Questions & Answers


Over the years, a number of questions have been sent to us concerning railway technical subjects.  Here are some of them and the answers.  Many of the answers have been provided by Dr Felix Schmid, Course Director for the Railway Systems MSc University of Birmingham

Email your question to

Transformer Cooling

From Sumit Rao, 23 February 2011:

I am a railfan  from India and I enjoy reading your pages. I wanted to ask if a 3-phase 25kv ac 50hz electric locomotive with ac-ac transmission requires cooling of the transformer and other parts? Does it gets cooling from motor blowers?

From Felix Schmid,  23 Febuary 2011:

Usually, the transformers are in an oil bath and the oil is cooled in radiators, normally with a separate fan or air cooled. There may be situations where the traction motor blowers are used.

Tramway Conduit in Washington DC

From Robert McCreary, 24 January 2011:

I saw a picture, taken by a friend, of a street car in Washington DC on a street where there were no overhead wires for current collection. The track, however, looked like a cable system, with a slotted rail in the middle. My friend said they collected the current through this slot and avoided unsightly wires.

From Piers Connor, 24 January 2011:

Your friend was correct. The conduit current collection system was common. There's a reference here, which includes a paragraph on the Washington DC system.

Cylinder Dimensions

Robert C. Helwig

You usually hear about the size of the steam cylinders as 22 x 24 or 16 x 22. Were the dimensions the ID or OD of the steam chest cylinders?

From Piers Connor, 19 January 2011:

The inside naturally.

3rd Rail Safety

From Tony Brandwood, 17 May 2002:

I travel to work every day on the Wirral electric line to the University of Liverpool.  Given that the 3rd rail is very dangerous (a teenager was electrocuted at our local station after touching the 3rd rail), why isn't the line littered with the bodies of animals e.g. foxes which live along the line and regularly cross it?

From  Felix Schmid, 17 May 2002:

Thank you for the question.  The voltage on the third rail is not very high at 750Vdc.  In order to be electrocuted it is necessary to touch both the third rail and either ground or one of the running rails at the same time.  Many small animals will jump onto the third rail and thus there is no path for the current to flow to earth.  That is why birds can sit on high voltage lines without risk.

It may be possible that foxes and badgers learn that the third rail is dangerous from older generations - just like hefted sheep know how far they can stray.  Not sure about that one though.

Rubber Tyred Vehicles

From Helen Tan, 15 May 2002:

I read your question & answer column with much interest on the "trainweb" website.   I am with a consulting engineering firm in Singapore.  Would you mind providing your opinion regarding the use of rubber-tyred vehicles for light rail systems compared to system of wheels on steel rails?  I think the rubber-tyred vehicles, used for lower passenger capacities, are usually shorter in length than the conventional train cars.   They are more maneouvrable around horizontal curves, and also able to climb steeper gradients.   Apart from these advantages, are there other benefits of using rubber-tyred vehicles for a light rail system?

Also, do rubber tyred vehicles require some form of tracking guidance system that protrudes from the ground?  My concern is that this would make them unsuitable for running in combined LRT and vehicle traffic on public roadways.

From Felix Schmid, 15 May 2002:

Thank you for your enquiry.  As you know, there is already a light metro with rubber tyres in Singapore (at Bukit Panjang) - this is effectively an airport people mover and is not really suited to the application for which it is being used - the reliability figures speak for themselves.  This would definitely not be suitable for operation on a public road.   The VAL system is also not suitable since it is geared to automatic operation and requires vertical guide-rails.

The only two rubber tyred systems similar to trams I know are the GLT system and the Translohr system.   The GLT (Guided Light Transit) system by Bombardier is operating (after huge difficulties) in Nancy.  Although savings come from being able to operate parts of its route without special track, with vehicle costs in the same order as a light rail vehicle it is likely only to be viable on fairly heavily trafficked routes.  The initial experience has been very bad.  The Translohr system has only been used on a test track in Paris.  Both GLT and Translohr use a central guide-rail in the road and require twin wire trolley bus overhead supplies.

