You all have excellent ideas, which I can not surpass.
Practicing PRMs (closely-spaced parallel approaches, possibly ILS and an LDA) might not be worth the time spent on them, at least not each year.
As for engine failures, we (most of us) all need practice on V1 cuts.
How about already coming down the glideslope with a quick flameout, or better yet, descending from the FAF at 1200'/min. on a non-precision? This could be excellent training for a rare but possible problem-maybe best not as checking item, but to make us much better pilots. For the laymen out there, a non-precison is a misnomer, and requires much more work than a precision, because there is no glidepath indicator (except on some new aircraft which have an artificial glideslope for all database approaches), and the descent rate must be managed with our hands and changing the power settings and various descent rates: we extend landing gear with full flaps before the final approach fix. It is very hard to remember how to figure out where a visual descent point should be (no DME, especially on the NDB). I forgot a year ago.
Already at the MDA with Maximum Drag and an engine goes bang?
You need lots of power when fully configured for level flight-how about level with all this drag and trying to push up the good engine as you think what flap setting you need to AVOID a descent into the cloud/fog-covered ground, but not risk a stall?
Memorize this life-saving "technique", along with about how much extra rudder input you need, before it is too late. We have NO automation (just altitude hold), but might have 122 passengers in the longest stretched series.
As for a random flight control failure, try one with a stuck (up) spoiler panel. This is quite awkward and challenging-you might roll up to 75-80* bank (in the simulator) before you recover.
