Proceedings of the 2014 Joint Rail Conference JRC2014 April 2-4, 2014, Colorado Springs, CO, USA
JRC2014-3818
A VARIABLE NO-LOAD VOLTAGE SCHEME FOR IMPROVING ENERGY EFFICIENCY IN DC-ELECTRIFIED MASS TRANSIT SYSTEMS Lars Abrahamsson School of Electrical Engineering Electric Power Systems KTH Royal Institute of Technology Stockholm, Sweden Email:
[email protected]
Álvaro J. López-López Comillas Pontifical University ICAI School of Engineering Institute for Research in Technology Madrid, Spain Email:
[email protected]
Ramón R. Pecharromán Comillas Pontifical University ICAI School of Engineering Institute for Research in Technology Madrid, Spain Email:
[email protected]
Antonio Fernández-Cardador Comillas Pontifical University ICAI School of Engineering Institute for Research in Technology Madrid, Spain Email:
[email protected] s.es
Paloma Cucala Comillas Pontifical University ICAI School of Engineering Institute for Research in Technology Madrid, Spain Email:
[email protected]
Lennart Söder School of Electrical Engineering Electric Power Systems KTH Royal Institute of Technology Stockholm, Sweden Email:
[email protected]
Stefan Östlund School of Electrical Engineering Electric Energy Conversion KTH Royal Institute of Technology Stockholm, Sweden Email:
[email protected]
ABSTRACT Railway mass transit systems like subways play a fundamental role in the concept of sustainable cities. In these systems, the amount of passengers strongly fluctuates along the day. Hence, in order to provide a proper service without incurring disproportionate energy consumption, operation at different traffic densities is required. The majority of underground systems are DC-electrified. Standard DC voltages in railway systems are low for historical and safety reasons. In the rush hours, the large number of trains demanding power of the system may lead to overloaded substations and voltage dips. This problem is partially mitigated by means of substation-transformer tap regulation, which allows operators to increase the no-load voltage.
High no-load voltage has a beneficial effect at all trafficdensity scenarios in terms of transmission losses. However, at the same time it effectively reduces the system’s capacity to absorb regenerated energy, which may lead to inefficient energy consumption figures during off-peak hours. In this paper, the sensitivity of system energy consumption to no-load voltage has been analyzed. Several traffic-density scenarios in a case-study system are explored. As a result, a scheduled no-load voltage scheme is proposed for the operation of the system. This operation strategy improves energy efficiency without incurring a high investment cost. The only costs related to this proposed method are the costs of wear-andtear in tap-changers. In case there are devices such as energy storage systems installed in the system, there would be
1
Copyright © 2014 by ASME
handful of times per day should not be harmful to the tap changers. This paper, proposes a variable no-load voltage scheme for improving energy efficiency in electrified mass transit systems.
additional operation costs related to a simultaneous update of the voltage limits for their operation.
