Department of Civil Engineering, Faculty of Engineering, Bayero University, Kano - Nigeria

Quantitative Techniques (CIV 8341) prepared by Engr. Dr. A. D. Rafindadi

E. Queuing Theory
Queuing theory, a branch of operation research that studies how lines form, how they function,
and why they malfunction. It can be applied in construction management to analyze and
optimize the flow of resources, such as labor, equipment, and materials, through construction
processes. Queuing theory helps in understanding and improving the efficiency of construction
operations by studying the characteristics of waiting lines and queues.

Fig. 1.
Figure 1: shows a single-channel, single-server queue, which has three customers waiting in
the queue (line) and one being served.
Characteristics of the queuing system:
(a) Input source: One characteristic of the input source is the size. The size is the total
number of units that might require service from time to time. It may be assumed to be
finite or infinite. The customer assumption is that they generate according to ‘Poisson
Distribution’ at a certain average rate. Therefore, the equivalent assumption is that
they generate according to exponential distribution between consecutive arrivals. To
solve the problems, use & assume customer population as ∞
(b) Queue discipline: A queue is characterized by a maximum permissible number of units
that it contains. Queues are called finite or infinite, according to whether number is
finite or infinite. The service discipline refers to the order in which number of queues
are selected for service.
(c) Service mechanism: This consists of one or more service facilities, each of which
contains one or more parallel service channel. If there is more than one service facility,
the arrival unit may receive the service from a sequence of service channels.
At a given facility, the arrival enters the service facility and is completely served by
that server. The time elapsed from the commencement of the service to its completion
for a unit at the service facility is known as service time usually, service time follows
as exponential distribution.

1
 Department of Civil Engineering, Faculty of Engineering, Bayero University, Kano - Nigeria
Quantitative Techniques (CIV 8341) prepared by Engr. Dr. A. D. Rafindadi

Key Concepts:
1. Arrival Rate (λ): Represents the rate at which entities (e.g., tasks, trucks, workers)
arrive at the construction process.
2. Service Rate (μ): Represents the rate at which entities are serviced or processed by the
construction process.
3. Number of Servers in Parallel (c): number of servers in parallel.
4. Utilization (ρ):
𝜆
• Represents the ratio of the service rate to the arrival rate (𝜌 = 𝑐𝜇).

• Indicates the degree to which the construction process is being utilized.

5. Probability (Pn): The probability of having ‘n’ entities in the system.
6. Queue Length (Ls): Represents the number of entities waiting in the queue.
7. Queue Length (Lq): Represents the mean number of entities in the queue.
8. Average Time in the System (Ws): Represents the mean waiting time in the system.
9. Average Time in the Queue (Wq): Represents the mean waiting time in the queue.
10. (C2): Represents the squared coefficient of variation of a random variable;
𝑉𝑎𝑟𝑖𝑎𝑛𝑐𝑒
𝑪𝟐 = (𝑀𝑒𝑎𝑛)2

11. (C2s): Represents the squared coefficient of variation of service time.

12. (C2a): Represents the squared coefficient of variation of inter-arrival time.
13. (𝛔𝟐𝐬 ): Represents the variance of service time.
Applications in Construction Management:
1. Resource Allocation: Queuing theory helps in optimizing the allocation of resources,
such as workers or equipment, to construction tasks by understanding the flow and
bottlenecks in the construction process.
2. Project Scheduling: Queuing models can be used to analyze the impact of varying
arrival rates and service times on project schedules. This helps in identifying critical
paths and potential delays.
3. Equipment Utilization: By analyzing the utilization of construction equipment,
queuing theory can guide decisions on whether to acquire additional equipment or
adjust work schedules to enhance efficiency.
4. Worker Productivity: Queuing models can be applied to study the impact of worker
availability and allocation on construction site productivity. This aids in making
informed decisions about labor management.

2
 Department of Civil Engineering, Faculty of Engineering, Bayero University, Kano - Nigeria
Quantitative Techniques (CIV 8341) prepared by Engr. Dr. A. D. Rafindadi

5. Material Handling: Queuing theory is useful in optimizing the flow of materials

through construction processes, including storage, transportation, and distribution, to
minimize waiting times and delays.
6. Traffic Flow on Construction Sites: For large construction sites, queuing theory can
be used to analyze and manage traffic flow, ensuring efficient movement of vehicles
and reducing congestion.
7. Construction Crew Scheduling: Queuing theory can be applied to schedule
construction crews for different activities based on arrival rates of tasks and the crew's
capacity to complete them. This helps optimize resource utilization and reduce idle
time.
8. Material Inspection Queue: Queuing models can be used to analyze the waiting time
of materials in the inspection queue. This helps in determining the appropriate number
of inspectors or optimizing the inspection process to avoid delays.
9. Equipment Maintenance Queue: Queuing theory can be applied to schedule
equipment maintenance tasks, considering the arrival rate of maintenance requests and
the capacity of maintenance crews. This ensures timely maintenance without disrupting
construction activities.
10. Construction Site Security Check: Queuing models can help optimize the security
check process by analyzing arrival rates of personnel and vehicles, determining the
appropriate number of security checkpoints, and minimizing waiting times.
11. Construction Project Task Allocation: Queuing theory can assist in allocating tasks
to available resources based on the complexity of the tasks, arrival rates, and resource
capabilities. This ensures a smooth flow of work through the project.
12. Concrete Mixing Queue: Queuing models can be applied to optimize the concrete
mixing process by analyzing the arrival rate of concrete orders, the capacity of mixing
equipment, and the time required for each mixing operation.
13. Construction Material Delivery Queue: Queuing theory can help manage the
material delivery process by analyzing the arrival rates of deliveries, the unloading
capacity at the receiving area, and optimizing the scheduling of deliveries to prevent
congestion.
14. Construction Site Cafeteria Queue: Queuing theory can be used to analyze the
waiting times in the cafeteria queue, optimize staffing levels during peak hours, and
improve the overall dining experience for workers.

