408
J. Druschke et al.
26.4.1 Methodology for the Hamburg Scenario
In order to calculate possible emission savings creating a reference scenario is crucial.
The server systems for the “Park&Joy” application are completely set up and in
operation. However, the sensors are not yet fully installed in any major German city
for comprehensive use. And it is important that these sensors are currently intended to
cover street parking lots exclusively without considering car parks or private parking
areas.
To scale the required number of sensors, planning values from Deutsche Telekom
need to be used. Only for the city of Hamburg these values are fully available. A total
of approx. 1300 sensors, covering approx. 11,000 parking lots, are intended to be
used here. This information in combination with the number of inhabitants enables
not only the calculation that each sensor covers 8.46 parking lots, but also that 0.0061
parking lots per inhabitant are available in Hamburg. The number of daily parking
operations per parking lot is assumed to be 6 as an average annual mean value for a
simplified calculation. This is an estimated value, as daily and seasonal differences
in parking behavior as well as parking time restrictions vary greatly in some cases
(Rikus et al. 2015). The server lifetime in the Hamburg scenario is set to 4 years
(Berwald et al. 2015). Depending on the application (active and sleep time) and
quality of the local mobile telecom network, the sensor lifetime for “Park&Joy” is
projected 4 years as well.
With these values from the Hamburg planning scenario, fast projections can also
be carried out for other cities in Germany. Of course, these projections have to
be considered differentiated, since the number of parking lots covered by a sensor
depends very much on the geographical environment and the condition of the corresponding street segment. Also, the scaling of the number of parking lots on the basis
of the number of inhabitants permits only rough calculation. When the rollout took
place in Hamburg and other cities and first real data are available, these assumptions
can be adjusted accordingly and the model can be refined.
A major challenge in modelling is the definition or mapping of the parking process.
In particular, the time at which a search process starts and the average speed driven
during the search for a parking lot are difficult to determine. Major disparities of
cities for example in commuter traffic or the road network have an influence on
the parking situation and the variance of parking search times (Cookson and Pishue
2017). For this reason, an easily modifiable probability model is used at this point
until a well-founded real data basis is available to specify the model properly.
According to its own information, Deutsche Telekom’s predication server for the
“Park&Joy” application provides a probability of success for finding a parking lot
of 96% in connection with the sensor data and 78% with exclusive use of the traffic
data without additional sensor support. With the conceding of a driving distance of
500 m per parking attempt, these probabilities of success can be used to determine
a total distance for a parking process.
The two most important values taken from Table 26.4 are 521 m for a parking
attempt with “Park&Joy” and 1,396 m for a conventional parking attempt without
J. Druschke et al.
26.4.1 Methodology for the Hamburg Scenario
In order to calculate possible emission savings creating a reference scenario is crucial.
The server systems for the “Park&Joy” application are completely set up and in
operation. However, the sensors are not yet fully installed in any major German city
for comprehensive use. And it is important that these sensors are currently intended to
cover street parking lots exclusively without considering car parks or private parking
areas.
To scale the required number of sensors, planning values from Deutsche Telekom
need to be used. Only for the city of Hamburg these values are fully available. A total
of approx. 1300 sensors, covering approx. 11,000 parking lots, are intended to be
used here. This information in combination with the number of inhabitants enables
not only the calculation that each sensor covers 8.46 parking lots, but also that 0.0061
parking lots per inhabitant are available in Hamburg. The number of daily parking
operations per parking lot is assumed to be 6 as an average annual mean value for a
simplified calculation. This is an estimated value, as daily and seasonal differences
in parking behavior as well as parking time restrictions vary greatly in some cases
(Rikus et al. 2015). The server lifetime in the Hamburg scenario is set to 4 years
(Berwald et al. 2015). Depending on the application (active and sleep time) and
quality of the local mobile telecom network, the sensor lifetime for “Park&Joy” is
projected 4 years as well.
With these values from the Hamburg planning scenario, fast projections can also
be carried out for other cities in Germany. Of course, these projections have to
be considered differentiated, since the number of parking lots covered by a sensor
depends very much on the geographical environment and the condition of the corresponding street segment. Also, the scaling of the number of parking lots on the basis
of the number of inhabitants permits only rough calculation. When the rollout took
place in Hamburg and other cities and first real data are available, these assumptions
can be adjusted accordingly and the model can be refined.
A major challenge in modelling is the definition or mapping of the parking process.
In particular, the time at which a search process starts and the average speed driven
during the search for a parking lot are difficult to determine. Major disparities of
cities for example in commuter traffic or the road network have an influence on
the parking situation and the variance of parking search times (Cookson and Pishue
2017). For this reason, an easily modifiable probability model is used at this point
until a well-founded real data basis is available to specify the model properly.
According to its own information, Deutsche Telekom’s predication server for the
“Park&Joy” application provides a probability of success for finding a parking lot
of 96% in connection with the sensor data and 78% with exclusive use of the traffic
data without additional sensor support. With the conceding of a driving distance of
500 m per parking attempt, these probabilities of success can be used to determine
a total distance for a parking process.
The two most important values taken from Table 26.4 are 521 m for a parking
attempt with “Park&Joy” and 1,396 m for a conventional parking attempt without