INTRODUCTION Railway mass transit systems like subways play a fundamental role in the concept of sustainable cities. This transportation mode exhibits good energy-efficiency figures and in addition it removes emission from polluted urban environments. The passenger flows in this kind of systems are variable, thus they have to be operated at different traffic conditions in order not to have huge and unnecessary energy consumption in the substations. It is common to have two strong peaks of passengers in working days, hence two peak-hour periods from the traffic point of view. Traffic density, which is expressed in terms of headway, is somehow related to the energy efficiency of the system. This is due to the fact that regenerative braking is responsible for huge energy savings when applied. These savings may go up to 40% of the total energy consumption in the substations ([1]). However, in DC-electrified systems, it is necessary to have trains powering when a train is regenerating. Hence, when the traffic is sparse, a certain proportion of regenerated energy may be wasted in on-board rheostats. Receptivity may be defined as the ability of a system to absorb regenerative braking energy. Among the different possibilities for improving energy efficiency some of them are related to traffic operation and others are based on improving the electrical infrastructure. In the former group, some authors try to optimize the power and speed profile ([2]), others are based on maximizing synchronization by designing the timetable [3], etc. In the latter, there are different options which are generally oriented to improve receptivity ([4]). Energy sent to rheostats and conductor losses may be regarded as the two main sources of losses which are affected by improvements in system infrastructure. There are many techniques and infrastructure parameters that may reduce these losses [5]. Among this list of factors, varying the no-load voltage in substations appears as a decision which may improve energy efficiency without causing large investment costs. However, deciding the optimum no-load value is not straightforward, since low values improve receptivity but spoil conductor losses and vice-versa. In the literature, there are not many studies dealing with this topic, being [6] one of the few examples. Nevertheless, in the literature no study has been found proposing a variable scheme for the no-load voltage that makes it possible to have high no-load voltage in substations when the traffic is dense (and so receptivity high), and low voltage when the traffic is sparse and benefits are expected from improving receptivity. This is technically possible by using tap changers available in substation transformers, and it is commonly accepted that changing tap position (and so varying no-load voltage) a
CASE-STUDY DEFINITION In order to have a case-study to assess the impact of varying the substation no-load voltage, a realistic mass transit line model has been generated. A graphical representation of this system is given in Figure 8, in Annex A. The following are the characteristics of the electrical and track infrastructure of this case-study line: Eighteen passenger stations distributed along 14.5 km. The average distance between the stations is 844 meters, where 511 meters and 1.41 km are the minimum and maximum inter-station distances. Nine electrical substations equipped with twelve-pulse diode rectifiers. The nominal power of each substation is 4.8MVA. The average distance between substations is 1.8 km. The minimum and maximum substation distances are 0.69 km and 2.65 km respectively. Substations are connected to 15 kV utility-company three-phase buses and they provide nominally 750 V no-load voltage (position 0 in tap changers). The line presents a weak point in the stretch between the substations at km 6.7 and km 9.3. This is a common case in actual lines and usually forces the operator fix the no-load voltage to values above the nominal voltage of the line. In 1.5 kV lines, some operators fix this no-load voltage to values as high as 1.75 kV. Trains are operated on a double track with rigid overhead conductor. The maximum slope is 4 % and the maximum speed is 70 km/h. There are no isolators to electrically subdivide the line and conductors are paralleled every 100 meters. Both rails carry the return current back to the substations. The resulting impedance is 35 m/km. The rolling stock material used in the simulations has the following features: Maximum traction power: 3 MW. Maximum braking power (in regenerative braking cases): 2 MW. Auxiliary power: 200 kWh/h. The following practical conventions have been used to model the different operation scenarios and energy-saving techniques: Flat out train speed profiles. These train movement profiles are defined as the minimum-time profiles, so maximum acceleration and deceleration are applied to the trains to meet the speed limits in the line. These speed profiles represent maximum consumption profiles and lead to strong traction and regeneration powers. Figure 1 shows a typical inter-station speed and power profile. Stop time at passenger stations has been set to 30 seconds. Simulation time for each traffic density has been selected to be the headway (e.g.: 240 seconds simulated for the
2
Copyright © 2014 by ASME
In the minimum-headway case (Figure 2), the number of trains in the line is maximized, and so the probability of having powering trains that absorb braking energy. As a result, rheostat losses effectively disappear for no-load voltages below 830 V. Hence, the minimum energy consumption corresponds to this no-load voltage value. No problems are observed with regard to the minimum voltage in the line until no-load voltage goes below 770 V.
four-minute headway traffic case), which is the period at which the performance of the line is repeated. Then, hourly consumption results have been obtained by extrapolation. A one-second sampling time has been chosen. A detailed electric railway simulator has been used to obtain the results presented. Further information of this simulator may be found in [5, 7]. Train speed and power profile 80
Results for 4-minute traffic density Loss Power (MWh/h)
60 40 20 0 0
10
20
30 40 Time (s)
50
60
4000 2000
Rheostat losses Conductor losses
0.4 0.2
720
740
760
780 800 820 No-load voltage (V)
840
860
880
900
780 800 820 No-load voltage (V)
840
860
880
900
780 800 820 No-load voltage (V)
840
860
880
900
9 Rectified Energy Optimum value 8.8
8.6
8.4 700
720
740
760
0 800 Minimum Voltage (V)
Power (kW)
0.6
0 700
Rectified Energy (MWh/h)
Speed (km/h)
0.8
-2000 -4000 0
10
20
30 40 Time (s)
50
60
Figure 1: Typical inter-station train speed and power profile.
