Have the rules for ring mains changed over the years?

[timhoward]

True, but unlikely to need a 32amp socket circuit for most kitchens, radial for sockets, radials, for appliances.
Less nuisance tripping easier to test and safer.

We seem to have a knack in this country of having limited space for consumer units, and the luxury of 3 RCBOs for a kitchen may not be a given.
Of late, for living rooms and bedrooms I’ve been doing radials, and kitchens usually an rfc.
I don’t mind testing rings - there is something nice about confirming the connections are all good on the points that will regularly see the most current draw.

 

[Dustydazzler]

Rings are great when first installed as a Ring with No spurs etc,
Trouble with rings it comes after years of DIY alterations , they become infected with multiple spurs , become spilt rings / spider rings , faults develop , you get broken legs / lost continuity.
Rings still have a function but I think sooner or later they will become obsolete and we will be using 20a radials for sockets in smaller bunches. Like kitchen radial , lounge radial , bedroom 1&2 radial , bedroom 3&4 radial , hall & landing radial. The days of an entire house being on 1 or maybe 2 rings will soon phase out

 

[pc1966]

Trouble with rings it comes after years of DIY alterations , they become infected with multiple spurs , become spilt rings / spider rings , faults develop , you get broken legs / lost continuity.

That is kind of a false argument, as DIY bodges can make the same issue out of a radial, for example missing CPC that no one notices, swapped L & N, etc.

If faults are in hidden junctions behind walls or under floors it matters not what the intended configuration is (RFC or radial), you still have a pain to find & fix it, or have to abandon that part and put in new cables to get around it.

 

[pc1966]

Less nuisance tripping easier to test and safer.

No, no no! The standard radial test is only easier because you are not doing as much fault coverage.

Do you check the N resistance matches L with a radial to the final socket, and verify that the CPC resistance is in ratio to the L/N? Or just do the R1+R2 look OK for the ADS aspect of the certificate?

You check that as part of the r1, r2, rN of the RFC testing and so have verified there are no high resistance joints along the way, most folks never think to check that with a radial.

Do you check on each socket on a radial that the R1+R2 is very close to the correct value for that (very probably) unknown length of cable?

Probably not, where as with the figure-of-eight test on the RFC you know that every socket's "R1+R2" check should be identical, and therefore any that are high are either a spur (which I agree are bad things for various reasons), or a socket with a dodgy switch or similar.

So while you might prefer "easier testing" on a radial as it saves you time and ticks the box for ADS requirement, you are not doing as comprehensive a test as the RFC permits, and expects, you to do which really does lead to safer installation (e.g. less fire risk from bad connection, etc)

 

[Dustydazzler]

I still believe a radial is a better more simplistic circuit with less potential for future issues.
As I say Rings served a purpose when cable was in very short supply and you need 1 circuit to supply all the socket outlets in the entire house.
In 2023 you can install 4 or 5 dedicated 20amp radials which offer better segrigation of circuits , reduces nuisance tripping and in event of a trip you only lose 3 or 4 sockets rather than the entire house etc

 

[7029 dave]

I still believe a radial is a better more simplistic circuit with less potential for future issues.
As I say Rings served a purpose when cable was in very short supply and you need 1 circuit to supply all the socket outlets in the entire house.
In 2023 you can install 4 or 5 dedicated 20amp radials which offer better segrigation of circuits , reduces nuisance tripping and in event of a trip you only lose 3 or 4 sockets rather than the entire house etc

You do raise some valid points, but just imagine if you radialed a big house the amount of ways you would need would mean multiple CU,s just to house the circuits, something we came away from years ago.

 

[Pretty Mouth]

True, but unlikely to need a 32amp socket circuit for most kitchens, radial for sockets, radials, for appliances.
Less nuisance tripping easier to test and safer.

That's quite a lot of socket circuits for the average kitchen. One for general purpose sockets, one for the dishwasher, one for the washing machine, one for the tumble drier. And there's nothing wrong with that, if that's how you like to install. But for me, I'd just install a ring final.

Actual nuisance tripping, where there is no fault present whatsoever, in my experience is very rare in domestic. It's true that more points will be lost if they are all on one circuit, but I think one circuit for the kitchen sockets and appliances is fine.

Safer? How so?

Not true, they cannot detect a series fault in a ring final circuit

Can you offer a reliable source for this please? I don't believe it to be true.