The big drawback of GLT, Translohr and conventional trolleybus is the fact that the road has to be reconstructed in almost the same way as for a light rail / tramway system: the vehicles are relatively heavy and guidance forces them to run over the same part of the road surface all the time.  It is therefore necessary to re-build the road.

Rubber tyred operation is helpful for climbing steep hills and because of the short accelerating and braking distances.  However, the latter problem can be easily remedied on a light rail vehicle by using track brakes.  Rubber tyred vehicles require more energy and allow less regeneration during stopping.

Locomotive Power

From Mike Lipscomb, 13 May 2002:

I'm working in Kazakhstan on a rough alignment for a potential single track heavy-freight route across undulating terrain.  I know that 1 in 100 is a rule of thumb maximum ruling gradient on heavy rail, but I'm going to try to limit it to 1 in 200 since we're talking about diesel locos hauling up to 5,000 tonnes - does that sound reasonable, or too conservative?  

That sounds like a sensible proposition.  If you work on the basis of 1 in 100 then you need a minimal tractive effort to overcome gravity of 0.01*9.81*5000kN = 500kN.  If you go at 20m/s (about 70km/h) this translates into P=F*v=500*20KW=10,000kW or about three heavy diesel locos, just to overcome gravity.  Then you have rolling resistance, aerodynamic resistance etc.

3rd Rail Gaps

From  Anawat, 28 March 2002

I am working at a company which is the supplier for a new metro's M&E equipment and is going to operate the metro system in the near future.

The rolling stock is based on a 3 car train (Motor Car - Trailer Car - Motor Car), fed by a 3rd rail using a bottom contact collector shoe.  Collector shoes are installed on all vehicles.  Only collector shoes in each vehicle are connected together.  Collector shoes on both motor cars supply only traction eqiupment on that vehicle, while the collector shoes on the trailer car supply the power for auxiliary equipment for the whole train.

The problem is that when the trailer car is passing a crossover section where, currently, the 3rd rail is not provided, the 3rd rail gap is larger than the trailer car shoe span which causes loss of power supply in that area.  Passenger discomfort might occur due to lighting flicker and A/C power cut off.

a) I would like to know what are the solutions for this problem on existing systems.

b) Normally, the battery supply controller will start supplying power for essential loads and shedding off unessential loads immediatly the power supply is lost.  Would it be feasible to extend the time delay of the battery supply controller so as not to shed off unessential loads for 3-5 seconds after the loss of power supply.  Lighting and A/C are considered as unessential loads.

From Felix Schmid and Piers Connor.

Thank you for your e-mail.  This is a rather interesting problem which we have not come across before.  Most modern metros do not have current collection on all cars and simply connect the power supply along the train with a busline to equipment, like converters and compressors, which requires it.  The expense of putting collector shoes on trailer cars is unnecessary.  Even if the loss of power at gaps was not an issue, it is much more cost effective to supply equipment on the trailer from a connection to the shoes on one of the motor cars.  Shoes are a serious maintenance problem and should be kept to a minimum. 

It will be worthwhile to add additional third rail sections in crossover and turnout areas if possible.  Careful design of the third rail at crossings can avoid loss of power for more than a second or two.

Another solution would be to provide a common DC power busline throughout the consists.  This may involve structural alterations and may therefore not be possible.  However, you might consider feeding the auxiliary bus from the traction collector shoes by means of a high power diode which prevents the auxiliary supply from feeding the traction system.  You would probably have two diodes in series with a monitoring unit to detect diode (short-circuit) failures.

However, a problem which can arise with a busline connecting shoes in a 3-car unit is the fuse rating.  It is difficult to find a rating which will protect the busline equipment and, at the same time, not risk rupturing over gaps. 

It is worth investigating why your supplier did not put in high current links between vehicles.   We would assume that the power cables between vehicles are rated at only the auxiliary load on the motor cars.

