Practice Problems on Queues

Problem 1.
Five executives share the services of a typist. The average number of jobs submitted for typing is six per
day per executive and a job takes 15 minutes on average to type. Compute the following performance
measures for the typist: the average utilization, the average waiting time for each job, and the average
number of jobs waiting to be typed. A day consists of 8 hours. Assume that arrivals are Poisson and that
service times follow a negative exponential distribution.

Problem 2.
Kristen offers lunch specials on weekends. (This has nothing to do with the cookie business.) She
manages the weekend business by herself. On a typical weekend she receives on the average about 6
orders per hour. It takes Kristen on the average 7.5 minutes to prepare and box an order (this time is the
service time; it does not include the waiting time of the customer.) Kristen's contribution per order is
$3.00. Assume that the arrival process of orders follows a Poisson distribution and that the time to fill an
order is exponentially distributed.

(a) What are the arrival rate and service rate in this example?

(b) What is the average time that a customer spends in the system (waiting time + time taken to work on
the order)?

(c) What is the probability that a customer's order will be filled without waiting in line?

(d) Kristen estimates that by advertising the lunch special she can increase the volume of orders to 14.4
orders per hour. She can hire additional persons to help. One key consideration is that the average
time taken to fill an order (Ws) should not exceed 30 minutes. Each extra person will be paid $8.50
per hour. Compute the number of additional persons needed to handle the extra business.
Recommend whether Kristen should go in for the advertising and higher volume. Assume that
Kristen can advertize free of charge in the campus newspaper. Show all calculations.

Problem 3.
Sharp Discounts Wholesale Club has two service desks, one at each entrance of the store. Customers
arrive at each service desk at an average of one every six minutes. The average service time at each
service desk is four minutes per customer.
(a) What percentage of the time is each service desk idle?
(b) What is the probability that both service clerks are busy?
(c) What is the probability that both service clerks are idle?
(d) How much time does a customer spend at the service desk (waiting + service time)?
(e) How many customers, on average, are waiting in line in front of each service desk?

Problem 4. (continuation of Problem 3)

Sharp Discounts Wholesale Club is considering consolidating its two service desks into one location,
staffed by two clerks. The clerks will continue to work at the same individual speed of four minutes per
customer.
(a) How many customers, on average, are waiting in line?
(b) How much time does a customer spend, on average, at the service desk (waiting + service)?

Vishal Gaur, Johnson Graduate School of Management, Cornell University, vg77@cornell.edu

1
 (c) What are the advantages to Sharp Discounts Wholesale Club from consolidating the service
desks?

Problem 5.
L. Winston Martin (an allergist in Tucson) has an excellent system for handling his regular patients who
come in just for allergy injections. Patients arrive for an injection and fill out a name slip, which is then
placed in an open slot that passes into another room staffed by one or more nurses. The specific injections
for the patient are prepared, and the patient is called through a speaker system into the room to receive the
injections. At certain times during the day, patient load drops and only one nurse is needed to administer
the injections.
Let’s focus on the case in which there is only one nurse. Patients arrive once every three minutes on
average. It takes the nurse an average of two minutes to prepare and administer the patient’s injection.
Assume that patient arrivals are Poisson distributed and service times have a negative exponential
distribution.
(a) What is the average number of people you would expect to see in Dr. Martin’s office?
(b) How much time on average does a patient spend at Dr. Martin’s office?
(c) What is the utilization of the nurse?
(d) Assume that three nurses are available on one particular day. Patients still arrive at the same rate,
and each nurse takes an average of two minutes to prepare and administer a patient’s injections.
What is the average total time of a patient in the system?

Problem 6.
A study aid desk staffed by a graduate student has been established to answer students’ questions and help
in working problems prior during the final exam week. The desk is staffed eight hours per day. The dean
wants to know how the facility is working. Statistics show that students arrive at a rate of four per hour,
and the distribution is approximately Poisson. The assistance time for a student has an average of 10
minutes, and follows a negative exponential distribution. Assume that the all assumptions are satisfied, so
that queuing formulas can be applied.
(a) Calculate the percentage utilization of the graduate student.
(b) Calculate the average number of students in the system.
(c) Calculate the average time in the system.
(d) Before a test, the arrival of students increases to 8.4 per hour on average. The course instructor
decides that the average waiting time of students in line must be less than 5 minutes. How many
study aid desks should be set up to meet this target?

Problem 7. (Optional. This is an advanced problem to convey the idea of pooling.)

