I looked for actual regulations some months ago, found none and indeed found some evidence that there are none, ie the rules are being devised up by the manufacturers.  This being the case its possible that both are right, in the sense that they have come to their own decision and take different approaches to the risk.

Propane sinks so an air brick is clearly a vulnerable point of entry.  That said we make a vast fuss about 1.5kg of propane in a heat pump, but can have a 13kg bottle more or less anywhere including in eg a boat.  How many points of ignition are there under the floorboards which are likely to trigger before the propane disperses?  It all feels a bit disproportionate to me!

I appreciate that the risk is probably proportionate to the amount R290 gas in the HP system. That said, gas is gas and once released it immediately becomes manufacturer agnostic. 

I have just looked at an Ideal R290 HP installer's guide which states the following:

'When selecting an installation area, it should be considered that

R290 has a higher density than air and therefore in the event of

leakage from the heat pump, will displace air and pool in low-

lying areas. A protective safety zone is defined around the area

to prevent explosive and asphyxiating atmospheres, by avoiding

the following within this area:

1. Building openings, e.g., doors, windows, air intakes,

exhausts, cellar entrances, etc.

2. Ignition sources, e.g., electrical switches, plugs sockets,

drills, heaters, etc.

3. Areas outside of the property lines e.g., public areas,

adjacent buildings.

4. Ditches, troughs, un-trapped drains, or depressions in

the ground.

5. Surfaces exceeding 400 °C in temperature.'

An underfloor void is in all but name a cellar.

Looking at HP R290 capacities, Valliant and Cosy  are nearly the same.

I discussed this topic with my installer late last year before the install of my Grant R290.  There is an air brick within the exclusion zone for the unit.  Given that there are about 16 air bricks in the house in total we agreed that he could seal the one in the zone with silicone.  Maybe not the most elegant of solutions, but simple, and given no history of damp/condensation losing one didn't seem like a big deal.  His view was that this was going to be an increasingly common discussion point with R290 and Grant tech support were adamant that it was a no-go with the air brick open.

@l8again this is an interesting topic. Thanks for starting it.

I'm going to enquire further with a few installers and see what the industry says on the ground, but I have to say it's staggering how pedantic some aspects of heat pump installs get treated (like this), yet when really big, obvious mistakes happen in other parts of the process that aer deemed 'safe' they're often just ignored or dismissed. Bananas, really.

@jamespa, you appear to be spot on that there's no actual regulation forcing this after a quick search... it's all manufacturer guidelines, and they can differ based on their own risk calls. The propane sinking point is valid, and yeah, the fuss over 1.5kg in a heat pump versus a 13kg BBQ bottle sitting happily in a shed (or boat!) does feel disproportionate.

But as @l8again says, once released it's just gas, manufacturer-agnostic and an underfloor void is basically a cellar, so I get why some are cautious in a health and safety focussed world.

@efftee, that's interesting about your Grant install... sealing the one air brick with silicone worked fine, especially with plenty of others and no damp history. Sounds like a practical workaround.

@l8again, who was the German manufacturer you referred to in your original post? I've reached out Octopus, Vaillant and Stiebel Elton for clarity, as well as to other industry people that will hopefully answers. I'll pop answers here when I get them.

@editor 

Whilst you are contacting installers you may also ask about R290 heat pumps below windows.  My reading of the diagrams from several manufacturers (some are clearer than others) is that siting an R290 heat pump below an opening window is OK, and indeed that is where mine is and what logic would dictate (propane sinks).  There are however several on this forum whose installers have said this is a no no.  I don't know if they have stuck to this when pushed mind you.

The attitude of my installer was initially to say it was doubtful, then when I pointed out that the diagrams, at least to my reading, clearly say otherwise he said - well as long as you are happy, I'm happy!

My view in this is to weigh the risk to the reward.  There’s no actual regulation I know of that limits siting of R290 plant.  

But, and this is the real issue, should there be some disaster befall the site, the insurer may well do the insurer type thing of wiggling out of it as it’s not installed to manufacturer guidelines. 

then you get to the sue the installer and all the shenanigans that goes with it. 

So I would take the simple route a look at an R32 unit.  There’s not a lot of difference in performance and in some situations, the R290 is actually worse than the R32.  

