You are on page 1of 12

Energy 225 (2021) 120269

Contents lists available at ScienceDirect

Energy
journal homepage: www.elsevier.com/locate/energy

Taxi Ride sharing in Kuwait: Econ-enviro study


Sharaf AlKheder
Civil Engineering Department, College of Engineering and Petroleum, Kuwait University, P.O. Box-5969, 13060, Safat, Kuwait

a r t i c l e i n f o a b s t r a c t

Article history: Traffic congestion had been the most important and sensitive problem facing Kuwait roads for decades.
Received 29 March 2020 The objectives of this study were to: optimize the traffic flow on road networks by implementing shared
Received in revised form rides, reduce vehicle emissions and minimize the ride cost. The concept concentrated on sharing the
14 February 2021
journeys starting and ending around the same place and time. The effect of ridesharing was studied
Accepted 27 February 2021
according to the data collected for 3 months (July 2018, October 2018 and January 2019) from a taxi
Available online 8 March 2021
company. The data was then filtered so that the start point of all trips was South Surra. Three different
scenarios were created: single passenger, two passengers, and more than two passengers. Java NetBeans
Keywords:
Traffic
was used to calculate the total distance for each month and the cost for each trip after applying the
Ridesharing concept. “My driving” was used to obtain the fuel consumption to calculate the gas emissions for the
Passengers three scenarios;CO, NOx and HC emissions. For the trip distance analysis, the results for July, October, and
Fuel consumption January decreased by 0.84%, 0.45%, and 1.25%, respectively. For gas emissions and by comparing the first
Gas emissions scenario with the second one, the CO, NOx and HC emissions were reduced by 9.072%, 9.069%, and
9.074%, respectively. Finally, by comparing the first scenario with the third one, CO, NOx and HC emis-
sions were lowered by 0.116%, 0.08%, and 0.108%, respectively.
© 2021 Elsevier Ltd. All rights reserved.

1. Introduction these solutions inefficient.


A taxicab is also known as a taxi or a cab, which is the sort of car
Traffic congestion is a result of the increasing number of cars on that a single passenger or a tiny group of travelers uses to lease with
roads as more people prefer individual transport transit over mass a driver. A taxicab transfers passenger between their preferred
transport [1]. Large numbers of commuters heading to the same places. In Kuwait, easy locomotion and a variety of transportation
place tend to use a lot of taxis causing a high congestion level. In facilities are available. The most distinguished transportation fa-
addition, mass transport in Kuwait is limited to two types: buses cility is a taxi. That is why too many taxi offices can be found in
and taxis unlike other countries that have trolleybuses, metro, Kuwait. There are also different types of vehicles that serve as a taxi
railways and light rails. According to the latest statistics of the such as on Demand Taxis, Under Request and Limousine, depending
Central Statistical Bureau [2]; there are around 16,626 taxi cars in on the market choice. There are around 17,700 Taxi cars in Kuwait.
Kuwait (see Tables 1e7). According to the statistics provided by the General Department of
A new concept was developed based on the idea of ridesharing, Traffic, the minimum daily income of these offices is around 500
which refers to any combined and simultaneous journey of at least thousand Kuwaiti Dinars [4].
one driver and one passenger by assigning individuals with similar
timelines and paths to a common vehicle. Traffic congestion is a 2. Literature review
result of many reasons especially the behavior of road users and
poor infrastructure [3]. The government agencies in Kuwait have The quick growth of GPS positions and wireless communication
endeavored to find engineering solutions to the problem of traffic techniques have led to the development and popularization of on-
congestion such as roads widening, building bridges and other demand ride-sourcing applications such as Uber and DiDi around
engineering projects to facilitate the movement of cars. However, the world. Although in many towns the legality of ride-sourcing
the number of cars has increased with such expansions, making facilities is still under discussion, ride-sourcing is an appealing
option to travel and change the way passengers travel [5]. Ride-
sharing is an evolving mode that combines demands for real-time
E-mail address: sharaf.alkehder@ku.edu.kw. transport to the same car to save road traffic [5]. The confidence

https://doi.org/10.1016/j.energy.2021.120269
0360-5442/© 2021 Elsevier Ltd. All rights reserved.
S. AlKheder Energy 225 (2021) 120269

Table 1
Sample of filtered Data for one driver in July, January and October.

Month Date From To Distance (km) pickup time Drop-off time duration cost (KD)

