Full text


B- EES -0006

Master thes is


Solution with three sided converter



Mikael Strom

Adtranz Fixed Installations

September 2000



Wind power and railway feeding Solution with three sided converter

In this thesis it is investigated if it would be technically and economically possible to combine wind power and railway feeding. As a case study Blekinge Kustbana (BKB), a railway section in the south of Sweden is chosen. The thesis includes an extensive power and energy flow analysis on the railway section where large-scale wind power production is directly connected.

When the work of this thesis started it was first of all a three sided converter solution that should be studied. Where three-sided converter consists of a PWM-converter with the wind power connected to the DC-Iink. But during the work it was concluded that this solution is not the best solution for the case. A better solution is the three-sided transformer, were the wind power is connected through a third winding to the railway transformer on the 50Hz side.

Also different railway feeding systems and their influence on how the wind power can be used are investigated.

From the conclusions it can be mentioned that it would be profitable to directly connect 4.5MW wind power to BKB.

Page 1 (57)


1.1 1.2




2.2 2.2.1 2.2.2 2.3 2.4 2.5

2.5.1 2.5.2 2.5.3 2.5.4


3.1 3.2

3.2.1 3.2.2 3.3

3.3.1 3.3.2 3.4

3.4.1 3.4.2

4.1 4.2




5.2 5.2.1 5.2.2 5.3 5.4

5.4.1 5.4.2 5.4.3 5.5

5.5.1 5.6

5.6.1 5.6.2 5.6.3

Wind power and railway feeding Solution with three sided converter









Wind power placement ...... 6

Wind statistics .. .. . . .. .. .. . . .. .. .. . .. . .. .. .. .. .. . .. . . . .. . .. . . . .. . .. .. . . .. . .. . .. . . .. .. .. .. .. . . .. . . .. .. . .. . . .. . . 7



TECHNICAL DESIGN ... 1 0 Mechanical components ... 11

Generators ... 11

Frequency converters ... 12

Transformer ... 12


GENERAL ... 13


Rotary converters ... 14

Static converters ...... 14


Contact line ............... 17

Return current ............. 17


Power and energy consumption ....... 18

Locomotives ................... 20 THE COMBINATION OF WIND POWER AND RAILWAY FEEDING ... 21



CASE STUDY ......... 23

GENERAL ... 23


Main program ... 23

Other programs ... 24



Passenger train ... 25

Freight train ... 26

Traffic ... 27


Comments on the electrical model ... 28


Wind statistics for Blekinge ... 29

Wind realisations ...... 31

Chosen wind power plants ... 32


5.7.2 5.7.3 5.7.4 5.7.5 5.7.6 5.7.7 5.8

5.8.1 5.8.2 5.9

5.9.1 5.9.2 5.9.3 5.9.4

6.1 6.2 6.3 6.4 6.5


6.5.1 6.5.2

7.1 7.2 7.3


Wind power and railway feeding Solution with three sided converter

Whole system with 15kV 16Hz single phase ... 33

Simplifications in the track layout. ... 33

Logistics and human contro/ ... 34

Chosen constellation of the wind power park ... 34

Some assumptions about the plant and the wind characteristics ... 35

The model for the three sided converter ... 36

RESULTS ... 37

Dimensioning simulations ... 37

Energy simulations ... 45


Production and prices ........... 49

Costs ... 50

Grants ... 61

Calculation ... 51



GENERAL ... 53





Recommendations if the AT-system is built.. ......... 54

Recommendations if the BT-system is built ......... 54

FURTHER WORK ............ 55




REFERENCES .. , ... 57

Page 3 (57)



Master thesis

Wind power and railway feeding Solution with three sided converter

Reference notes is written like [x,py] and should be read as reference number x page y. Some reference's are taken from the web site and they are written like [4,x,py] and should be read as\tour chapter x page y. A problem with this site is that it is updated frequently. Therefor the used headlines as they were when this was written is listed below.












[4,4,p1 0]





Wind Obstacles Wind Shade

Guide to the Wind Shade Calculator Wind Shade Calculator

Wake Effect Park Effect

Speed Up Effects: Tunnel Effects Speed Up Effects: Hill Effects Selecting a Wind Turbine Site Betz Law

Wind Turbine Components Wind Turbine Generators Synchronous generator

Changing Generator Rotational speed Asynchronous (Inductor) generator

Variable Slip Generator for Wind Turbines




Wind power and railway feeding Solution with three sided converter

The idea to feed an electrical railway system with wind power may look odd in the beginning, but in fact they have quite a lot in common. Electrical trains have a fluctuating power need and wind power is a fluctuating power source. The trains power need is some MW, which is about the same size magnitude as generated by a large-scale wind power plant. But most important in this thesis is the fact that wind power and railway systems frequency converters could be built together. This is due to the fact that the frequency converters for some wind power plants uses the same principle as the PWM converters used in railway systems.

If wind power and railway feeding should be reasonable an electrified (or electrifiable) railway section with sufficient traffic load that runs though an area with good wind conditions must be found. A railway section that matches these conditions is Blekinge Kustbana. Therefore Blekinge Kustbana is chosen for a case study.


The purpose with this thesis is to make power flow simulations for Blekinge Kustbana including the three-sided converter and large-scale wind power.

From these simulations it can be found out what kind of equipment should be used to construct a technically well working system. When this is done one can make an economical calculation to find out if, and under what circumstances it will be profitable to combine wind power with railway systems.

Page 5 (57)


Wind power and railway feeding Solution with three sided converter



Mankind has used wind energy for a long time, in windmills and for water pumping. For electricity producing it has been used mostly in areas far from a common electrical grid. However, during the last decade's large-scale wind power production for national grids has started to be something usual.

The reasons for this are basically bigger environmental consciousness, technological progresses and higher oil price that has made the price for wind power electricity competitive.

The principal components of a modern wind power plant are the tower, the rotor and the nacelle, which accommodates the transmission mechanisms, the generator and the yaw mechanism. Switching, protection and control systems, lines and transformers, will also be required for supplying the common grid.


