I would be inclined to use an ambient temperature of 30 degrees for an outdoor  installation. That in my view represents the highest temperature that is regularly achieved. I do not feel that the presumed ambient temperature has to reflect the highest temperature ever achieved, but the highest likely/regularly achieved temperature.

For a large or complex installation, it would in some cases be reasonable to apply engineering judgement and consider not just the load in amps, but also the nature of the load. If the load is primarily street lighting or decorative lighting, then a lower ambient could reasonably be presumed on the grounds that operation during the heat of the day in very unlikely.

On the other hand, if the load is primarily air conditioning or refrigeration, then I might be more cautious and allow for a 40 degree ambient on the grounds that the load is likely to be greatest during an exceptional heatwave.

For a large or long hour load. consider the cost of losses in a marginally sized cable, this may suggest significant upsizing.

For a domestic outbuilding, loads might include space heating and a clothes dryer, significant use of which is improbable in a heatwave.

Finally, it might not matter much ! on a small and simple job, going up cable size from say 4mm to 6mm is affordable so as to remove any doubt or concern. On a large job, it is likely that cables will sized according to voltage drop rather than current rating.

Thanks for the considered response. That's the approach I would usually take, but the Regs seem to be unambiguous:

Where the ambient temperature in the intended location of the non-sheathed or sheathed cables differs from the reference ambient temperature, the appropriate rating factors given in Tables 4B1 and 4B2 are to be applied to the values of current-carrying capacity set out in Tables 4D1A to 4J4A. 

Where the ambient temperature in the intended location of the non-sheathed or sheathed cables differs from the reference ambient temperature, the appropriate rating factors given in Tables 4B1 and 4B2 are to be applied to the values of current-carrying capacity set out in Tables 4D1A to 4J4A. 

That though is from Appendix 4 - which is only Informative rather than normative - i.e. it's just giving advise on some possible ways of achieving compliance with the main parts of BS 7671 - it shouldn't be taken as a definitive part of BS 7671 (as it's isn't). Other methods are often used - e.g. using manufacturer's data for cable rating or MCB characteristics, which can differ from the examples given in BS 7671 appendices - and it's quite legitimate to do so.

   - Andy.

Thanks Andy. So, even though the maximum recorded ambient may be, say, 38ºC in a particular area, it's ok to use 30ºC as an "average" so to speak (for general outdoor installations in the UK).

Also in section 5 (emphasis mine):

522.1.1 Ambient temperature (AA) A wiring system shall be selected and erected so as to be suitable for the highest and lowest local ambient temperatures and so that the limiting temperature in normal operation (see Table 52.1) and the limiting temperature in case of a fault (see Table 43.1) will not be exceeded.

I wouldn't want to make a sweeping statement that it's always OK to calculate on the basis of 30 degrees - but rather it's legitimate to use engineering judgement do decide which factors apply and how much weight to give them. As broadgage mentioned, you might want to think twice if the cable was feeding air con.

Cable calculations always involve a whole stack of assumptions many of which won't be accurate for any specific situation - but there is a lot of margin in there - e.g. we can't guarantee that someone will distribute loads evenly around a ring circuit so there will occasions  where >>20A flows along one of the legs, yet we know from experience that such cables survive perfectly well in practice. Generally occasional small overloads of short duration are perfectly acceptable (the resulting reducing in cable insulation lifespan is very small) - so it's down to judgement is your situation falls into a similar category (even if the overload is due to a reducing in cable capacity rather than an increase in actual load).

Also keep in mind that higher ambient temperature will de-rate protective devices as well as the cable (both the thermal element of MCBs and fuse links will heat up more at higher ambient temperatures for the same current) so the cable might not be quite as unprotected as first appears - although the likelihood of the load being unwantedly disconnected might be higher too!

  - Andy.

522.1.1 Ambient temperature (AA) A wiring system shall be selected and erected so as to be suitable for the

Indeed. I'm not suggesting that taking an engineering judgement approach would mean the cable was unsuitable for the conditions at all - but rather that methods other than appendix 4 might be used to demonstrate it that is suitable.

   - Andy.

Seems to me that the general approach is to use 30ºC unless the cable is installed in a particularly hot/cold zone and, of course, thermal insulation etc.

or more prosaically, try 30C and if it looks about right stick with it, unless the economics or mechanics of the situation are such that it justifies the extra effort (analysis/consultancy fees etc) required to prove that another approach will be OK in  that particular situation, Also be aware that '7671 is used in other countries - Malta comes to mind, and parts of the middle east, where 40 degrees is very much an unwise assumption in the other direction.

Mike

or more prosaically, try 30C and if it looks about right stick with it, unless the economics or mechanics of the situation are such that it justifies the extra effort (analysis/consultancy fees etc) required to prove that another approach will be OK in  that particular situation, Also be aware that '7671 is used in other countries - Malta comes to mind, and parts of the middle east, where 40 degrees is very much an unwise assumption in the other direction.

Mike