𝑀𝜇

iii. The average number of customers waiting for service:

𝑀
𝝀
𝝀𝜇
𝜇
𝐿𝑞 = 2 𝑃0 (3)
𝑀−1 ! 𝑀𝜇 −𝝀

iv. The average number of customers in the system:

Figure 6: Coding Environment of the decision support
𝐿𝑠 = 𝐿𝑞 + (𝜆 𝜇 ) (4) system displaying the results of the queue performance
when average arrival rate (𝜆) = 1.000, and average service
rate (pr server) (𝜇 ) = 0.38553 cars/minutes with = 2 – 12
servers in NNPC Mega Petroleum Station Owerri.
209
IJSTR©2014
www.ijstr.org
INTERNATIONAL JOURNAL OF SCIENTIFIC & TECHNOLOGY RESEARCH VOLUME 3, ISSUE 12, December 2014 ISSN 2277-8616

7. Simulation Results of various Arrival Rate adopted in any establishments but only applicable to where
of Customers/Mins single line multi-server queues exist.

Summary of Simulation Results at various Average Arrival Rates of Customer/Mins at

9. Reference
NNPC Mega Petroleum Station Enugu [1] Blanc, J.P.C. (2011). Queueing Models: Analytical
/Mins Best No.
and Numerical Methods (Course 35M2C8),
Servers M vs ʎ/Mins
y = 7.213x5 - 67.07x4 + 242.9x3 - 427.5x2 + 367.5x - 120.0 Department of Econometrics and Operations
1 3 9 R² = 0.964

1.1 4 8 Research Tilburg University, pp 30-57.

7
1.2 4
Number of Servers (M)

6
1.3 4
1.4 4
5
4
[2] Chinwuko E.C and Nwosu M.C (2014), Analysis of
M
1.5
1.6
5
5
3 Poly. (M) a Queuing System in an Organization (A Case
2
1.7 5
1
Study of First Bank PLC). 2014 National
1.8 5
1.9 6
0
0 0.5 1 1.5 2 2.5 3
Conference on Engineering for Sustainable
2 6 Average Arrival Rate/Minutes
Development, Nnamdi Azikiwe University Awka,
2.1 6
2.2 7
Figure 4.47: Scatter Plot of Number of Servers (M) vs. Average
Arrival Rate/Minutes
Anambra State Nigeria: Paper F29, pp 256-267
2.3
2.4
7
7
Manufacturing and Industrial Infrastructure Issues.
2.5 7 The Recommended Best Number of Servers is given by:
2.6
2.7
7
8 M = 7.213 5 - 67.07 4 + 242.9 3 - 427.5 2 + 367.5 - 120.0 [3] Egolum C.C (2001), Quantitative Technique for
Management Decisions. Nnamdi Azikiwe
University Awka, Anambra State. Pp 91-131.
Figure 7: Summary of Simulation Results at various
Average Arrival Rates of Customer/Mins at NNPC Mega [4] Hiller S.F and Lieberman J.G. (2005) Introduction
Petroleum Station Enugu to Operations Research, Boston: McGraw Hill,
Eight Edition.
Summary of Simulation Results of various Average Arrival Rates of Customer/Mins at
NNPC Mega Petroleum Station Owerri [5] Lapin L.L. (1981), Quantitative Methods for
/Mins Best Server
Business Decision, Harcourt: B. Jouanaurch, 2nd
Utilization
9
M vs ʎ/Mins
y = -9.228x5 + 79.09x4 - 266.5x3 + 440.9x2 - 354.6x + 114.2 Edition. pp 182 – 201.
R² = 0.966
1 4 8
1.1 4 7
[6] Mgbemena C.E., 2010. Modeling of the Queuing
Number of Servers (M)

1.2 4 6

1.3 5
5

4 M
System for Improved Customer Service in Nigerian
1.4 5 3 Poly. (M) Banks, M.Eng Thesis, Nnamdi Azikiwe University
2
1.5 5
1 Awka, Nigeria.
1.6 6
0

1.7 6 0 0.5 1 1.5 2 2.5

1.8 6
Average Arrival Rates/Minutes
[7] Nain P. (2004). Basic Elements of Queuing Theory
1.9 7 Figure 4.48: Scatter Plot of Number of Servers (M) vs. Average Application to the Modeling of Computer Systems
Arrival Rate/Minutes
2 7
(Lecture notes), Pp 16-31
2.1 7

2.2 8 The Recommended Best Number of Servers is given by:

2.3 8 M = -9.228 5 + 79.09 4 - 266.5 3 + 440.9 2 - 354.6 + 114.2

[8] Obamiro, J.K.,(2003) ―Application of Queuing
Model in Determining the Optimum Number of
Service Facilities in Nigerian Hospitals‖, M. Sc.
Project Submitted to Department Business
Figure 8: Summary of Simulation Results at various
Administration, University of Ilorin.
Average Arrival Rates of Customer/Mins at NNPC Mega
Petroleum Station Owerri.
[9] Ohaneme C.O, Ohaneme L.C, Eneh I.L and
Nwosu A.W (2011) ―Performance Evaluation of
8. Conclusion Queuing in an Established Petroleum Dispensary
The evaluation of queuing system in an establishment is System Using Simulation Technique‖ International
very essential for the betterment of the establishment. Most Journal of Computer Networks and Wireless
establishments are not aware on the significance of Communications (IJCNWC), Vol.2, No.2, April
evaluating their queue performance. The implication of this 2012.
is that they are not able to tell the minimum number of
servers that can service their customers at peak periods. [10] Singh V., (2007) ―Use of Queuing Models in Health
This results to customers spending longer time than Care‖, Department of Health Policy and
necessary before receiving service. In respect of this, a Management, University of Arkansas for Medicals
decision support system has been developed using the Sciences.
application of Microsoft Excel which can be used in
evaluating queue performance. The simulation of average [11] Sztrik, J. (2011). Basic Queuing Theory, University
arrival rates of customers in NNPC Mega Station Enugu of Debrecen, Faculty of Informatics, pp 17-57.
and Owerri respectively using the developed decision
support system gives the best server utilization at various [12] Takagi, H (1991) ―Queuing Analysis, A Foundation
arrival rates. This will assist in determining the best number of Performance Evaluation‖, Vacation and Priority
of server that can serve both maximum and minimum Systems, 1, 10-19.
demand periods. The developed support system can be
210
IJSTR©2014
View publication stats
www.ijstr.org