July July 6, 2018 shuhada jabriya 10 20:03:40 20:19:41 0:16:1 4.00


July 13, 2018 zahra hawally 13.2 3:42:44 4:06:24 0:23:40 4.00
July 14, 2018 salam salwa 14.7 3:04:10 3:19:33 0:15:23 3.50
July 14, 2018 shuhada sabhan 15.3 4:17:16 4:29:02 0:11:46 3.50
July 14, 2018 shuhada sabhan 15.3 23:02:38 3:16:58 0:14:20 3.50
July 15, 2018 zahra yarmouk 13.6 4:44:42 4:59:14 0:14:32 3.50
July 17, 2018 salam sabhan 19.8 0:47:18 1:00:54 0:13:36 3.50
July 18, 2018 salam sabhan 19.8 22:04:54 22:22:37 0:17:43 3.50
July 19, 2018 shuhada eqaila 21 1:22:16 1:48:50 0:26:34 5.00
July 25, 2018 salam salmiya 15 01:05:51 1:22:13 0:16:22 5.50
July 26, 2018 hitten airport 11.4 5:59:11 06:16:34 0:17:23 7.00
July 26, 2018 hitten sharg 17.4 6:40:32 7:00:58 0:20:26 4.00
July 29, 2018 salam hawally 11.6 22:35:46 2:45:32 0:9:46 3.50
July 30, 2018 shuhada jaber alahmed 53.3 0:45:13 1:15:38 0:30:25 7.00
January January 1, 2019 zahra airport 11.4 23:48:27 23:59:29 0:11:02 10.00
January 3, 2019 salam airport 11.4 23:32:21 23:45:30 0:13:09 10.00
January 3, 2019 Shuhada salam 6.8 5:48:11 5:57:12 0:09:01 3.00
January 13, 2019 siddeq airport 11.4 19:12:08 19:26:26 0:14:18 7.00
January 16, 2019 salam gurain 20.1 19:32:40 19:44:22 0:11:38 4.50
January 16, 2019 salam eshbeliya 13.5 20:29:16 20:37:31 0:08:15 4.00
January 17, 2019 shuhada hitten 3.6 0:45:53 0:57:13 0:11:20 4.00
January 17, 2019 salam salmiya 15 1:40:57 1:54:24 0:13:27 4.00
January 17, 2019 salam salwa 14.7 21:41:35 21:59:26 0:17:51 4.00
January 18, 2019 hitten rumaithya 12.4 0:55:33 1:11:52 0:16:19 4.00
January 18, 2019 hitten salmiya 14.7 18:38:25 19:01:27 0:23:02 4.00
January 20, 2019 hitten rumaithya 12.4 0:52:07 1:06:55 0:14:48 4.00
January 20, 2019 Shuhada hitten 3.6 2:21:55 2:33:10 0:11:15 3.50
October October 1, 2018 zahra salmiya 20 18:21:11 18:49:43 0:28:32 4.50
October 2, 2018 zahra Airport 11.4 4:11:21 4:14:24 0:3:3 7.00
October 5, 2018 zahra abdullah almubarak 21.4 2:44:51 3:03:15 0:18:24 4.50
October 5, 2018 hitten yarmouk 10.2 0:45:48 2:02:10 0:16:23 4.00
October 7, 2018 hitten adan 13.3 20:37:05 20:55:48 0:18:43 4.00
October 11, 2018 salam salam 4 0:53:45 0:57:13 0:03:28 0.00
October 14, 2018 salam faiha 13.3 2:11:49 2:30:58 0:19:09 3.50
October 14, 2018 salam Airport 11.4 3:33:23 3:49:58 0:16:35 7.00
October 14, 2018 salam hawally 11.6 19:13:35 19:26:15 0:12:40 3.50
October 16, 2018 hitten hawally 14.2 21:13:24 21:27:57 0:14:33 4.50
October 17, 2018 hitten adan 13.3 20:11:18 20:31:48 0:20:30 4.00
October 19, 2018 salam sabhan 19.8 18:15:27 18:35:36 0:20:09 3.50
October 20, 2018 salam salwa 14.7 2:07:24 2:21:19 0:13:55 3.50
October 20, 2018 shuhada sabah alsalem 8 3:40:55 3:57:50 0:16:55 6.50
October 20, 2018 salam siddeq 4.1 5:32:06 5:41:15 0:09:09 2.50
October 21, 2018 salam faiha 13.3 20:24:48 20:37:34 0:12:46 4.00
October 21, 2018 shuhada rawda 15 22:05:13 22:35:45 0:30:32 3.00
October 23, 2018 salam sharg 18.8 22:20:50 22:39:20 0:18:30 3.50
October 25, 2018 zahra mushrif 13 18:50:02 19:10:30 0:20:28 3.50
October 27, 2018 shuhada salmiya 16 1:05:54 1:24:26 0:18:32 4.00

Table 2
Summary description of the study sample.

Nationality Kuwaiti Non-Kuwaiti


92% 8%

Gender Male Female


18.7% 81.3%
Age Cohort 15e24 25e34 35e44 45e54 55e64 65e74
66.7% 19.3% 5% 6% 2.7% 0.3%
Employment Status Student Employed Not Employed Retired
62.7% 29.7% 5.7% 2%
Marital Status Single Married Divorced Widowed
73% 23.7% 2.3% 1%
Monthly Income <200KD 200-500 KD 501-1000 KD 1001-2000KD >2000KD
28.7 32.7% 13% 18.7% 7%

in ridesharing is cited as a major obstacle to people’s acceptance of Sidecar and Flywheel especially after the success of Ridesharing
ridesharing in several researches [6e10]. In a ridesharing scheme, idea [12]. Social networks will emerge from travel behaviors due to
Wang et al. [11] showed that the exclusive allocation of travel re- the co-presence of activities [13]. In view of the comfort of sitting
quests and friendship distribution between the social networks next to an alien in a limited area, without the presence of a typical
(e.g. friends, colleagues) is not necessarily achievable. Many ride- trusted taxi driver, acceptance of shared, autonomous vehicles is
sharing apps appeared with different features such as Lyft, even more ambiguous [14].

