A Discrete-Event Analysis by Vitalis Okafor

Introduction : The Car Wash Problem

Simulation Model
Performance Analysis
Operating Cost Analysis
Observations
Questions: #1 to #4
Processes: Markovian (exponential distribution)
Arrival rate, (average incoming cars): 12 cars per hour
Service rate, (serving average) 6 cars per hour
Mean Time to Failure (MTTF) 100 hours
Mean Time to Repair (MTTR) 4 hours
Resources: 2 Car Wash Machines
Queue capacity (parking space/ line) 4 cars
Operation:
8 hours per day; 300 days per year
Maintenance Costs: $16,000 per year per machine
Customer time value: $20 per hour
Loss in Customer Good will: $50 per lost sale
Repair costs: $500 per repair

The Car Wash problem was modeled using Arena as shown below








Modules:
Create (Dirty Cars): Random (Expo) w/value 1; 12 entities per arrival
Decide (Is there Parking Space?): 2-way by Condition; NQ(Washing Dirty Cars.Queue) < 4)
Process (Washing Dirty Cars): Seize Delay Release Action; Resource, Car Wash Machines, 1; Delay: EXPO
(1/6) Hours
Record (Lost Sales): Count; Value: 1
2 Dispose Modules: Clean Exit and Dirty Exit ( = lost sales)
Resource: Car Wash Machines; Fixed Capacity; Capacity = 2; Failures 1 Row (Car Wash Machine Failure)
Failure: Car Wash Machine Failure; Time; Up Time EXPO(100) hours; Down Time EXPO(4) hours
Replication Parameters: Length 2400 hours; Hours Per Day 8 hours
Results:









Total number of cars through carwash in 1 year (300 days): 28,932
Total number of cars washed: 10,366 (~36%)
Total number of lost sales: 18,566 (~64%)
System Metric
Value Half-width Minimum Maximum
Average Total Processing
Time (TIS) 0.1319 0.00642345 0 14.159
Average WIP 1.5902 0.089335264 0 18
Average Total Washing
Time per Car 0.3682 0.025650106 0.00007157 14.159
Accumulated VA Time 1728.27
Accumulated Wait Time 2088.27
Average Wait Time in
Parking lot (queue) 0.2015 0.023967654 0 14.0586
Average Number Waiting in
Parking lot (queue) 0.8701 0.084888612 0 4
Average Machine
Instantaneous Utilization 0.3601 0.015533451 0 1
Avg Machine Number Busy 0.7201 0.031066901 0 2
Avg Time Machine Failed 5.5827
Number of Failure
Observations 24
% Busy 40.47
%Failed 5.58
% Idle 53.95
Machine Maintenance:
$16,000/yr/machine 2 machines 1 yr = $32,000
Total Customer Time Valuation (for washed cars):
$20/hour Avg Washing Time per car Cars Washed
=$20/hr 0.3682 hr 10,366 = $76,335.22
OR
$20/hour (Accum VA Time + Accum Wait Time)
= $20/hr (1728.27 + 2088.27) hours = $76,330.80
Total Loss in Customer Good Will (for lost sales):
$50/lost sale Lost Sales
=$50/lost sale 18,566 lost sales = $928,300
Repair Costs:
$500/repair Number of Failure Observations
=$500/repair 24 = $12,000
After several iterations, some common trends were noted:
Increase in number of car wash machines (resource capacity) leads to a significant increase in percentage of washed
cars.











Increase in parking lot (queue) size leads to a gradual increase in percentage of washed cars.
0
20
40
60
80
100
2 3 4 5
P
e
r
c
e
n
t
a
g
e

o
f

C
a
r
s

W
a
s
h
e
d

No. of machines
Percentage of Cars Washed vs Number of Machines
Percentage of Cars
Washed vs Number of
Machines
0
20
40
60
80
100
3 4 5 6
P
e
r
c
e
n
t
a
g
e

o
f

C
a
r
s

W
a
s
h
e
d

Parking Lot Size
Percentage of Cars Washed vs Parking Lot Size
Percentage of Cars
Washed vs Parking Lot
Size
After several iterations, some common trends were observed:
The wait time (or time in parking lot/queue) significantly decreases with and increase in the number of machines.
The Average Machine Instant Utilization decreases with addition of car wash machines and a greater percentage of
idle time.
With the constant number of machines (resource capacity), increase in queue size led to an increase in wait times
but also an increase in Value Added (VA) Time (and Customer Time Valuation). However, simultaneous increase in
both factors led to an overall improvement in system performance.
Cost of machine maintenance is proportional to the number of machines available.
VA Time and Customer time valuation decreased with an increase of number of machines but was maximized at 5
machines.
Loss in customer good will decreases significantly with an increase in number of machines and gradually with an
increase in the number of parking lot spaces (queue size).
Number of Failure Observations were random amongst the iterations. However, the repair costs did not weigh
much compared to the other values in the cost analysis.

