I’ve had this before and I guess I can sympathise with the ATCOs who are probably trying to make it all work while keeping a bunch of Airbus and 737 low cost inbounds flowing!
I would offer that sometimes people try too hard to perfect everything at once – sometimes it is better to accept a slight localiser deflection, get the glide slope profile sorted and stable, then come back and sort the localiser. All the while just set a known good power setting and let speed stabilise. I know on my IR reval last night I started to chase the needles a bit on my asymm RNAV app and had to consciously tell myself to slow down and let the corrections stabilise!
We have all had similar situations happen. In a high workload situation where you have expected some automation to automate, it doesn’t. In some ways disconnecting the autopilot is dangerous in such a case as it significantly increases your workload. It is different from being in a stable configuration ahead of the aircraft and then deciding to hand fly.
And the unstable hand flown decent is of course the thing that will kill you.
One way to handle it is in the moment it failed to intercept when you were already close and high. Tell the controller that you are unable to accept the approach this close and high and request vectors for a longer final. Ideally notice it earlier and point out you need more track miles to commence the approach.
I once had the same thing, at Oxford, in your plane. I tried to save it and had to go around. I should never have started down. It was also caused by a failure to intercept.
I would offer that sometimes people try too hard to perfect everything at once – sometimes it is better to accept a slight localiser deflection, get the glide slope profile sorted and stable, then come back and sort the localiser. All the while just set a known good power setting and let speed stabilise. I know on my IR reval last night I started to chase the needles a bit on my asymm RNAV app and had to consciously tell myself to slow down and let the corrections stabilise!
I actually think that can become a bit of a training/check ride technique and can in fact be negative training. In training with a safety pilot you can get away with some experimenting and playing to recover. In real difficult IMC as in the OP, if you are properly outside parameters just go around. I don’t mean one dot here or there but a truly unstable approach is just not worth trying to save. Go missed and start again.
Unless you are on fire, it is only a small amount of fuel involved and some time.
JasonC wrote:
In real difficult IMC as in the OP, if you are properly outside parameters just go around
On simulator it is easy to pause, switch IMC off and look at the view outside when approach needles go wild, usually, the mental picture on the starting point is 10 times wrong but on rare occasions when that mental picture is right (usually by looking at GPS) you only need one or two inputs on stick/power to bring all of it in order without much chasing
I think one has zero chance to “save it” starting from the wrong place, we all need de-zoom to avoid over-navigating on GPS while CDI needles don’t have a similar function unfortunately 
Thank you very much everyone for the suggestions.
The key things I take away from all the comments are 1) the need to be more thoughtful about how the approach will evolve under unusual and challenging conditions and 2) being more forceful with ATC.
Thinking back over the situation, it was made even more challenging by the fact that in the phase I was being vectored and descended onto the localiser I had an effective tailwind component of about 40 kts. I had been asking approach control for descent which had been refused until way too late. Had I been more on my game I would have realised that I had no chance of getting stabilised on the localiser from the position I was in. Several of you suggested asking ATC for vectors to a specific place on the localiser "8 DME etc) – that is something I had not thought of before but will do in the future.
Lastly Jason commented on the danger of disengaging the autopilot at the moment it fails to capture the localiser, which is of course exactly what I did! He is absolutely right, at that point of surprise the stress level jumps through the roof, making an effective recovery even harder. Better to continue on the autopilot heading, level out and only then calmly ask ATC for vectors around and back.
Thank you again.
One other suggestion @Buckerfan if, let’s say, you were further out and had a bit more time and hypothetically you forgot to hit APPR or something.
Turn inbound on the heading bug in heading mode. See if you can work out what the problem is and perhaps you have time to sort it out before the glide slope comes it.
Disconnecting an autopilot is always an option, and sometimes necessary, but single pilot can often lead to task saturation.
Some airlines don’t use Control Wheel Steering (Airbus doesn’t have it, and BA may not have got their head around it), but CWS is there precisely for this situation.
@Peter may think differently 
, but my experience with Spanish ATC is that they are very good (at least in PMI TMA) at vectoring you to the ILS in a CDA and will typically place you below the GS before the FAF so you can intercept it from below. They are however expecting a certain descent angle above 5% which is a bit high for my liking . If you follow a stable path, they are usually good at issuing stepped descent clearances without you having to stop descent.
Thanks for sharing, @Buckerfan
The situations where I found myself non-stabilized were usually with a late change of runway, but it always happened in VMC so I did not take evasive action and just tried to make it to the threshold within parameters, rather than changing autopilot, Aspen and ipads in the last minute…I have done it for practice a few times but the workload is very high. If in IMC I think I would ask extended vectoring or holding to reconfigure.
Antonio wrote:
my experience with Spanish ATC is that they are very good (at least in PMI TMA) at vectoring you to the ILS in a CDA and will typically place you below the GS before the FAF so you can intercept it from below.
The only time I have ever been vectored to intercept a glideslope from above was approaching Palma. Luckily in VMC but far too high, fast, and close for an ILS to have worked. A Visual approach was OK
Antonio wrote:
but my experience with Spanish ATC is that they are very good
My experience with Spanish ATC is great. I didn’t have any problem with their ELP (very often stressed by @peter ) – it seems that non-native speakers understand each other better, probably because of not having heavy British accent and using simplified version of English
BTW few days ago ATC at approach to LDZA asked me to keep high speed (160-165 KIAS) until outer marker to be number one on landing. At the end I landed 7 minutes before Turkish Airlines which was the cause of this request.
Hello
This approach report is very interesting from the point of view of the human factor : most of them have lived this kind of approach which ended up in a landing without any big problem but of which we were not very proud to the issue!
The main remarks are :
