Earthing and Bonding Design for 690V AC Railway Tunnels

Mansour Abdelwahed said:

On the telecom side, we’re following TIA-607-E, which actually allows telecom systems to be bonded to the main power earthing system — as long as it’s done properly.

ISO 30129 (EN 50310) along with IEC 61000-5-2 would be the European approach. These would be required for EMC conformity, including the ETSI approaches. There are some key differences between the IEC approaches (IEC 60364 series, IEC 61000 series, and EN 50310/ISO 30129) vs the TIA standards - choose only the approaches that are compatible with both systems.

There is also a potential issue with electrical safety if certain approaches to "grounding" in TIA are chosen without appropriate consideration of the safety approaches in IEC 61140 (and IEC 60364 series).

This could affect the safety of individual items of equipment to relevant EU/UK standards.

DC systems like London underground use voltages of order 630, with an offset earth. I'm not a railway guy either, but I have friends who are involved. My first observation is that any railway of any weight will consume  a lot of current at that sort of voltage. Have you made estimates for the likely traction current? One reason that  LV systems (sub kV) are usually live rail , and not overhead live wire, is voltage drop, and its also the reason the feeder transformers are less than a km apart on busy sections.

You say your tunnel if 5km long ! 

For example an 8-car train can use up to ~3600 MW of power (4200 amperes at your voltage more or less )  for motor currents  when accelerating hard. 
Now if we consider how much voltage drop we might like at peak load mid-span, and indeed how much uplift if you have some kind of  regenerative braking, you very quickly need a substation every KM or so, or hopelessly large conductors, or more sensibly a higher traction voltage.
(Again using London underground as an example, for a nominal 630V line, overshoot to 790 during braking was common, as were dips below 500V. In recent time, there has been a program of voltage uplift to 750V nominal and 890 during braking all new motors are qualified to run between  500 and 1000V Compared to normal voltage excursions seen in power distribution, where 10% droop is bad, the expected perturbations  are very large.)

I'm worried that before getting tangled with the finer details of the earthing there are wider issues of sensible voltages and currents and the no. and spacing of the feed points to decide.

In summary, 690V wont really go 5km, or even half of it, with any sensible conductor choices. You will need HV or at least MV , and suitable step-down  transformers, in the tunnel. These of course will each need their own earthing arrangements.

Mike.

Edit
PS
some facts and figures for track impedances in this Australian paper here https://railknowledgebank.com/TrackAsset
- R, Rail Resistance: ~ 0.03 ohm/km (each steel rail)
- L , Rail Inductance: ~ 2mH/km (1 mH = 1 x 10-3 Henry)
- RE, Rail to Earth Resistance: ~ 100 ohm.km
- RT, Track Resistance, rail to rail: ~ 200 ohm.km
- C , Track Capacitance, rail to rail, ~ 1,000 pF/km

I understand the figures in the UK at least are comparable.



PPS
As an old rule of thumb, maximum distances for economic power distribution balancing cable costs and losses against transformation costs  and losses set in at about 1 volt per metre. So while a 12V battery is fine on a car a few metres long, but a big lorry with trailer etc might be better with a 24V system.
In terms of AC 230V single phase will go a few hundred metres, and 400V and split phase more or less double this as the drop in the neutral conductor can be greatly reduced, but to double that distance again needs really out-size wiring for the current and it quickly becomes cheaper to drop in a second substation.  11kV networks tend to have a span of no more than 10s of km, so 33kV is used in more rural systems, while 400kV and higher is needed to reach across a small country.

Its not a perfect rule - one can of course throw copper (or aluminium)  at the problem to extend the reach slightly  further, and for small loads that is often done, and systems where we are only signalling but not really delivering power, like phones (50V, few km) and Ethernet (4Vp-p 100m, POE 50V/100m ), this is estimator method is way out  but this  does not look like a small load ;-) 

Hi Mansour,

Thanks for sharing all the details — I really learned a lot just by reading your post and the discussion that followed.