2
S. AlKheder Energy 225 (2021) 120269

ensuring that taxis will attract passengers while at the same time
Table 3 the income of taxi drivers is guaranteed. The study sought to reduce
The effect size guidelines for Phi and Cramer’s V by Cohen (1988). operating expenses and maximize satisfaction for passengers. The
Magnitude of effect size for Phi/Cramer’s V results showed that by ride-sharing taxi, 19% travel mileage and
66% taxi demand can be spared. Dial a ride issue has been
Small 0.1
Medium 0.3 researched by Santos and Xavier [19]; which models circumstances
Large 0.5 in which individuals are prepared to share rides to save cash. . Both
instances of riding and sharing were regarded. A modern GRASP
heuristic with route connections was suggested and a highly
2.1. Environmental benefits of ride sharing effective algorithm to calculate shortest routes, based on contrac-
tion hierarchies, was used for solving the issue. The simulation
Ridesharing has been one of the many successful solutions to findings showed that on average travelers paid nearly 30% less than
protect the environment from damages caused by cars, such as gas what they would have paid for personal rides. Nourinejad and
emissions. According to a study in Dublin [15], the rapid growth of Roorda from the University of Toronto in Canada [20] have also
private car ownership between 1996 and 2006 had caused a implemented an agent-based model for dynamic ridesharing
chronic urban congestion and increased the greenhouse gas (GHG) testing. In this study travelers participated while ensuring that the
emissions. . The Irish government studied the effect of rideshare on costs were spread between them. The results showed higher user
the environment. A logit model was used to investigate the rela- savings on cost.
tionship between those who rideshare and those who didn’t.
COPERT 4, a computer model that is used to estimate the emissions
2.3. Reducing fleet size and decreasing the total travel distance
from road transport, was also used. Results showed that individual
ride-sharing savings of 12,674 t of CO2 emissions per year was
A case study in New York [21] was conducted to assess both
achieved. Biying Yu et al. [16] performed a joint research in Beijing
traditional and shared independent taxi benefits and disadvan-
with DiDi Chuxing, a Chinese transportation network company, to
tages. The study proposed an Agent-Based Model (ABM) to study
study the environmental impact of ridesharing. The direct effects of
the impact on the two distinct kinds of taxis of ridesharing. The
the mode change were calculated using the fuel life cycle analysis.
findings showed that the fleet can be reduced by 59% without
They gave positive results as ridesharing was able to reduce energy
significantly increasing the waiting time or any extra travel dis-
consumption by 26.6 thousand tce, CO₂ emissions by 46.2 thousand
tance. Fiedler D et al. [22] also made a case study in the city of
tons and NOx emissions by 235.7 tons. A similar study [17] in the
Prague to measure the potential of ridesharing on traffic. Results
city of Lisbon developed an agent-based model to determine the
showed that the average overall travel distance in MOD systems
impacts of ridesharing on two examples: a self-driving shared taxi
without driving sharing and present conditions were lowered by
and a self-driving shared bus. The study demonstrated that with
35% and 60% when ridesharing was implemented. Another case
the full adoption of ridesharing services, carbon emissions would
study in New Jersey [23] made a scenario shift from a system of
be decreased by almost 40%.
private vehicles to a similar one with a shared ride taxi. They
employed empty car repositioning methods for testing fleet size,
2.2. Operation cost reduction and client satisfaction maximization car miles, and service level. The results indicated that using shared
taxi services can lead to a great decrease in fleet size while keeping
A routing optimization model study was applied in China [18] a high level of service.

Table 4
Trip cost based on trip distance (km) from AlAreej Company sample calculation & total cost.

Trip distance (km) Cost

From To

1 5 3
5 10 3.5
10 15 4
15 25 4.5
25 30 5
30 35 5.5/6
35 40 7/8
Khairan/Bnaider/Alzoor (From any area) 20
Khairan/Bnaider/Alzoor (From Almantaqa Alashra) 15
Salmi/Alabdli/Scrap 25

Sample of cost calculation for the first scenario


Date From To Distance (km) pickup drop-off Duration Cost (KD)
January 2, 2019 zahra airport 11.4 10:31:24 10:47:41 0:16:17 10
Sample of cost calculation for the second scenario
Date From To Distance (km) pickup drop-off Duration Cost (KD)
January 2, 2019 zahra airport 11.4 10:31:24 10:47:41 0:16:17 10
January 2, 2019 zahra airport 11.4 10:29:04 10:47:50 0:18:46 10
Total cost for each month for the existing condition
Month Total cost (KD)
July 1249
October 1105
January 997

3
S. AlKheder Energy 225 (2021) 120269

Table 5
Average environmental emissions for the first, second & third scenarios.

Environmental emissions for the first scenario from Hitten to Khaldiya

From To During Fuel Consumption (L) CO Emissions(g) NOx Emissions(g) HC Emissions(g)

Hitten Khaldiya AM Peak 1.67 31.059 6.043 7.198


Hitten Khaldiya Afternoon Peak 1.51 28.083 5.464 6.509
Hitten Khaldiya PM Peak 1.46 27.154 5.283 6.293

Environmental emissions for the first scenario from Alzahra to Khaldiya

From To During Fuel Consumption (L) CO Emissions(g) NOx Emissions(g) HC Emissions(g)

Alzahra Khaldiya AM Peak 1.69 31.431 6.115 7.284


Alzahra Khaldiya Afternoon Peak 1.49 27.712 5.392 6.422
Alzahra Khaldiya PM Peak 1.33 24.736 4.813 5.733

Environmental emissions for the first scenario from Alsiddeq to Khaldiya

From To During Fuel Consumption (L) CO Emissions(g) NOx Emissions(g) HC Emissions(g)

Alsiddeq Khaldiya AM Peak 1.53 28.455 5.536 6.595


Alsiddeq Khaldiya Afternoon Peak 1.42 26.407 5.138 6.121
Alsiddeq Khaldiya PM Peak 1.22 22.690 4.415 5.259

Average Environmental Emissions for the First Scenario

From To Average CO Emissions(g) Average NOx Emissions(g) Average HC Emissions(g)

Hitten Khaldiya 28.765 5.597 6.667


Alzahra Khaldiya 27.960 5.44 6.480
Alsiddeq Khaldiya 25.851 5.030 5.992

Average Environmental Emissions for the Second Scenario

From To To Sum of average CO Emissions(g) Sum of average NOx Emissions(g) Sum of average HC Emissions(g)

Hitten Alzahra Khaldiya 56.725 11.037 13.147

Average Environmental Emissions for the Third Scenario

From To To To Sum of average CO Emissions(g) Sum of average NOx Emissions(g) Sum of average HC Emissions(g)

Hitten Alzahra Alsiddeq Khaldiya 82.576 16.067 19.139

Table 6
Trip cost Analysis.