For a modern auxiliary converter, load shedding for lighting and control circuits will not occur within a 3-5 second range.  If the converter also supplies the air conditioning, the equipment load would have to be shed immediately.  If the AC units are fed directly from the DC supply, it will cut out anyway.  Most passengers would not notice a few seconds loss of AC. They would notice a loss of lights.

Simply providing section gaps which are longer than a vehicle may not protect staff working on the line: I assume that your metro uses regenerative braking on its trains.  If a train enters a section while braking it will create voltages which can easily kill maintenance staff.   No work on electrical equipment on a railway should be carried out without fully grounding (earthing) the third rail.  Diodes between trailer and motor cars could also cause risks.

Our correspondence indicates that there have been some design decisions on your metro which may not be as appropriate as intended.  You may have to revisit the general design of the system

Control Circuit Earth Faults

From  George Brooks, 9 August 2001

I am doing a research paper on the effects of low voltage direct current grounds on commuter trains.  I am interested in how other railroads troubleshoot and repair these problems.  I would like to see a maintenance procedure on any railroad you could find or possibly point me in the right direction to find this information.   The information I am looking for is in the control circuits.  I am interested in low voltage short circuits to ground on GE Silverliner IV trains, I.E. wheelslip circuits, 38 volt circuits, cab make-up circuits, communication circuits, etc.  I would like to know how other railroads treat this problem.  I would also like to know what the safety considerations might be since this problem sometimes activates relays in other circuits.  I have been trouble-shooting low voltage DC grounds for 15 years, however I have never researched the problem.  I am having trouble finding infomation, so if you could point me in the right direction it would be greatly appreciated.


Dear Mr. Brooks,

Thank you for your e-mail.  I agree, this is a difficult problem. I am not able to point you to any literature in particular but a first port of call may be trolleybus design. Trolleybuses present great problems in electrical circuit design because of the risk of the "negative" pole leaving the wire.  All power systems have to be double insulated with an automatic system to supervise the insulation quality.

When I was at GEC many years ago we used to have a (single) common earthing point for power and control circuits.  Control wiring would normally have return (common) wire loops, allowing double sided feeding.  However, these could be problematic if the out and return (common) wires were not run close together (EMI!).  Electronic circuits were generally earth free (i.e., the negative or return wire was not grounded) but the screens or shield were taken back to a common earth point which was also the chassis earth point of the electronic systems.   Some people would earth the shields at both end and would then wonder why there were spikes on the electronic circuits.

I guess, one of the problems today is the use of auxiliary inverters which tempt people into having floating common rails for the control and electronic systems.  Such systems are much more likely to allow dangerous back-feeding than earthed / grounded systems.  They should have at least a low-ohmic connection to the overall ground so that an earth leakage breaker can detect stray currents.

In terms of textbooks you may wish to look at standard textbooks on measurement.  There was also an excellent book on grounding and shielding published in the 1970s or 1980s.  Another source may be trouble shooting manuals for computers. There is a web site in German:

Sorry I cannot provide more useful information.  With this publication on the site, there maybe someone else who can offer their views.

From Piers Connor

In my experience, all safety circuits have to be hard wired individually with separate return circuits and relay contacts.  Any attempt to use a common ground would raise the risk of leakage faults and "back stairs" circuits.  I recall one e.p. brake control circuit where it was possible to release the brake from an isolated cab if the driver had applied the brake from the front cab.

Jerk Rate

From  Lin Chao Ming, 31 July 2001

Regarding the ride quality over a turnout and return curve, I am interested in the jerk rates, which are referred to as point heel or in the body of  transition rate and cant deficiency; how do you calculate these?   And what is the relationship between them ? 