I'm not an expert in traction systems or tunnel design, and I'm still learning myself, but I’d like to share my thoughts based on my understanding and what makes sense to me from a safety perspective.

From what I know, having earth connections only at both ends of a 5 km tunnel sounds a bit risky — mainly because in case of an earth fault somewhere in the middle of the tunnel, the return path impedance might be high, and that could lead to a voltage rise along the PE conductors. That could create unsafe touch voltages, especially if someone is near metallic equipment or structures.

So, in my opinion, adding intermediate equipotential bonding bars or supplementary earth electrodes (even if they have a few ohms of resistance) could help reduce the risk — especially at access points or areas where people might be present. It may not reduce fault current a lot, but it can help equalise potentials and improve safety.

Also, I believe the standard EN 50122-1 is the best fit here, especially since it directly covers railway and tunnel earthing. I’ve read that BS 7671 doesn’t apply to traction systems, so it might only help for auxiliary systems like lighting or SCADA, as you mentioned.

Again, I’m still learning and may not have the full picture, but I hope this perspective adds something useful to the conversation.

Best regards,
Mudassar

Mudassar said:

From what I know, having earth connections only at both ends of a 5 km tunnel sounds a bit risky — mainly because in case of an earth fault somewhere in the middle of the tunnel, the return path impedance might be high, and that could lead to a voltage rise along the PE conductors. That could create unsafe touch voltages, especially if someone is near metallic equipment or structures.

Have a look at BS EN IEC 61936-1 and BS EN 50522 for earthing of the HV systems ... the combined system may well have to achieve much lower earthing resistances than you are currently considering to avoid problems simply bonding the two systems together.

Yes, EN 50122-1 for traction, but also consider there are other standards for signalling that may apply. BS 7671 (along with local national requirements) could be used for LV power for auxiliary supplies, power to SCADA, telecomms, security, drainage pumps and other utility supplies, etc.

There are particular EMC requirements for equipment in a certain distance of the rails, even if they are not part of the rail systems. 

Really appreciate your detailed response — and especially the references to BS EN IEC 61936-1 and BS EN 50522. I have to admit, those weren’t on my radar before, and they seem quite relevant, particularly given the complexity of bonding multiple systems in a tunnel environment. I’ll definitely be digging into those.

Your point about achieving lower earth resistance than just “less than 5 ohms” is well taken. In a 5 km tunnel, with critical systems and potential high fault currents, relying solely on endpoints for earthing might be technically compliant in some cases — but from a risk and safety point of view, I’m starting to think it might fall short, especially when considering touch and step voltages during fault conditions.

Also, I find your mention of EMC considerations quite thought-provoking. We often think about earthing and bonding primarily from a safety and fault-clearing perspective, but EMC often sits in the background — and in a rail tunnel where you've got traction power, sensitive SCADA, telecoms, and maybe signaling all coexisting, ignoring EMC could introduce operational headaches or worse. I suppose that’s where IEC 61000-5-2 and EN 50121 family might start to come into play?

And yes, BS 7671 is sometimes underestimated in these scenarios — while not intended for traction systems, its role in auxiliary system design is critical. Perhaps the challenge is more about integration — how we ensure clean separation where needed, but also safe and intentional bonding where required?

In short, this is proving to be a lot more layered than just sizing the PE and throwing in a couple of rods at each end!

Curious to hear how others have approached similar long tunnel systems — especially with mixed-use voltage systems and shared infrastructure. Has anyone come across specific mitigation strategies that worked well (or didn’t)? Maybe distributed earth mats, or even zoned equipotential sections?

Thanks again for your insight. These contributions are really helping shape my thinking.