Saying that, the arc is unlikely to be at a dangerous level, doesn't instil confidence.

We can show this by calculation.

First a radial:

Let's put a 32A load on it, and disconnect the circuit N conductor from the N bar in the CU. On energising the circuit, the potential difference between the disconnected N conductor and the N bar will be the full 230V. There is a potential for an arc to form, given just the right gap between them.

Next a ring, about 100m in length:

Absolute worst case scenario: let's put a 32A load a very short distance along one leg, and disconnect the N conductor of that leg from the N bar, leaving the N of the other leg in place. On energising the circuit, the current will flow the long way through the N conductor back to the N bar. The voltage drop across the 100m of N conductor would be:

V=IR
V= 32 X 0.89
V= 28.5V
(R based on 100m of 2.5mm conductor at 70deg. In reality, this would likely be running hotter as the conductor is overloaded, so the voltage drop would be a little more)

Now we have a potential difference between the disconnected N conductor and the N bar of just 29V. The potential for an arc to form between them is there, but it is much, much lower than for the radial based on the much lower V between them.

 

[pc1966]

In 2023 you can install 4 or 5 dedicated 20amp radials which offer better segregation of circuits , reduces nuisance tripping and in event of a trip you only lose 3 or 4 sockets rather than the entire house etc

At one time the argument for the RFC was saving in copper cost, but going forward it might be savings in AFDD costs!

I fully agree that if you only need a handful of sockets in one place then a radial is the best choice, but generally for a flat or floor of a house where you naturally loop round anyway the RFC makes a lot more sense, and what you lose in terms of future DIY bodgability you gain in terms of redundant-path CPC safety.

EDIT: Just to add "infected with multiple spurs" is my phrase of the day! I might prefer infested instead, but the same sentiment.

 

[nicebutdim]

I still believe a radial is a better more simplistic circuit with less potential for future issues.

It's a simplistic circuit in its basic form but, as already pointed out, there's no requirement that radials must be installed in that manner and nor is there any requirement for future additions to follow it.

A good part of the argument against ring circuits seems to be based on the assumption that radials won't be butchered in the same manner rings often are. Such an assumption is, at best, optimistic. I'd wager that as time passes, and radial socket circuits become more commonly found in older houses, electricians will have similar arguments about them many decades hence.

In 2023 you can install 4 or 5 dedicated 20amp radials which offer better segrigation of circuits , reduces nuisance tripping and in event of a trip you only lose 3 or 4 sockets rather than the entire house etc

Imagine, if you will, that we're in 2073 and those circuits serve multiple additional points, including garage, outdoor socket, outdoor lighting etc. and all those additions have been added from the nearest available point. There may even be a ring or two lollipopped into the equation. Poor workmanship is poor workmanship, whether it's carried out by professional or amateur, and there'll always be people bodging additions to any circuit, when a few pounds are waved under their nose. I fail to understand how not installing ring circuits will make any difference to those following in their wake.

 

[pc1966]

The only difference in radial versus RFC for future butchering is the risk of a spur infestation allowing overload of some 2.5mm due to a 32A OCPD.

That of course could be done by DIY mods of a 32A radial, if using 2.5mm to extend instead of 4mm.

Also the risk is not a massive one, as you would need to have a few heaters or similar on the spurs. True, quite possible in a kitchen (due to heater loads in dishwasher, washing machine, etc) but for most homes (and especially in the light of current electricity costs) the chance of several heaters on max on the spurs is fairly low.

Really this is a point for fixing a butchered system: You either try to rid it if the spur infestation to get a proper RFC back, or you establish they are connected OK and drop the OCPD to 20A. You might argue for splitting the RFC to two 20A breakers, but unless there is an easy point about halfware I doubt it is worth it. Many flats, etc, have been OK with just 20A for sockets for years.

Basically back to the issue of how easy is it to exceed 20A on sockets, allowing for diversity, in any sensibly planned flat/floor of a house.

 

[mainline]

No, no no! The standard radial test is only easier because you are not doing as much fault coverage.

Like I said, it's easier

 

[mainline]

That's quite a lot of socket circuits for the average kitchen. One for general purpose sockets, one for the dishwasher, one for the washing machine, one for the tumble drier. And there's nothing wrong with that, if that's how you like to install. But for me, I'd just install a ring final.

Actual nuisance tripping, where there is no fault present whatsoever, in my experience is very rare in domestic. It's true that more points will be lost if they are all on one circuit, but I think one circuit for the kitchen sockets and appliances is fine.

