I just discovered that footnote 13 in the article incorrectly states that the AFCI that I tested was powering 40 “Watt” bulbs – not so. They were small nightlights drawing about 40 mA, not 40 Watts.

In this situation would it help to add a second neutral and make a parallel splice at the circuit breaker.

Mike Holt and Randal Andress: Thanks for the great info on GFCIs and multi-wire branch circuits. I update online programs for NAED based upon the NEC and your info was helpful in understanding NEC2017 422.5.

A local Real estate agent had a Home Inspected, and found some Plugs in the home were not grounded, the home inspector said just put a GFCI Plug in and that would take care of the problem. My theory would be to Bond the Neutral to the ground on the plug. What do you think?

What I do not understand I why we drag technology of the 19th century into the 21st century and then try to use modern technology caused by outdated technology we should not be using in the first place. The first example is the neutral. The transformer and generator have a neutral. There is no need to wire that neutral anywhere but to an earth ground. We do not need a stinking neutral wired to our house, marina or business. Europe uses 240 volt or sometimes 220 volt domestic wiring. This may rattle some people because this voltage is a little dangerous for a home handy man to mess with. Fine, modern homes are already switching to low voltage lighting using an isolated transformer. Just continue that trend. In the past, and I mean past, like almost 50 years ago I worked with high cycle hand tools ( Electrostoom in Rotteram) and even on a 200 HP high cycle motor (American Standard R-12 refrigerant compressor). Why are we not using 300 Hz systems? They are highly efficient and reduce material costs. Do not give me the inductive reactance excuse. When I was a kid we used to say a one farad capacitor would be the size of the Empire States Building and cost hundreds of millions. Today you can buy one for under ten bucks and carry it away in your pocket. We can easily cut impedance to resistive only. There is Romex like cable manufactured though not widely distributed that has a conductor to carry 24 control voltage. Let us put aside the class2 NEC issue. Another can of worms. Using control voltage and solid state power diodes and transistors we can make all outlets controllable. No power to them without the 24 volt signal.

Great study, can you provide a basic electrical math example on how the GFCI won't trip on a multi-wire branch circuit application.

I x E = GF (ground faults) +/- 5ma = L1 I x E = GF (ground faults) +/- 5ma = L2 I x E = GF (ground faults) +/- 5ma = N

Thanks elaborate if possible ...... ?

My first thought in reading this report is Hot tubs. These installations have used 2 pole GFCI protection with a shared neutral installations for many years. Are we now saying the protection offered is minimal??

Wow! Great article; and comments by everyone. Really very enlighting. Thanks for sharing!

MWBC the breaker are suppose to trip when there is a fault in either line.

Thanks for the article and link to the full article Mike. Randal, Appreciate your efforts and work in the outstanding presentation. I'm studying the figures and rereading the article for the third time.

What does this have to do with a marina? Wouldn't a MWBC do the same on land?

Mike, Thank you for presenting this very important electrical engineering "discovery" about Multi-Wire Branch circuits and the GFCI circuit breakers in current use. *** One possible solution may be to have a double-pole standard double-pole circuit breaker feeding an external set of Two GFCI circuit breakers, which feed the Multi-Wire Branch Circuit. The set of Two GFCI circuit breakers might be a pair of GFCI Dead-Front devices, located adjacent to the primary load center. *** In a Marina situation, this is a critical problem, unless all Marina MWBC are outlawed and replaced. *** I will be watching for comments on this topic.

Reading the article I can't say the results are surprising; the simple math bears out what must happen in a MWBC given the scenarios presented.

What I find troubling is that MWBC's continue to be installed with frequency. Neither the voltage drop or the marginal increase in wire cost justify the litany of potential and even likely long-term problems with a MWBC. This GFCI issue is just one more reason to abandon the MWBC practice - I find the MWBC as the tube and k-nob of our generation...

Why would one protect a MWBC with a 2-pole GFCI? It seems the proper method would be to protect each 'branch', where required, with GFCI receptacles.

Thanks for the great article, and very informative.

I have always viewed GFCI as a last ditch effort towards protection when proper grounding cannot be assured such as allowing users to plug equipment of unknown quality into a receptacle. Many people seem to feel that GFCI is the "best" protection available. My feeling has always been that GFCI is better than nothing but certainly not a substitute for proper grounding. This research seems to agree with my feeling.

I realize that there are most applications where the GFCI functions reliably but in many applications it should not be used. Understand that the GFCI reads current indirectly, by induction as does a clamp on amp meter. It is calibrated for a certain frequency, that is to say you should make sure,if you work internationally, that you do not install a 50 Hz GFCI in a 60 HZ application. Also, I do not believe it is possible to convert current detected by induction in a CT or the related differential transformer to true RMS current. For example, the GFCI can not detect true RMS current if there are harmonics, as there often are in the neutral. Large breakers 600 amp plus use a ground fault relay which can detect true RMS current and detect that current directly. I would test for ground faults with a good RMS voltmeter. The ground test meters are great but you are going to lay out 600 bucks for a Fluke and it is over kill. To do this you are going to responsibly break the ground and take a direct voltage drop or shunt current reading. But if you do this often and have the bucks the ground test specialized meter is safer and has more options. As for problems with shared neutrals we can write a book. Also, be it understood there ground loops and ground neutral loops. They are not the same. But that is another book.