A further iteration (this is exploratory) ...
Assumptions:
- the tricycle is initially at rest
- friction is negligible except between wheel and floor
- all wheels are identical
- tension in string is small?
Let:
- R be the radius of the wheel.
- r be the distance from the wheel axle to the pedal axle.
- I be the moment of inertia of each wheel
- m be the mass of each wheel
- M be the mass of tricycle without wheels
For the first picture, the forces acting on the wheel:
- Weight of wheel itself acting vertically downward through axle and wheel/floor contact point.
- Contact force from floor.
- Tension in string acting horizontally, line of action R + r above floor.
- Force of axle acting on wheel, line of action unknown.
Resolving horizontally:
- No horizontal component.
- Friction acting horizontally, line of action at floor level.
- Tension in string acting horizontally, line of action R + r above floor.
- Horizontal component of axle acting on wheel.
Resolving vertically:
- Weight of wheel itself acting vertically downward through axle and wheel/floor contact point.
- Normal component
- No vertical component.
- Vertical component of axle acting on wheel.
Perhaps it is useful to consider the wheel as a lever hinged at its contact point with the floor. When the string has moved a small distance {\delta}s in short time {\delta}t it has pulled the wheel axle forward {{\delta}s R} / {(R + r)} and rotated the wheel sin^{-1}({\delta}s / r) (the latter using small angle approximation).
During the short time {\delta}t the accelerations (translational and rotational) from rest are constant so the final velocity is twice the average velocity.
Final translational velocity of wheel axle is
2 \times \frac{{\delta}s R}{{\delta}t(R + r)}
Translational acceleration of axle ({\Delta v} / {\Delta t})
2 \times \frac{{\delta}s R}{{\delta}t^2(R + r)}
Similarly the final rotational velocity and the rotational acceleration are
2 \times \frac{sin^{-1}({\delta}s / r)}{\delta t}
and
2 \times \frac{sin^{-1}({\delta}s / r)}{\delta t^2}
Mmm ... that can't be right; in the limit, as {\delta}t approaches zero, the accelerations become very small.