Effects of Eco-Driving Training: A Pilot Program in Belgrade Public Transport

This paper shows an eco-driving pilot program that has been implemented in the public transport company “JGSP Belgrade”, Serbia, in order to assess the possibilities of using eco-driving for an entire car fleet of the assessed company in the future. Eco-driving training and education of thirteen drivers were conducted in real driving conditions on the route length of 14 km and consisted of three phases. The results of eco-driving training of thirteen bus drivers confirm the findings of the previous researches that eco-driving has got many benefits. After the training, all drivers saw fuel consumption reduced by 8.61% on average, and consequently the average CO2 emission reduced by 8.61%. The implementation of eco-driving training in the assessed company and the attained fuel economy, leads to significant annual savings. However, some driving parameters were not significantly improved after training indicating a driver’s slow adaptation and application of new driving techniques.


INTRODUCTION
Road transport has a negative impact on fuel economy and ecology. In order to identify the factors that influence fuel consumption and CO 2 emissions, researchers around the world have developed methods for fuel consumption estimation [1,2] and CO 2 emissions estimation in road transport [3]. However, many researchers believe that driver behaviour has a great influence on fuel-use and CO 2 emissions [5][6][7]. Eco-driving program seems like a promising strategy that can positively change the driver's behaviour and improve driving fuel efficiency. In several European countries, eco-driving is applied as a manner to promote economic driver behaviour. In practice, ecodriving has attracted attention among automakers, policy makers, and researchers because its implementation is not expensive and can be cost-effective.
Eco-driving advices are well defined and comprehensible for drivers. They include the following: avoiding harsh stops and starts; anticipating traffic situations, smooth acceleration, using the engine braking, maintaining steady speed, eliminating excessive idling, driving at or below the speed limit. Besides, Barkenbus [8] noted that some eco-driving adherents involve automobile maintenance measures. The recommendation is a monthly tyre pressure checking and regular changing of air filters. This paper shows ecological and economical advantages of eco-driving dealing solely with driving behaviour. This eco-driving training pilot program for bus drivers is a preliminary step to, possibly, future implementation of eco-driving training program for all drivers in the public transport company "JGSP Belgrade", Serbia. This pilot program should assess the possibility of using eco-driving training in the aforementioned company and its 841 vehicles in the future.

Advantages of Eco-Driving Program
There are numerous advantages of using the ecodriving programs. Primarily, these programs can be connected to safety driving training for existing drivers. Furthermore, the programs can be applied to any vehicle on the road regardless of its age, size and type. The results of eco-driving are visible shortly after their implementation for the entire car fleet. Savings in fuel consumption and reduction of emissions are visible from day one [8].
A study for the bus drivers conducted in Greece showed the potential fuel savings of 10-15% [9]. Basarić et al. [10] also indicated the positive fuel economy effects of eco-driving on bus drivers. They established it was possible to achieve savings in fuel consumption and CO2 emission by approximately 11,71%. In addition, Ho, Wong and Chang [11] have shown that the eco-driving training had the potential to reduce fuel consumption and CO 2 emission by 10%. Rutty, Matthews, Andrey and Del Matto [12] showed the favourable results on the impact of ecodriving training on CO 2 emissions in the period immediately after the training, comparing with CO 2 emission one month before training. Analysing the vehicle operating parameters such as idling, idle fuel consumption, CO 2 emission and harsh acceleration, they determined a reduction in CO 2 emission by 1,7 kg per vehicle per day. In addition, some papers indicated the safety benefits. Namely, Haworth and Symmons [13] in their report indicated a reduction of traffic accidents by 35% using ecodriving training. SenterNovem [14] showed that the accident damage rate decreased by 40% after eco-driving training in Germany. Beusen et al. [15] have defined the key safe driving behaviours which represent the basis of eco-driving.

