RSM 1340

Term 2 2017
Problem Set 1: Queueing

SOLUTIONS
[These questions use specific numbers – yours may have been slightly different.]

1. Pearson Airport has two main runways, one of which is only used for takeoffs and the other
one is only used for landings at any given time. A single runway has a capacity of at most 40
landings per hour. There are several air traffic controllers at Pearson. When a plane is on its
final approach to Pearson, one aircraft controller is assigned to it until it lands. This controller
cannot be assigned to another flight during this time. Final approach is considered to be the final
45 kilometers of a flight. Planes cover this distance in exactly six minutes.

a) What is the minimum number of air traffic controllers needed to handle landings at the airport
when running at full capacity?

By Little's law:
Throughput rate at full capacity is 40 planes per hour.
Flow time is 6 minutes = 1/10 hour.
So, inventory, which is the number of airplanes in final approach, is 40 * 0.1 = 4 airplanes. Thus,
we need at least 4 air traffic controllers.

b) According to Canadian Air Transport Security Authority (CATSA) regulations, air traffic
controllers must have a 15 minute break for every 45 minutes they are directing air traffic.
During peak periods, what is the true minimum number of air traffic controllers needed to handle
landings?

The regulations imply a utilization of at most 75%, so an air traffic controller can only handle
0.75 * 10 = 7.5 landings/hour.
Assuming that there are 40 landings per hour during peak times, 40 / 7.5 = 5.33, or six air traffic
controllers are needed.

c) If too many flights arrive into the Pearson airspace, they are required to fly in a holding
pattern (“circle”) prior to beginning their final approach. Suppose that airplanes arrive randomly
into the Pearson airspace to queue up for landing at a rate of 35 aircraft per hour.
What is the average wait time for an airplane before it can start its final approach (in minutes,
please keep two decimals)?
This is a M/D/1 queue (CVa=1, CVp=0 since final approach takes exactly 6 minutes), with
lambda = 35 airplanes per hour, mu = 40 airplanes per hour, and m = 1 runway.
The wait time = 0.0875 hours = 5.25 minutes.

G/G/s
Parameters
Arrival Rate (1/hr) 35
Var. Coef. Arrivals 1
Service Rate (1/hr) 40
Var. Coef. Service 0
Nr of Servers 1
Results
Load Factor 0.875
Fraction Not Served 0
Thruput (1/hr) 35
Utilization 0.875
Remaining Results for All Customers
Avg Nr in System 3.9375
Avg Nr in Queue 3.0625
Time in System (hr) 0.1125
Wait in Queue (hr) 0.0875
Pr{Wait=0} 0.125

Time in System (min) 6.75

Wait time (min) 5.25

d) On average 250 kg of CO2 emissions per flying hour are generated. What is the total amount
of CO2 produced in one hour by the delays alone before entering the final approach at Pearson
assuming the above arrival rate?

On average, an airplane waits for 5.25 minutes before landing and there are 35 airplanes per
hour arriving. So, 35 * 5.25 / 60 = 3.0625 hours is the total wait time of all arrivals per hour.
Then 250 * 3.0625 = 765.625 kg of CO2 emissions per hour due to delays before entering the
final approach.
2. The Toronto Public Library currently offers 3D printing in three branches (Toronto Reference
Library (2 printers), Fort York, and Scarborough Civic Centre). The library plans to add one
additional 3D printer in a fourth branch. It expects that the demand rate will be 1.5 users per
hour on average at this location.

There are two different options for the 3D printer they could purchase. These differ in
manufacturer and the resin the machine uses. Under option A, prior to printing there is a setup
time of two minutes, and the average printing time for a typical object is 20 to 40 minutes
(uniformly distributed). Under option B, the setup time is longer, at 10 minutes, though the
printing time is shorter, between 5 and 35 minutes (uniformly distributed).

1
The standard deviation for a continuous uniform distribution is � × (𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀 − 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀)
12
where MaxValue is the largest possible value and MinValue is the smallest.

After the printing is completed (under either option), the 3D object has to be placed in a UV light
curing box for 36 minutes to increase the strength of the object. This time depends on the
intricacy of the object and varies with a standard deviation of 6 minutes. There is only room for
one object in the box.

a)
i. What is the service rate under option A in users per hour?
Mean service time, p = (40+2 + 20+2)/2 = 32 minutes, service rate = 1/p = 60/32 = 1.875
users/hr.
ii. What is the CVp under option A?
standard deviation = sqrt(1/12) * (42-22) = 5.77
CVp = st dev / mean = 5.77/32 = 0.18
iii. What is the average wait time in queue for a user (in minutes)?
Demand = 1.5 customers/hr = 1.5/60 = 0.025 customers/hr.
G/G/1, CVa = 1.
66 minutes, see QMacros below.

b) What is the wait time for using the printer using option B?
i. What is the service rate under option B in users per hour?
Mean service time, p = (35+10 + 5+10)/2 = 30 minutes, service rate = 1/p = 60/30 = 2 users/hr.
ii. What is the CVp under option B?
standard deviation = sqrt(1/12) * (35-15) = 8.66
CVp = st dev / mean = 8.66/30 = 0.289
iii. What is the average wait time in queue for a user (in minutes)?
G/G/1, CVa = 1.
49 minutes, see QMacros below.
c) What is CVa and the average wait time (in minutes) at the curing stage if
i. Option A is used?
CVa2 propagation formula with rho1 = 0.8, CVa1 = 1, and CVp1 = 0.18.
(CVa2)2= 0.82 * 0.182 + (1- 0.82) * 1 = 0.381
CVa2= 0.617

FTq = 66 minutes, see QMacros below.

ii. Option B is used?
CVa2 by propagation formula with rho1 = 0.75, CVa1 = 1, and CVp1 = 0.289.
(CVa2)2 = 0.752 * 0.2892 + (1- 0.752) * 1
= 0.484
CVa2 = 0.696

FTq = 83 minutes, see QMacros below.

d) Which option (A or B) should be chosen to minimize the average total time in system?
Total time (A) = 98 + 102 = 200 minutes.
Total time (B) = 78.6 + 118.9 = 197.5 minutes.
So, choose B.