Dear Mr. Lin,

As you will know, jerk is the rate of change of acceleration.  This can be applied to accelerations in all axis directions.  However, you are probably interested in lateral jerk, i.e., the jerk which is experienced when entering any curve.  In order to calculate the jerk you would have to find out the lateral acceleration at the beginning and end of a curve transition (normally, this will be OMEGA*Rsquared) and then find out how long the train takes to cover the transition length (LENGTH divided by LINEAR SPEED).   The change in acceleration divided by the time then gives you the jerk over that length.

When calculating the jerk experienced by passengers you will need to base your calculation on the net lateral acceleration, that is, you have to take into account the cant of the track (and thus that of the train minus suspension relaxation) when you calculate the lateral acceleration values.

Ruling Gradients and Rail Weights

Questions from  John Duncan, 28 July 2001

I am researching the 19-mile Peebles Railway, 1855-1876, whose engineer was Thomas Bouch.   It was reported that the ruling gradient was about 1 in 90.  I should be grateful if someone could explain what a ruling gradient is and how it is calculated.

Answers from Roger Viggers, 29 July 2001

The ruling gradient is simply the steepest gradient on a line where the gradient varies.  Thus if a line starts level, climbs at 1 in 150, followed by 1 in 70, 1 in 65, 1 in 30, 1 in 100, 1 in 450 and finishes on the level at a higher elevation then the stretch at 1 in 30 will be the ruling gradient and will determine the weight of train for a given locomotive or the power requirements for a given weight of train.

In C.J.A. Robertson's 1974 article on the St Andrews Railway in the Journal of Transport History, he mentioned 'relaying the track with 75 lb. rails to take heavier locomotives.'   How long is a 75 Ib. rail, or indeed, any other rail quoted in lbs.?

Rail weight quoted in 'lbs' is a shorthand and refers to its weight per yard, i.e. 75lb rail is 75lb per yard of rail.  A rail can be as short or as long as is needed for the job, subject to the material of which it is made.  Originally made of cast and later wrought iron rails were as short as 3 feet.  Later steel was introduced and lengths extended to 45ft and then 60ft.  They are now laid in lengths of a quarter mile or more.  The reason for them being relatively short previously is due to transport and handling limitations.

Electric-Diesel MU-Cable

From  Winfried Mörs, 21 June 2001

In some areas in South Africa and under certain operating conditions, it is economically viable to reduce the number of electric locomotives in a multiple electric locomotive consist by replacing one or more of them with diesel locomotives.  This is the case in areas where the route topology is predominantly flat and the reduced number of electric locomotives are capable of providing adequate power to haul the train most of the way.  In the rare events where inclines are encountered that require more power, the extra diesel locomotive(s) are temporarily used to "top-up" the tractive (and braking) effort requirements of the train.   Using fewer electric locomotives releases scarce electric locomotives for service elsewhere, as well as improves their efficiency.

This implies a combination of electric and diesel locomotives in a consist.  The question is if you know of suitable remote control technology currently available to accomplish the control of the trailing diesel locomotives.  Spoornet was involved in the development of a rather expensive MU-cable technology system, dedicated to the interfacing of the class 9E-electric and 34-500 diesel locomotives on the OREX iron-ore export line.  I was wondering if something else is available off-the-shelf to accomplish this control from some of the other classes of electric locomotives.  I have tried web searches, but with limited success.


Dear Mr Mors,

Thank you for your interesting enquiry.  First a question: presumably, you do not allow the additional diesel locomotives to "idle" while they are not helping to pull the train. Any remote control equipment will therefore have to be able to start up the diesel locomotive as well.  After all, we do not want more pollution than absolutely essential.

As regards remote control equipment I would suggest that you look at GE Harris' Locotrol Equipment which should do the job quite happily.  See their web-site:

I should imagine that they will be able to provide a voltage free contact to start the diesel unit(s).

Maximum Gradients

From Tristan Neagle, 11 Jun 2001

I am an engineering student at the University of Oxford and am currently producing a report on transport safety.  I would like to ask if you know of the maximum gradient up which it is possible for a normally railed train to travel.  I assume that it is closely related to the coefficient of friction between two parts, both made of steel but also imagine that it can vary quite by quite a large amount?