Minimum voltage Minimum allowed
700 600 500 400 700
720
740
760
Figure 2: Voltage scan results. 4-min headway.
RESULTS The first step in the study consists of carrying out a voltage scan in a set of traffic-density scenarios. These traffic scenarios, defined by the headway (time between consecutive trains), are intended to explore the different behaviors that might be observed in the system when the total number of trains vary. The headways, selected to properly scan the different traffic scenarios that can take place in this line, are 4, 6, 8, 10, 12 and 14 minutes. For each traffic scenario, a no-load voltage scan from 700 to 900 V with 10 V steps has been carried out at each traffic density scenario to characterize the effect of this parameter in system losses (and in energy consumption at substations) and in the minimum voltage observed on the line. Figure 2 and Figure 3 show the results for the minimum and maximum headways. In both figures, the top graph represents conductor and rheostat losses. Then, in the central graph, rectified energy at substations is given, and the minimum energy point (subject not to have voltages below the minimum allowed) is marked. Finally, the bottom graph presents the minimum voltage of the line, where the dashed curve represents the minimum allowed voltage of the system.
Figure 3 shows the results for the maximum headway studied (14 minutes). Results are qualitatively different in comparison with the short headway case. The lower number of trains in the line leads to lower receptivity. Actually, rheostat losses do not disappear even for the minimum no-load voltage studied. The minimum energy point corresponds to 760 V, which is close to those voltages that lead to forbidden minimum voltages. For both headways, minimum voltages are similar, which is due to the restrictions of the case-study line. The most disadvantageous point is around the station placed in km 7.3 for all the headways studied.
3
Copyright © 2014 by ASME
Receptivity vs. traffic density Results for 14-minute traffic density
100
0.6
90
0.4 0.2 0 700
Rectified Energy (MWh/h)
95
Rheostat losses Conductor losses
Receptivity (%)
Loss Power (MWh/h)
0.8
720
740
760
780 800 820 No-load voltage (V)
840
860
880
900
80 75 70
3 65
Rectified Energy Optimum value
60
2.9
4
6
8
10
12
14
Traffic density (trains/min)
Figure 4: Line receptivity vs. traffic density (V0 = 750V).
2.8
2.7 700
720
740
760
780 800 820 No-load voltage (V)
840
860
880
900
780 800 820 No-load voltage (V)
840
860
880
900
This effect may be observed in Figure 5, where the green curve represents the optimum voltage for each traffic scenario. High optimum voltages for short headways confirm that when receptivity is high there is less margin for the no-load voltage to increase energy efficiency than when this parameter is low. The red-dashed curve is included in the figure as an appropriate voltage to separate peak-hour scenarios from off-peak-hour scenarios, hence grouping traffic densities into two clusters.
800 Minimum Voltage (V)
85
700
Minimum voltage Minimum allowed
600 500 400 700
720
740
760
Optimum no-load voltage vs. traffic density 900
Figure 3: Voltage scan results. 14-min headway.