2.2.1 Wind power placement

Wind power plants are much more dependent on their location than other power plants. The energy in the wind decreases very fast when it meets obstacles of any kind. These obstacles can for example be trees, hills and buildings. The energy admission on a spot can be calculated by quite complex methods [4,2,p20-28]. but approximate values for the relative energies in different kinds of topologies will be given in table 2.1 Table 2.1 Relative energies for different topologies.

Roughness class Description Relative energy

0 0.5 1



Seas and lakes 1.0

Completely open terrain with a smooth surface. 0.7 Open agricultural area with fences and only very 0.5 scattering buildings. Only softly rounded hills.

Agricultural land with some houses and 0.4 hedgerows with a distance of approximately 500m

Villages and towns 0.3-0.1

As you can see the wind energy decreases very fast when the roughness of the landscape increases. An offshore wind power plant will give

approximately the double energy amount than one placed in an agricultural


2.2.2 Wind statistics Distribution

Master thesis

Wind power and railway feeding Solution with three sided converter

The wind variations on a location are usually described using the Weibull distribution. The Weibull probability density distribution is:

x~O {2.1)

0 otherwise

This distributions tells us how likely it is to have a wind speed over a certain value. The parameter cis called the shape parameter and a is a scale factor. The Weibull probability density distribution for different shape parameters is given in figure 2.1.




0 2 3 X

Fig 2.1 Weibull probability density distribution for a=1 and different c's

In most areas strong gale force winds are rare, while moderate and fresh winds are common, therefore the shape parameter often lies around 2. If the shape parameter is exactly 2 the distribution is called Raleigh, which is also a widely used distribution for wind statistics.

For the Wei bull (Raleigh) distribution the mean value and variance is:

Expected value m


ar(l + 11 c) (2.2)

Variance (2.3)

Where r(x) is the Gamma function defined as:


Page 7 (57)



Wind power and railway feeding Solution with three sided converter

From the definition of natural wind, turbulence is the fluctuation with higher frequencies than the mean wind speed variations. Turbulence is therefore, the deviation of the instantaneous wind speed U(t) from the mean wind speed m.



U(t) - m (2.5)

The variability is best described with the variance

clu ,



a; = i f[u c t) -m]

2 dt (2.6)


Through measurements it has been shown [1,14-22] that au== mlln(zlz0)

gives an estimate of the wind variance for levels near ground level, which can be used if no measured au is available. Here


is the height above ground and


0 is the so-called surface roughness length, which depends on terrain type, shown in table 2.2.

T a e. bl 2 2 T YPICa va ues or su ' I f rf ace roug1 ness h en~r th Zo

Type of terrain ZO(m)

Mud flats, ice 1


to 3*1


Smooth sea 2*10-4 to 3*10


Sand 2*1 0


to 1


Snow 1


to 6*1


Mown grass 1 0"3 to 10-;l Low grass, steppe 1

o -z

to 4*1

o -z

Fallow field 2*1


tO 3*1


Hiqh qrass 4*1


to 10"1

Palmetto 1


to 3*1


Forest and woodland 10·1 to 1

Suburb 1 to 2

City 1 to 4

Observations of the natural wind shows that the probability density function for u(t) agrees well with the Gaussian distribution

p(u) = 1 exp(.::.!i_) 2 (2.7)


51i 20',;

and this has been found to be satisfactory when calculating wind loads.

Information of the turbulence structure in the time domain is given by the auto-correlation function r('r).

1 fto+T/2


= - -

2 u(t) · u(t +T)dt

T*O' II t0-TI2 (2.8)


Wind power and railway feeding Solution with three sided converter

The maximum value of r('r) is achieved when r= 0 and r(r)decreases to zero when


increases. Physically the explanation for this is that two values separated with a small time step are more correlated than two values with a large time step. The auto-correlation function tells one how fast the

correlation between values decreases as a function of their separation.

Though the auto-correlation function gives information about the turbulence it is not commonly used, instead the frequency spectrum




is used to describe the turbulence in the frequency domain. There are some standard frequency spectrums (for example Kaimal and Von Karman) that can be calculated if the parameters

m, z



0 are known.

The conclusion from the above discussion is that even if only the


m, z



0 are known, one can get quite far in describing the wind turbulence, for different frequencies or in the time domain. But one must keep in mind that these are statistical methods with many

assumptions and all results given by them must be taken under consideration if they are reasonable.

When the frequency spectrum and the mean wind speed is known (or estimated) there are several different methods to simulate the wind. The one used in this thesis is derived in [5).


The blades of a wind turbine rotor extract some of the flow energy from air in motion, and convert it into a mechanical torque on an axis. By definition the turbine stands in a free air flow and therefore it's no use in slowing the wind to much, because then the wind starts passing on the outside of the turbine instead. The German Betz showed that the greatest energy extract that can be established is 59.3% or 16/27 of the wind energy and the wind speed behind the plant should be 1/3 of the origin wind speed [4,3,p5]. The energy in airflow is proportional to the wind speed in cube, but the energy production differs from the cubic increase on account of different

efficiencies for different wind speeds and maximum/minimum power limits.


Two or three blades?

All public grid connected wind turbines today is off the horizontal type with 2-3 turbine blades. For some years ago the 2-bladed was quite usual for large plants, but nowadays the 3-bladed turbine has almost the whole market. The reason for this is that 3-bladed turbines has a smother movement and is easier to calculate and construct. The tip speed ratio,

Page 9 (57)


Wind power and railway feeding Solution with three sided converter

which is defined as the speed of the blade tip divided by the wind speed, is also smaller for the 3-bladed version, which decreases the noise.

Power rating

Because the power increase so fast with the wind speed it is necessary to limit the power extraction at higher wind speeds. The power rating for wind power plant is reached at the rated wind speed, which usually lies around 13(m/s). The power rating has during the last decades increased heavily, and this is one of the major reasons for wind power to be economically competitive.

Variable speed?

Most wind turbines today run with almost constant speed or alternatively two speeds. This in combination with the fast fluctuating wind causes high peak torque on the gearbox and generator and high fatigue loads on the tower and blades. Moreover the energy from the wind must immediately be converted to electrical energy. If the surrounding grid is weak this can cause voltage flicker.