Sample of cost calculation for the first scenario after Ride-Sharing is applied
Date From To Distance (km) pickup drop-off Duration Cost (KD)
January 2, 2019 zahra airport 11.4 10:31:24 10:47:41 0:16:17 10
Sample of cost calculation for the second scenario after Ride-Sharing is applied
Date From To Distance (km) pickup drop-off Duration Cost (KD)
January 2, 2019 zahra airport 11.4 10:31:24 10:47:41 0:16:17 5
January 2, 2019 zahra airport 11.4 10:29:04 10:47:50 0:18:46 5
Total cost for each month after applying Ride-Sharing concept
Month Total cost (KD)
July 1238
October 1102
January 971

2.4. Ride share in Middle East Other similar promising ideas were also introduced in Egypt such
as “Goo and Back” [26].
Rideshare was also introduced in the Middle East and specif-
ically in the city of Riyadh, KSA. A group of students [24] at King 3. Methodology and data analysis
Saud University from the computer science department conducted
a study based on ridesharing cars. A smart mobile application called 3.1. Study area
‘where are you’ was designed to serve the needs of ridesharing. The
system of the application searches for the nearest driver in the way South Surra and its affiliated areas; Alsalam, AlShuhada, Hutten,
of requester based on the GPS position. The application also pro- AlZahra and AlSaddiq were selected as the study area; it has been a
vides a ‘female only’ option to respect the specificity of Saudi thriving area in recent years and many buildings were erected there
Arabia. In Egypt also a group of Egyptian students launched a as shown in figure (1).
similar project. A group transport company was launched with a
similar idea of Uber and Careem through a taxi. However, they used
the tour buses which stopped due to the revolutionary events in 3.2. Data collection
Egypt to get everyone satisfaction. The application of the company,
named SWVL, had reached 50 thousand users in just 3 months [25]. The data of taxi trips were collected from TAXI ALAREEG Com-
pany for three months; July (2018), October (2018) and January
4
S. AlKheder Energy 225 (2021) 120269

Table 7
Environmental Emissions for both Second & Third scenarios.

Environmental emissions for the second scenario after applying ride-sharing concept

From To To During Fuel Consumption (L) CO Emissions (g) NOx Emissions(g) HC Emissions (g)

Hitten Alzahra Khaldiya AM Peak 2.81 52.261 10.168 12.112


Hitten Alzahra Khaldiya Afternoon Peak 2.78 51.703 10.060 11.983
Hitten Alzahra Khaldiya PM Peak 2.73 50.773 9.879 11.767
Average gas emissions 51.579 10.036 11.954

Environmental emissions for the third scenario after applying ride-sharing concept

From To To To During Fuel Consumption (L) CO Emissions (g) NOx Emissions(g) HC Emissions (g)

Hitten Alzahra Alsiddeq Khaldiya AM Peak 4.50 83.693 16.284 19.397


Hitten Alzahra Alsiddeq Khaldiya Afternoon Peak 4.42 82.205 15.994 19.052
Hitten Alzahra Alsiddeq Khaldiya PM Peak 4.39 81.647 15.885 18.922
Average gas emissions 82.515 16.054 19.118

Fig. 1. The study area.

(2019). This data includes information on the date and place of taking into account that the maximum time difference between any
destination, the travel time and the price of the journey. two requests for a taxi was 15 min. Finally, it was assumed that all
company cars were capable of carrying four passengers. Following
3.3. Tools and method the three scenarios that were created, trips were conducted for
each scenario from the selected study area to Khaldiya, College of
Different kinds of software were used in this study. Excel was Engineering and Petroleum (Kuwait University), specifically during
used to calculate the total distance for the existing condition. Java the three peaks: AM, afternoon and PM for a week. For the first
NetBeans was used to analyze the condition where ride sharing was scenario, single driver/single passenger, it was assumed that three
applied. “My Driving” was used to analyze both conditions. Finally, passengers from three different areas: Hitten, Alzahra, and Alsid-
IBM SPSS was used to analyze the questionnaires. deq were heading to the same destination, College of Engineering
and Petroleum (Kuwait University), individually with a maximum
3.4. Procedure of 15 min time difference between their trips during the three
peaks. For the second scenario, single driver/two passengers, it was
3.4.1. Model assumptions assumed that two passengers heading to the same destination were
Three explicit assumptions for building the model were sharing their trips during the three peaks. The trip was started from
designed. First, the collected data was categorized so that the start Hitten, picking another passenger from Alzahra then heading to the
point of all trips on which the ride-sharing concept was applied was College of Engineering and Petroleum (Kuwait University). For the
one of these five areas: Al Salam, Al Shuhada, Hutten, Al Zahra and third scenario, single driver/three passengers, it was assumed that
Al Saddiq. Secondly, all trips were heading to the same destination three passengers heading to the same destination were sharing

5
S. AlKheder Energy 225 (2021) 120269

Fig. 2. Driver route before and after applying ride-sharing concept.

Fig. 3. Algorithm code.

6
S. AlKheder Energy 225 (2021) 120269

their trips during the three peaks. The trip started from Hitten, start point of all trips on which the ride-sharing concept was
picking another passenger from Alzahra then the third passenger applied was one of these five areas: Al Salam, Al Shuhada, Hutten,
from Alsiddeq was picked heading to the College of Engineering Al Zahra and Al Saddiq. All trips were heading to the same desti-
and Petroleum (Kuwait University). nation considering that the time difference between any two re-
quests for a taxi had a maximum of 15 min. Also, it was assumed
that all company cars were proper for four persons. Figure (2)
3.4.2. Algorithm analysis by Java NetBeans software
shows the Ride-Sharing concept that was applied in this study.
An algorithm is a procedure based on conducting a sequence of
Two passengers A and B from different starting points but near to
specified actions to solve a problem. It usually means a small pro-
each other were heading to the same destination D at almost the
cedure that resolves an ongoing problem in mathematics and IT
same time. The real driver’s route was to pick each passenger
[27]. The algorithmic tool in this study was Java NetBeans software.
individually to their destination so there were two routes. After
NetBeans can provide a software development product that is
applying the concept, a single driver from origin C had to pick the
needed by developers, businesses, and users. It also enables them
two passengers A and B to their destination D; so there was one
to develop the products quickly, efficiently and easily by getting
route.
more benefits from the strength of the Java platform and the other
In the developed code, the needed inputs from the company
standards. For the current situation of the selected network in this
data for the three months were inserted. These inputs were trip
study, data for three months were chosen; July since it encounters
start point location, trip endpoint location, time of pick up and drop
summer vacation, October since it encounters the beginning of
off for each trip, trip distance (km) and trip cost. A database for the
workdays and January. The company data was filtered so that the