Safer? How so?


Can you offer a reliable source for this please? I don't believe it to be true.

Hager's take on it:
Contrary to common belief, AFDD’s do offer protection against arc faults in ring final circuits and to the equipment being fed from this circuit. A series arc fault in one leg however, is unlikely to be at a dangerous level so will not be detected.

We can show this by calculation.

First a radial:

Let's put a 32A load on it, and disconnect the circuit N conductor from the N bar in the CU. On energising the circuit, the potential difference between the disconnected N conductor and the N bar will be the full 230V. There is a potential for an arc to form, given just the right gap between them.

Next a ring, about 100m in length:

Absolute worst case scenario: let's put a 32A load a very short distance along one leg, and disconnect the N conductor of that leg from the N bar, leaving the N of the other leg in place. On energising the circuit, the current will flow the long way through the N conductor back to the N bar. The voltage drop across the 100m of N conductor would be:

V=IR
V= 32 X 0.89
V= 28.5V
(R based on 100m of 2.5mm conductor at 70deg. In reality, this would likely be running hotter as the conductor is overloaded, so the voltage drop would be a little more)

Now we have a potential difference between the disconnected N conductor and the N bar of just 29V. The potential for an arc to form between them is there, but it is much, much lower than for the radial based on the much lower V between them.

It could happen, though, therefore Afdds may not detect an arc fault on a RFC.

 

[mainline]

Found a bit of a write up.


Have we got it right or is this yet another UK outdated insular custom? In this paper I intend to show you the disadvantages of using ring circuits. Opinions I have formed as a result of problems experienced during many years of inspecting and testing electrical installations and training people to do it. David has pointed out that the original thinking behind the development of the 13 amp plug and socket system was for domestic premises – economy homes. Having read David’s paper, it seems to me that the introduction of ring circuits was almost an afterthought and that the original intention was for a socket to be used on a radial circuit. In my view it should have stopped at domestic premises. Other premises were only mentioned once in the history paper. However, over the years people have been brainwashed into believing that 13A sockets mean ring circuits.I have just recently even found a ring circuit supplying a single socket for a heating boiler. Hardly dangerous but demonstrates a complete lack of understanding by the installer, who incidentally was Part P registered.
Usage. Ring circuits are used almost everywhere in this country, and some others:Schools - laboratories and workshops Offices both large and small Hospitals – wards and surgical/treatment areas Retail premises, although some will not have ring circuits because of additional dangers and costs. Public buildingsAs well as Domestic At this point I would like to make it clear that I am not against ring circuits, there is aplace for them in modern installations provided they are properly designed inaccordance with BS 7671, carefully installed and tested as detailed in IEE Guidancenotes 3 or the On-Site Guide. If all three were properly applied some of the disadvantages would disappear.

Main Issues.
Safety is the main issue and safety being important becomes one of the main disadvantages. Ring circuits are misused and abused. They are installed without proper consideration as to their purpose and loading, additional points are frequently added as spurs without considering the existing layout of the circuit.They are used for heating circuits and IT circuits again, without considering the load or the need for secure protective conductor connections or reinforced cpcs.The ring circuits cost more to install than two radial circuits. Regulations – 433-02-04BS 7671. There are only four regulations that state requirements for ring circuits.The critical regulation is 433-02-04 which is probably largely ignored because it isoften impractical to apply. This regulation requires the load to be distributed aroundthe circuit so that the current in any part of the ring does not exceed the installedrating of the cable. This means that in a circuit intended to supply a washingmachine, tumble dryer and a dishwasher the points need to be wired so that the loadcurrent in both legs of the ring is shared as equally as possible. More often than not if you look around a kitchen you will find the washingmachine, dryer and dishwasher grouped around the sink, for obvious reasons. And if the sink does not happen to be more or less in the centre of the ring, one leg will carry more current than the other.