With variable speed a gust wind will only accelerate the rotors and storing the energy as rotary mass and all the above problems will decrease.

Theoretically also the utilisation will increase, but this advantage is so small that it is hardly worth mentioning. If the variable speed is used you must have special generators or frequency converters, also the demand on the control systems increase.

To summarise one can say that constant speed give high demands on mechanical components while variable speed give high demands on electrical and which is best varies from case to case.


A wind power plant consists of the wind turbine (blades and hub) with accompanying machinery [4,4:1). A tower on a fundament carries the machinery and the machinery is protected from wind and weather by the nacelle.

Figure 2.2 shows which components a modern large-scale wind power plant consists of.


Wind power and railway feeding Solution with three sided converter

A:r<or Oil He:ilt Comrol

9->aft C:>oi;:r




. ..._

~3:;:-:::::::::~ ~.-..,._

-- &:- _:'~ ~~

I _ I --


Yaw :'<11..11d Malr.

o,;...., Tower Proofiog R'al'rle

F'itoh S;;:;:rn1J

Dri"'= Bradat


Figure 2.2 Components of a wind power plant

2.5.1 Mechanical components

The rotor blades capture the wind energy and transfer it to the rotor hub.

When the wind speed is above the rated wind speed one has to dissipate its power somehow, this is usually done by changing the blade angle with the pitch drive. This blade angle control can also be used for better utilisation and for braking.

The rotor hub is attached to the rotor shaft of the wind turbine, which transmit the torque to the gearbox. The gearbox function is to transform the relatively slow speed of the wind turbine {20-40rpm) into the high speed of the generator (1 000-2000rpm).

The yaw drive turns the machinery so that the blades become perpendicular to the wind direction.

The wind wane measures the wind direction and speed and sends signals to a system that controls the pitch and yaw drive.

To be able to brake the wind turbine in emergency cases a mechanical brake is often placed on either the rotor shaft or the high speed shaft before the generator.

2.5.2 Generators

There are two types of generators that are in use in usual wind power plants, synchronous and induction generators [4,4: 1 0-13].

Practically all electrical power is generated by synchronous generators, but in wind power plants the induction generator is the dominating type. This is mostly because of the sturdiness, simplicity and low cost of the generator.

However a usual induction generator has the disadvantage of being a sink of reactive power and, unlike the synchronous generator, cannot provide the reactive power requirements of consumers and grids.

Page 11 (57)


Wind power and railway feeding Solution with three sided converter

Variable speed options

The synchronous generator by definition must have a rotational speed that is exactly the same for all power output if it's directly connected to a

common grid. This fixed rotational speed gives that the forces on the plant at gust winds gets very high if a synchronous generator is used.

With an induction generator the rotational speed can change and follow the wind gusts so that the forces on the plant decreases. This change in

rotational speed is however very small so even the induction generator must run at almost constant speed.

To be able to run the induction generator at variable speed a so-called variable slip generator is used [4,4,p15]. Here the excitation current on the rotor is controlled in a manner that the generator current has the same frequency, though it has different rotational speeds. This excitation control can also be used to make the induction generator produce reactive power.

2.5.3 Frequency converters

When a synchronous or usual induction generator is used the only possible way to have variable speed is to use frequency converters [4,4,p12]. These converters are built up with power electronics with the same principle as the static converters for electrical railways, see paragraph 3.2.2. This possible synergy was one of the basic reasons to make this investigation.

2.5.4 Transformer

The generated voltage lies in the region of 0.5-1kV and can therefore only be transmitted some hundred meters. Therefore a transformer must be placed in the absolute vicinity of the plant


Wind power and railway feeding Solution with three sided converter



For historical reasons the electricity type used in railway systems in Europe is different from country to country. The electricity used differs from 50Hz three phase in the public grid by frequency and/or number of phases.

Figure 3.1 shows the distribution between different systems in west and central Europe.

~3kVdc •

1111111 =

l.SkVdc ITIID25 kV. 50 Hz 1- ac


LS kV, 16 2J3 Hz 1- ac

Figure 3.1 Railway feeding in Europe

As the figure shows the railway feeding in Sweden is 15kV, 16 213Hz single phase. To connect the railway feeding with the 50Hz three phase grid, one use frequency converters and transformers. The converters are of two major types namely rotary and static. The contact line then feeds the train locomotives, which are of two major types; Thyristor and Asynchronous.

Page 13 (57)


3.2 CONVERTERS 3.2.1 Rotary converters

Wind power and railway feeding Solution with three sided converter

A rotary converter consists of two shaft coupled synchronous machines as shown in figure 3.2.

50Hz 3phase



h Hz lphase

Figure 3.2 Rotary converter

On the 50Hz side there is a three phase motor with 12 poles, and on the 16Hz side a single phase generator with 4 poles. The number of revolutions of the shaft is 500rpm and the frequency conversion occurs automatically.

On the common shaft there is a de exciter for field generating to generator and motor.

The benefit with rotary converters is that the load on the feeding grid is always symmetrical. They can also produce reactive power and an opposite power flow (when the trains brakes) is handled automatically.

With the rotary converter the contact line system and the feeding grid is separated from each other. The power fluctuations of the single phase with double feeding frequency 331/ 3 Hz is compensated by the rotary masses of the motor and generator. One disadvantage is the long start up time and the complicated synchronisation [6,p11-12].

3.2.2 Static converters

Nowadays most of the converters taken into operation in Sweden are static;

i.e. they are built up with power electronics [7,p67-71]. Two types are used;

cycle-converters and converters with an intermediate D.C. link; PWM converter. Cycle-converters create the low frequency voltage directly from the 50Hz grid. PWM converters first rectify the voltage to a D.C. voltage, which is then converted into 16Hz voltage. Due to the fact that static

converters have no mechanical parts they are much simpler to start up and synchronise than rotary converters. They also need less maintenance and are often cheaper to install. On the other side they always produce

harmonics both to the three phase and the single phase side, which causes demands on the feeding grid. To be able to handle an opposite power flow, they also need some extra electrical circuits.



