RCD used for ADS on a TN system: what verifies L-N circuit impedance?

I think we may still be talking about two different issues.

I fully accept the requirements concerning maximum prospective fault current, breaking capacity and the suitability of the protective device. That is not the issue I am trying to explore.

My question is specifically about the minimum prospective L-N short-circuit current at the remote end of the circuit, because that is the value on which the designer's fault disconnection time and Chapter 43 conductor thermal-withstand calculation depend.

Where an RCD is deliberately relied upon for ADS on a TN system under Regulation 411.4.204, what part of initial verification confirms that the as-installed L-N loop impedance at the remote end remains sufficiently close to the designer's calculated value for the minimum L-N fault current and Chapter 43 calculation to remain valid?

Historically, where ADS depended upon the MCB and Table 41.3, the maximum Zs requirement provided a useful indirect safeguard. If the L-CPC loop, often the higher impedance loop because the CPC may be smaller than the neutral, had sufficiently low impedance to operate the MCB within the required time, there was considerable confidence that the L-N fault path would also provide sufficient fault current.

Once an RCD and Table 41.5 are used for ADS, that relationship largely disappears. A relatively high L-E loop impedance can still satisfy the ADS requirement because only sufficient residual current to operate the RCD is required.

But the RCD does nothing for an L-N fault. The validity of the Chapter 43 calculation therefore still depends upon the actual L-N impedance being low enough to produce the minimum fault current assumed in the design.

This leads to what I think is the key issue:

There does not appear to be an equivalent BS 7671 table giving a maximum permissible ZL-N in the way that Tables 41.3 and 41.5 give maximum Zs values for ADS.

The maximum permissible L-N impedance is therefore design-specific. It depends upon the protective-device characteristic, conductor size, k value and the resulting permissible fault-clearing time.

Consequently, unless the designer provides the installer/verifier with either:

• the calculated minimum L-N fault current, or
• the corresponding maximum permissible ZL-N at the end of the circuit,

what numerical acceptance criterion does the verifier actually have?

The verifier can demonstrate continuity and polarity, and may establish that the circuit appears reasonable. But "reasonable" is not the same thing as demonstrating that the as-installed circuit satisfies the assumptions within the designer's Chapter 43 calculation.

For example, if the designer calculated an end-of-circuit L-N loop impedance of 0.6 Ω but the completed installation actually had 1.5 Ω due to additional conductor resistance or a resistive connection, the conductor could still have continuity and correct polarity. The L-E Zs could also comfortably satisfy the much higher Table 41.5 value because the RCD provides ADS.

Yet the actual L-N minimum fault current would now be much less than the designer calculated, potentially moving the MCB into a very different part of its time/current characteristic. The original Chapter 43 thermal-withstand calculation could therefore no longer be valid.

That is the verification gap I am asking about.

So I am not asking how initial verification establishes maximum PFC or whether the protective device has sufficient breaking capacity.

I am asking:

How does initial verification demonstrate adequate minimum L-N fault current at the remote end of an RCD-protected TN circuit?

And, specifically:

If BS 7671/GN3 does not provide a tabulated maximum ZL-N, doesn't the designer need to state the design-specific maximum ZL-N or minimum L-N fault current on the design information so that the installer has something against which the completed circuit can actually be verified?

Otherwise, I struggle to see how the verifier can demonstrate that the as-installed Chapter 43 condition matches the design calculation, rather than simply demonstrating continuity, polarity and compliance with an RCD-derived maximum Zs that may be almost entirely unrelated to the required L-N fault current.

That is the distinction I am trying to establish.

I understand the calculation you are describing, including the need to consider the range of prospective fault currents, Cmin⁡C_{\min}Cmin​, Cmax⁡C_{\max}Cmax​, protective-device characteristics and, where applicable, the manufacturer's I2tI^2tI2t data.

I completely understand the design aspects. However, I still don't think that addresses the point I am trying to establish.

I am not questioning how the designer carries out the Chapter 43 calculation. My question is what happens after that calculation, when the circuit has actually been installed and has to be verified. And RCD max Zs may introduce a different approach to verification. 

The designer has necessarily made an assumption about the impedance of the L-N fault path at various points along the circuit. That impedance determines the prospective L-N fault current:

That prospective fault current then determines the operating behaviour of the OCPD and hence whether the conductor is adequately protected against the thermal effects of the fault.

Indeed, electrical design software such as ProDesign does precisely this. It does not simply perform a single adiabatic check at the origin. It calculates fault conditions along the circuit and checks the prospective phase fault current, protective-device disconnection time and conductor thermal withstand at the relevant points.

In simplified terms, the design process is therefore:

Whether the protective device is a fuse or circuit-breaker, and whether the calculation ultimately uses a time/current characteristic or manufacturer's energy let-through data, does not alter my fundamental question.

The calculation depends upon the prospective fault current, and that fault current depends upon the impedance of the installed L-N fault path.

There is also an important distinction here between Chapter 41 and Chapter 43.

The maximum Zs values and disconnection times in Chapter 41 are concerned with automatic disconnection of supply for protection against electric shock. An L-N fault between live conductors does not necessarily have to satisfy the Chapter 41 ADS disconnection time applicable to an earth fault.

It is therefore quite possible for a phase fault to have an OCPD operating time outside the familiar Chapter 41 disconnection times, potentially even beyond 5 seconds, provided that the applicable Chapter 43 requirements for protection against fault current and conductor thermal withstand are nevertheless satisfied.

That is precisely why the designer's calculated L-N fault current can matter independently of ADS.

For example, suppose the design calculation assumes at the remote end of a circuit:

which, simplistically at 230 V, gives:

The designer's software uses the resulting fault current and the appropriate protective-device data to establish that the conductors remain adequately protected.