Dear Mr. Neagle,

Steel wheel on steel rail operation is possible without special devices (rack and pinion, Fell rail etc.) up to a gradient of about 10% (Sheffield Supertram 10%, Uetlibergbahn 7%). Beyond about 7% you need all axles powered, beyond about 5% it is essential to fit track brakes. Adhesion coefficient can vary between 0.02 and 0.3 or so but you would not normally assume that it is worse than 0.05 over more than a few metres.

Cable Meggering

From Philip Rose, 10 May 2001

In regards to cable testing, on multicore cables, although one does a loop resistence test on a cable, what is the technical reasoning behind sending a 1000v through each individual core of the cable to the rest of the cores, when in fact if there is a problem with the individual cores, would not the loop resistence test acknowledge it?


Dear Mr Rose,

Low voltage resistance measurement does not detect flaws in the insulation BETWEEN conductors.  The high voltage test will show up potential future leakage paths which could lead to (dangerous) energising of wires which belong to different circuits and which may have safety critical roles.

See also this link to notes on meggering procedures.

LRT Power Rails

From Jaclyn Eng, 2 May 2001

Please advise if the third rail  (power rail) utilized on most light rapid transit systems is considered a electrical conductor.

Dear Ms Eng,

Yes, the third rail or power rail is a conductor for electricity and can be of several types: underrunning (normal today), top running (most frequent) and side-running (not such a good idea).  Rails can be steel or aluminium with a (stainless) steel running surface.  The latter construction comes in two varieties, bonded or shape fitted.

Transient Short Circuit Calculation

From Renato Peres Vio, 13 March 2001:

Let me introduce myself, my name is Renato Vio and I'm an electrical enginer working at Siemens, Brasil in a calculus division for the power generation and distribution systems.   We are working in a project of rectifier substations (electric traction power supply) and we have to calculate the short-circuit in a direct current.  This calculus is divided in two parts, steady state and transient.  One important parameter to the calculus of the transient short circuit is the inductance of the catenary/rail way; so my question is:  How can I estimate the inductance of the catenary/rail way?  Is there a typical value for this parameter?


Thank you for your question.  I did a calculation like this about 18 years ago so it is not all fresh.  You basically have to look at the rate of rise of the current (L/R) which determines whether the breakers on the train or in the substation will cut the supply.  I have looked up some values in SACHS (the bible of electric traction) and have found the following information:


For 45kg rails you would have to assume 0.022 ohm/km for the return current.  Depending whether you have 3rd rail or overhead operation you would have to add the respective ohm value.  With a cross section of  240 square mm, for example, the resistance of the overhead line would be 0.1 ohm/km for single track lines.


Overhead, single track: 1.5mH/km
Overhead, double track: 1.3mH/km
Third Rail, double track: 3.2mH/km

Do not forget the substation inductance and, depending on the location of the fault, the line inductor/transformer primary on the train.

Do not hesitate to send students on our MSc course in Railway Systems Engineering which would equip you to solve questions like this, of course!

Axle Design

From Tong U Lipp, 6 Feb 2001:

Why are the axles of high speed trains hollow, in comparison to, say, Singapore MRT trains which have solid axles ?

Also, we are currently looking at automating the ultrasonic examination of our axles.  This examination is currently being manually carried out.  What is your view on this automation? Is it feasible?


Thank you for your e-mail. Axles of high speed trains are hollow for several reasons:

(1) the material in the centre of a shaft does not do much work (see your mechanical engineering textbooks)

(2) the unsprung mass of high speed rolling stock should be as low as possible to minimise impacts on the track - combine with (1) to reduce mass

(3) a hollow axle allows ultrasonic inspection from the inside without taking off the bearings.  This is already done on some administrations.

(4) automatic ultrasonic inspection is feasible and has been automated for NEW axles by Fraunhofergesellschaft in Stuttgart, Germany.