Disadvantage:
Not easy to achieve. Regulation 543-02-09Regulation 543-02-09. This regulation requires the protective conductor of a ring circuit to be wired in the form of a ring, unless it is formed by metal covering or ametal enclosure. Most people ignore the metal covering part and run separate cpcs for each circuit. Hence metal trunkings become half filled with green and yellow cables that are unlikely to ever see an amp in their whole existence. Disadvantage: Waste of cable and labour.Safety Many rings are wired incorrectly particularly by DIY persons. Sometimes however electricians can get it wrong. I have had electricians say to me “I can’t believe I did that” Even competent people make mistakes at times. Another disadvantage.A lack of understanding of the system is another problem. Unless a ring circuit is wired correctly with spurs restricted to 1 double point per spur,there is an increased fire risk due to overheating of cables and connections.If there are breaks in the conductors or loose connections in terminals there are both fire and shock risks.Testing The safety of a ring circuit relies on proper testing. It is a vital part of the installation process. If the correct testing method is not fully applied defects with the circuit are unlikely to be identified and corrected. This applies to both initial testing as well as periodic inspection and testing.Testing is however a time consuming and expensive operation, hence it is very often not done fully as prescribed in GN 3.History IEE Wiring Regulations - 13th Edition 1955 Regulation 505A test shall be made to verify the continuity of all conductors of every ring circuit installed in accordance with Regulation 114 (b)IEE Wiring Regulations - 14th Edition 1966 Regulation D 10A test shall be made to verify the continuity of all conductors (including the earth-continuity conductor) of every ring circuit.No test methods givenIEE Wiring Regulations – 15th Edition 1981 Regulation 613-2A test shall be made to verify the continuity of all conductors (including the protectiveconductor) of every ring final circuit. See Appendix 15.Appendix 15 showed a very detailed test method.WHY?There must have been a reason to introduce a specific test method in Appendix 15.Could it have been that the industry and consumers were having safety problems? It is clear to me that in the 25 – 30 years following the introduction of the ring circuit there must have been safety problems that were referred to the IEE for resolution andbecame the driver for the test we have today. I am sure that it could not have been simply 'a good idea at the time’ IEE Wiring Regulations – 16th Edition 1991 Regulation 713-03A test shall be made to verify the continuity of all conductors (including the protectiveconductor) of every ring final circuit.The test method was transferred from Appendix 15 to Guidance Notes 3.Test Methods The method introduced into the 15th edition called for a resistance measurement to be made at every outlet point first between phase and neutral and then between phase and cpc, with the conductors joined together at the distribution board. The text said that the resistance at the centre point of the ring would be equal to the sum of the phaseloop resistance and the neutral or cpc loop resistance, divided by four.This sent everybody running around like headless chickens looking for the mid-points of ring circuits. Consultants were marking the mid-points on drawings or instructing contractors to label the socket at the mid-point, or marking the mid-point on the ‘as installed’ drawings.What a lot of nonsense.If the text had said that the highest value of resistance measured between phase and neutral, or cpc, with the conductors joined at the distribution board, should be a quarter of the sum of the conductor resistances added together, and all other points would be of lesser value, it would have saved the industry a great deal of unnecessary work time and cost. Happily this method was changed for the 16th edition.The 16th Edition Method.The recommended and only proven method of testing involves breaking the ring,separating the conductors at either the distribution board or at a point, doing the tests and re-assembling the circuit after completing the tests. How can one be sure that the ring is complete after reassembly? Still a funny way of doing things! Is this a disadvantage? In many instances, probably most, ring circuits are not properly tested.Most people testing will test the ring ‘end to end’, many cannot be bothered to do the‘interconnected conductors’ test. Electricians freely admit this because testing eachpoint twice takes too long.None of the test results schedules that I have seen provide for the ‘interconnectedconductors’ test value to be recorded. This is an important record that demonstrates:a) the test has been done and b) the circuit is correctly wired.If suitable provision were made in the schedules of tests results there is a chance that the testing would be done properly.