Fig. 4. Significant Results: A) The number of times you request a taxi per month.B) The possibility of requesting a taxi with likeable look. C) Usage of Smartphone application for
requesting a taxi if available. D) Usage of ride-sharing option. E) Usage of “females only” option.

7
S. AlKheder Energy 225 (2021) 120269

code was created that contains trips distances in km and trips costs
from the selected study area, South Surra (Al Salam, Al Shuhada, NOx ¼ F * 13:6 g=gal (2)
Hutten, Al Zahra, and Al Saddiq), to almost all of Kuwait areas. After Hydrocarbons Emissions (g)
the request was accepted, based on the defined conditions, and the
ridesharing was applied, the cost for each passenger was divided HC ¼ F * 16:2 g=gal (3)
into half from the cost of each passenger start’s point to the final
destination and then the total distance was calculated. Fig. 3
(Annexure) shows the algorithm code that was used to apply the
idea of shared ride based on the defined constraints. For each 3.4.4. Questionnaire
month file, the date of each ride was read by the code. Then after Two individual online surveys were done to attain a different
sorting the trips with the same date, the trips starting within a time specific aim.
difference of a maximum of 15 min between them and heading to
the same destination were shared. 3.4.4.1. Ride sharing questionnaire. The aim was to know what the
In this study, three scenarios were created for calculating the public prefers and if they like the idea of ‘ride-sharing’ option, and if
cost and gas emissions for the trips before and after applying the they would like to use it or not. It was divided into two types of
Ride-Sharing concept: a single driver (single passenger), a single questions: General questions and Personal questions.
driver (two passengers) and a single driver (more than two 3.4.4.1.1. Frequency distributions. The online survey that was
passengers). completed for this project had a population of 300 people, of whom
A sample of the filtered data for one driver during the three 94% were Kuwaiti and 86% were female. The majority of those who
months is shown in Table (1) where each of them consists of the completed the survey were in the age group of 18e24 years rep-
date, the time, the location of origin and destination, the length and resenting 70%.
the cost of trips. 3.4.4.1.2. Significant Results. The following section represents
the frequency distributions of the measured variables obtained
from the online survey. Figure (4-A) shows that only 8.6% of the
3.4.3. “My driving” analysis total that requested a taxi, 3.6% of them asked for a taxi once a
“My Driving” is an application used for IOS and Android sys- month and 5% used them two times or more in a month, while the
tems; it is used to keep the driving history and big-data. The de- remaining did not. According to Figure (4-B), 84% of the partici-
velopers are obsessed with the exceptional high data quality and pants will use a taxi if it looks likable; 42.7% of them said that they
precision. The application aims to analyze the driving behavior, to may use it, and 16% will not use it. According to Figure (4-C), 72.3%
have a better driver with the help of machine learning and to of the participants would ask for a taxi in case of the availability of a
provide tips on how to improve driving and save on both gas and smartphone application. According to Figure (4-D), most of the
insurance [28]. In this study, “My Driving” was used to find the fuel participants would not use the rideshare option as the percentage
consumption. Following the three scenarios that were created, trips was 77.3% for No, and 22.7% would use it. Finally, according to
were conducted for each scenario from the selected study area to Figure (4-E), 40.8% of the females would use the female rideshare
Khaldiya, College of Engineering and Petroleum (Kuwait Univer- option while the rest will not use it.
sity), specifically during the three peaks: AM, afternoon and PM for
a week. For the first scenario, single driver/single passenger, it was 3.4.4.2. Weekday morning trips to work questionnaire. The aim was
assumed that three passengers from three different areas: Hitten, to obtain the most common mode of transport used by people to
Alzahra, and Alsiddeq were heading to the same destination, Col- reach their works in the morning, the duration needed to reach
lege of Engineering and Petroleum (Kuwait University), individu- their work and their times of departure. The questionnaire was
ally with a maximum of 15 min time difference between their trips divided into two types of questions: General questions and Per-
during the three peaks. For the second scenario, single driver/two sonal questions.
passengers, it was assumed that two passengers heading to the 3.4.4.2.1. Frequency distributions. The online survey, which was
same destination were sharing their trips during the three peaks. completed for this project, had a population of 300 people, of
The trip was started from Hitten, picking another passenger from whom 92% were Kuwaiti and 81.3% were female. The majority of
Alzahra then heading to the College of Engineering and Petroleum those who completed the survey were between the ages of 15e24
(Kuwait University). For the third scenario, single driver/three years, with a percentage of 66.7% and most of them were students
passengers, it was assumed that three passengers heading to the representing 62.7%. Also, 73% of the population were unmarried,
same destination were sharing their trips during the three peaks. and the highest monthly income of the population was between
The trip started from Hitten, picking another passenger from 200KD and 500 KD representing 32.7%. Table (2) summarizes the
Alzahra then the third passenger from Alsiddeq was picked heading different characteristics of the population in the survey.
to the College of Engineering and Petroleum (Kuwait University).
The environmental emissions before and after the Ride-Sharing
3.4.4.3. Sample size calculation. The required sample size (SS) for
concept were measured using equations (1)e(3) from Traffic ware
the survey was calculated using equation (4).
(2013). CO, NOx, and HC were the most common vehicle emissions,
which were calculated by multiplying the emissions rate in g/gal by
SS ¼ ½Z^2  ðpÞ  ð1  pÞ=c2 (4)
the fuel consumption in L.
Carbon Monoxide Emissions (g) where, Z ¼ 1.96 for 95% confidence level.
c ¼ confidence interval ¼ ±0.05,
CO ¼ F * 69:9 g=gal (1)
p ¼ 0.5 for maximum sample size required, because the function
Nitrogen Oxides Emissions (g) p (1- p) ¼ p - p2 is maximum when its derivatives (1-2p) is equal to