Difficulties in Adaptation and Application of Eco-Driving
For over 15 years, eco-driving has been an initiative with a widespread adoption, but the level of accomplished results and its long-term usage depend on certain limitations. Insufficient exposure of drivers to eco-driving information affects their behaviour and sustainability of the changed driving style over a longer period of time. Barkenbus [8] confirmed that there was a need for more initiatives on eco-driving in order to show positive application results, and to persuade the drivers that they should change their habits and driving style.
Before the implementation of eco-driving, the gaps in the public's limited comprehension of the nature of ecodriving and deeply rooted driving behaviour, should be filled. The researches show that the drivers have a tendency to forget the given instructions during their training, and that they fall back to their driving habits. Beusen et al. [15] showed that the majority of drivers made the immediate improvements in fuel consumption after the training, and some drivers had a tendency to return their old driving habits. S. Siero, Boon, Kok and F. Siero [16] noted a decline in positive attitude of the post-office drivers six months after the fuel economy training. However, it should be noted that the driving habits formed through experience are hard to break. Dogan, Steg and Delhomme [17] showed that the eco-driving style could hardly become a driving habit because it depended on certain factors, such as: traffic conditions, environment and personal motives.
Cultural and educational barriers slow the rate of adoption and sustainability of eco-driving from occurring. A study in France, that included 1200 car drivers, showed that despite an intense motivation to reduce fuel consumption, majority of drivers ignored the eco-driving instructions [18]. However, implementation of eco-driving feedback devices in the vehicles could be a good practice that would help the drivers to make a decision when driving. Yang, Li and Zheng [19] showed that these devices have a great role in reduction of CO2 emission and fuel consumption. When utilising these devices, one must take into account the type, complexity, contents and presentation of information from the perspective of ergonomy.
The bus drivers have a difficulty in maintaining a balance among the goals to be achieved. They have to adhere to the bus timetable, on the other hand, need to drive economically. Dogan et al. [17] has shown that performances in fuel economy decline when the drivers have to balance the reduction of fuel consumption and travel time.
Some researchers proposed their solution for the aforementioned problems. Barkenbus [8] believes that the eco-driving should become a norm and suggests the measures, such as regulation, training, education, feedback devices, fiscal stimulus and strengthening of social norms that can be used to influence the driving behaviour of individuals. Sullman, Dorn, and Niemi [20] believe that in order to motivate the drivers to apply the eco-driving principles and improve their driving performances, the transport company should pair up the eco-driving training with the behavioural change techniques. Some researchers suggest that the eco-driving program should be included in the national climate change strategies [21,22]. Van den Hoed et al. [21] claim that encouraging professional, individual drivers and fleet owners through eco-drive program to drive in a more ecological way can lead to reduction of CO2 emissions in the transport sector. Also, several IEA countries have included the eco-driving activities as part of their CO 2 emission reduction strategies and national climate change programs [22].

RESEARCH METHODOLOGY 2.1 Participants
Eco-driving training was conducted on July 5th, 6th, 12th and 13th, 2016 in the public transport company "JGSP Belgrade", operating in Belgrade, Serbia, and included thirteen bus drivers who voluntarily agreed to be monitored for a period of time. The average age of the participants was 42, with a standard deviation of 2,62. The average driving experience was 12 years (SD = 2,19). The study participants were selected using purposive sampling.

Test Route and Test Vehicle
In order to eliminate the difference in road conditions for the drivers, training was conducted on the same predetermined route, from the public transport company "JGSP Belgrade" depot in Novi Beograd to the bus 85 last stop in Banovo Brdo. Total length of the route in both directions was around 14 km.
The same bus, MAN SG313 with automatic transmission, was used for each day of testing so that the drivers could drive the same bus each time. It was equipped with a Controller Area Network (CAN) bus enabling the reading of driving parameters. In case of each measurement, the test vehicle parameters that affect fuel consumption were approximately the same.