Question1:
a) What number of car wash bays do you recommend?
In order to meet the loading demands of 12 cars per hour, the car wash would require at least 4
car wash machines. Any less than 4 machines, there will be a greater percentage of lost sales
than cars that make it through to get washed, which is an unfavorable outcome in a competitive
market. In general, the more machines available, the more cars can be washed.

b) What are the performance characteristics of your recommended design?
My recommended design will be a 5 car wash machines 5 car parking lot system. This
recommendation is made due to cost and resource utilization constraints as shown previously.
The performance analysis and cost analysis are attached to the next couple of slides

c) How will performance be impacted if your recommendations are not followed?
As mentioned in part a, the more machines available, the more cars can be washed. If the system
is left as it currently is, the car wash would be loosing almost $1 million a year to lost sales, or
loss in customer good will, as it was aptly termed. This net loss will most likely go to the
competition and this would lead to loss in more customers as a more successful business is
always more attractive to the customer - and consequently, a possible business closure.
5 car wash machines and 5 car parking lot
Total number of cars through
carwash in 1 year (300 days):
28,344
Total number of cars washed:
18,793 (~66%)
Total number of lost sales:
9,551 (~34%)

Cost Analysis
Machine Maintenance $ 80,000.00
Total Customer Time Valuation (Cars
Washed) $ 91,709.84
Total Loss in Customer Good Will $ 477,550.00
Repair Costs $ 13,000.00
System Metric
Value Half-width Minimum Maximum
Average Total Processing
Time (TIS) 0.1594 Correlated 0 12.5685
Average WIP 1.8825 0.07794983 0 22
Average Total Washing
Time per Car 0.244 0.012308638 0.00003178 12.5685
Accum VA Time 3171.11
Accum Wait Time 1346.23
Average Wait Time in
Parking lot (queue) 0.07163448 0.012017539 0 11.8991
Average Number Waiting in
Parking lot (queue) 0.5609 Correlated 0 5
Avg Machine Instant
Utilization 0.2643 0.010831435 0 1
Avg Machine Number Busy 1.3215 0.054157176 0 5
Avg Time Machine Failed 2.9943
Number of Failure
Observations 26
% Busy 39.92
%Failed 3.24
% Idle 56.84
Question 2
a) What is the economic value of an additional parking space for the queue?
This value was calculated by considering the difference in total customer time valuation and total
loss in customer good will in the iterations with 4 car parking lot 2 machines and 5 car parking lot
2 machines, as shown below:
Economic value: $(928,300 841,650) + $(93,638.52 76,335.22) = $103,953.30

b) What is the value of the last parking space in the default design (4
th
space)?
Applying the same logic as in part a, the economic value of the 4
th
space we calculated to be
$50,383.31.

c) Shouldnt the 4
th
and 5
th
spaces have the same value?
No they should not, because the improvement in system performance with additional parking
spaces was not a linear progression. The improvements cause a gradual increase in customer time
valuation and a significant reduction in loss in customer good will which add up to make a very
impactful net gain.
Question 3
Car wash bays tend to break down with a MTTF of 100 hours of use time and take 4
hours (MTTR) to repair for $500. MTTF and MTTR are exponential distributed. How does
this feature impact your optimal design?
This feature led to 26 failure observations in the 300 day-1year cycle. This led to a failed time
percentage of 3.24%; with the machines busy 39.92% of the time, and idle 56.84% of the time. These
failures cost a total of $13,000 to repair for the year.

Question 4
How sensitive are your recommendations to the current load estimate of 12 cars per
hour? Our demand may go up if we successfully prepare for a competitor.
The recommended design was selected especially with the anticipated increase in load demands in
mind. As is seen in the question above, the resources still experience a bit of idle time which means
they have the capacity to handle more load. To verify this, the system was run with an average
incoming of up to 15 cars per hour and was still able to output favorable system performance values.
(i.e. Cars washed > Lost sales)