3
 Department of Civil Engineering, Faculty of Engineering, Bayero University, Kano - Nigeria
Quantitative Techniques (CIV 8341) prepared by Engr. Dr. A. D. Rafindadi

Factors Affecting Queuing:

Several factors influence the dynamics of queues. Here are some key factors in queuing:
1. Arrival Rate (λ): The rate at which customers or entities join the queue is known as
the arrival rate (λ). This factor represents the frequency with which new arrivals enter
the system.
2. Service Rate (μ): The service rate (μ) is the rate at which the service provider processes
customers or entities in the system. It reflects how quickly tasks are completed or
services are delivered.
3. Queue Length: The number of customers or entities waiting in the queue at a given
time is the queue length. It is a dynamic variable that changes as arrivals join and
departures leave the queue.
4. Queue Discipline: Queue discipline refers to the rules governing how entities are
served from the queue. Common disciplines include first-come-first-served (FCFS),
last-come-first-served (LCFS), and priority-based.
5. Service Time Distribution: The distribution of service times for each entity in the
queue is a critical factor. The variability in service times can impact the overall system
performance.
6. Queue Configuration: The physical layout and design of the queue can influence how
entities move through the system. Factors include the arrangement of waiting areas, the
number of service points, and the layout of the service facility.
7. Balking: Balking occurs when potential customers decide not to join the queue due to
its perceived length or other factors. Understanding balking behavior is essential in
managing customer satisfaction.
8. Reneging: Reneging happens when a customer who has joined the queue chooses to
leave before being served. This may be influenced by factors such as wait time and
changing priorities.
9. System Capacity: The maximum number of entities that the system can accommodate
at a given time is the system capacity. Exceeding this capacity may lead to congestion
and increased wait times.
10. Queue Management Policies: Policies and strategies employed by the system manager
to control and optimize the queue. This includes decisions related to opening or closing
service points, adjusting staffing levels, and implementing priority rules.

4
 Department of Civil Engineering, Faculty of Engineering, Bayero University, Kano - Nigeria
Quantitative Techniques (CIV 8341) prepared by Engr. Dr. A. D. Rafindadi

11. Customer Behavior: The behavior of customers, including their tolerance for waiting,
preferences for service options, and reactions to delays, can significantly impact
queuing dynamics.
12. Arrival Pattern: The pattern or distribution of arrivals over time can affect queuing
behavior. For example, sporadic bursts of arrivals may lead to temporary congestion.
13. Service Variability: Variability in service times and the unpredictability of how long
each entity will take to be served contribute to the complexity of queuing systems.
14. Information Availability: The availability of information regarding wait times, service
status, and other relevant details can influence customer decisions and behaviors in the
queue.
15. Technology and Automation: The use of technology, such as self-service kiosks or
online queuing systems, can impact the efficiency and management of queues.
Two Moment Approximations:
The basics of two-moment approximations are very popular in industry for obtaining queueing
estimates, i.e., the mean waiting time in a queue and the mean length of a queue. These
approximations can usually only provide means of outputs, i.e., waiting times and queue
lengths, based on three inputs in a standard queue:
• the mean and variance of the inter-arrival time,
• the mean and variance of the service time, and
• the number of servers.
This situation arises frequently in factories, airports, and hospitals, where limited data, i.e.,
only means and variances of the inputs, are available.
Note that the ‘mean’ is the so-called first moment. Thus, if a random variable is denoted by X,
the first moment, E[X], is the mean, while the variance is E[X2] - (E[X])2, where E[X2] is the
so-called second moment. Thus, the variance is not the second moment, but rather the second
moment minus the square of the mean.
Basic Queueing Formulas:
Little's rule provides the following results:
𝐿𝑞
𝐿𝑠 = 𝜆𝑊𝑠 ; 𝐿𝑞 = 𝜆𝑊𝑞 = 𝑊𝑞 =
𝜆
The first of the above applies to the system and the second to the queue, which is a part of the
system. Another useful relationship in the queue is:
1
𝑊𝑠 = 𝑊𝑞 +
𝜇

5
 Department of Civil Engineering, Faculty of Engineering, Bayero University, Kano - Nigeria
Quantitative Techniques (CIV 8341) prepared by Engr. Dr. A. D. Rafindadi

The above equation says the mean wait in the system is the sum of the mean wait in the queue
1
and the service time (𝜇).