G/G/s
Parameters
Arrival Rate (1/min) 0.025 0.025 0.025 0.025
Var. Coef. Arrivals 1 1 0.617116953 0.695970545
Service Rate (1/min) 0.03125 0.033333333 0.027777778 0.027777778
Var. Coef. Service 0.180421959 0.288675135 0.166666667 0.166666667
Nr of Servers 1 1 1 1
Results
Load Factor 0.8 0.75 0.9 0.9
Fraction Not Served 0 0 0 0
Thruput (1/min) 0.025 0.025 0.025 0.025
Utilization 0.8 0.75 0.9 0.9
Remaining Results for All Customers All Customers All Customers All Customers
Avg Nr in System 2.451058626 1.966265857 2.553080698 2.972582724
Avg Nr in Queue 1.651058626 1.216265857 1.653080698 2.072582724
Time in System (min) 98.04234502 78.65063426 102.1232279 118.9033089
Wait in Queue (min) 66.04234502 48.65063426 66.12322792 82.90330894
Pr{Wait=0} 0.2 0.25 0.200808699 0.168001559
Option A B Cure | A Cure | B
3. Toronto Helicopter (TH) offers sightseeing tours of Toronto, every day from 9am to 5pm.
They have five helicopters available for tours starting and landing at the Toronto Island Heliport.
Tour lengths vary between 15 and 35 minutes depending on the sightseeing tour itself and air
traffic, with the average tour being 20 minutes in duration. After each tour, TH must check the
helicopter and refuel. This turnaround time depends on the number of staff members working.

TH has found the turnaround time to be 20 minutes less one minute for each ground crew
member working, e.g., if five ground crew members are working the turnaround time would be
15 minutes. For simplicity, assume that if needed all of the helicopters can land at the Heliport at
the same time and the turnaround time does not depend on the number of helicopters currently
turning around, i.e., all five helicopters could turnaround in 15 minutes if 5 ground crew
members are working. Ground crew members are paid $40 per hour.

Assume the coefficient of variation of the flight plus turnaround time is 0.5.

While many tourist groups have booked their flights in advance, others arrive randomly, leading
to an average demand for eight requests per hour. The coefficient of variation of the demand is
0.5.

TH’s operating margin (not including ground crew costs) is $100 per flight. However, they
believe they lose $2 in future business for each minute of waiting time.

a) What is the maximum number of flights the company can offer per day if 10 ground crew
members are working?

Service times = 30 minutes in total, so mu = 2 flights per hour. With 5 helicopters and 8 hours
per day, 2 x 5 x 8 = 80 flights per day.

b) How many ground crew should they hire to maximize their net operating margin? What is the
expected net operating margin per hour?

This is a G/G/s queue. See QMacros below.

Using trial and error, you can find the number of ground crew should be 8, .i.e, a turnaround
time of 12 minutes resulting in a service rate of 1.875 flights per hour.

This yields an expected net operating margin of:

8 groups/hr x $100/group – 8 groups/hr x 6.25 min. wait time per group x $2 per min.
– 8 crew members x $40/hr per crew member = $380 per hour.
 G/G/s
Parameters
Arrival Rate (1/hr) 8
Var. Coef. Arrivals 0.5
Service Rate (1/hr) 1.875
Var. Coef. Service 0.5
Nr of Servers 5
Results
Load Factor 0.853333333
Fraction Not Served 0
Thruput (1/hr) 8
Utilization 0.853333333
Remaining Results for All Customers
Avg Nr in System 5.099821097
Avg Nr in Queue 0.833154431
Time in System (hr) 0.637477637
Wait in Queue (hr) 0.104144304
Pr{Wait=0} 0.587579008

c) Suppose there are 8 ground crew members. Further assume that 8 groups of tourists arrive
per hour randomly without the need of booking in advance and the tourists can choose their
own flight durations (leading to a CVp=1). Under these changed conditions, the company
decides to recommend groups of tourists to come back at a later time if too many groups of
tourists are waiting.

i. What is the maximum number of groups the company should allow to wait if they want to
maximize their net operating margin (including crew and waiting costs)?

This is a M/M/s/N queue. See QMacros below.

Using trial and error, you can find the number of places in the system, N = 7, so at most 2
groups should be kept waiting.

ii. What is the expected net operating margin?

The expected net operating margin is

7.04 groups/hr x $100/group – 7.04 groups/hr x 3.24 min. wait time per group x $2 per min.
– 8 crew members x $40/hr per crew member = $338 per hour.
 M/M/s/N
Parameters
Arrival Rate (1/hr) 8
Service Rate (1/hr) 1.875
Nr of Servers 5
Nr of Places in System 7
Results
Load Factor 0.853333333
Fraction Not Served 0.11993454
Thruput (1/hr) 7.040523676
Utilization 0.750989192
Remaining Results for Served Only
Avg Nr in System 4.135363331
Avg Nr in Queue 0.38041737
Time in System (hr) 0.587365872
Wait in Queue (hr) 0.054032539
Pr{Wait=0} 0.653147032