8
S. AlKheder Energy 225 (2021) 120269

zero. Therefore 1-2p ¼ 0; 2 p ¼ 1; p ¼ 0.5. 4. Data analysis


Applying those values in the SS equation had resulted in a
required sample size of 385. The collected and analyzed sample of The data that was collected from Java NetBeans were used for
300 would give results with some error. both the analysis of the existing condition and the condition in
which the Ride-Sharing concept was applied.

3.4.4.4. Coding. In this study, IBM SPSS version 25 (Statistical 4.1. Existing condition analysis
Package for the Social Science) software was used for statistical
analysis. SPSS has been around since 1968 and is considered the This includes three different perspectives: the trip distance, the
standard statistical software to analyze data for the social sciences. trip cost in the three scenarios and the gas emissions.

4.1.1. Trip distance analysis


3.4.4.5. Chi-Square Test. Chi-Square (c2) is a test used to determine This section shows the total distance for each month, which is
if a substantial distinction in one or more categories existed be- calculated by adding the distance for each trip for each month. This
tween the anticipated frequencies and the observed frequencies. was done by Excel. The total distance for July, October and January
Chi-Square Test is illustrated based on two hypotheses, which are was 4654.8 km, 4160.1 km and 2735.6 km, respectively.
the null hypothesis (H₀) and the alternative hypothesis used (H₁),
where;
4.1.2. Trip cost analysis
H₀: independent, “implies that the anticipated and observed
Based on the data that was collected from the company, all trips’
frequencies have no important difference”.
costs estimation was based on the trip distance (km). Some costs
H₁: dependent, “implies that the anticipated and observed fre-
depend on the popularity of the area like the airport. For the first
quencies have an important difference”.
scenario, single driver/single passenger, the cost calculation was
Chi-Square is a test for independence; it is a correlation or a test
based on the passenger trip distance as shown in Table (4). For the
for an association. Dependency (H) is considered valid if the p-value
second scenario, single driver/two passengers, the final amount
was found to be < 0.05. In such a case, a connection occurs, and an
was the sum of the cost of each passenger’s trip from his/her start
important connection exists between the responses considered
point to the final destination. The same was for the third scenario,
(variables). For p-value >0.05, the test is not significant and the null
single driver/more than two passengers. Each passenger had to pay
hypothesis (H₀) is expected.
his/her trip cost even they were heading to the same place, and
The Chi-Square formula used on the collected data is shown in
then finally, the total cost for the existing condition for each month
equation (5):
was calculated by adding each trip cost.
X ðO  EÞ2
x2 ¼ : (5) 4.1.3. Environmental emissions analysis
E
Carbon monoxide (CO), nitrogen oxides (NOx) hydrocarbon (HC)
where, x2 : Chi Square. emissions were determined and analyzed as environmental emis-
O: The observed Frequency in each category. sions before applying the Ride-Sharing concept, which is the first
E: The Expected Frequency in the corresponding category. scenario. The emissions were calculated using equations (1)e(3)
Since Chi-Square Test only indicates if the association is signif- based on the fuel consumption values, which were figured out by
icant but does not measure the association level, an effect size “My Driving” as shown in Table (5).
called Phi or Cramer’s V is used depending on the size as shown in
Table (3). For a 2  2 table, Phi is used. While Cramer’s V is applied 4.2. Ride-sharing condition analysis
for tables where at least one variable had only two categories, for
example, 2  3 tables, 2  4 tables, and so on. For the Phi effect size, The condition in which Ride-Sharing is applied was analyzed
0.1 is a small effect, 0.3 is medium and 0.5 is a large effect (Cohen, from several aspects; the trip distance, the trip cost in three sce-
1988). In addition, as per Cohen, the guidelines for Phi divided by narios and the gas emissions as was done in the existing condition
the square root of dG* are used to determine the magnitude of the analysis. The data for analysing the condition of the Ride-Sharing
Cramer’s V effect size, as shown on the following table. The effect was collected from the algorithm code in Java NetBeans.
size Phi and Cramer’s V are defined as shown in equations (6) and
(7). 4.2.1. Java NetBeans simulation results
This section represents the code simulation outputs; the trip
sffiffiffiffiffi
distance and the trip cost.
x2
4¼ (6)
n
4.2.1.1. Trip distance analysis. This section shows the total distance
n ¼ the number of observations for each month after applying the Ride-Sharing concept, which was
x ¼ degree of freedom (df) calculated by adding the distance of each trip for each month. The
sffiffiffiffiffiffiffiffiffiffiffi distance for July, October and January was 4615.7 km, 4141.2 km
x2 and 2701.4 km, respectively.
V¼ (7)
n:df *
4.2.1.2. Trip cost analysis. The cost for each passenger trip was
df * ¼ min (r-1, c-1) divided into half from the cost of each passenger start’s point to the
r ¼ number of rows final destination. A sample of cost calculation is shown in Table (6)
c ¼ number of columns in the contingency table. for the different scenarios after the concept was applied. There was