Classification of queuing models using Kendal and Lee notations:

Generally, any queuing models may be completely specified in the following symbolic form
a / b/ c: d / e
a: Type of distribution of inter – arrival time
b: Type of distribution of inter – service time
c: Number of servers
d: Capacity of the system
e: Queue discipline
M: Arrival time follows Poisson distribution and service time follows an exponential
distribution.
▪ Model I: M / M / 1: ∞/ FCFS
Where M Arrival time follows a Poisson distribution
M: Service time follows a exponential distribution
1: Single service model
∞: Capacity of the system is infinite
FCFS: Queue discipline is first come first served
▪ Model II: M / M / 1: N / FCFS
Where N: Capacity of the system is finite
▪ Model III: M / M / 1: / SIRO
Where SIRO: Service in random order
▪ Model IV: M / O / 1: / FCFS
Where O Service time follows a constant distribution or is deterministic
▪ Model V: M / G / 1: / FCFS
Where G Service time follows a general distribution or arbitrary distribution
▪ Model VI: M / Ek / 1: / FCFS
Where Ek: Service time follows Erlang distribution with K phases.
▪ Model VII: M / M / K: / FCFS
Where K Multiple Server model
▪ Model VIII: M / M / K: N / FCFS
Model I: M / M /1: / FCFS
1. Utilization factor traffic intensity/Utilization parameter / Busy period

6
 Department of Civil Engineering, Faculty of Engineering, Bayero University, Kano - Nigeria
Quantitative Techniques (CIV 8341) prepared by Engr. Dr. A. D. Rafindadi

𝜆
𝜌=
𝜇
Where λ = Mean arrival rate; μ= mean service rate
Note: μ > λ in single server model only
2. Probability that exactly zero units are in the system
𝜆
𝑃0 = 1 −
𝜇
3. Probability that exactly ‘n’ units in the system
𝜆 𝑛
𝑃𝑛 = 𝑃0 ( )
𝜇
4. Probability that n or more units in the system
𝜆 𝑛
𝑃𝑛 𝑜𝑟 𝑚𝑜𝑟𝑒 =( )
𝜇
more than ‘n’ means n should be n + 1
5. Expected number of units in the queue/queue length
𝜆2
𝐿𝑞 =
𝜇(𝜇 − 𝜆)
6. Expected waiting time in the queue
𝐿𝑞
𝑊𝑞 =
𝜆
7. Expected number of units in the system
𝜆
𝐿 = 𝐿𝑞 +
𝜇
8. Expected waiting time in the system
1
𝑊 = 𝑊𝑞 +
𝜇
9. Expected number of units in queue that from time to time – (OR) non - empty queue
size
𝜇
𝐷=
𝜇−𝜆
10. Probability that an arrival will have to wait in the queue for service
Probability = 1 – Po
11. Probability that an arrival will have to wait in the queue more than w (where w > o),
the waiting time in the queue

7
 Department of Civil Engineering, Faculty of Engineering, Bayero University, Kano - Nigeria
Quantitative Techniques (CIV 8341) prepared by Engr. Dr. A. D. Rafindadi

𝜆
𝑃𝑟𝑜𝑏𝑎𝑏𝑖𝑙𝑖𝑡𝑦 = ( ) 𝑒(𝜆−𝜇)𝑤
𝜇

12. The probability that an arrival will have to wait more than v (v > o) waiting time in the
𝑣
system is 𝑒(𝜆−𝜇)
13. Probability that arrival will not have to wait in the queue for service = Po

Information required to solve the queueing problem:

Example 1: Arrivals of trucks at a construction site are considered to be Poisson at an average

time of 8 minutes between one arrival and the next. The time taken to unload a truck is
distributed exponentially, with a mean of 4 minutes.
Determine:
(a) Expected fraction of the day that the unloading bay will be in use.
(b) Expected number of trucks in the queue.
(c) Expected waiting time in the queue.
(d) Expected number of trucks in the system.
(e) Expected waiting time in the system.
(f) Expected number of trucks in the queue from time to time.
(g) What is the probability that a truck will have to wait in the queue for unloading?
(h) What is the probability that exactly 3 trucks are in the system?
(i) What is the probability that a truck will not have to wait in the queue for unloading?
(j) What is the probability that there are 3 or more trucks in the system?
(k) What is the probability that a truck will have to wait more than 6 minutes in the queue for
unloading?
(l) What is the probability that there are more than 5 trucks in the system?

8
 Department of Civil Engineering, Faculty of Engineering, Bayero University, Kano - Nigeria
Quantitative Techniques (CIV 8341) prepared by Engr. Dr. A. D. Rafindadi