Typical Faults Found

The most dangerous fault:
Cross connections between two ring circuits or a ring and a radial so that the over-current and fault current protection is compromised becoming as much as 60 or 64 amps, disconnection times are completely blown and circuit isolation relies on 2 devices rather than a single device. Interconnections occur usually in distribution boards but can easily occur when ring circuits cables are installed in trunkings.
In one hospital, sockets mounted in dado trunking were intended to be connected alternatively to essential and non-essential supplies distribution boards. The circuit cables were inter-connected between the two boards. Would have had an interesting result if the circuits had been connected to different phases.
Other Faults
Incomplete ring on one or all circuit conductors – broken loops Part of a ring missing, a link cable having been left out, resulting in two 2.5mm2
cables being protected by a single 32 A protective device
Loose Connections due to conductors crammed into back boxes that are too small, especially for spurs, one cable not secured and overcrowded distribution boards.
Too many spurs on a ring, and spurs on spurs - risk of over-heating
Spur cables too long. A ‘ring’ wired as a figure of eight, risk of overloading 1 leg of the ring
Break or loose connections in the live conductors, 3 conductors in one terminal, one loose. Overheating likely to cause a hot spot at a termination that may eventually burn out or cause a fire.
Break or bad connection in the cpc due to loose screws or over zealous tightening, thus increasing Zs of the circuit so that the limiting value is
exceeded and the 0.4 second disconnection time is not achieved.
Incorrect polarity. All these could cause danger and are therefore serious disadvantages. They would be eliminated by applying the correct testing methods.
Testing ring circuits can take 5 or 6 times longer than testing radial circuits, and if any of the above defects are present fault finding can take a considerable time and become very expensive. Fault finding on radial circuits is relatively simple and quick.
Big disadvantage to the installer. Who pays in the long run?
Disadvantages galore, can’t happen with radials Installation
Consider the disadvantages with circuit wiring:
A 32A ring circuit serving 100m2 uses more cable and therefore takes longer to install
than 1 x 32 A radial circuits serving 100m2
A 32A ring circuit serving 100m2 uses more cable and therefore takes longer to install
than 2 x 20A radial circuits each serving 50m2 the latter having a higher loading
capacity of 40A.
Ring circuits wired with 3 single core 2.5 mm2 cables drawn into a straight run of
conduit or trunking take much longer to install than radial circuits wired with 3 single
core 4.0 mm2 cables.
Each of these situations use less of the worlds resources of copper.
To my mind, in offices, workshops, classrooms and laboratories the only justification
for installing a ring circuit is where a single circuit is run completely around the room.
If it is necessary to install all 6 conductors in a single run of conduit or trunking then 2
radial circuits are much more practical and cost effective.
I have discussed this with many engineers who all agree with this philosophy. I know
that some engineers will not consider using ring circuits in commercial installations.
Additional points. Domestic and commercial consumers have a multitude of low-current
appliances. New installations need many sockets and flexibility is needed to
allow furniture to be moved around and for future alterations and additions.
Extending or breaking into a ring circuit is not a straight forward exercise.
Many domestic ring circuits have been modified incorrectly by DIY persons
and are no longer a ring and are probably unsafe.
More often than not, particularly in domestic premises, additional points are installed
as spurs from the ring or spurs from spurs, with total disregard for the existing load
and usage. This can, depending on the load, change the balance of the circuit.
I am sure that nobody ever tests the ring continuity and layout prior to installing an
additional point. I am equally sure that very few people install an additional point by
diverting the ring cables to include it in the ring. Furthermore I am certain that very
few people, especially DIY, ever apply the ring test after installing the additional
point.
Unless thorough testing is carried out on a new or particularly a modified ring
circuit, wiring faults may go undetected and invalidate the basic safety
principles of the system.
Another potential danger and disadvantage.
Training
It has been said many times that if electricians are trained properly the problems
would not exist. I do not disagree with that. An apprentice who is brought up with the
system should understand the correct installation methods, however testing is a
different issue. I have found that some electricians, who may be exceedingly good
tradesmen, have great difficulty in grasping the test method and the benefits of doing
the test, and are likely to give up. Others swallow it whole and become very
competent testers.
A big disadvantage for some.
Other Options: Radial and Tree Circuits
There are good reasons for considering the use of other types of circuits
IEE Guidance Notes show radial circuits in the conventional circuit arrangements.
• 32 A ring – 7 kW – 100 m2
• 32 A radial – 7 kW – 100 m2
• 20 A radial – 4.5 kW – 50 m2
In my view
• 2 x 20A radials better than 1 x 32A ring
A 20 A circuit to serve 50 m2 floor area and a 32 amp circuit, 100 m2. These are
based on the maximum anticipated load in these areas not exceeding 5 kW or 7 kW
respectively.
The limiting factor in such areas is the cable length - voltage drop and the earth loop
impedance of the circuit. Voltage drop is unlikely to be a problem neither will earth
loop impedance because in the near future all such circuits will require RCD
protection. The limiting factor need only be the maximum anticipated load that would
be used in the area. It is now recommended that kitchens are treated as a separate entity and have at
least one ring circuit. 2 x 20 A radial circuits in a kitchen will use less cable than a ring circuit and
provide greater capacity as long as care is taken to ensure that fixed loads such
as washing machines, driers etc are not all on one circuit.
Tree Circuits
A tree circuit is simply a radial circuit with branches. A 20 A tree circuit wired with
2.5 mm2 cables would be far more versatile than a straight radial circuit and probably
far more practical. Points could be placed economically wherever they may be used,
the limitation would still be the maximum load likely to be used in the area, not the
number of sockets.
Controls Ring circuits do not readily facilitate separate control of groups of socket
outlets. Radial and Tree circuits do.
This added bonus gives an opportunity to control sections of the circuit separately
with switches and timers.
In Commercial buildings by Building management systems
In domestic buildings – Smart Homes – Home Bus Systems, automatic and telephone
control.
Applications
Typically a standard 3 bedroom domestic property could be adequately served by
2 x 20 A 2.5 mm2 radial or tree circuits, and
1 x 32 A 4.0 mm2 radial or tree circuit in the kitchen.