Wind power and railway feeding Solution with three sided converter

As mentioned before the Cycle-converter directly build up the single phase low frequency voltage from the 50Hz grid. This is done, by controlling the gate signal on several thyristors as Figure 3.3 shows.

Main curcuit

Curve shape

Figure 3.3 Principle of the cycle-converter

The result is something that looks like a sine wave, but it still have a lot of harmonics. To improve the sine wave and increase the output power one uses a so-called 12-pulse connection [7,p165). This connection has two converters who are connected serially. One of the converters is fed from a transformer with .i-connected secondary winding and the other from an Y- connected secondary winding. To decrease the harmonics that still exists there are filters on both the 50Hz and the 16Hz side. The filters on the three phase side are also used to generate reactive power as compensation for the converter's consumption of reactive power.


PWM-converters have a D.C. voltage system that separates the feeding grid from the single phase grid, this decreases the harmonics compered to the cycle-converters a lot. Here the three phase voltage is converted to a D.C voltage in a 12-pulse rectifier. The voltage is then chopped in the inverter in a manner that the pulses time mean value is the wanted single phase voltage. Figure 3.5 shows the principle of a PWM-converter.

Just like the Cycle-converter there are harmonic filters on both the low and high frequency side of the converter. Each inverter can only handle limited power so several inverters have to be connected in parallel. This is also used to improve the curve shape on the single-phase side. Figure 3.6 shows a main circuit for a typical PWM-converter where the filters and three phase transformers are included.

Page 15 (57)



/ / 1/

Wind power and railway feeding Solution with three sided converter

Main curcuit

1'-- I'-


~ /


r--... _,.. 1/

Curve shape Figure 3.5 Principle of the PWM-converter

22kV Three phase 50Hz

..._ "T""I--~-f



+: l_ tdlJ

LJ ~

- 1 ~



Inverter Rectifier DC-Link +



Overvoltage Limiter



'---- -- -- - '


+-Q-IZI- !1

_ _ p- g]j]


Figure 3.6 Main curcuit for the PWM-converter


16.5 kV One phase

16 213Hz

In contrast to the Cycle-converter the PWM-converter raises no special demand on the feeding grid and the reactive consumption is relatively low.

To be able to handle an opposite power flow the PWM-converter must be provided with an inverter on the three phase side. These inverters are however not commonly used, instead the regeneration that can not be taken care of by other trains or other types of converters is simply burned away in the Overvoltage Limiter.


Wind power and railway feeding Solution with three sided converter 3.3 LOW FREQUENCY GRID

After the converters there are two main alternatives to feed the low frequency grid [7, p 160-161]. One is to transform the voltage to 15kV and directly feed the contact line. The other alternative is to transform the voltage to 130kV and use a feeding line that goes in parallel with the railway. This parallel line then feeds the contact line through transformers.

The advantage of the second alternative is that fewer converters are needed.

The locomotives are provided with voltage only through the contact line, but there is a lot of ways of handle the return current. In Sweden the most common way to handle the current is to use a system with booster transformer, but also auto transformers are used.

3.3.1 Contact line

The contact line provides the locomotives with voltage. The thickness of the contact line is limited by the fact that it can't be to heavy, therefore the voltage drop fast becomes troublesome if the distance between feeding points becomes to far. With a system with booster transformers the

maximum distance between feeding points is about 1 OOkm while it for auto- transformers can be longer because of it's higher transmission ability.

3.3.2 Return current

System with booster transformers

In a system with booster transformers (BT-system), an inductor connection through a current transformer with a ratio of 1 : 1 is used to force the current into the return conductor. The primary winding is in series with the contact line and the secondary winding is in series with the return conductor. This forces the current in the return conductor to be the same as in the contact line. The current transformer is called Booster transformer. Figure 3.6 shows an example of the usage of booster transformers.

Booster transformer Return condL tor

Contact line post

Figure 3.6 System with Booster transformers

System with auto transformers

In a system with auto-transformers (AT-system) a negative feeder whose voltage is in opposite phase with the feeding voltage is used. The

Page 17 (57)


Wind power and railway feeding Solution with three sided converter

transformer is placed between the negative feeder and the contact line with the middle point connected to the rail according to figure 3.7.

Uneg Negative feeder

~ tu~~ns -


r ..

I--~ Contact







contact ~


( ) (

z ~ r

I ( )


( )


Figure 3. 7 System with auto-transformers

Usually the negative feeder voltage (Uneg) is -15kV, which means that the transmission voltage (Utrans) becomes 30kV. But also other values such as -25kV on the negative feeder can be used. Though the transmission voltage increases the voltage on the contact line (Ucontact) is still the same and the benefit is higher transmission voltage.

In figure 3.7 the thick lines shows how the currents ideally is distributed. As the figure shows the return current will run in the negative feeder except on the section where the locomotive is. Due to the leakage inductance in the transformers the current distribution will not be ideal, which means that there will exist rail currents along the whole track section (thin lines in figure 3.7).


3.4.1 Power and energy consumption

A traction vehicle energy and power consumption is a quite complex story [7,p13-35]. However a brief description of the main parts is derived below.

The draft or braking force for a traction vehicle is:




is the train dynamic mass, a is the acceleration and F, is the running resistance.

Fr can be divided up into four parts 1. rolling resistance (Frr) 2. air resistance (Fra) 3. curve resistance (F,k) 4. incline resistance (Frg)



Wind power and railway feeding Solution with three sided converter

Now suppose that the track is relative straight with small inclines, then Frk


Frg= 0 and equation 3.1 becomes

(3.2) Here


is the vehicle acceleration or retardation,




can be expressed as


where m and vis the vehicle mass and speed respectively. The dynamic mass ma is often only some percent larger than the mass, so

approximately mn


m and equation (3.2) becomes:


The parameters Ct. c2and


3 do not depend on the mass and have the order of magnitudes of c1 = 10"2(m/s2) D.!= 10-4(s-1) and


3 = 1 O(kg/m).