(m) What is the probability that a truck will have to wait more than 8 minutes in the system?
(n) The construction company will install a second unloading bay when convinced that a truck
would have to wait at least 6 minutes in the queue for unloading. By how much should the flow
of arrivals be increased to justify a second unloading bay?
Solution:
The mean arrival rate = λ = 1/8 x 60 = 7.5/hour.
The mean service = μ = 1/4 x 60 = 15/hour.
(a) Expected fraction of the day that the unloading bay will be in use: 0.5 (or 50%)
7.5
𝜌= = 0.5
15
(b) Expected number of trucks in the queue: 0.5 trucks
7.52
𝐿𝑞 = = 0.5 𝑡𝑟𝑢𝑐𝑘𝑠
15(15 − 7.5)
(c) Expected waiting time in the queue: 4.0 minutes
0.5
𝑊𝑞 = = 0.066ℎ𝑟𝑠(4 𝑚𝑖𝑛𝑢𝑡𝑒𝑠)
7.5
(d) Expected number of trucks in the system: 1.0 truck
𝜆
𝐿 = 𝐿𝑞 + = 0.5 + 0.5 = 1 𝑡𝑟𝑢𝑐𝑘
𝜇
(e) Expected waiting time in the system: 8.0 minutes
1 1
𝑊 = 𝑊𝑞 + = 0.066 + = 0.133ℎ𝑟𝑠(8 𝑚𝑖𝑛𝑢𝑡𝑒𝑠)
𝜇 15
(f) Expected number of trucks in the queue from time to time: 2 trucks
𝜇 15
𝐷= = = 2 𝑡𝑟𝑢𝑐𝑘𝑠
𝜇 − 𝜆 15 − 7.5
(g) Probability that a truck will have to wait in the queue for unloading: 0.5 (or 50%)
𝑃𝑟𝑜 = 1 − 𝑃0
𝜆 7.5
𝑃0 = 1 − =1− = 0.5
𝜇 15
𝑃𝑟𝑜 = 1 − 0.5 = 0.5
(h) Probability that exactly 3 trucks are in the system: 0.0625 (or 6.25%)
𝜆 𝑛 7.5 3
𝑃𝑛=3 = 𝑃0 ∗ ( ) = 0.5 ∗ ( ) = 0.0625
𝜇 15
(i) Probability that a truck will not have to wait in the queue for unloading: 0.5 (or
50%)

9
 Department of Civil Engineering, Faculty of Engineering, Bayero University, Kano - Nigeria
Quantitative Techniques (CIV 8341) prepared by Engr. Dr. A. D. Rafindadi

𝜆 7.5
𝑃0 = 1 − =1− = 0.5
𝜇 15
(j) Probability that there are 3 or more trucks in the system: 0.125 (or 12.5%)
𝜆 𝑛 7.5 3
𝑃𝑛=3 𝑜𝑟 𝑚𝑜𝑟𝑒 = ( ) = ( ) = 0.125
𝜇 15
(k) Probability that a truck will have to wait more than 6 minutes in the queue for
unloading: 0.236 (or 23.6%)
7.5 (7.5 6
𝑃𝑟𝑜𝑏𝑎𝑏𝑖𝑙𝑖𝑡𝑦 = ( ) 𝑒 −15)60 = 0.236
15
6
𝜔 = 6 𝑚𝑖𝑛𝑢𝑡𝑒𝑠 = ℎ𝑟𝑠
60

(l) Probability that there are more than 5 trucks in the system: 0.015625 (or 1.5625%)
𝜆 𝑛 7.5 6
𝑃 𝑚𝑜𝑟𝑒 𝑡ℎ𝑎𝑛 5 𝑡𝑟𝑢𝑐𝑘𝑠 = ( ) = ( ) = 0.015625
𝜇 15
(m) Probability that a truck will have to wait more than 8 minutes in the system: 0.3679
(or 36.79%)
8
𝑉 = 8 𝑚𝑖𝑛𝑢𝑡𝑒𝑠 = ℎ𝑟𝑠
60
8
(7.5−15)
𝑃𝑟𝑜𝑏𝑎𝑏𝑖𝑙𝑖𝑡𝑦 = 𝑒 60 = 0.3679

(n) Increase in the flow of arrivals to justify a second unloading bay: To justify a second
booth should be increased from 7.5 to 9 per hour.
6
𝑊𝑞 = ℎ𝑟𝑠 = 0.1 ℎ𝑟𝑠
60
𝐿𝑞 𝜆2 𝜆 𝜆
𝑊𝑞 = = = =
𝜆 𝜇(𝜇 − 𝜆)𝜆 𝜇(𝜇 − 𝜆) 15(15 − 𝜆)
𝜆
0.1 =
15(15 − 𝜆)
𝜆 = 9 𝑝𝑒𝑟 ℎ𝑜𝑢𝑟
Example 2: Consider the following single-server queue: the inter-arrival time is exponentially
distributed with a mean of 10 minutes and the service time is also exponentially distributed
with a mean of 8 minutes, find the following:
A. mean wait in the queue,
B. mean number in the queue,
C. the mean wait in the system,
D. mean number in the system and

10
 Department of Civil Engineering, Faculty of Engineering, Bayero University, Kano - Nigeria
Quantitative Techniques (CIV 8341) prepared by Engr. Dr. A. D. Rafindadi

E. proportion of time the server is idle.

Solution: Let's denote the following parameters:
• λ: Arrival rate (customers per minute)
• μ: Service rate (customers per minute)
Given that the inter-arrival time is exponentially distributed with a mean of 10 minutes, we can
1
calculate the arrival rate (λ) using the formula 𝜆 = mean inter−arrival time =
1
𝑐𝑢𝑠𝑡𝑜𝑚𝑒𝑟𝑠 𝑝𝑒𝑟 𝑚𝑖𝑛𝑢𝑡𝑒
10