9
S. AlKheder Energy 225 (2021) 120269

Fig. 5. Java NetBeans output for: A) July, B) October and C) January.

no cost analysis for the third scenario due to the nature of the data. October, and January; it was decreased by 0.84%, 0.45%, and 1.25%
The total cost for the condition where ride-sharing was applied for respectively. Moreover, when comparing the cost of the three
each month was calculated by adding each trip cost as shown in scenarios in both conditions, the cost was divided into half from the
Table (6). cost of each passenger start’s point to the final destination. Also, by
Java NetBeans outputs for each month contained the rides that comparing the total trip for the three months of July, October, and
have been shared, the total distance and the total cost as shown in January; it was decreased by 0.889%, 0.272%, and 2.678% respec-
Figure (5). tively. Finally, when comparing the environmental emissions for
the three scenarios between the average gas emissions and by
4.2.2. Environmental emissions analysis comparing the first scenario with the second scenario, the CO, NOx
The environmental emissions analysis for the condition where and HC emissions were reduced by 9.072%, 9.069%, and 9.074%,
Ride-Sharing concept was applied (the second and the third sce- respectively; and by comparing the first scenario with the third
nario) during AM, afternoon and PM peaks are presented in Ta- scenario the CO, NOx and HC emissions were reduced by 0.116%,
ble (7). The environmental emissions calculations were the same as 0.08%, and 0.108%, respectively.
for the environmental emissions analysis for the existing condition.
4.3.1. Correlation analysis
4.3. Results Comparisons Correlation analysis was conducted to examine the in-
terrelationships of questions, which reveal the dependence of
The results of the existing condition were compared with the behaviour on other conducts among the people surveyed. Using the
results of the condition in which the Ride-Sharing concept was Chi-Square test, the correlations between the two chosen issues
applied to understand the effect of the ride-sharing concept. From were evaluated according to the following hypothesis:
Figure (6), comparing the trip distance for the three months July, H0: independent (the two questions/behaviours are
10
S. AlKheder Energy 225 (2021) 120269

Fig. 6. Results Comparisons: A) Total Trip distance for July 2018, October 2018 and January 2019 before and after applying Ride-Sharing Concept. B) Total Trip cost for July 2018,
October 2018 and January 2019 before and after applying Ride-Sharing Concept. C) Environmental emissions (First with Second scenario) D) Environmental emissions (First with
Third scenario).

independent of each other) effect size that was 0.605 at level p ¼ 0.000 for the effect of the
H1: dependent (the two questions/behaviours are dependent on number of licensed drivers per household on the number of cars
each other) per household,
Dependency (H) is applicable when p-value has been deter-
mined to be less than 0.05. In this situation, the responses studied x2 ð12; N ¼ 300Þ ¼ 262:604; p ¼ 0:000 < 0:05:
against each other are linked. If the p-value is less than 0.02, then
there is a strong relationship. For p-values greater than 0.05, the
relationship is not considered significant.
4.3.5. The effect of the time of departure on the duration of the
morning work trip
4.3.2. The effect of the monthly income on the mode of transport
Based on the likelihood ratio of the Chi-square Test, there was
Based on the likelihood ratio of the Chi-square Test, there was
no significant relation between the time of departure and the
no significant relationship between the monthly incomes and the
duration of the morning work trip, x2 (21, N ¼ 300) ¼ 27.090,
mode of transport used by participants,
p ¼ 0.196 > 0.05.

x2 ð28; N ¼ 300Þ ¼ 35:686; p ¼ 0:141 > 0:05:


5. Conclusions

Ridesharing has been one of the successful solutions in


4.3.3. The effect of the mode of transport on the distance travelled improving traffic and transportation efficiency in addition to its
per day great impact on the environment and cost. The effect of applying
Based on the likelihood ratio of the Chi-square Test, there was ridesharing was studied according to the data collected for 3
no significant relationship between the modes of transport and the months (July - 2018, October - 2018 and January- 2019) from a taxi
distance travelled per day by participants, company, which was then filtered so that the start point of all trips
is from South Surra. Three different scenarios were created for this
x2 ð28; N ¼ 300Þ ¼ 36:572; p ¼ 0:067 > 0:05: study: single passenger, two passengers, and more than two pas-
sengers. Java Net Beans was used to calculate the total distance for
each month and the cost of each trip after applying the concept.
“My driving” was used to get the fuel consumption so that the gas
4.3.4. The effect of the number of licensed drivers per household on emissions for the three scenarios of CO, NOx and HC emissions can
the number of cars per household be calculated. For the trip distance analysis for the 3 months, it was
Since the assumption of the Chi-square Test has been violated noticed that it was decreased by 0.84%, 0.45%, and 1.25%, respec-
(7 cells 35% > 20%) then the likelihood ratio was read. The analysis tively. For the gas emissions, by comparing the first scenario with
showed that there was a strong significant relationship due to the the second scenario, the CO, NOx and HC emissions were reduced
11
S. AlKheder Energy 225 (2021) 120269