The absolute value of a is for low speeds far higher than c1 and for higher speeds the last two terms in equation (3.4) is dominating therefore


can be set to zero.

By multiplying equation (3.4) with the speed and use the above order of magnitudes an approximate value of the power demand becomes.

P= amv


10-4 mv2


10v3 (3.5)

The mass of a train lies from some hundred tonnes for high-speed trains up to some thousand for freight trains and therefore the power consumption peak values are typically some MW.

If one consider how a train behaves on a typical track, with stations and different speed limits, the conclusion from equation (3.5) is that the power consumption fluctuates very fast. On an electrical railway section there are often several vehicles on the track and the power consumption fluctuations becomes even worse. This gives that electrical railway and their vehicles must be constructed to tolerate these variations and therefor also tolerate a fluctuating power source such as wind power.

Though the absolute power consumption can be very high for traction vehicles the relative energy consumption per mass unit is much lower than for most other transporting methods. This is mostly because the c2

parameter (friction) is much smaller for trains than for busses or trucks and that the speed is lower than for aircraft. With regeneration the main part of the consumption that depends on the first term in equation (3.5) can be reused, which decreases the energy consumption even more.

Page 19 (57)


Master thesis

Wind power and railway feeding Solution with three sided converter

The low mean energy to peak power ratio often means that electrical railway have to pay more for the electrical energy than a more stable consumer does.

3.4.2 Locomotives

In Sweden the most common electrical locomotive types are the Thyristor and the Asynchronous locomotive. The Thyristor type had for decade's almost the whole market, but in recent years the Asynchronous locomotive has become the most manufactured type. The Thyristor locomotive has D.C. motors and the name comes from the thyristor rectifier it uses [7,p49- 57]. As the name indicates the second type have 3-phase asynchronous motors fed from PWM-converters [7,p59-67]. This engine has many

benefits compered with the D.C. engine. It is robust and has not the need of moving parts as commutators. The asynchronous engine can also be made smaller and lighter and asynchronous locomotives have less influence on the feeding grid.


Wind power and railway feeding Solution with three sided converter



There are a lot of different thinkable system solutions for the combination of wind power and railway feeding appendix 1 . For example you could

directly generate the type of voltage needed by single phase or De- generators or have a low frequency generator and a phase converter. In this thesis the purpose was to investigate the solution with a three sided converter, which has the public grid, the railway and the wind Power Park on each side as figure 4.1 shows.

Public grid

DC-linked (PWM) converter 50Hz

16,7 Hz

Figure 4.1 System with three sided converter

The reasons to choose this solution instead of some other AC-solution were basically:

1) Both the wind power and the railways use frequency converters with a DC-Iink. So instead of having one converter for the wind power plant and one for the railway they could be built together.

2) The variations of the difference between generation and demand can be neutralised by the public grid.

3) The principal of this thesis has obtained a patent for this solution.

The solution above may seem quite simple to design, but during the work with this solution some problems with it occurred

1) The railway converter must be provided with an inverter on the three phase side, to be able to handle the opposite power flow at high wind power production. These inverters are not commonly used today, see paragraph 3.2.2.

2) The voltage level and quality on the DC-Iink is today not the same for the wind power and railway converters.

3) During failure or maintenance of the converter the wind power must be shut down.

Page 21 (57)


Wind power and railway feeding Solution with three sided converter

Problem 2 could be solved with wind power plants custom-made for railway feeding, while on problems 1 and 3 there is not much to do.


A more straightforward system, but with lower efficiency could be to have a three-sided transformer instead. Here the wind power plant is connected with the railway at the railway transformer station as figure 4.2 shows.

Public grid

Converter static or rotating

50Hz 16,7 Hz

Three sided transformer

Figure 4.2 System with three-sided transformer

The benefits with this solution are that it can be utilised using only standard products available today. Therefore it could be a suitable system

configuration to study how wind power and railway systems behaves together. Also it hasn't any special demands on what kind of plants and converters that is used.

On the other hand there is no direct connection between the technologies and therefor no synergies (such as the use of PWM-converters) can be achieved reducing the investment cost.




Wind power and railway feeding Solution with three sided converter

The object chosen for a case study is "Biekinge kustbana" (BKB). BKB lies in the south of Sweden and runs from Kristianstad to Karlskrona a distance of about 130km. The main reasons for choosing BKB is:

1) Railway electrification under discussion.

2) Excellent wind conditions.

3) Long distance to any large-scale production i.e. nuclear or hydro 4) Sufficient traffic load.

5) Suitable size for pilot project.

The purpose for the case study was to simulate the power and energy flow in the wind power, railway system and main grid under different

circumstances. It were basically three questions the simulations should answer:

1. What extra equipment will be needed to connect large-scale wind production to BKB?

2. How large should the wind Power Park be to cover a large part of the consumption of BKB without to much overproduction?

3. Is it economically reasonable to combine wind power with railway feeding at BKB?

The model for the simulations included the three-side converter and the problems with it unfortunately occurred after the simulations were done [12]. But the simulated model was quite general and the results from it can with some reasonable adjustments be used for other system solutions such as the three-sided transformer.


5.2.1 Main program

The main program used was SIMTRAC, which uses the power system simulation program SIMPOW® as calculation kernel, where SIMPOW is a simulation an analysis program developed and sold by ABB Power

Systems. SIMTRAC is a program for simulations of railway systems, where consideration is taken of the trafficking vehicle's movements in the system.

The program can handle:

• Different kinds of AC and DC railway feeding

• An arbitrary feeding grid

• An arbitrary large part of a distribution net

Page 23 (57)


Wind power and railway feeding Solution with three sided converter

• Free placement of stations

• An arbitrary traffic situation

• An arbitrary number of traction vehicles of different types

• The vehicles reaction on high and low voltage

• Regeneration of power

• The influence of curves and uphill and downhill slopes

• Different speed profiles

• Different speed limits for different vehicles types

An example that visualises the form of input and output data that SIMTRAC uses is given in Appendix 5.