Similarly, given that the service time is exponentially distributed with a mean of 8 minutes, we
1
can calculate the service rate (μ) using the formula 𝜇= =
mean service time
1
𝑐𝑢𝑠𝑡𝑜𝑚𝑒𝑟𝑠 𝑝𝑒𝑟 𝑚𝑖𝑛𝑢𝑡𝑒
8

Once we have λ and μ, we can use the following formulas for an M/M/1 queue:
𝐿𝑞 𝜆 0.1
Mean Wait in the Queue (Wq): 𝑊𝑞 = = 𝜇(𝜇−𝜆) = 0.125(0.125−0.1) = 32𝑚𝑖𝑛𝑢𝑡𝑒𝑠
𝜆

𝜆2 0.12
Mean Number in the Queue (Lq): 𝐿𝑞 = 𝜇(𝜇−𝜆) = 0.125(0.125−0.1) = 3.2
1 1
Mean Wait in the System (Ws): 𝑊𝑠 = 𝑊𝑞 + 𝜇 = 32 + 1/8 = 40𝑚𝑖𝑛𝑢𝑡𝑒𝑠
1
10
𝜆
Mean Number in the System (Ls): 𝐿𝑠 = 𝜆𝑊𝑠 = 𝐿𝑞 + = 3.2 + 1
=4
𝜇 8
𝜆 8
Proportion of Time the Server is Idle (P0): 𝑃0 = 1 − 𝜇 = 1 − 𝜌 = 1 − 10 = 0.2

Example 3: Consider the following single-server queue: the inter-arrival time is exponentially
distributed with a mean of 10 minutes and the service time has the uniform distribution with a
maximum of 9 minutes and a minimum of 7 minutes, find the (i) mean wait in the queue, (ii)
mean number in the queue, (iii) the mean wait in the system, (iv) mean number in the system
and (v) proportion of time the server is idle.
1
Solution: We have an M/G/1 system. We also have: 𝜆 = 10; the mean service time will be (7 +
1 (9−7)2 1 8
9)/ 2 = 8, i.e., 𝜇 = 8. The variance of the service time, σ𝟐𝐬 will equal = 3. Also, 𝜌 = 10.
12

Then:
𝜆2 σ𝟐𝐬 + 𝜌2
𝑁𝑢𝑚𝑏𝑒𝑟 𝑖𝑛 𝑡ℎ𝑒 𝑞𝑢𝑒𝑢𝑒, 𝐿𝑞 = = 1.608
2(1 − 𝜌)
𝐿𝑞
𝑊𝑎𝑖𝑡 𝑖𝑛 𝑡ℎ𝑒 𝑞𝑢𝑒𝑢𝑒, 𝑊𝑞 = = 16.08𝑚𝑖𝑛𝑠
𝜆
1
𝑊𝑎𝑖𝑡 𝑖𝑛 𝑡ℎ𝑒 𝑠𝑦𝑠𝑡𝑒𝑚, 𝑊𝑠 = 𝑊𝑞 + = 24.08𝑚𝑖𝑛𝑠
𝜇
11
 Department of Civil Engineering, Faculty of Engineering, Bayero University, Kano - Nigeria
Quantitative Techniques (CIV 8341) prepared by Engr. Dr. A. D. Rafindadi

𝑁𝑢𝑚𝑏𝑒𝑟 𝑖𝑛 𝑡ℎ𝑒 𝑠𝑦𝑠𝑡𝑒𝑚, 𝐿𝑠 = 𝜆𝑊𝑠 = 2.408

𝜆
𝑃𝑟𝑜𝑝𝑜𝑟𝑡𝑖𝑜𝑛 𝑜𝑓 𝑇𝑖𝑚𝑒 𝑡ℎ𝑒 𝑆𝑒𝑟𝑣𝑒𝑟 𝑖𝑠 𝐼𝑑𝑙𝑒, 𝑃0 = 1 − = 1 − 𝜌 = 0.2
𝜇
Example 4: Consider the following single-server queue: the inter-arrival time has a gamma
distribution with a mean of 10 minutes and a variance of 20 min2. The service time has the
normal distribution with a mean of 8 minutes and a variance of 25 min2, find the following:
A. mean wait in the queue,
B. mean number in the queue,
C. the mean wait in the system,
D. mean number in the system and
E. proportion of time the server is idle. Simulation results indicate Wq to be about 8.1
minutes.
1
Solution: We have a G/G/1 system. We also have: 𝜆 = 10; the variance of the inter-arrival time
1
is 20. The mean service time will be 8, i.e., 𝜇 = 8. The variance of the service time, σ𝟐𝐬 s is 25.
8
Also, 𝜌 = 10. Then,

Where:
• Ca2 is the squared coefficient of variation of the inter-arrival time.
• Cs2 is the squared coefficient of variation of the service time.
Calculations:
σ2a Variance of inter − arrival time 20
𝐶𝐚𝟐 = = = 2 = 0.2
1 2
( )2 (mean of inter − arrival time) 10
𝜆
σ2s Variance of service time 25
𝐶𝐬𝟐 = = = = 0.3906
1 2 (mean of service time)2 82
(𝜇 )

