Folks,
For may years, I worked with an operator with a well deserved reputation for risk minimization ( inaccurately aka "safety") and I have always been happy with their logic, and applied it to my own operations.
ALL their aircraft were certified with whatever were the DCA/CAR/FAR transport standards of the day, which all included a certified "stop/go" capability in the event of engine failure.
NOT a feature, generally, of FAR 23 aircraft.
The training rules were really quite simple:
(1) The only time you shut and engine down (by the approved method) and feathered was above 3000'minimum, 6000'+ preferred AGL, daylight, VMC.
(2) All engine failures on takeoff, and any related operation below 3000', simulate the failure by SLOWLY retarding the throttle, sufficient to ensure the student/candidate correctly identified the failed engine, and then manipulating the manifold pressure/RPM to simulate a feathered prop.
(3) NO simulated asymmetric at night.
(4) Going beyond the above was a dismissal offence, such was the risk analysis of any training value versus the increased opportunity for disaster, almost always fatal.
Works for me, and for anybody who works for me.
The FAA, some years ago, published an interesting analysis of outcomes of in-flight engine failures in singles versus twins, as always most instructive.
In short, subsequent to the failure, the fatal accident rate for the landing heavily favored the single, and where the twin resulted in a major accident was almost always because of asymmetric mishandling, resulting in loss of control in flight (there is some sobering in-flight genuine footage on YouTube, illustrating how sudden and violent it can be getting below Vmca in a FAR 23 twin), whereas most singles arrived back at ground level more under control than less.
In other words, a FAR 23 twin is not just a single with twice the chance of engine failure, it is worse than that.
Tootle pip!!