5.2.2 Other programs

The time series for the wind and wind power production was made in Mathcad® a mathematical calculation program developed and sold by Math Soft. The presentation graphs and some subsequent treatment of the SIMTRAC output was made in Origin® a data analysis program developed and sold by Microcal.


The railway section investigated here is BKB, but to make the model more accurate we also include the already electrified sections from Kristianstad to Hassleholm and Karlskrona to Emmaboda and connect them electrically.

The reason to include these two sections is that they are connected to the nearest existing feeding points to BKB. A sketch of the railway sections with stations and some important distances is shown in figure 5.1.



82,8km )( 47,2km )

Figure 5.1 Railway sections

Traffic routes

On these sections there is two valid passenger traffic routes namely, Hassleholm to Karlskrona and vice versa and Emmaboda to Karlskrona and vice versa. The freight traffic route is assumed to be Emmaboda to Hassleholm and vice versa.


Every passenger trains stops at each station in figure 5.1 while the freight trains does not stop until they reaches the end station on the route.


Wind power and railway feeding Solution with three sided converter 5.4 CHOSEN TRAINS AND TRAFFIC

5.4.1 Passenger train

The passenger train chosen for simulations was the so-called "Flight train"

which is the same type as the trains, which runs on the railway to the airport in Oslo. This train is a modern train with an Asynchronous locomotive with high performance. Reasons for choosing this train were basically:

1. Probably this train or something like it will run the "Biekinge Kustbana"

after the Electrification [13].

2. A plausible model is available in Simtrac.

Some important data for this train is:

1. Mass= 358 tonnes

2. Max speed= 210km/h (max speed on track= 160km/h) 3. Power factor-1 for all speeds and effort

In figure 5.2 the chosen passenger train tractive effort curve [7,20-21) at nominal voltage and the maximum allowed power consumption as a function of pantograph voltage is shown.

en Qi

Q) .t::




~ 150








~ Ill 50

~ 0


1\ " "'-.. '



.Q c

a. 6

E ::>






0 0.

E ::>


)( 2


kmlh 0


v I



0 50 100 150 20() 250 10 12 14 16

Speed Voltage at Pantograph

Figure 5.2 a) Tractive effort curve for the passenger train at nominal voltage b) Maximum allowed power consumption as a function of pantograph voltage for passenger train.

From figure 5.2 it can be noticed that the allowed power consumption and therefore also the tractive effort starts to decrease below 15kV and below 1 O.SkV the train can not operate at all.

Page 25 (57)


5.4.2 Freight train

Wind power and railway feeding Solution with three sided converter

The chosen freight train was a train of large size with a thyristor locomotive often called RC-class locomotive. This train, which is quite heavy and old consumes a lot of both reactive and active power and therefore is

interesting both for dimensional and energy calculation purpose. Though it compromises a type of locomotive not being manufactured anymore, locomotives of this type are running railways in Sweden and will do so for long time and therefore probably also BKB [13]. Also for this train there is a plausible model in SIMTRAC and some important data for the chosen train is:

1. Mass=1200 tonnes 2. Maxspeed=120knVh

3. Power factor varying with speed and effort.

In figure 5.3 the chosen freight train tractive effort curve at nominal voltage and the maximum allowed power consumption as a function of pantograph voltage is shown.






(1) 200






5 iii 150


~ 0

1\ '\

(1) 100

(1) >




I- 50


0 20 40 60 80 100 120



c 0 10

"" c. 8 E =>






(1) 0 ~


E :J

·x E

~ 2 km/h o






12 14 16

Voltage at Pantograph

Figure 5.3 a) Tractive effort curve for the freight train at nominal voltage b) Maximum allowed power consumption as a function of pantograph voltage for the freight train

From figure 5.3 it can be noticed that the allowed power consumption and therefore also the tractive effort starts to decrease below 13kV and below 1 0.5kV the train cant operate at all.




5.4.3 Traffic

Wind power and railway feeding Solution with three sided converter

Two different traffic situations were studied. One is an extreme traffic situation for dimensional purpose and the other is a normal situation for energy calculation.

Extreme situation

Here the simulated time was 1 0000 seconds and the traffic situation was as table 5.1 shows.

T bl 51 T a e ra ff IC d unnq e xt reme s1 ua 1on

iTrains from ~0 Departure time (s) ~ype

Hassleholm Karlskrona 200 and 3800 Passenqer

Karls krona Hassleholm 1200 and 4500 Passenger

Emmaboda Karlskrona 500 and 2900 Passenger

Karlskrona Emmaboda 1500 and 4950 Passenger

Hassleholm Emmaboda 1900 Freight

Emmaboda Hassleholm 3200 Freight

Normal situation

Here the simulated time were 86400s i.e. a hole twenty-four hour period.

The passenger traffic situation was taken straight from the valid timetable in the area. Shortly this means that 1 train departs from each end stations every hour during daytime (Sam to 11 pm).

The freight traffic is assumed to be one train from Hassleholm to Emmaboda at ?am, and one train in the other direction at 3pm.


Two different models for the simulations was used, namely one system with booster transformers (BT-system) and one with auto transformers (AT- system) with -25kV on the negative feeder. The model for the system is shown in figure 5.4.

Emmaboda Brakne


Rotating converter @-8wind power park (negative load)


Trains (loads)


I r· 'L b

~ ) n m1 e us

Figure 5.4 Electrical model for the systems

Page 27 (57)


Wind power and railway feeding Solution with three sided converter In figure 5.4 the contact line impedances are

Z1= 0,3+j0,23 Z2=0,21+j0,2 Z3=0,15+j0,16 (Q/km) For the BT system.


Z1=Z2=Z3= 0,037+j0,032 (Q/km) For the AT system.

5.5.1 Comments on the electrical mode l


The fixed nodes in the system are, Hassleholm, Kristianstad, Brakne-Hoby, Karlskrona and Emmaboda. The trains are modelled as a varying load connected to a moving node.


In Hassleholm there is installed two PWM-converters of 13,4MVA each.

