Reply from: Nathaniel August 12 2019, 2:42 pm EDT
Thank you for the response. Please note, my comment was intended to focus on the safety aspects of different choices, not on how they fit with present NEC rules.
I know that the proposed type of system isn't used in the US, and is likely to be treated harshly. It seemed worth noting that the proposal might have some technical benefit by avoiding one class of dangers (dangerous potentials from the system being transmitted to a dock by the equipment grounding/bonding conductor), in spite of admittedly being able to introduce another class (ground fault together with GFCI failing to operate). I leave aside the important further questions of whether a state can make its own change to the Code, whether other laws limit what methods can be used, or whether there would be further dangers due to misunderstanding and wrong implementation of a changed rule. Having seen your article several years ago about a town that installed ground-rods for street-lights, but had no metallic path back to the source and no sensitive ground-fault detection, I do appreciate that keeping to one principle may seem safer than confusing people by having a choice of several principles that's found in the international standard I mentioned.
"There is no such thing as 'local grounding.'"
I realise it's not a method permitted by the present NEC! I was trying to invent a suitable US-style term for what I'd call 'independent earthing'. I mean connection of equipment grounding conductors in the installation (dock) to a ground electrode local to that installation, without metallic connection to anything outside. This is common for ground-fault protection in many countries, e.g. France, Italy, some rural parts of Germany and UK. It of course almost always needs more sensitive devices than plain overcurrent protection, to assure disconnection of ground faults. Unfamiliarity may make it seem worrying, but I've not managed to find statistics that suggest it to have a disadvantage compared to the systems that have a metallic path back to the grounded conductor of the supply source. Each has its own mixture of advantages and disadvantages.
"The system in the US requires an effective ground-fault current path from the point of a fault to the power supply. Removing the effective ground-fault path (equipment grounding conductor) would create a condition where the ground fault could not be removed."
I appreciate that this is the requirement in the present Code. It is still interesting whether it would be better to have a different choice, at least in special situations where people are very sensitive to potentials of just a few volts. The "could not be removed" applies only if the sensitive protection devices are not used or fail to operate (see next point).
"The USA does NOT depend on an electronic device (GFCI or RGD) to provide protection 'only' from a ground fault."
If this means that the Code currently doesn't permit this dependence, I of course agree, but it then comes down to the earlier question of which danger is greater in the docks (I do not have a confident answer - I just think it's not obvious, and would choose a TT system for myself!).
If it instead means that - even outside the Code - there is a law or feeling against such dependence on electronic devices, then I have plenty of sympathy with that feeling! The countries that make extensive use of TT systems tend to have simpler electromechanical devices for the ground-fault protection, rated typically at 30 mA or 100 mA. Several large studies over the past few decades have shown that the reliability of these devices is also not perfect. I tend to have two cascaded: first a slightly slower and less sensitive one, then a rapid sensitive one. But that's just me - hardly anyone else worries so much. It's not a nice thought what could happen to someone in contact with grounded equipment in an outdoors environment outside the bonded zone, if there's a ground fault and a failure to operate; but it's also not a nice thought what can happen in such an environment with the metallically connected system prescribed by the NEC, even without an equipment ground-fault, if there is a medium-voltage fault or an open supply-neutral, or just a few tens of volts of drop along a medium-voltage feeder's neutral.
An alternative proposal, which removes the dependence on sensitive protection devices, removes the connection to the supply neutral (EGC to dock), but I believe also violates the present NEC by not bonding equipment grounding conductors on both sides of a transformer, is the following. Its short summary in IEC terminology is "make the dock a separate TN-S system" for the dock, in contrast to the Virginia proposal's "make the dock a separate TT system". It is sometimes done for marina supplies in Europe, and for building sites in the UK where open-neutral concerns prohibit NEC-like bonding of EGCs to a supply neutral.
Take an isolating transformer with good insulation from primary to secondary ('double insulation'). Keep the secondary isolated from the primary. Ground one pole of the secondary to an independent electrode away from the house. Connect dock EGCs to the grounded pole. Ensure that the impedances permit overcurrent protection to operate adequately during ground faults. Add a GFCI on the secondary output for additional protection. Possibly include a GFCI on the primary supply just in case of trouble with the transformer insulation. At the location of the transformer, EGCs of at least one system (primary or secondary side) should be insulated, to avoid accessible voltage; the choice will depend on where the transformer is located.
| 