 

[mainline]

 

[pc1966]

Found a bit of a write up.

That is mostly rubbish I'm afraid.

On-Site Guide. If all three were properly applied some of the disadvantages would disappear.

That applies to EVERY DAMN CIRCUIT.

In one hospital, sockets mounted in dado trunking were intended to be connected alternatively to essential and non-essential supplies distribution boards. The circuit cables were inter-connected between the two boards. Would have had an interesting result if the circuits had been connected to different phases.

Again that is simple incompetence and shows a complete lack of any form of proper testing.

Break or bad connection in the cpc due to loose screws or over zealous tightening, thus increasing Zs of the circuit so that the limiting value is
exceeded and the 0.4 second disconnection time is not achieved.
Incorrect polarity. All these could cause danger and are therefore serious disadvantages. They would be eliminated by applying the correct testing methods.

Exactly the same for a radial. In fact on the CPC front worse.

Testing ring circuits can take 5 or 6 times longer than testing radial circuits, and if any of the above defects are present

Only because you are testing more than R1+R2 for a radial.

Much of our labour comes from agencies and you get what you are sent. In London
you hardly ever hear English spoken on construction sites. Electricians trained in EU

Is that you Farage?

countries other than Ireland will not have heard of ring circuits. They may be very
good competent tradesmen in their own countries but never-the-less are not competent
to install socket circuits in this country.

Again, this is people who are not competent to do the job. If you do not know and understand UK wiring regs and practice you have absolutely no job doing it. Same for UK tradespeople working in USA not knowing the NEC, or in EU, etc.

Europeans do not understand ring circuits. This also applies to Australians, New
Zealanders and South Africans many of whom come to this country to make a

Of course not because they don't use them. Why, because they do not have fused plugs. That is fundamental to the use of 32A supply for end appliances in the UK. It is why we can have a socket off a 40-50A cooker supply, etc.

The limiting factor in such areas is the cable length - voltage drop and the earth loop
impedance of the circuit. Voltage drop is unlikely to be a problem neither will earth
loop impedance because in the near future all such circuits will require RCD

That is written as if VD will be solved by the use of RCDs, either incompetence on the behalf of the writer, or weasel words at best. just compare cable length limits of RFC with radial of same cable.

Controls Ring circuits do not readily facilitate separate control of groups of socket
outlets. Radial and Tree circuits do.

How often do you need to control a group of sockets?

 

[westward10]

Did you actually read read it.

 

[littlespark]

Did you actually read read it.

I couldn’t be arsed arsed

 

[7029 dave]

I couldn’t be arsed arsed

Same same

 

[nicebutdim]

I couldn’t be arsed arsed

I tried, and God knows I like information, but found it tough going.

From what I did read, the main issues were a result of poor workmanship or circuit design.

 

[Pretty Mouth]

It could happen, though, therefore Afdds may not detect an arc fault on a RFC.

A low-level serial arc fault (below the threshold needed to operate the AFDD) can occur in a radial circuit too. So 'AFDDs may not detect an arc fault in a radial circuit' is also true.

As I understand it:
AFDDs should operate where there is arcing at dangerous levels, whether the circuit is a ring or radial. They shouldn't operate where there is an absence of arcing at dangerous levels. The nature of ring finals means that arcing at dangerous levels is less likely. This doesn't mean that AFDDs are any less effective when installed on a ring final.

(whether or not AFDDs work as intended is another discussion)