by 9.072%, 9.069%, and 9.074%, respectively. By comparing the first States Am 2014;111(37):13290e4.
[10] Gargiulo E, Giannantonio R, Guercio E, Borean C, Zenezini G. Dynamic ride
scenario with the third scenario, the CO, NOx and HC emissions
sharing service: are users ready to adopt it? Procedia Manuf 2015;3:777e84.
were reduced by 0.116%, 0.08%, and 0.108%, respectively. [11] Wang Y, Winter S, Ronald N. How much is trust: the cost and benefit of
ridesharing with friends. Comput Environ Urban Syst 2017;65:103e12.
Declaration of competing interest [12] Chaudhry B, Yasar A, ElAmine S, Shakshuki E. “Passenger safety in ride-
sharing services “. Procedia Computer Science 2018;130:1044e50.
[13] Hackney J, Marchal F. A coupled multi-agent microsimulation of social in-
We authors declare that we have no conflict of interest teractions and transportation behavior. Transport Res Pol Pract 2011;45(4):
regarding our submitted article. 296e309.
[14] Krueger R, Rashidi TH, Rose JM. Preferences for shared autonomous vehicles.
Transport Res C Emerg Technol 2016;69:343e55.
Acknowledgment [15] Caulfield B. Estimating the environmental benefits of ride-sharing: a case
study of Dublin. Transport Res Part D 2009;14:527e31.
[16] Yu Biying, Ma Ye, Xue Meimei, Tang Baojun, Wang Bin, Yane Jinyue, Weia Yi-
The author would like to thank Dana AlHouli, Fatmah AlSultan, Ming. Environmental benefits from ridesharing: a case of Beijing. Appl Energy
and Noura Faraj for their efforts in data collection and analysis. 2017;191:141e52.
[17] Martinez L, Viegas J. Assessing the impacts of deploying a shared self-driving
urban mobility system: an agent-based model applied to the city of Lisbon,
Appendix A. Supplementary data Portugal. International journal of Transportation science and Technology
2017;6:13e27.
Supplementary data to this article can be found online at [18] Lin Yeqian, Li Wenquan, Qiu Feng, He Xu. Research on optimization of vehicle
routing problem for ride-sharing taxi. Procedia Social and Behavioral Sciences
https://doi.org/10.1016/j.energy.2021.120269.
2012;43:494e502.
[19] Santos D, Xavier E. Taxi and ride sharing: a dynamic Dial-a-ride problem with
Credit author statement money as an incentive”. Expert Syst Appl 2015;42:6728e37.
[20] Nourinejad M, Roorda M. Agent based model for dynamic ridesharing.
Transport Res C Emerg Technol 2016;64:117e32.
Sharaf AlKheder: Conceptualization; Data curation; Formal [21] Lokhanadwala M, Cai Hua. Dynamic ride sharing using traditional taxis and
analysis; Funding acquisition; Investigation; Methodology; Project shared autonomous taxis: a case study of NYC. Transport Res C Emerg Technol
administration; Resources; Software; Supervision; Validation; 2018;97:46e60.
[22] Fiedler D, Certick M, Alonso-Mora J, Cap M. The impact of ridesharing in
Visualization; Writing e original draft; Writing e review & editing. mobility-on-demand systems: simulation case study in Prague. ResearchGate;
2018.
References [23] Zhu S., Kornhauser A. The interplay between fleet size, level-of-service and
empty vehicle repositioning, strategies in large-scale, shared-ride autono-
mous taxi mobility-on-demand scenarios. Transportation Research Board
[1] Mahmud K, Gope K, Chowdhury S. Possible causes & solutions of traffic Jam
96th Annual Meeting Location: Washington DC, United States Date: 2017-1-8
and their impact on the economy of Dhaka city. J Manag Sustain 2012;2(2).
to 2017-1-12, Pagination: 15p, Report/Paper Numbers: 17-05960.
[2] Central Statistical Bureau. Land transport. Ministry of Interior; 2017.
[24] Chorfi H, AlAtee A, AlOqeely A, AlMutairi L, Akihriji L, AlDossari S. A smart
[3] Mhandu E, Kazembe T. Urban myths pertaining to road accidents in
real-time ride-share for Riyadh city. Procedia Social and Behavioral Sciences
Zimbabwe: the case of chinhamo service centre along seke road linking harare
2015;195:1932e7.
and chitungwiza. Africana 2013;6(2):170e86.
[25] Mustafa A. “SWVL” Online Bus. Young people inaugurate a bus company with
[4] Deyab M. 182 million dinar for the taxi market annually and the counter does
an online application that attracts the world’s attention. Company “Careem”
not work. AlQabas; 2016. https://alqabas.com/80275/.
announces cooperation with it. 40 lines for transporting citizens inside Cairo
[5] Wang Y, Kutadinata R, Winter S. The evolutionary interaction between taxi-
and Giza. The world’s most important newspapers are published". AlYoum
sharing behaviours and social networks. Transport Res Part A 2019;119:
AlSabea 2017. https://regular.swvl.com/affordable_buses_in_egypt_en?&
170e80.
utm_medium¼search&utm_source¼Google&utm_content¼Bra_En&utm_
[6] Amey AM. Real-time ridesharing: exploring the opportunities and challenges
campaign¼Google_BRA_EG-Cairo_R_REG_BRA_ALL_MCLICK_G_LP_B_G_Bra_
of designing a technology-based rideshare trial for the MIT community.
En&gclid¼EAIaIQobChMIk9Kgvtmh7wIVkr7tCh3rIQitEAMYASAAEgIyyPD_
Master’s thesis. Massachusett Institute of Technology; 2010.
BwE.
[7] Chaube V, Kavanaugh AL, Pe rez-Quin
~ ones MA. Leveraging social networks to
[26] Saleh S. “The way "Uber and Careem “.”GOO and BACK”, new application for
embed trust in rideshare programs. In: Proceedings of the 43rd Hawaii in-
mass transport, "From door to door" “. AlWatan. 2016. https://www.
ternational conference on system sciences. Honolulu, HI; 2010.
elwatannews.com/news/details/1570829.
[8] Wessels R. Combining ridesharing & social networks. In: ITS World congress;
[27] Rouse M. Algorithm”. TechTarget WhatIs.com; 2019.
2009. https://www.utwente.nl/ctw/aida/education/ITS2-RW-Pooll.pdf.
[28] Sharewiz. My driving app. 2018. https://www.sharewiz.ca/my-driving-app.
[9] Santi P, Resta G, Szell M, Sobolevsky S, Strogatz SH, Ratti C. Quantifying the
benefits of vehicle pooling with shareability networks. Proc Natl Acad Sci Unit

12

You might also like