880
It may be stated that the receptivity of the line conditions the reduction in energy consumption that may be achieved by varying the no-load voltage. In this study, receptivity has been defined as:
∙ 100
860
840
(1)
No-load voltage (V)
%
Poor performance Good performance Optimum values
Where EREGENERATED represents the total amount of regenerated energy by the whole set of trains; and EACCEPTED represents the total amount of energy actually fed back to the line. Figure 4 shows receptivity results for the whole set of traffic densities studied. The no-load voltage used for obtaining these values has been the nominal voltage of the line (750 V). It may be observed that as the headway is increased, receptivity drops steadily: the expected trend. In the highest headway case (14 min) receptivity drops to 62%, which is mainly due to the sparse traffic. If receptivity is high for a given traffic scenario, the possible benefits of tuning substation voltages will be low. When this parameter exhibits poor values, reducing no-load voltage, in general will improve energy efficiency figures.
820
800
780
760
740
720
700
4
6
8
10
12
14
Traffic density (trains/min)
Figure 5: Optimum voltage results. Figure 5 also shows a voltage interval (yellow) around the optimum voltage for each traffic scenario. These intervals represent no-load voltages which, despite not being optimum, still lead to good energy efficiency. Rectified energy increase lower than 50 kWh/h has been decided to be the criterion for including a no-load voltage inside the good-performance interval.
4
Copyright © 2014 by ASME
No-load voltage switching scheme
Once the energy consumption in the line for each no-load voltage and traffic scenario has been studied, it is necessary to define three operative decisions in order to assess achievable energy savings: the no-load voltage before applying the variable no-load voltage scheme, the traffic timetable and the no-load voltage values and switching times. For the first one, taking into account that the line has been defined to have a weak point around km 7.5, it has been set to a value higher than the nominal voltage for this electrification system. It is common when voltage dips take place in a railway line that operators set no-load voltage to very high values. To represent that behavior, 850 V has been selected as the no-load voltage before applying the variable scheme. For the second one, a subset of traffic-density scenarios has been selected to represent 4 operation modes: Super-peak scenario: 4-minute headway. Peak scenario: 6-minute headway. Off-peak scenario: 10-minute headway. Sparse scenario: 14-minute headway. Then, two different typical days have been defined from the operation point of view: working days (from Monday to Friday) and weekend days. Table 1 shows the amount of hours assigned to each operation mode for the two different types of day, assuming the line is operating during 16 hours per day.
Super-peak Peak
Sparse Open S1
Peak
2
10
S2
S3
PEAK ZONE
S4
Close
PEAK ZONE
Ferial day
Off-peak Sparse
Open
S1
S2
S3
S4
Close
Figure 6: Variable no-load voltage switching scheme. With all this information, it is already possible to evaluate the increase in energy efficiency associated to the variable noload voltage scheme. Figure 7 shows hourly energy savings for each operation mode. As expected, the longer the headway, the higher the saving, even when the total amount of energy involved in this cases is smaller. Results vary from 10 kWh/h in the super-peak mode (in which receptivity is close to 100%) to over 60 kWh/h in the sparse mode. The energy saving for the off-peak, which is the most frequent operation mode in the line, is close to 60 kWh/h as well. Energy saving vs. traffic density SPARSE 60
Energy saving (kWh/h)
0
PEAK ZONE
Working day
Off-peak
Table 1: Amount of hours for each operation mode and type of day S-PEAK PEAK OFF-PEAK SPARSE WORKING 3 3 7 3 DAY FERIAL DAY
PEAK ZONE
4
Regarding the no-load voltage values and switching times, after analyzing the results in Figure 5 traffic densities have been divided into two groups based on their optimum no-load voltage (up to 8-minute headway and from 10-minute headway on). For the four operation modes defined, this division allows to operate with only two different no-load voltages: one for the peak scenarios and another for the off-peak ones. 810 V has been selected as a reasonable value for the peak scenarios and 770 V has been selected to play safe in the off-peak cases. Since they are linked to working hours, the income of passenger into a mass transit line usually exhibits two peaks, and the operation of the system is, of course, adapted to those. The application of a variable no-load voltage operation under these circumstances would be possible with only four switching events per day, which is not supposed to be harmful for the transformer tap changers. Figure 6 shows the distribution of the four operation modes along both working days and weekend days. Switching instants are represented from S1 to S4, and since the operation is sparse both at the beginning and at the end of the day, no further switching operations are required.