Structure Gauge

Dear Felix,

I have a question and would appreciate if you could answer it.  Kinematic Gauging is derived from the Structure Load Gauge.  How is this done?  I understand that you first define your structure load gauge.  Secondly, you work out your kinematic load gauge.  Finally, you add in track variations and you obtain your kinematic gauge.  But how is all this done?  Do we use computers or is it plain mathematics?


Dear Ajmer,

You should not always ask complicated questions <g>!   The exact detail escapes me too but here is my attempt at a description:  The structure gauge is defined in the standard/specification for a particular railway line and is a clearly defined two dimensional space, effectively a sort of imaginary tunnel running along the railway line.  No structure of any kind (platform, bridge, overhead line support etc) must project into this space.

The static envelope of the train is defined by the largest physical cross section of the train at the nominal suspension level with nominal wheel diameters.  This must include manufacturing tolerances.  The kinematic or dynamic envelope is based on the static envelop and the enlargement results from a combination of wheel wear, lateral motion, suspension movement and track irregularities (top and line).  Suspension movements (both vertical and lateral) are therefore limited by bump stops, sideways travel is limited by the flanges etc.

Once you have defined the dynamic envelope you have to take it along the track to check whether it infringes the structure gauge.  In doing this you have to look at the largest excursions of the body when the train goes around a curve, e.g. the kinematic envelope in the middle of the vehicle will moves sideways with respect to the centre line of the track. You can now derive the swept path of the train.   This must never touch the structure gauge.

Yes, a great deal is done by computer but the final check often has to be done by running a train along the track. You then often discover that you have not got it quite right.  Bring out the hammer and chisel!

That was easy, was it not?  I am not sure whether it is all totally accurate as regards the terminology but it is technically correct.


Harmonic Imbalance?

Good Day from Mahomed,

I am an engineer working in a distribution utility in South Africa.  While surfing your site, I came across some very valuable information regarding railways.  From a utility point of view my concern is the unbalance and harmonic injection from these railways.  Is there another site that explains this or maybe you could assist? I would like to know which has a greater impact on the unbalance and harmonics-AC/DC tractions.

Dear Mahomed,

DC traction generally uses three phase feeders to rectifier substations. All three phases are loaded equally unless there is a fault in the equipment. Harmonics can be filtered out although this makes the substation more expensive. The DC powered railways in SA are quite well behaved, I think. On DC powered railways the biggest problem are stray currents.

AC traction is single phase virtually everywhere in the world (the exceptions are three or four mountain railways). Sometimes there are dedicated railway (single phase supply) power stations, particularly in the case of 16.7Hz electrification.  Where mains frequency electrification is concerned (50Hz or 60Hz) the supply is usually provided to the railway by a utility in the form of a three phase HV link. There are virtually always problems with imbalances and harmonics.  The latter can be eliminated to some extent by including load factor correction in substations (capacitative). This is expensive.

Also by looking at a particular railway line is possible to identify whether it is AC or DC? - since they both use the overhead as a positive and one rail as a return.

DC electrification nowhere in the world exceeds 3000V (because of the risk of arcing).  AC supplies on 95% of railways are 11kV or more and the easiest way to differentiate between the two is by looking at the insulators: if there are more than two or three discs per insulator then it is likely to be AC.



Current Measurement in Rails

From Graham Hicks:

Do you know what equipment would typically be used to measure the running rail and conductor rail currents?  We have been asked to look at a spectrum from dc up to 150 kHz.  Whilst we have suitable receivers/analysers, we are unsure of the transducers normally employed to interface with the high currents, say 11,000 A.


Just a quick and off the cuff reply.  The question means current and there can only be two or three methods for measuring the conductor rail currents: low impedance shunts with direct measurement are only suitable for the RMS component.  Current transformers would be the normal solution but could have bandwidth problems, although should be OK to around 200kHz.  The third option is Hall effect probes.  In the latter two cases they would need data recording equipment.