𝜌2 (1 + 𝐶𝑠2 )(𝐶𝑎2 + 𝜌2 𝐶𝑠2 )

𝑁𝑢𝑚𝑏𝑒𝑟 𝑖𝑛 𝑡ℎ𝑒 𝑄𝑢𝑒𝑢𝑒, 𝐿𝑞 = = 0.8010
2(1 − 𝜌)(1 + 𝜌2 𝐶𝑠2 )
𝐿𝑞
𝑊𝑎𝑖𝑡 𝑖𝑛 𝑡ℎ𝑒 𝑄𝑢𝑒𝑢𝑒, 𝑊𝑞 = = 8𝑚𝑖𝑛𝑠
𝜆
1
𝑊𝑎𝑖𝑡 𝑖𝑛 𝑡ℎ𝑒 𝑠𝑦𝑠𝑡𝑒𝑚, 𝑊𝑠 = 𝑊𝑞 + = 16.01𝑚𝑖𝑛𝑠
𝜇
𝑃𝑟𝑜𝑝𝑜𝑟𝑡𝑖𝑜𝑛 𝑜𝑓 𝑇𝑖𝑚𝑒 𝑡ℎ𝑒 𝑆𝑒𝑟𝑣𝑒𝑟 𝑖𝑠 𝐼𝑑𝑙𝑒, 𝑃0 = 1 − 𝜌 = 0.2
Using the following equation:
𝜌2 (𝐶𝑎2 + 𝐶𝑠2 )
𝐿𝑞 ≈ 𝑔 = 0.9450𝑔
2(1 − 𝜌)

12
 Department of Civil Engineering, Faculty of Engineering, Bayero University, Kano - Nigeria
Quantitative Techniques (CIV 8341) prepared by Engr. Dr. A. D. Rafindadi

−2(1−𝜌)(1−𝐶𝑎2 )2
𝑔 = 𝑒𝑥𝑝 ( ) when 𝐶𝑎2 ≤ 1, we use the given equation.
3𝜌(𝐶𝑎2 +𝐶𝑠2 )

−2(1 − 𝜌)(1 − 𝐶𝑎2 )2

𝑔 = 𝑒𝑥𝑝 ( ) = 0.8348
3𝜌(𝐶𝑎2 + 𝐶𝑠2 )
𝐿𝑞
Then, Lq = 0.9450(0.8348) = 0.7889, which implies Wq = = 0.7889(10) = 7.889 mins, which
𝜆

is also reasonably close to the simulation estimate of 8.1 mins.

Example 5: A construction project involves pouring concrete for structural elements. Concrete
trucks arrive at the construction site to deliver concrete, and a team of workers is responsible
for pouring the concrete into formwork. The pouring process takes an average of 30 minutes
per truck, and trucks arrive at the site at an average rate of 2 trucks per hour.
Key Variables:
• Arrival Rate (λ): 2 trucks per hour
• Service Rate (μ): 1 truck every 30 minutes (or 2 trucks per hour)
Queuing Model Notation: This scenario can be represented as an M/M/1 queue:
• M: Stands for a Markovian arrival process (Poisson process).
• M: Stands for a Markovian service process (exponential service times).
• 1: Indicates a single server (pouring crew).
Queuing Model Equations:
𝜆 2
1. Utilization (ρ): 𝜌 = 𝜇 = 2 = 1

• The system is fully utilized (ρ = 1).

𝜆 2
2. Average Number of Trucks in the System (L): 𝐿 = 𝜇−𝜆 = 2−2 = −∞

• Since ρ = 1, the average number of trucks in the system is infinite. This implies
that the system is not in a steady state.
1 2
3. Average Time a Truck Spends in the System (W): 𝑊 = 𝜇−𝜆 = 2−2 = −∞

• Similarly, since ρ = 1, the average time a truck spends in the system is infinite.
Interpretation: In this scenario, the pouring crew is fully utilized, indicating that they are
continuously occupied with concrete pouring. However, the queuing model suggests that the
system is not in a steady state, and the average number of trucks in the system and the average
time a truck spends in the system are infinite. This might imply that the crew is unable to keep
up with the incoming rate of trucks, leading to congestion and an unmanageable queue.

13
 Department of Civil Engineering, Faculty of Engineering, Bayero University, Kano - Nigeria
Quantitative Techniques (CIV 8341) prepared by Engr. Dr. A. D. Rafindadi

Action: Construction managers may need to address the situation by considering additional
resources, such as increasing the number of pouring crews or optimizing the pouring process,
to bring the system into a more manageable state.
Example 6: Let's consider a simple queuing system, often referred to as the M/M/1 queue,
where entities arrive according to a Poisson process with rate λ and are served according to an
exponential service time distribution with rate μ. The notation M/M/1 indicates a single server
(1), with exponential arrival and service processes (M stands for memoryless or exponential).
Formulation:
• Arrival Process (Poisson): λ = 4 entities per hour.
• Service Process (Exponential): μ = 5 entities per hour.
• Queue Discipline: First-come-first-served (FCFS).
• Queueing System Capacity: Unlimited (M/M/1).
Performance Measures:
𝜆
1. Utilization Factor (ρ): The fraction of time the server is busy. 𝜌 = 𝜇
𝜆
2. Average Number of Entities in the System (L): 𝐿 = 𝜇−𝜆
1
3. Average Time an Entity Spends in the System (W): 𝑊 = 𝜇−𝜆