General Hospital is a large urban hospital. Its management is considering the consolidation of two of
their surgical services, cardiac surgery and thoracic surgery, into one unit since patients in both services
experience chest surgery and require similar nursing skills. The admissions and average length of stay
(ALOS) profiles of the two types of patients are quite different. Cardiac patients are more prevalent with
an average arrival rate of 1.11 per day versus 0.32 for thoracic patients. Cardiac patients also stay, on
average, more than twice as long with an ALOS of 7.7 days versus 3.8 for thoracic patients. In addition,
cardiac patients generally have a greater risk of death if not treated within a few days. While the fees that
the hospital charges are comparable for both types of patients, the costs associated with cardiac patients
are higher. The hospital feels that it will be able to improve the level of patient service and reduce cost at
the same time by consolidation.

a) General Hospital uses bed utilization as a measure of efficiency. Their management has traditionally
determined the number of required beds in each unit based on a target utilization of 90%. (Assume
that when the target cannot be met exactly, a higher utilization is preferable to one lower than 90%.)
Given this target, how many beds will General Hospital save by consolidating these two services?
Vishal Gaur, Johnson Graduate School of Management, Cornell University, vg77@cornell.edu
2
b) Dr. Good, the new CEO of General Hospital, believes that, in the light of increased competition,
patient service should be emphasized. Given this new focus and the observation that long delays for
surgical beds can result in backups in the emergency room, he has decided that the expected delay for
a surgical bed should not exceed 2 days. How many beds can be saved through consolidation given
this new standard?

Vishal Gaur, Johnson Graduate School of Management, Cornell University, vg77@cornell.edu

3
 SOLUTIONS TO PRACTICE PROBLEMS ON QUEUES

(In the questions with more than one server, your answer may not match mine exactly due to rounding
errors and interpolation. There will be no penalty for such differences in the test.)

Problem 1: Queuing Systems

Arrival rate = 6/day/executive x 5 executives = 30 jobs per day

Service rate = 4 jobs per hour x 8 hours per day = 32 jobs per day

With one typist in the typing pool, this is a single channel single phase system. Thus, M/M/1 system
applies.

Utilization = 30/32 = 93.75%.

The average number of jobs waiting to be typed, Lq = 14.06 jobs.
The average waiting time for each job, Wq = Lq /  = 0.47 days.
The remaining two performance measures that we discussed in class are: Ls = Lq + / = 15 jobs, and Ws
= Ls /  = 0.5 days.

Problem 2: Queuing Systems

(a) Arrival rate,  = 6 orders/hour. Service rate,  = 60/7.5 = 8 orders/hour.

(b) Ws = 1 / ( – ) = 0.5 hours

(c) This is the probability that Kristen is idle = 1 –  = 0.25

(d) A minimum of two servers will be needed to serve the customers if the arrival rate is 14.4/hr (because
utilization must be less than 1). The following table gives the performance characteristics of different
number of servers. Note that this is a multi-channel system (M/M/S), so that we must use the table to
compute performance measures.

  Number of Utilization
(orders/hr) (orders/hr) servers, S  = /(S) Lq Ls Ws (hours)
14.4 8 2 0.9 7.67 9.47 0.66
14.4 8 3 0.6 0.53 2.33 0.16

In order that the average time taken to fill an order, Ws, is less than 30 minutes, Kristen should hire two
more people. This will cost 2x8.5 = $17 per hour.

Increase in contribution = (14.4 – 6) x 3 = $25.2 per hour

Since the increases in contribution exceeds the hiring cost, Kristen should go in for advertising and higher
volume.

Vishal Gaur, Johnson Graduate School of Management, Cornell University, vg77@cornell.edu

4
Problem 3.
Since the two service desks are located separately, this system has two servers with independent queues.
Thus, we can treat each service desk as a single-channel single-phase system. (This is similar to the
scenario in First City National Bank in which each teller has a separate queue).

 = 10/hr  = 15/hr
(a) Idle time = 1 – / = 33.33% of the time.
(b) Probability that both clerks are busy = (/)2 = (0.6667)2 = 0.4444.
(c) Probability that both clerks are idle = (1 – /)2 = (0.3333)2 = 0.1111.
(d) Time in the system, Ws = 1/( – ) = 12 mins.
(e) Waiting time, Wq = Ws – 1/ = 8 mins.
From Little’s Law, Length of queue, Lq = Wq = 1.33 persons.

Problem 4.
Now there is a single queue for the two service desks. Thus, we have a two-channel (M/M/2) system.

 = 20/hr  = 15/hr M = 2.
(a) We have / = 1.33. To look up in the table, we round it up to 1.4. There are two servers. Thus,
from the table, we see that the length of queue, Lq = 1.3449.
(b) We apply Little’s Law to get Wq = Lq/ = 0.0672 hours.
Thus, the total time in the system is given by Ws = Wq + service time = 0.0672 hr + 1/ = 0.134
hours = 8.03 mins.
(c) Due to risk pooling, the total time spent by a person at the service desk is reduced by about 4
mins. The # of people waiting in queue is also reduced from 2*1.33 (for two queues) to 1.3449
persons.