This is a proportionately large capacity and is therefore modelled as an infinite node. In the neighbourhood of this node there is also some hard trafficked railways that can consume some of the produced wind power.

This has small effect when the BT-system is used, but with the AT-system the wind power production could be transmitted here and used in stead of being transferred to the public grid.

Kristianstad and Karlskrona

The section between these nodes is not electrified yet. In the electrical model they are nodes just because the line impedance change here, see below.


In Emmaboda there is installed two rotary converters with power rating 5,6MVA each. This is modelled as one rotary converter of 5.0MVA. Here it is assumed that the rest of the installed power must provide neighbour sections. These sections can also consume some of the produced wind power. Rotary converters can during a short time (some minutes) be overloaded as much as 1 00% and therefore the modelled rotary converter can produce as much as 1 OMVA peak levels.

Brakne-Hoby (wind power)

In Brakne-Hoby it is assumed that wind Power Park and the three-sided converter will be placed. This node will from now on be called "Wind Power". The reason for choosing this spot is simply because it lies abaut half way between Hassleholm and Emmaboda.

The converter is assumed to be a PWM converter with power rating of 15MVA that can handle up to 17MVA peak levels. So far the converter has nothing to do with the wind power i.e. it would be needed anyway for power supply to BKB. To create the three-sided converter the wind power plant will be connected to the DC-Iink on the PWM-converter. To be able to handle opposite power flow the converter must be provided with an inverter against the public grid, which is assumed to have a power rating of about 5MVA.


Wind power and railway feeding Solution with three sided converter

In the simulations the three-sided converter has no special model, instead the power flow between the public grid and the railway is handled by a rotating converter and the wind power park is directly connected to the contact line, see paragraph 5.7.

The wind Power Park is assumed to be three 1.5MW wind power plants, see paragraph 5.7.5. Here it is assumed that the frequency converter on the wind power plants can keep the power factor to 1, therefore the wind Power Park is modelled as a negative active load. This load get its values from the power curve of the chosen wind power plant multiplied with 3, see figure 5.5. The wind time series is calculated with the program derived in [5], see paragraph 5.6.2.


The trains are modelled as varying loads connected to a node that moves along the contact line. The load value and the node position in the system is varying with speed, effort, voltage levels etc.

Rotating converters

A rotating converter is modelled as an ideal synchronous generator behind a reactance (Xr) were Xr=j0.765Q. An infinite bus drives the generator. The active and reactive production of this generator is varying with the system demand and can be negative during opposite power flow.


The lines and their impedance's are an equivalent for the contact line booster or auto transformers and return conductor.

In the BT-system Z1 and Z3 is the equivalent for the existing system while Z2 is the equivalent for an assumed contact line standard on the not electrified section.

In the AT -system the impedance's is not the one for a 40kV system.

Instead an equivalent for a 15kV system is used. For a 40kV system the line equivalent is Z= 0.26+j0.23, which in a 15kV system becomes 0,26+j0,23*(15/40)2= 0.037+j0.032(Q/km) [13].

5.6 MODEL FOR WIND POWER 5.6.1 Wind statistics for Blekinge

The institute of meteorology at Uppsala University has measured the wind during the period of June 1990 to January 1993 on a spot called Utlangan on the south-east coast of Blekinge. From these measurements the following information of the wind characteristics at 25m height was given.


The wind is Weibull distributed with shape parameter


= 2.21 and scale factor a= 9.32m/s. This distribution is plotted in figure 5.5

Page 29 (57)


% 10


Wind power and railway feeding Solution with three sided converter



·---l. ___ ll_

1)' c

Q) ::l


: I ~

r-···--·-r · --- T·- ·---·

I '


- - ;·-- i -1-- ---r-,--

, -- -- ---[- ---, ___ t ___ -- -

g 4



0 +---.--t--..--t--...---+-.----+=---ir-~-T-m/s

0 5 10 15

Wind speed (U)

Figure 5.5 Distribution Utlangan at 25m.

The mean value and the variance is:

Mean value m = 8.25 m/s Variance

c::l =

15.6 (m/s)2

Variations during 24 hours

20 25

In figure 5.6 the daily variations at Utlangan is plotted.

0 5

- -Variations during 24 h ----Sine approximation


Hour 15

Figure 5.6 variations during 24 hours

20 30

Where the sine approximation is made to establish a function that easily can be used in the wind realisation program, see paragraph 5.6.2.


Yearly variations

Wind power and railway feeding Solution with three sided converter

The yearly variations are plotted in figure 5.7.



2. 9,0


Q) 8,5

~ (/)


c: 8,0




0 2 6 8 10 12

Month (Jan= 1)

Figure 5. 7 Yearly variations.


From the same meteorologists as above a wind spectrum was given.







0,4mlln(zl zo) Where


= nzlm

Dimensionless frequency

z =

Height over ground m

zo =

Roughness length m m


mean wind speed m/s

5.6.2 Wind realisations


The above data for the wind was used in a wind simulation program derived in [5]. The program uses a wind spectrum to generate a time series for the wind speed. The input to and output from the program is:


1) The mean value of the wind (m). Different values of m were used in different simulations, see results. In some simulations m is changed during the 24h according to the sine wave change in figure 5.6.

2) The wind spectrum S(n), were the spectrum in equation (5.1) was used.

The chosen parameters where z=50m and Zo different for different simulations see results.

Page 31 (57)


Wind power and railway feeding Solution with three sided converter

3) Limits for the output frequency contents (w, and w2). Here w1=0 and w2=1 (rad/s) was chosen because it was assumed that the outage from the wind power park would have approximately this frequency contents.

4) A parameter N that decides the accuracy for the output meaning that the spectrum for the output data converts to the real spectrum as 1/N2.

Here N=1 00 was assumed to be accurate enough and was therefore chosen

5) Time step between output values. This time step was chosen to be 60sek.


A stochastic time series vector with mean value::::m and spectrum:::::S(n).

5.6.3 Chosen wind power plants

The chosen wind Power Park was three 1.5MW plants manufactured by TACKE Windenergie GmbH.