4. Probability of Having n Entities in the System (Pn): Pn = (1−ρ) * ρn

Solution Steps:
4
1. Calculate Utilization Factor (ρ): 𝜌 = 5
4
2. Calculate Average Number of Entities in the System (L): 𝐿 = 5−4 = 4
1
3. Calculate Average Time an Entity Spends in the System (W): 𝑊 = 5−4 = 1ℎ𝑜𝑢𝑟

4. Calculate Probability of Having 0 Entities in the System (P0): Pn = (1−0.8) * 0.80 = 0.2
Interpretation:
• The server is busy 80% of the time, indicating a high utilization factor.
• On average, there are 4 entities in the system.
• An entity spends, on average, 1 hour in the system.
• There is a 20% chance that there are no entities in the system.
Multiple-Server Queues:
We will only consider the identical (homogenous) server case in which there are c identical
servers in parallel and there is just one waiting line (i.e., the queue is a single-channel queue).
Let c denote the number of identical servers.

14
 Department of Civil Engineering, Faculty of Engineering, Bayero University, Kano - Nigeria
Quantitative Techniques (CIV 8341) prepared by Engr. Dr. A. D. Rafindadi

Model I: M/M/K: ∞/FCFS

𝜆
𝜌=
𝑐𝜇
For the M/M/c queue (Ross, 2014),
𝜆 𝑐
𝑃0 (𝜇 ) 𝜌
𝐿𝑞 =
𝑐! (1 − 𝜌)2
Where:
𝑐−1
(𝑐𝑝)𝑚 (𝑐𝑝)𝑐
𝑃0 = 1/ [ ∑ + ]
𝑚! 𝑐! (1 − 𝜌)
𝑚=0

Or
1
𝑃0 = 𝑛
1 𝜆 1 𝜆 𝐾 𝑘𝜇
[∑𝐾−1
𝑛=0 ( ) ]+ ( )
𝑛! 𝜇 𝐾! 𝜇 (𝑘𝜇 − 𝜆)
Note that P0 denotes the probability that there are 0 customers in the system.
Hence, Wq can be obtained as follows:
𝐿𝑞
𝑊𝑞 =
𝜆
1
𝑊 = 𝑊𝑞 +
𝜇
𝜆
𝐿 = 𝐿𝑞 +
𝜇
Then, for the G/G/c queue, we have the following approximation (Whitt, 1976; Medhi, 2003):

𝐺/𝐺/𝑐 𝑀/𝑀/𝑐 𝐶𝑎2 + 𝐶𝑠2

𝑊𝑞 ≈ 𝑊𝑞 ,
2
𝐺/𝐺/𝑐
where 𝑊𝑞 denotes the waiting time in the queue for the A/B/c queue. The above works
well for M/G/c queues but does not always work well when the inter-arrival time is not
exponentially distributed. For multi-server queues, it has been shown that data on two moments
is usually not sufficient to generate good approximations for the mean waiting time or queue
length.
Example 7: A construction site has 3 unloading bays for trucks. Trucks are found to arrive in
a Poisson fashion at an average rate of 6 trucks per hour for unloading materials. The unloading
time is found to have an exponential distribution with a mean of 18 minutes. The trucks are
processed on a First-Come-First-Served (FCFS) basis. Calculate:
(a) Average number of trucks in the system

15
 Department of Civil Engineering, Faculty of Engineering, Bayero University, Kano - Nigeria
Quantitative Techniques (CIV 8341) prepared by Engr. Dr. A. D. Rafindadi

(b) Average time a truck spends in the system

(c) Average queue length
(d) How many hours a week can an unloading bay be spent with the trucks.
Solution:
a) Average number of trucks in the system
𝐾=3
𝜆 = 6/ℎ𝑟
1
𝜇= ∗ 60 = 3.33/ℎ𝑟
18
1
𝑃0 =
1 𝜆 𝑛 1 𝜆 𝐾 𝑘𝜇
[∑𝐾−1 ( ) ] +
𝑛=0 𝑛! 𝜇 𝐾! (𝜇 ) (𝑘𝜇 − 𝜆)
1
𝑃0 = 0 1 = 0.145
1 6 1 6 1 6 2 1 6 3 3 ∗ 3.33
( ) + ( ) +
0! 3.33 1! 3.33 2! (3.33) + 3! (3.33) 3 ∗ 3.33 − 6
𝜆 𝐾
𝜆𝜇 (𝜇 )
𝐿𝑞 = ∗𝑃
(𝐾 − 1)! (𝐾𝜇 − 𝜆)2 0
6 3
6 ∗ 3.33 (3.33)
𝐿𝑞 = ∗ 0.145 = 0.532
(3 − 1)! (3 ∗ 3.33 − 6)2
𝜆 6
𝐿 = 𝐿𝑞 + = 0.532 + = 2.334
𝜇 3.33
b) Average time a customer spends in the system
1 𝐿𝑞 1 0.534 1
𝑊 = 𝑊𝑞 + = + = + = 0.388
𝜇 𝜆 𝜇 6 3.33
c) Average queue length
𝐿𝑞 = 0.532
d) Assuming 5 days a week and 8 hrs a day the number of hrs in a week the cash paying
assistant spends with the customers.
𝜆
𝜌=
𝑐𝜇
𝜆 6
=𝜌∗5∗8=( )∗5∗8=( ) ∗ 5 ∗ 8 = 24.02 ℎ𝑟𝑠
𝑐𝜇 3 ∗ 3.33
Example 8: Consider the following scenario: the inter-arrival time has an exponential
distribution with a mean of 10 minutes. There are two servers, and the service time of each