Problem 5.
We have  = 20/hr and  = 30/hr. Questions (a)-(c) consider a single channel single phase (M/M/1)
system and Question (d) considers a multi-channel system.
(a) Average number of patients in the system, Ls = 2.
(b) Average time in the system, Ws = 6 mins.
(c) Utilization of the nurse = 66.7%.
(d) From the table, Lq = 0.0112. Hence, Wq = Lq/ = 0.0112/20 = 0.00056 hrs = 0.0336 mins. Therefore,
Ws = Wq + service time = 2.0336 mins.

Problem 6.
 = 4/hr  = 6/hr

(a)  = / = 0.667.

(b) Ls = 2 students.
(c) Ws = 0.5 hr = 30 mins.
(d) We now have a multi-channel, single-phase system. Thus, we should use the table to compute
performance measures. This question asks us to back calculate the required number of servers given
arrival rate, service rate and a waiting time target. We have
 = 8.4/hr, µ = 6/hr.
Thus, /µ = 8.4/6 = 1.4.
The target is that Wq ≤ 5 minutes. Converting into hours, we get Wq ≤ 1/12 hr.
By applying Little’s Law, we see that this target is equivalent to having Lq less than or equal to
*[1/12 hr] = 0.7.

Vishal Gaur, Johnson Graduate School of Management, Cornell University, vg77@cornell.edu

5
 From the table, in the row for /µ = 1.4, we see that 2 servers gives us Lq = 1.34 and 3 servers gives
us Lq = 0.178. Therefore, the answer is 3 servers.

Problem 7.
(a) Arrival rate of cardiac surgery patients, c = 1.11 patients/day
Arrival rate of thoracic surgery patients, t = 0.32 patients/day
Service rate of cardiac surgery patients, c = 1/7.7 = 0.13 patients/day
Service rate of thoracic surgery patients, t = 1/3.8 = 0.26 patients/day

Total arrival rate at the consolidated unit, cons = 1.11 + 0.32 = 1.43 patients/day
Average service time at the consolidated unit, 1/ cons = (1.11(7.7) + 0.32(3.8)) / 1.43 = 6.8273 days
 cons = 0.1465 patients/day

Each bed is equivalent to one server. We want  = /S ≥ 90%.

Number of beds in the cardiac unit, Sc  1.11/(0.90*0.13) = 9.4872  Sc = 9 beds

Number of beds in the thoracic unit, St  0.32/(0.90*0.26) = 1.3675  St = 2 beds

(Note : St = 1 gives a utilization > 1. Therefore, St = 2 for stability.)

Total number of beds = 9 + 2 = 11 beds

Number of beds in the consolidated unit, Scons  1.43/(0.90*0.1465) = 10.8457  Scons = 10 beds

 Hospital can save one bed by consolidation.

(b) The new standard is that Wq  2 days.

This implies that for the cardiac surgery unit, Lqc = c * Wq  1.11 * 2 = 2.22
For the thoracic surgery unit, Lqt = t * Wq  0.32 * 2 = 0.64
For the consolidated surgery unit, Lqcons = cons * Wq  1.43 * 2 = 2.86

For cardiac patients:

Sc ≥ 9 for stability ( = /S < 1).
Sc = 10,  = 85.38%, from M/M/s Lq Table with Sc = 10,  = 86%, Lqc = 3.42 > 2.22.
Sc = 11,  = 77.62%, from M/M/s Lq Table with Sc = 11,  = 78%, Lqc = 1.22 < 2.22.
Thus, the answer is Sc = 11.

For thoracic patients:

St ≥2 for stability ( = /S < 1).
St = 2,  = 61.54%, from M/M/s Lq Table with St = 2,  = 62%, Lqt = 0.77 > 0.64.
St = 3,  = 41.03%, from M/M/s Lq Table with St = 3,  = 42%, Lqt = 0.11 < 0.64.
Thus, the answer is St = 3.

Sc + St = 14.

For consolidated case:

Scons ≥ 10 for stability ( = /s < 1).
Scons = 11,  = 88.74%, from M/M/s Lq Table with Scons = 11,  = 88%, Lqcons = 4.36 > 2.86.
Sc = 12,  = 81.34%, from M/M/s Lq Table with Sc = 12,  = 82%, Lqcons = 1.90 < 2.86.
Vishal Gaur, Johnson Graduate School of Management, Cornell University, vg77@cornell.edu
6
 Scons = 12.

Therefore, the hospital can free 2 beds by consolidation.

Vishal Gaur, Johnson Graduate School of Management, Cornell University, vg77@cornell.edu

7