Some date for the plant is:

• Generator= Induction with variable slip and reactive power control

• Sweep area = 3318m2

• Tower Height= 65m

• Power rating = 1.5MW

The power curve for this plant is plotted in figure 5.8.



~ 1,0

a.. ~



!!: 0



o .o +-~,...:::~-...---.---.--,---.---.--.--+---u

0 5 10 15 20 25

Wind speed (m/s)

Figure 5.8 Power curve of the chosen wind power plant This power curve has the function:



0,17 -0,04U -0,015U2 +0,0047U3 -0,0002U4 3


U < 13 (S.

2) P(U) = 1,5 13$ U $ 25 P(U) = 0 Elsewhere


Wind power and railway feeding Solution with three sided converter

The reactive power generation/consumption of the plant is excluded in the model.

5.7 SIMPLIFICATIONS AND ASSUMATIONS MADE IN THE MODELS This is far as is known the first study of large-scale wind power and

electrical railway feeding and a complete model of the system would be far to complex, so some simplifications has to be made. The most important simplifications are listed below.

5.7.1 Disconnection of system from surrounding grid and other railways

The system is disconnected from the public grid and other railway systems without any concern about short-circuit power, impedance or loads at the feeding points. This means that the injected power in these points can deviate a little from the "real values".


The total power consumption for the model will be unchanged and at least an estimation of the power in every feeding point can be calculated. During the simulations the injected powers were observed so that they lied in reasonable regions and no extraordinary values was observed.

5.7.2 Whole system with 15kV 16Hz single phase

As figure 5.4 show, the rotary converters, the wind power plant and the infinite busses all work with 15kV, 16Hz and single phase. So all coefficient of efficiency is set to unity for transformers and converters and the

impedance of transformers to zero.


Compared with the losses in contact wires and trains this kind of losses can be neglected.

5.7.3 Simplifications in the track layout

The whole track is modelled like a completely straight track with no gradients, curves or tunnels, but the valid speed limits and distances between stations are included.


It was basically three reasons why the gradients was ignored:

1) Blekinge kustbana runs through a rather flat landscape.

2) The total energy consumption for several trains travelling back and forth on railway track is only marginally influenced by gradients. This is due to the fact a train that travels uphill stores potential energy that can be used then travelling downhill. The only thing that happens is that the power consumption will be moved a little in time. If the landscape had

Page 33 (57)


Wind power and railway feeding Solution with three sided converter

been mountainous the gradients have had to be included because the power demand should change too much, but this is as mentioned above not the case.

3) It is not certain that the simulated results become more accurate due to the fact that gradients are included. This is due to the fact that human control can not be included in the simulation program. Take for an example a train that travels over a small hill. A usual locomotive driver would not increase the power consumption just to keep exactly the right speed. Instead he would let the speed decrease a little uphill and increase downhill and therefore have constant power consumption.

With the same situation the simulation program will keep the right speed and therefore change the power consumption.

Through experience it is known that curves and tunnels influence on power consumption can be neglected [13].

5.7.4 Logistics and human control

The only care about the logistics has been that the trains leaves the start station according to time-table and arrives to the end station on about the right time. This means for example that several trains can be near each other and that two trains can meet each other on a single track. The human control is neglected totally i.e. the trains travels as fast as the track and train allows without any care about the power demand or the voltages levels.


This has very small impact on the energy consumption. For the power consumption it always gives higher peaks than it would in reality and the voltage drops will be lower in reality.

5. 7.5 Chosen constellation of the wind power park

The wind power park was chosen to be a park consisting of three 1.5MW plants to start simulations with. After these starting simulations it was concluded that there was no use in simulations with other constellations.


To establish a starting value for the size of the wind Power Park the following observations was made:

1) Introductory simulations showed that the mean active power demand in node Wind Power was 2-3MW.

2) The mean production of a wind power park will according to paragraph 5.6.1. be 1/3 of the installed

3) To minimise the power need from the public grid and the transmission losses the wind Power Park must cover the main part of the power demand

4) The production that can not be used in the railway must be kept low on order not to increase the transmission losses


Wind power and railway feeding Solution with three sided converter

5) The load and generation is fluctuating and not correlated at all which makes it impossible to fulfil point 3 and 4 so the park size must be a compromise.

With these observations it concluded that 4-6MW installed power should be reasonable.

Then a co-operation with a wind power manufacturer started and one of the manufactured plants that could be suitable was a 1.5MW plant and

3*1.5=4.5MW lies in the above region, therefore it was decided that three 1.5MW plants was a reasonable starting wind power park.

Also some criterions whether to reconsider the choice was set-up, these criterions were:

1) At least 25% of the energy demand in the selected railway system should.

2) At least 75% of the produced energy shall be used within the railway And after the starting simulations it turned up that these criterions were fulfilled so no reconsideration had to be made.

5.7.6 Some assumptions about the plant and the wind characteristics The plant is placed so near the feeding point that the transmission losses can be neglected. The wind characteristics at this location (Brakne-Hoby) are the same as the measured ones at Utlangan.


This could be a questionable assumption because it is unknown if a wind power park can be placed near Brakne-Hoby and if the wind parameters are as good as the measured. One the other hand the measured wind is all known data and to get other measurement will take too much work and time. The mean wind speed also increases with the height, which could be of benefit because the plant tower is more than twice as high as the height for the measured wind parameters.

Alternatively it is unknown how large the transmission impedance will be if the park is placed elsewhere, but sure is that it will be low if the

transmission voltage has a normal value (1 0-30kV) and the distance is lower than some 1 O:th of km.

Also it should be mentioned that choosing the specific plant was a mistake if it should be used with the three-sided converter. This is due to the fact that it doesn't use a PWM converter at all. The variable speed ability is controlled directly in the generator. The cause of this mistake was that at the start of this thesis it was believed that variable speed was the same as using PWM converters. However this plant can without any (known) problems be used in the three-sided transformer solution. Also the power curve for different wind power plants with the same power rating doesn't deviate so much from each other. The reactive power can due to the reactive power control be set to zero.

Page 35 (57)





Relaterade ämnen :