Now suppose the completed circuit has an actual L-N impedance of:

The available L-N fault current is then only:

The OCPD may now be operating at a quite different point on its time/current characteristic. If that moves the device out of its instantaneous region, the reduction in fault current could result in a substantially increased operating time.

I am not saying that 153 A necessarily means the conductor fails the Chapter 43 requirement. That would require assessment using the particular protective-device and conductor data.

My question is more fundamental:

How does the person carrying out initial verification establish that the installed L-N loop is 1.5 Ω rather than the 0.5 Ω upon which the designer's calculation was based, and against what acceptance value is that measured or derived value assessed?

This becomes particularly significant where an RCD is deliberately relied upon for ADS on a TN system under Regulation 411.4.204.

If ADS is demonstrated using the RCD and Table 41.5, the permitted L-E Zs can be dramatically higher than the value that would have been required for the OCPD itself to provide ADS.

A satisfactory RCD-derived Zs therefore demonstrates the required condition for ADS for an earth fault, but it does not demonstrate that the L-N fault path has the impedance assumed in the Chapter 43 calculation.

The RCD cannot assist with an L-N fault because, absent another path, there is no residual imbalance to cause it to operate.

Historically, where the OCPD itself provided ADS and the circuit was assessed against the relevant Chapter 41 maximum Zs, there was also a useful indirect safeguard. If the L-CPC loop, potentially involving a CPC smaller than the neutral, had sufficiently low impedance to produce the required operation of the OCPD, there was considerable confidence in the integrity of the L-N path as well.

When an RCD-derived Table 41.5 value is used instead, that useful relationship largely disappears.

And this is where I see the practical verification issue.

As far as I can establish, there is no equivalent generic table providing a maximum ZL−NZ_{L-N}ZL−N​ for the installer/verifier in the same way that the Chapter 41 tables provide maximum Zs values for particular ADS conditions.

Nor could there necessarily be a simple universal value, because the acceptable L-N impedance is inherently dependent upon the particular:

• conductor size and material,
• protective device,
• protective-device characteristic,
• prospective fault current,
• required operating time, and
• thermal withstand of the conductors.

The designer's Chapter 43 calculation must therefore establish an acceptable operating envelope and, implicitly, a limiting L-N impedance or minimum L-N fault current.

So this brings me back to the original question.

Unless the designer communicates either:

• the maximum permissible ZL−NZ_{L-N}ZL−N​ at the relevant point, or
• the minimum permissible prospective L-N fault current,

what numerical acceptance criterion does the person carrying out initial verification actually have on site?

Continuity testing can demonstrate continuity.

Polarity testing can demonstrate correct polarity.

A Zs measurement can demonstrate the appropriate condition for ADS.

An RCD test can demonstrate the operating characteristics of the RCD.

But none of those, as far as I can see, necessarily compares the actual installed L-N fault path against the design-specific limit implicit in the designer's Chapter 43 calculation.

That is the particular link between design and initial verification that I am trying to identify.

Put another way:

Design demonstrates that the proposed circuit is safe. Initial verification should demonstrate that the circuit actually installed retains the electrical characteristics upon which that safe design depends. Where is the designer's limiting L-N impedance or minimum L-N fault current translated into an acceptance criterion for the person verifying the completed installation?

If the answer is that BS 7671 does not require a direct comparison against a stated maximum ZL−NZ_{L-N}ZL−N​, and instead compliance is inferred from continuity measurements, inspection, polarity and the other initial-verification procedures, then that may well be the answer.

But it would also appear to mean that, unless the designer voluntarily provides the installer with a limiting ZL−NZ_{L-N}ZL−N​ or minimum IfI_fIf​, the person carrying out initial verification has no design-specific numerical L-N impedance criterion against which to verify that particular assumption in the Chapter 43 calculation.

That is the point I am trying to establish.

I am not asking how to perform the Chapter 43 calculation. I am asking how the electrical parameter upon which that calculation depends is practically given and verified in the completed installation. As historically other than RCD the Max Zs value for protective devices automatically kept Ipfc within manageable limits.

I completely agree, Mike, and I think that gets to the heart of my question.

If we accept that the minimum prescribed tests do not necessarily verify the installed L-N impedance against the value assumed in the design, what additional test would you prescribe, and importantly, what maximum value would you assess it against?

With earth-fault loop impedance we have a clearly understood verification process and maximum Zs values against which the measured result can be assessed.

For L-N impedance, particularly where the RCD is relied upon for ADS and the OCPD-derived Zs limit is therefore no longer providing that indirect safeguard, I cannot find an equivalent clearly defined acceptance criterion in BS 7671 or the associated guidance.

Presumably the meaningful benchmark would have to come from the design, either as a maximum permissible ZL−NZ_{L-N}ZL−N​ or a minimum prospective L-N fault current.

That is really the point I have been trying to establish: if we decide that L-N impedance should be verified, what do we test, and what numerical value determines pass or fail?

I completely agree, Mike, and I think that gets to the heart of my question.

If we accept that the minimum prescribed tests do not necessarily verify the installed L-N impedance against the value assumed in the design, what additional test would you prescribe, and importantly, what maximum value would you assess it against?

With earth-fault loop impedance we have a clearly understood verification process and maximum Zs values against which the measured result can be assessed.

For L-N impedance, particularly where the RCD is relied upon for ADS and the OCPD-derived Zs limit is therefore no longer providing that indirect safeguard, I cannot find an equivalent clearly defined acceptance criterion in BS 7671 or the associated guidance.

Presumably the meaningful benchmark would have to come from the design, either as a maximum permissible ZL−NZ_{L-N}ZL−N​ or a minimum prospective L-N fault current.

That is really the point I have been trying to establish: if we decide that L-N impedance should be verified, what do we test, and what numerical value determines pass or fail?