Procedure
The study was conducted between 9 a.m. and 4 p.m. As an initial hypothesis, approximately the same number of vehicles and network conditions were selected in order to characterise the time periods and the driving behaviour in the tested interval. The test drive was taken twice by each driver: first before eco-driving training (normal driving) and second after eco-driving training with the professional eco-driving instructions. Eco-driving training has two sessions; a theoretical approach and a practical driving under instructor. The whole pilot program included three stages (in Fig. 1): • Before eco-driving training -D1 (Drive cycle 1) -This drive was completed by all the participants in the study so that the driving style could be recorded before the implementation of eco-driving tips. The first seven drivers (driver 1 -driver 7) conducted the initial drive (D1) on 5th July 2016 and the other six drivers (driver 8 -driver 13) on 12th July 2016 between 12 a.m. and 4 p.m. This serves as the control point in comparison to drive cycle 2 (D2) so that the potential fuel economy, advancement in driving and vehicle parameters achieved with eco-driving, can be assessed.
• Classroom training session -Theoretical eco-driving training was undertaken by the licensed instructors from RICo Training Centre (Accredited Training Institute of the World's Road Transport Organisation -IRU from Geneva) in duration of 120 minutes. Classroom training session took place on 6th July 2016 (driver 1 -driver 7) and 13th July 2016 (driver 8 -driver 13) before drive cycle 2. The training program was a static approach which aimed at urging drivers to apply eco-driving techniques after learning. Some of the advices given to the drivers include: driving at the speed limit, slow motion, decelerate smoothly, anticipate traffic flow, maintain a steady speed at low RPM (revolutions per minute) etc.
• Practical driver training -D2 (Drive cycle 2) -Drive cycle 2 combines the techniques learnt during the classroom teachings, as well as the real-time guidance from the instructor during the drive cycles. This drive cycle was conducted on 6th July 2016 (driver 1 -driver 7) and on 13th July 2016 (driver 8 -driver 13) between 12 a.m. and 4 p.m. After the completion of drive cycle 2, the obtained results were presented to drivers in order to evaluate the effects and identify the key needs for future development. The logged parameters are as shown in Tab. 1.

Data Collection
Data was collected after connecting CAN network with the software Key Driving Training System produced by Belgian company Key Driving Competence, using the diagnostic cable SAE J1939. Using two interfaces (Kvaser and Squarell) the data was translated into computable and readable data that could be used on a PC.
The data collected were statistically evaluated to measure the effectiveness of the Eco-driving training program. Statistical evaluation of the driving parameters was investigated using paired t-test (at 5% significance level) in statistical program MINITAB 17. Paired t-test was used to compare whether the results of driving parameters after the eco-driving training were significantly improved to the results before training.

RESULTS AND DISCUSSION
Analysis of the driving quality parameters before (D1) and after eco-driving training (D2) as per Tab. 2 show an average speed deceleration of nine drivers between range of 2,92% -14,67% while four drivers increased the average speed after training between range of 0,73% -23,70%. Similar values were obtained in the analysis of the average speed while the vehicle was in motion. The results among thirteen drivers revealed that there was no significant effect of the eco-driving training in both parameters, p = 0,756 and p = 0,548, respectively.
After completing the second Drive cycle (D2), and when compared to the test drive before the training (D1), almost all drivers reduced the fuel consumption while the vehicle was still in motion. Driver 2 achieved the least savings in fuel consumption (0,99%), whereas Driver 11 achieved the most (20,95%). Driver 12 only achieved an increase of 0,30%. Also, a lower overall fuel consumption was determined for all the drivers after the training, with an average saving in fuel consumption by 8,61%. The results showed that eco-driving techniques contributed to reducing parameters FCM (fuel consumption -on the move), TFC (total fuel consumption) and AFC (average fuel consumption) considerably and values of these parameters before training were significantly greater than those after training, p = 0,000, p = 0,001 and p = 0,001, respectively. There are previous evidences which have also reported eco-driving training to result in an increase in fuel savings for bus drivers [9,10,23,24,25]. These researchers showed that eco-driving could increase the fuel economy up to 27%.
Consequently, the average CO2 emission in tested drivers was reduced from 0,13% to 19,95% in relation to the emission before training. Average reduction in CO 2 emission for all thirteen drivers was 8,61%. Results showed that Eco-driving program significantly affected the reduction of average CO 2 emission after training, p = 0,001. Barić, Zovak and Periša [26] showed it was possible to achieve even higher emission reduction, of up to 32%.
Result analysis after the eco-driving training in tested drivers (in Tab. 3) showed that nine drivers reduced the parameter APGP (the average accelerator pedal position) avoiding strong accelerations indicating that eco-driving training affected a significant improvement of this parameter after training, p = 0,009. Pushing the throttle not more than 50% can be beneficial for fuel economy [27]. After eco-driving training, nine drivers increased the time of driving without throttle while the vehicle was in motion.
The improvements are between range of 0,32% -49,49% while four drivers decreased between range of 3,38% -38,13%. Post-training improvements were no statistically significant, p = 0,155. In addition, there was no significant effect of the eco-driving training on increasing total distance with driving without throttle, p = 0,479.  After applying advices on reducing the risky traffic situations the drivers were given during their training, ten drivers reduced their brake time during driving after ecodriving training from 3,53% to 70,15%. After the training these ten drivers used the brake less between range of 36,84% -86,60% which consequently led to an increased fuel economy. Eco-driving program was effective in these two parameters, BT (brake time) and BC (brake count), so it was statistically shown that their values before training were greater than those of after training, p = 0,004 and p = 0,003, respectively. These findings were justified by previous researches [5,28], claiming that a large number of sudden breakings could be responsible for the increase of the fuel consumption by around 30-44% in comparison to vehicles that were driving smoothly and moderately.
Although the improvements of the average engine speed were achieved to nine drivers when compared to the drive before the training, between range of 0,76% -5,19%, the results among thirteen drivers revealed that there was no significant effect of the eco-driving training, p = 0,080.
After eco-driving training seven drivers reduced the idling time from 12,94% to 64,80% while six drivers increased between range of 0,92% -51,70%. The result among the thirteen drivers showed that there was no significant effect of the eco-driving training on reducing the vehicle's idle time, p = 0,300. Researchers have shown that idling time is the most inefficient use of fuel [12]. Natural Resources Canada [30] have concluded that the idling time of the vehicle of more than 10 seconds has a higher fuel consumption than turning the engine off and starting it again.
Tab. 4 summarized the outcome of the eco-driving training and it disclosed that there was a general improvement in the driver's manner of driving.