Mudassar said:

Your point about achieving lower earth resistance than just “less than 5 ohms” is well taken. In a 5 km tunnel, with critical systems and potential high fault currents, relying solely on endpoints for earthing might be technically compliant in some cases — but from a risk and safety point of view, I’m starting to think it might fall short, especially when considering touch and step voltages during fault conditions.

Yes ... also perhaps worth considering traction return and signalling system needs though.

Is the traction power AC or DC, and what voltage level?

Mudassar said:

but EMC often sits in the background

Not on a railway - as you will have noticed, the BS EN 50121 series is quite extensive in its coverage, and covers more than just traction systems, but also lineside services and equipment, communications etc.

Mudassar said:

zoned equipotential sections?

Often used for large infrastructure, even inside large buildings (like airport terminals).

I would also throw BS EN 62305 into the mix for lightning protection, especially if the railway has overhead traction power, and/or any overhead cables. It will certainly cause issues between two ends, and end and centre, of the tunnel. Technicians have been killed working on telecomms cabling in tunnels during lightning storms!

From what I know, having earth connections only at both ends of a 5 km tunnel sounds a bit risky — mainly because in case of an earth fault somewhere in the middle of the tunnel, the return path impedance might be high, and that could lead to a voltage rise along the PE conductors.

Have a play plugging some numbers into Ohm's Law. Often, unless the PE conductor has a significantly higher impedance than the line conductor, the voltage difference along the PE doesn't change that much.

For (greatly simplified) example, for a fault close to the transformer - say L and PE are both 0.01Ω each, so 0.02Ω earth fault loop. With Uo of 690V you'd have a fault current of about 34.5kA - sticking that into V=IR, you end up with 345V dropped along the line conductor and 345V rise along the PE.

For a fault much further away, say we're looking a 1Ω each for L & PE - so 2Ω loop, the fault current would now be just 345A - plugging that in Ohm's Law for each of the 1Ω conductors gives (yet again) 345V across both L and PE conductors!

Of course if the L and PE are of differing c.s.a.s or differing materials, the balance can change - if the PE happens to have a lower impedance, the "problem" of voltage along the PE can actually get better with distance rather than worse.

   - Andy.

AJJewsbury said:

the voltage difference along the PE doesn't change that much.

Depends how long the tunnel actually is ... what's good for a short underpass doesn't hold true for the Channel Tunnel.

Also, what's good for DC and 50 Hz doesn't hold true for higher frequencies (for EMC), even over distances under 5 m at MHz.

Good Morning,

As for the lightning protection,

Thank you for your valuable input regarding lightning protection and equipotential zoning.

With regard to lightning protection, we would like to clarify that the train enters a tunnel that is entirely enclosed within the mountain, meaning the track and associated systems are fully shielded from direct atmospheric exposure once inside. As such, the risk of direct lightning strikes within the tunnel environment itself is minimal.

However, we fully agree that transient overvoltages due to lightning—especially induced surges from overhead lines, earthing differentials at tunnel portals, or nearby aboveground infrastructure—remain a concern. These risks become particularly relevant for longitudinal systems such as telecoms, signaling, and power cables that extend from outside environments into the tunnel, potentially introducing dangerous potential differences during storm conditions.

In this context, the application of BS EN 62305 and zoned equipotential bonding strategies remains highly relevant—not for direct strike protection inside the tunnel, but to mitigate internal transients, control potential rise, and protect sensitive systems and maintenance personnel.

We will ensure these factors are accounted for in the earthing design strategy, with special attention to:

• Equipotential bonding at tunnel portals and intermediate points,

• Surge protection devices (SPDs) at entry and termination points of communication and power systems,

• Differential bonding measures between traction and auxiliary systems.

Thanks Andy,

Exactly — that’s precisely what led me to consider the introduction of intermediate earthing rods along the tunnel to reduce the effective length of the protective earth (PE) conductor and mitigate potential differences.

The critical question now is whether we need to separate the earthing system of the traction power from the other auxiliary systems such as lighting, ventilation, SCADA, and emergency systems.