Assumption: Some mechanism exists to sense RPM, either through an encoder attached to the motor shaft, or via back-emf sensing.
An approach to achieve better results that what the question describes, for low RPM operation of a motor, is to use a PID controller algorithm thus:
- Motor is provided maximum rated power at start-up, as specified in the motor datasheet
- As the RPM sensed approaches the desired set-point, the power is systematically reduced my modifying the triac trigger phase
- Once the RPM set-point is achieved, the PID controller continues to sustain that RPM by increasing or decreasing power to compensate for loading effects
- If load drives the motor to stall, or beyond acceptable power ratings, the controller initiates a controlled fail-safe spin-down of the motor, and also triggers an alarm indication.
From the Wikipedia article linked above, this graph might help explain this process visually:
Depending on the acceptability of overshoots (or not) and desired system behavior, the Proportional (P), Integral (I) and Derivative (D) values of the PID algorithm need to be tuned. The diagram above specifically covers tuning the I value: For absolutely no overshoot of RPM, but with greater time to reach set-point, the red line on the graph shows the preferred behavior, as achieved with a Ki = 0.5.
On the other hand, the black trace, with Ki = 2, achieves (and overshoots) the set-point fastest, and then over/undershoots the set-point in diminishing cycles till it settles down.
There exist excellent motor controller ICs which incorporate both back-EMF sensing (if applicable to your type of motor) and PID controlling, in one package.
Also, assuming the OP has limited experience in designing such systems, off-the-shelf PID controllers for motors are available for a variety of power ratings. These allow a set-point to be interactively set, along with tuning parameters or constraints.
There are also several projects by hobbyists out there, designing and implementing PID AC motor control using microcontrollers or microcontroller boards. For instance, see this YouTube video, for one such Arduino-based PID controller for AC motors.
Links to details are provided in the description text of that video.
There's a good solid guide to motor power curves here: http://lancet.mit.edu/motors/motors3.html
Basically there are two ends of a quadratic curve at which power is zero. At zero RPM, the motor is stalled and exerts maximum torque force (and will draw a large current, the "stall current" listed in the datasheet). It doesn't actually deliver any power to the load because it's not moving, therefore no work is done.
At the other end, there is no load on the motor and it spins freely at its maximum RPM. It doesn't deliver any power either, as there is no load.
There is a maximum power delivered at half of the no load speed, in the middle of the torque curve.
As you say, the ability to deliver high torque at zero and near-zero speed is extremely useful, and is why motors have been used for power conversion in diesel-electric trains for 50 years.
Best Answer
It is possible to build a direct-coupled alternator/generator. Both the alternative energy industry and US Department of Defense have been investing in this area for several decades. What's attractive about direct drive turbogenerators is the extremely low power/weight ratio, even though efficiencies are awful at this scale.
Capstone Corp has been doing this for awhile in the 60kW class, mostly for backup power (natural gas-fired):
Others are working on smaller scale units - in the 300W to 5kW range, also direct-drive. I have not seen much traction in the industry at these sizes, however.
http://www.azmark.aero/az.nsf/html/Product+Gallery+-+Power+Systems
Is it possible? Yes.
Is it practical? It depends on the application's ability to trade mass for cost.