Eco-Driving
Assessment of the fuel economy provides the calculation of the economic benefits for the public transport company "JGSP Belgrade", from implementing eco-driving education among its drivers. In this company, the overall structure costs of 32% go to fuel expenses. Every bus consumes around 22 829 l of diesel annually, which indicates that an average fuel saving of 8,61% results in an annual saving of 1966 l of diesel fuel per bus. If an average price for 1l of diesel fuel is €0,90, than an annual saving for each bus would be €1769 amounting to €1 487 729 for the 841 bus fleet. Symmons et al. [29] established that if a car fleet consuming around 1,5 million litters of diesel fuel per year saved just 1%, it would made the savings of $15 000 per annum.

CONCLUSIONS
In this paper, the effectiveness of Eco-driving training in the public transport company "JGSP Belgrade", Serbia, was investigated. The results confirm the findings of the previous researches that Eco-driving training reduces fuel consumption, CO2 emissions, and improves driver behaviour short-term. In the present pilot program fuel economy was enhanced on average by 8,61% after the training and hence the average CO 2 emission was reduced by 8,61%. Furthermore, the findings of this pilot program also demonstrated that eco-driving training has the potential to significantly reduce fuel-related costs for transport companies.
Eco-driving effects have shown a general improvement in the driver's manner of driving. However, some driving parameters (D, AS, ASM, DWTM, DWTTD, IT, AES) were not significantly improved after training indicating a driver's slow adaptation and application of new driving techniques. Therefore, eco-driving program needs to examine how the guidelines for these driving parameters could be improved.
Although our results were enticed under realistic conditions with potentially a great validity, this study had limitations that should be accepted in the context of the results. One limitation was the small sample size (N = 13) and it may not be possible to generalize the findings to other companies without restrictions. However, these findings can provide a preliminary access into what level of influence an eco-driving training and education can have on bus drivers' fuel efficiency. Furthermore, the primary goal of drivers is not to save fuel but to take passengers to the destination site; drivers might not react to the interventions as drivers with other primary goals in different settings. There may be other external influences outside of our control which may distract the effects.
Future researches should test if, and to what extent, these results could be sustainable over a longer period, and whether they could be transferred to other domains of energy-efficiency behaviour.