OFF-PEAK
PEAK 40
20 SP-PEAK
4
6
10
14
Traffic density (trains/min)
Figure 7: Energy saving results. Finally, these results have been aggregated to obtain annual energy savings, which have been obtained by extrapolating the weekly results (composed of 5 working days and 2 weekend days) to 52 weeks (one year). Under these assumptions, a 285.5 MWh yearly energy saving has been obtained, which corresponds to 1.1% of the total energy consumption. CONCLUSIONS AND FUTURE WORK It has been observed that energy consumption in substations is sensitive to the no-load voltage at which a line is operated. Since they are related to low receptivity to regenerative braking, the achievable savings by tuning the no-load voltage in substations are greater in off-peak hour scenarios. For this reason, the best no-load voltage in terms of energy consumption is different for peak and off-peak traffic-density scenarios.
5
Copyright © 2014 by ASME
Congress on Railway Research - WCRR 2011. Lille, France, 22-26 May, 2011.
Taking into account that passenger income in a line usually exhibits two peaks, it would be possible to apply a variable noload voltage scheme that improves energy efficiency with a handful of switching times per day, which is not expected to be harmful for the tap changers. The results show that energy saving in the substations is 1.1% of the total energy consumption. It is important to state that this saving may be achieved by incurring only low investment costs. However, this result is particular for the studied system. The same analysis presented in this paper should be replicated to validate the results for another system. In the future, traffic noise (stochastic dwell times, etc.) will be included at each traffic-density scenario to better characterize receptivity with the aim of having more significant results. REFERENCES [1] W. Gunselmann, "Technologies for increased energy efficiency in railway systems" in 2005 IEEE 11th European Conference on Power Electronics and Applications. 2005. [2] M. Dominguez, A. Fernandez, A. P. Cucala and P. Lukaszewicz, "Optimal design of metro automatic train operation speed profiles for reducing energy consumption". Proceedings of the Institution of Mechanical Engineers, Part F (Journal of Rail and Rapid Transit), vol. 225, pp. 463-473, 09/01, 2011. [3] A. Ramos, M. Peña, A. Fernández-Cardador and P. Cucala, "Mathematical programming approach to underground timetabling problem for maximizing time synchronization". CEPADE, vol. 35, pp. 88-95, 2008. [4] T. Koseki, "Technologies for Saving Energy in Railway Operation: General Discussion on Energy Issues Concerning Railway Technology". IEEJ Transactions on Electrical and Electronic Engineering, vol. 5, pp. 285-290, 05, 2010. [5] Á J. López-López, R. R. Pecharromán, A. FernándezCardador and A. P. Cucala, "Assessment of energy-saving techniques in direct-current-electrified mass transit systems". Transportation Research Part C: Emerging Technologies, vol. 38, pp. 85-100, 1, 2014. [6] S. Açikbas and M.T. Söylemez, "The effects of no-load voltage level of traction power supplies on energy consumption of a mass rail transit system". IET Conference on Railway Traction Systems (RTS 2010). Birmingham, UK. 2010. [7] A. J. López López, R. R. Pecharromán, E. Pilo, A. P. Cucala and A. Fernández-Cardador, "Analysis of energy-saving strategies in railway power supply systems". 9th World
6
Copyright © 2014 by ASME
ANNEX A CASE-STUDY ELECTRICAL INFRASTRUCTURE Electrical infrastructure in the case study 3-phase 15kV (AC)
4.8MVA 12-pulse rectifier 750V (DC) TRACK 1
TRACK 2
STOP
0
STOP
STOP
STOP
2.6
STOP
STOP
4.9
STOP
STOP
STOP
STOP
STOP
6.7 Position in the line (km)
9.3
STOP
10
STOP
STOP
11.5
STOP
STOP
12.9
STOP
STOP
14.5
Figure 8: Case study electrical infrastructure.
7
Copyright © 2014 by ASME
