Академический Документы
Профессиональный Документы
Культура Документы
Data
The statistics gathered are the average amount of passengers on the buses, the passenger miles
per gallon, the average and maximum wait time, the average travel time, the calculated
simulation time, the maximum and average queue length, and the number of passengers that
were created. We chose the passenger miles per gallon because this is what the company running
the buses will want to maximize. The average travel time for the passengers was collected
because this is what the company wants to minimize. The average amount of passengers on
buses, average wait and queue length all provide a visual on what most passengers experience for
the case, and show whether it is reasonable. The maximum wait time and max queue length are
the most obvious indicators of whether a case is at equilibrium; they grow the most when they
are not (though other statistics do too).
These are also useful for comparing cases and seeing if the statistics stop improving at a certain
number of buses when they are added to the route. Knowing how many passengers there are and
how long the simulation ran lets us see if the business of the stops or buses is reasonable.
Tables 1-6 and graphs 1A-6B--shown below--compare only the longest simulation time; shorter
times that are compared can be found in the Codes Appendix. We have indicated which
simulation did not reach equilibrium by shading it red in the tables.
All data can be referenced below. When the bus capacity is 40 and the arrival rate is 60, the best
case is 10 buses on the route, which has a PMPG (passenger miles per gallon) of 135 and a
average travel time of 700 seconds. Ten buses and above reach equilibrium, but the PMPG
decreases up to 30 mpg while the travel time only improves by at most 20 seconds. The average
queue length stays the same for 10 or more buses. This trend continues in every case. When the
same rate is simulated with a bus capacity of 60, the best case is 6 buses on the route with a
PMPG of 130 mpg and a travel time of 700 seconds. When the arrival rate is 120 seconds and the
bus capacity is 40, the best case is 5 buses on the route with a PMPG of 100 mpg and travel time
of 650 seconds. When the same load is run at a bus capacity of 60, the best case is 4 buses on the
route with a PMPG of 100 and a travel time of 720 seconds. With a load of 30 and a capacity of
60 passengers, the best case is 13 buses on the route with a PMPG of 50 mpg and a average
travel time of 705 seconds. When the same load is simulated with a bus capacity of 40, the best
case is 16 buses on the route. This case has a PMPG of 145 mpg and an average travel time of
680 seconds.
Graphs 1A-6B have a negative nonlinear correlation. They show how once a specific number of
buses on the route reaches equilibrium not much changes in the data as buses increase. Also, the
number of buses before that point on the route are not viable options. This tells us that from that
point onward, adding more buses is costly for the PMPG. That is why the best case tends to be
the first point equilibrium is reached. Average travel time tends to not improve much past that
point.
Measurement
A case that is not at equilibrium cannot be chosen; as time goes on the statistics increase
uncontrollably and the maximum and average statistics inflate. This is a worst case scenario; it
would not be sustainable throughout the day as the buses are running. In theory, all cases could
reach equilibrium, but when there are a lot of passengers arriving with only a few buses in
comparison, it often does not. In all of the cases ran, a point was reached where all buses above
that number on the route would all get to equilibrium. Of course, some reach equilibrium faster
than others. The longest we ran all of the simulations was 1,000,000,000 seconds. This is the
most extreme case and lets us know whether equilibrium was reached. An equilibrium example
is shown in Graph 7. For our graph, we used the example of passenger miles per gallon over
time, but many more data points could be used to show this. Table 7, located directly above
Graph 7, contains more demonstrative statistics. These statistics all show very low variation; as
the time increases to extremes, the numbers stay more or less the same. For example, our
average queue length is consistently between 7.19 and 7.22 passengers. Our average amount of
passengers on buses is between 23 and 24. Passenger miles per gallon is between 140 and 147
mpg (the 147 number is from the simulation with the lowest time which is more subject to
randomness). Passenger wait time is between 89.9 and 9.4 seconds.
When a case does not reach equilibrium, these numbers increase at a rapid rate. This can be
observed in the Appendix or in the unsustainable numbers seen in the highlighted red sections of
Tables 1-6. To see the rest of the data analyzed (at equilibrium or not, and simulated for different
amounts of time) see Appendix in the Code.
Above, we mentioned statistics seeming reasonable. This means that the numbers one would
expect or predict to be the outcome for each category were in the ballpark of our generated
numbers. For example, our average amount of passengers on each bus for all the test cases is
between 15 and 30. This is a reasonable range because in any operational busing system, buses
will not run at an average full capacity all day unless there is a shortage. Our average queue
length is between two and seven --- another reasonable range because passengers will be
distributed randomly throughout stops. We had a max queue length range from roughly 30-80
people. These numbers make sense because certain downtown stops will have higher congestion,
though our average queue length shows this is not the case for each station. The average travel
time is usually around 12 minutes which seems a very reasonable amount for a 10 stop route.
The average max waiting time is roughly in the range of three and seven minutes which, based
on personal experience, is an accurate estimation of a waiting time to get on a bus.
Conclusion
When the average inter-arrival rate is 120 seconds, the bus capacity only makes the best case
change by 1 bus; four (at capacity 60) or five buses (at capacity 40). Since the arrival rate is low,
the capacity of the buses is not a big factor. Here, the number of buses is what should be
considered. The difference between the amount of buses on the route is three for an arrival rate
of 30 and four for an arrival rate of 60 seconds. From this we can conclude that the bus capacity
is more important when passengers arrive at a higher rate. It is more likely that the buses would
fill closer to capacity in this instance.
Since this route is increasing in ridership and popularity, it makes more sense to buy larger buses
than to increase the quantity. By doing so, the bus company should be able to keep costs down
while providing customers with a satisfactory travel time.
Data Reference:
Table 1: Graph A
Table 1: Graph B
Table 2: Graph A
(below)
Table 2: Graph B
Table 3: Graph A
Table 3: Graph B
Table 4: Graph B
Table 5: Graph A
Table 5: Graph B
Table 6: Graph A
Table 6: Graph B