b) What is the expected total runtime of a program?

c) What is the average number of programs waiting in the system queue?

Solution. The system is M/M/1 with λ = 10 jobs per minute and µ = 12 jobs per minute. It
It will assume that the system is open and that the capacity is infinite. As ρ = 10/12 < 1, the system
it will reach the steady state and the formulas obtained in class can be used.

a) The server will be unoccupied 1−5/6 = 1/6 of the total, that is, 10 seconds every minute (already
The computer is busy 5×10 = 50 seconds per minute.
b) Total average time is W= 1
µ(1−ρ) = 1
12(1−5/6)
1/2 minute per program.
ρ2 4.16 jobs.
c) The average number of programs waiting in the queue is Lq= 1−ρ

4. The bank window completes transactions in an average time of 2 minutes. The clients
they arrive at an average rate of 20 customers per hour. If it is assumed that the arrivals follow a process of
Poisson and service time is exponential, determine

a) The percentage of time that the cashier is unoccupied.

b) The average waiting time of customers in line.
c) The fraction of customers that must wait in line.

Solution. M/M/1 system with λ = 20 and µ = 30.

a) P(idle cashier) = p01−ρ= 1/3. The cashier is idle 33% of the time.
b)Wq1/15 = 4 minutes.
c)L= 2,Lq= 4/3, therefore the fraction of customers who must wait in line isq
2/3 is approximately 66.6%.

5. A grocery store is attended by one person. Apparently, the pattern of arrivals of

customers during Saturdays behave according to a Poisson process with a arrival rate of
10 people per hour. Customers are served in a FIFO order and due to the prestige
from the store, once they arrive they are willing to wait for the service. It is estimated that the time that
The time taken to serve a customer is exponentially distributed, with an average time of 4 minutes.
Determine:

a) The probability of there being a queue.

b) The average length of the waiting line.
c) The average time a customer spends in line.

Solution. M/M/1 System with λ= 10 and µ= 15.

a) P(waiting line) = 1−p0−p1= 4/9.

b)Lq4/3 people in line.
c)Wq2/15 hours = 8 minutes on average in line.

2
6. In a factory there is a Social Security office that workers have access to during
the working hours. The personnel manager, who has observed the influx of workers at the window,
It has been requested that a study be conducted regarding the operation of this service. A person is appointed to a
specialist to determine the average waiting time of workers in line and the duration
average of the conversation that each one maintains with the window employee. This analyst
arrives at the conclusion that during the first and last half hour of the day the influx is
very reduced and fluctuating, but during the rest of the day the phenomenon can be considered
stationary. From the analysis of 100 periods of 5 minutes, successive or not, but situated in the phase
stationary, it was deduced that the average number of workers attending the window was 1.25 per
period and that the time between arrivals followed an exponential distribution. A similar study on
the duration of the conversations led to the conclusion that they were distributed exponentially with
average duration of 3.33 minutes. Determines:

a) Average number of workers in line.

b) Average waiting time in line.
c) Compare the time lost by the workers with the time lost by the office worker. Calculate the
cost to the company, without an hour of inactivity of the clerk is worth 250 euros and one hour of
Worker 400 euros. Would it be profitable to set up another window?

M/M/1 system with λ = 0.25 and µ = 0.3.

a)Lq4.166 workers.
b)Wq 16.66 minutes.
c) During each hour, there is, on average, Lq 4.166 customers queuing. That is, the hourly cost
for idle workers is 4.166×400 = 1666.66 euros. On the other hand, 1−ρ = 0.166, so
the cost of the time the office worker is idle is 250×0.166 = 41.5 euros per hour,
which is much inferior.
If another window were added, the system would be M/M/2. In that case, p0= 0.411yp1= 0.34, of
the time of clerical staff that would be lost each hour would be, on average, 2p0+p1= 1.166
hours. This implies a cost of 291.5 euros per hour. On the other hand, each hour there would be,
media,Lq1.01 workers in line. Thus, the time lost by the workers would have a
cost of 400×1.01 = 404 euros per hour.
The sum of the two costs is much lower in this second case, so it would be profitable.
put another window.

A banking entity is considering the possibility of installing an ATM network in one of its branches.
Since the attendance of the public demanding this service is unknown, place a single
cashier for a month. Daily data is collected on the arrival times of customers,
just as from the service times. Assuming that the branch is located in a neighborhood
since there is no other similar service, the customer who arrives prefers to wait to use the ATM,
when this is busy.
After the timely analysis of the collected data, it is estimated that: (i) the arrivals follow a process of
Poisson; (ii) the distribution of service time is exponential; (iii) the average elapsed time

3
 The time between two consecutive arrivals is 7.5 minutes; (iv) the average service time is 5 minutes.
per client.
Calculate:

a) Average waiting time that each customer must endure in line.

b) Average queue size and probability that when going to the cashier there is already someone there
cola.

Solution. M/M/1 system with λ = 1/7.5 and µ = 1/5.

a)Wq10 minutes.
b) The average number of queues isq1.33 people, and the probability that there are at least
two people in the system is 1−p0-p1= 4/9.

The workers of a factory have to take their work to the quality control department.
before the product reaches the final stage of the production process. There are a large number of employees
and the arrivals are approximately 20 per hour, following a Poisson process. The time for
Inspecting a piece follows an exponential distribution with a mean of 4 minutes. Calculate the number.
means of workers in quality control if there are:

2 inspectors.
b) 3 inspectors.

Solution.

a) M/M/2 system with λ= 20 and µ= 15. Therefore, L= 2.4 employees.

b) M/M/3 system with λ= 20 and µ= 15. Then L= 1.87 employees.

