This is an interesting article. The voltage gradients are quite low compared to the normal EMC fare, only a few volts per km (millivolts per metre) across the surface of the planet., but unlike most EMC/EMP problems that are transient surges or bursts of oscillation, while this sort of effect is so slow as to be effectively a near static DC offset across the earth surface, 

( For mental  calibration purposes. with EMC immunity tests, we look to illuminate the test object with  RF signals of whole volts per metre, sometimes tens to hundreds of them before things stop working. By contrast a few microvolts per meter are fairly sensible radio signal magnitudes to expect  at the limit of range of a radio telemetry link. )

The effect of this such that if you hammered in a pair of earth spikes a km apart and ran your meter leads between them you would see a few volts, probably a DC level changing slowly during the day. as the earth rotates. I am not sufficiently expert in railways to comment if the assumed thresholds or mechanisms for various fault types are credible or not.

 I can imagine problems with larger LV installations and  certain odd N-E situations and  RCDs however.

(The earth is moving both round the sun, and round it own axis, in a significant but more or  less static but non-uniform magnetic field, and acts rather like a 'wire' moving in the magnetic field within a generator - now if the current in the wire and in the planet followed the same spatial path perfectly, there would be no voltage, nor if the field was non-divergent, Neither or these  is true so you see the difference in the voltage seen in the surface wire, and the voltage seen in the average of all ground paths.).

Radio things are not affected by this sort of near DC effect, but some systems are well and truly nobbled by the change in density of the earths ionosphere that occur with solar storms, and  that does usually occur at the same time.
(From the perspective of the radio it is as if the top of the sky turned  silver and reflective  at frequencies where it used to be transparent, and at some frequencies where it used to be reflective, it turns pitch black. Those of us with ham radio licences rely on that shifting reflective layer to bounce signals round the planet on HF, but it only works at the right time of day and then with the right ionospheric conditions - and is very sensitive - in some ways HF comms is the canary in the coal mine, and measuring propagation by firing RF up and looking at the echoes (ionosounding) is one very sensitive way these events are monitored.(These results drive some predictions http://www.infotechcomms.co.uk/propcharts/december/ select a 24 hour time- UTC  and a frequency, and see where signals get to change by the hour...)

From a wiring perspective, probably insulated line pairs and transformer coupling is your friend, although there is precedent in the US and Canada for their grid transformer cores being saturated by the DC flowing in the ground-neutral path  - much like the blinding of an RCD by DC Then large currents flow, and fuses operate. The UK has  shorter transmission line distances and also does  not t distribute its HV neutral which reduces the severity of this sort of effect.

Mike.

This is an interesting article. The voltage gradients are quite low compared to the normal EMC fare, only a few volts per km (millivolts per metre) across the surface of the planet., but unlike most EMC/EMP problems that are transient surges or bursts of oscillation, while this sort of effect is so slow as to be effectively a near static DC offset across the earth surface, 

( For mental  calibration purposes. with EMC immunity tests, we look to illuminate the test object with  RF signals of whole volts per metre, sometimes tens to hundreds of them before things stop working. By contrast a few microvolts per meter are fairly sensible radio signal magnitudes to expect  at the limit of range of a radio telemetry link. )

The effect of this such that if you hammered in a pair of earth spikes a km apart and ran your meter leads between them you would see a few volts, probably a DC level changing slowly during the day. as the earth rotates. I am not sufficiently expert in railways to comment if the assumed thresholds or mechanisms for various fault types are credible or not.

 I can imagine problems with larger LV installations and  certain odd N-E situations and  RCDs however.

(The earth is moving both round the sun, and round it own axis, in a significant but more or  less static but non-uniform magnetic field, and acts rather like a 'wire' moving in the magnetic field within a generator - now if the current in the wire and in the planet followed the same spatial path perfectly, there would be no voltage, nor if the field was non-divergent, Neither or these  is true so you see the difference in the voltage seen in the surface wire, and the voltage seen in the average of all ground paths.).

Radio things are not affected by this sort of near DC effect, but some systems are well and truly nobbled by the change in density of the earths ionosphere that occur with solar storms, and  that does usually occur at the same time.
(From the perspective of the radio it is as if the top of the sky turned  silver and reflective  at frequencies where it used to be transparent, and at some frequencies where it used to be reflective, it turns pitch black. Those of us with ham radio licences rely on that shifting reflective layer to bounce signals round the planet on HF, but it only works at the right time of day and then with the right ionospheric conditions - and is very sensitive - in some ways HF comms is the canary in the coal mine, and measuring propagation by firing RF up and looking at the echoes (ionosounding) is one very sensitive way these events are monitored.(These results drive some predictions http://www.infotechcomms.co.uk/propcharts/december/ select a 24 hour time- UTC  and a frequency, and see where signals get to change by the hour...)

From a wiring perspective, probably insulated line pairs and transformer coupling is your friend, although there is precedent in the US and Canada for their grid transformer cores being saturated by the DC flowing in the ground-neutral path  - much like the blinding of an RCD by DC Then large currents flow, and fuses operate. The UK has  shorter transmission line distances and also does  not t distribute its HV neutral which reduces the severity of this sort of effect.

Mike.