16
 Department of Civil Engineering, Faculty of Engineering, Bayero University, Kano - Nigeria
Quantitative Techniques (CIV 8341) prepared by Engr. Dr. A. D. Rafindadi

server has uniform distribution with a maximum of 20 minutes and a minimum of 10minutes,
find the
A. mean wait in the queue,
B. mean number in the queue,
C. the mean wait in the system,
D. mean number in the system and
E. proportion of time the server is idle.
Results from discrete-event simulation, which are known to be very accurate, show that the
mean waiting time in the queue is 9.5693 minutes. Compute the error in the G/G/c
approximation.
Solution:
This is an M/G/2 system. We have 𝜆 = 1/10; the 𝐶𝑎2 = 1 as a result. The mean service time
will be (10 + 20)/2 = 15, i.e., 𝜇 = 1/15. The variance of the service time, σ2s will equal
(20−10)2 15
= 8.33. Also, 𝜌 = 2∗10 = 0.75. Then:
12

σ2s 8.33
𝐶𝐬𝟐 = = = 0.03
1 2
(𝜇 )2 15

Using the G/G/c approximation, we first assume the queue to be an M/M/c queue and compute
its Lq: Now using the following formula.
1
𝑃0 = = 0.1453
(𝑐𝑝)𝑚 (𝑐𝑝)𝑐
[∑𝑐−1
𝑚=0 + ]
𝑚! 𝑐! (1 − 𝜌)
𝜆 𝑐
𝑃0 (𝜇 ) 𝜌
𝐿𝑞 = = 1.929
𝑐! (1 − 𝜌)2
𝐿𝑞
𝑊𝑞 = = 1.929 ∗ 10 = 19.29
𝜆
Now, we need to transform this to a G/G/2 queue using the approximation equation.

𝐺/𝐺/𝑐 𝑀/𝑀/𝑐 𝐶𝑎2 + 𝐶𝑠2

𝑊𝑞 ≈ 𝑊𝑞 = 19.29((1 + 0.03)/2)) = 9.93
2
𝐺/𝐺/𝑐 𝐺/𝐺/𝑐 1
Then, 𝐿𝑞 = 𝑊𝑞 ∗ 𝜆 = 9.93 ∗ 10 = 0.993. The error in the approximation is:
[9.9376 − 9.5693]
∗ 100% = 3.07%
9.9376
1
Wait in the System = 𝑊 = 𝑊𝑞 + 𝜇 = 9.93 + 15 = 24.93𝑚𝑖𝑛𝑠:

Number in the System = 𝐿 = 𝜆𝑊 = 2.493.

17
 Department of Civil Engineering, Faculty of Engineering, Bayero University, Kano - Nigeria
Quantitative Techniques (CIV 8341) prepared by Engr. Dr. A. D. Rafindadi

1. Wayne L Winston, Operations Research: Applications and Algorithms, 2nd edition, PWS-Kent
Publishing, Boston, 1991.
2. Kahraman and A. Gosavi. On the distribution of the number stranded in bulk-arrival, bulk-service queues
of the M/G/1 form. European Journal of Operational Research, 212(2):352–360, 2011.
3. W. Kraemer and M. Langenbach-Belz. Approximate formulae for the delay in the queueing system
GI/G/1. In Proceedings of the 8th International Telegraphic Congress, volume 2(3), page 235/1 235/8,
Melbourne, Australia, 1976.
4. V. Gupta, M. Harchol-Balter, J.G. Dai and B. Zwart. On the inapproximability of M/G/K: why two
moments of job size distribution are not enough. Queueing Systems, 64(1): 5{48, 2010.
5. W.G. Marchal. An approximation formula for waiting times in single-server queues. AIIE Transactions,
8: 473, 1976.
6. J. Medhi. Stochastic Models in Queueing Theory. Academic Press, Amsterdam, Second edition, 2003.
7. S. M. Ross. Introduction to Probability Models. Academic Press, San Diego, CA, USA, Eleventh edition,
2014.
8. W. Whitt. The queueing network analyzer. Bell System Technical Journal, 62(9):2779-2815, 1983.
9. Taha H A, Operation Research - An Introduction, Prentice Hall of India, 7th edition, 2003
10. Ravindran, Phillips and Solberg, Operations Research: Principles and Practice, John Wiely & Sons, 2nd
Edition
11. D.S.Hira, Operation Research, S.Chand & Company Ltd., New Delhi, 2004

18