9. An airplane takes about 4 minutes on average to land from the moment the control tower
gives the landing signal. If the arrivals of the planes occur on average, at a rate of 8
For now and following a Poisson process, how long will the pilot expect to wait circling the airport?
before receiving the land signal?
Solution. M/M/1 System with λ = 8 and µ = 15. Therefore, Wq 4.56 minutes.

10. A computer company has a mainframe that clients can access through
about some terminals (of different types) that are rented. A client wants to determine the speed
optimal terminal that I should rent. The client's jobs are generated according to a process of
Poisson with a rate of 50 programs per day of 8 hours. The average size of a program is
1000 sentences. It is known that the reading time of sentences is exponential. The client estimates
10 euros is the cost of delaying a program by one day. The company estimates a speed of 100
sentences per minute, and any similar increase raises the daily rental price of
terminal at 100 euros. Determine the optimal speed of the terminal.
M/M/1 System with λ = 50 programs sent per day and µ = ? programs executed
per day (decision variable). The cost of delaying a program by one day is 10 euros. Like 100

4
 sentences per minute equals 48 programs per day, then the cost per program per unit
The daily increase is 100/48 = 2.083 euros.
Therefore, the total cost is equal to
10L+ 2.083µ
What is maximized is µ = 52.19 programs read per day.

11. A railway company paints its own cars as they are needed, in its own
workshops where painting is done by hand one by one at a speed that is distributed according to an expo-
The average revenue every 4 hours and an annual cost of 4 million euros has been determined.
Wagons can arrive according to a Poisson process with an average of one every 5 hours. Furthermore, the cost per
Each inactive carriage costs 500 euros per hour.
Two other possibilities are proposed. One is to commission this work to a painting company that
spraying with aerosol with the consequent time saving. However, the budget for this second
The alternative is 10 million euros annually. In this case, the process approximates that of a Poisson.
at a rate of one every 3 hours. The other option is to set up another workshop exactly like the one that exists.
Currently, with the same service rate and annual cost that allows painting two cars at the same time.

In all cases, the work is considered uninterrupted, that is, one works 24×365 = 8760 hours.
annual. Which of the three procedures is preferable?
Solution. Workshop with hand painting: system M/M/1 with λ = 1/5 and µ = 1/4 wagons/hour. By
so much
4 times 106
CT= 500L+ = 2456.62 euros per wagon.
8760
Workshop with spray paint: system M/M/1 with λ = 1/5 and µ = 1/3 wagons/hour. Therefore,

10×106
CT= 500L+ 1891.55 euros per wagon.
8760

Two workshops with hand painting: system M/M/2 with λ = 1/5 and µ = 1/4 wagons/hour. Therefore,

8x106
CT= 500L+ 1389.43 euros per wagon.
8760

Coste anual por hora =2456.62. Coste con aerosol =1891.55. Coste con dos servidores =1389.43.
Therefore, it is preferable to hold two workshops.

12. A computer repair company receives an average of 10 repair requests per day,
which are distributed according to a Poisson process. It is assumed that µ is the repair rate of
the repair person for computers/day, and the repair time is exponential. Each unit of
repair speed incurs a cost of 100 euros per week. Additionally, it has been estimated that the
the cost of having unrepaired computers is 200 euros per computer per week, this cost being
proportional to time. Assuming that a week has five working days, it is requested:

a) What determines the optimal repair speed.

b) What determines if it would be more economical to have two people, each at half the speed
determined in the previous section.

5
 Solution. M/M/1 System with λ= 10 and µ=?.

a) The total cost is 20µ + 40 euros daily. Therefore, the optimal rate is 14.47 computers.
day, with a daily cost of 378.43225 euros.
b) In this case, it is an M/M/2 system. Therefore, the daily cost of 395,184 euros, which is
worse than the previous one.

13. An aircraft maintenance base has resources to review only one engine of
airplane at the same time. Therefore, in order to return the airplanes as soon as possible, the policy followed consists of
to postpone the inspection of the 4 engines of each airplane. In other words, only one is inspected.
airplane engine every time a plane arrives at the base. With this policy, planes arrive according to
a Poisson distribution with an average rate of one per day. The time required to check an engine (a
once the work starts) has an exponential distribution with a mean of 1/2 day. It has been made
a proposal to change the review policy so that the 4 engines are reviewed in a manner
consecutively every time a plane arrives at the base. Even though this would mean quadrupling the
expected service time, each aircraft would need to be checked only with a frequency of 4
times smaller. Use queue theory to compare the 2 alternatives.
Solution. In both cases, we are dealing with M/M/1 queues, since both the times between arrivals and
Service times are random variables with an exponential distribution.
In the current situation, the arrival rate is λ = 1 airplane per day and the service rate is μ = 2 airplanes.
On the day. With these parameters, ρ= 0.5<1, and therefore a steady state exists. The values of
the interest amounts are:

L= 1airplane, Lq= 1/2 airplane, W = 1 day, Wq= 1/2d´ıa

If the proposal to change the policy were followed, the queue would still be an M/M/1, but now
the arrival and service rates would be λ = 0.25 airplanes per day, and µ = 0.5 airplanes per day, respect-
actively. In this case, ρ remains 0.5 and therefore the steady state still exists.
the values of the interest amounts are:

L= 1 airplane, Lq= 1/2 airplane, W= 4 days, Wq2 day

With the proposed configuration, every time a plane is to be inspected, it will go through the system.
the quadruple of the time spent with the previous system, but since each plane will go with a
frequency four times lower, the lost time in the workshop in the long term will be the same. In this case,
The decision between one configuration and the other should be made based on operating costs.