I know we might be choosing based on GWP or newest tech, but don’t get too hung up on it.  

view the entirety of the system not just the source. 

@l8again, as an update, I’ve been in touch with person from Octopus Energy you’ve referenced above (he’s the author of the manual), and he’s pointed out that the version floating around online (the one above) was an old draft from about 2 years ago that somehow got uploaded to a PDF site, hence why it explicitly allowed air bricks (with notes like not directly behind the unit and optional shrouds). That’s no longer current.

If you’re still pursuing Cosy, I’d suggest reaching out to Octopus installers directly for a fresh site survey they’ll apply the latest manual and can confirm on air bricks for your setup.

@saenergy, thanks Simon. That’s very helpful. Can you also give us your take on R290 heat pumps below windows?

Professional risk assessment (inc some common sense)

Multiple foundation air bricks offering 30% free air ventilation for block and beam type floating floors as defined as a well ventilated area acts as a plenum permanently moving air not as a pooling zone, this means even on a still day, due to the temperature difference under the floor the plenum will naturally vent all the time this could mean the risk of gas build up is minimal.

We must also stop assuming that every bit of gas leaked would poor through the very small holes and not be partly carried away by air along the side of the wall while the gas is boiling off, so no longer the full volume of gas, if air is moving the gas and gas is sinking some gasses will spread out away from the vents, nothing is to say that all of the gas would leak at once, 90% of refrigerant leaks on DX systems are very slow and often noticed within days due to performance of the system dropping. The exclusion zone is a manufacture recommended guide line to help prevent risk at the worst case scenario.

If you were to Risk asses this correctly you would have to measure, likelihood = 1 - Severity = 5  L x S= 5 Medium. 

R290 Refrigerant also boils off very rapidly so the only way to pool it is to confine it or reduce its surface area to slow the boiling:- example, if 1.5kgs leaked next to a sump or drain pipe, the sump and or drain pipe could fill quickly reducing its boiling surface and in some small way confining it for a short while, this adds to some of the risk if there is an ignition, if the gas leaks to a larger open area it boils off much faster, if it leaks in to a well ventilated area the boil of or dispersion is even faster. 

I have yet to see any R410A, R32 and R290 monoblock loose any refrigerant so they are normally very well built. (I have seen a lot of units -Likelihood is very low) 

Argument 1

Now think of the rules for NG to be plumbed in flats (which is an endless supply BTW of M3), all risers must have a free air grilles of 25% H/L ventilation so in the event of the gas leaking it can disperse on ignition and be smelt before ignition to evacuate the area. 

Argument 2

Most caravans have the small cylinders in a chamber inside the side of the van, it is recommended for safety while in use stand the cylinder externally under the caravan in a  well ventilated area. like a well ventilated void under the living area (13kgs). Same as the well vented area under a house.

IMHO with regards to the exclusion zone specifically 

We should always follow the manufacturers guid lines as they have spent a lot of money and time making sure they cannot be prosecuted for any damage or harm their equipment may cause should it fail, in the event we cannot obaide by these rules rigidly then you should always risk asses and take the so far as is reasonably practicable solution (this does mean if it is safer to place the unit in the middle of the garden, then this is really where it should go), H&S is a about preventing injury but not at the expense of not moving forward, other wise they would not place R290 in the machines as that would be the safest course of action.  

I think as the home owner you can sign a weaver to say you refuse to have the unit in safer location as it will be undesirable and I also think the installer can say I am not installing it near your air vents. 

Conclusion

It is not the law so there is no legal requirement to follow this rule, it is however at this moment best practice and a risk and should something however slim go wrong and someone got hurt, it would be insurance and professional bodies as well as RIDDOR H&S exec that would decide yours and the installers fate.  

For the sake of the fate, the risk your house would be damaged by 2 less vents out of 20 verses an explosion is indisputable and I would get your agreement to block the vents.

@editor apart from the question of noise, as the argument is that propane is heavier than air, I see no reason not to have the heat pump under a window.

again though, I would be far more concerned about insurance risk if manufacturer instructions are not adhered to.

my view is that you use the right kit for the right job.  Just because a heat pump can produce high temps, doesn’t mean it’s better for the job.  The real target is to drop flow temps.  Designing out LT is not the way forward.