Summary and result


1 Introduction

When a rower starts the recover at a “high” speed at the sliding and decelerates towards the next catch, the maximum boat speed (within one cycle) is reached immediately after the start of the recover. Because the resistance of the boat is proportional to the square of the boat speed, a relatively high penalty (resistance, loss of energy) must be paid in this phase of the recover. An obvious idea is then to start the recover at low sliding speed and to accelerate towards the end. The boat speed will be smaller at the start of the recover and higher at the end. A more equal distribution of the boat speed during the recover will be the result and less energy will flow out of the system. A quantitive analysis of this idea is unknown to me and therefore a simple simulation model has been constructed and run.


2 Model

The model is presented in Fig 2.1




















This model is the same as in Simulation of Rowing and is based on the single scull. For a discussion of the equation of motion reference is made to this webpage and to Recover, Details and Model.




In a model for analysing the recover phase only, there is no need for a kinematic  coupling between sculls and seat motion. This results in a simpler model, the sliding length and the mass distribution are more directly related to the situation in the boat. The sliding length has been taken  s = 0.9m, m1=45kg (part of the rower’s mass that is considered fixed to the boat, hull of the boat plus hydrodynamic mass), m2 = 65kg (the mass moving with respect to the boat),  total mass m =m1+m2=110kg.

The hull resistance F= -0.5*C*(dx/dt)2, with C = 2.5Ns2/m2.


The seat motion (assumption: all parts of m2 move with the same speed over the same distance) takes place in T = 1s for all modes of motion to be considered. The mean seat velocity is vm = s/T = 0.9m/s.


Three modes of seat motion are considered:

1 uniform (black line)

2 uniformly decelerated (red)

3 uniformly accelerated (blue)
















From Fig 2.2 follows that the value of the seat acceleration for mode 1, 2 and 3 is: 0,  -2vm*/T,  2*vm/T respectively (in fact the signs are to be  reversed because the seat velocity is negative).



3 The simulation

At the end of the previous pull through, t = 0, the recover motion has not yet begun, the hull has a speed Vb = 4m/s for all modes.

The start of the simulation is the start of the recover, also t=0. An instantaneous change of hull speed occurs, calculated with the principle of conservation of momentum. This delivers 3 different initial values of the hull speed at the start of the simulation.

At the end of the simulation, seat in the end position but in mode 1 and 3 still with non-zero speed, the principle of conservation of momentum is applied again to obtain the speed at the begin of the next pull through.



4 Simulation results

 The simulation results in a graph of the hull speed (=boat speed) during the recover. See Fig 4.1.


















The graph shows very different developments of the hull  speed during the recover. But it is obvious that the area under the various curve (they are not straight lines!) are not so different. By direct calculation with the simulation model yields the results in Table 4.1






hull speed at the

start of the next pull


energy outflow

during recover















Table 4.1

Main results of simulation


5 Conclusion

Mode 3, the accelerated recover yields indeed the greatest distance covered during the recover but the difference with the most unfavourable mode 2 is only 0.033m, or 8...9m advantage on a 2000m course. This advantage is not free of charge, because the outflow of energy is 1.2J  (a very small quantity indeed) more than in mode 2. This energy loss is also reflected in the rest speed, at the start of the next pull. The simulation procedure could have been adjusted as to find such a seat speed that results in a maximum distance during the recover. As the practical significance of such an exercise is questionable, it was not executed.

In practice such sharp differences in seat motion as in the simulation can never be obtained. It will be a smooth curve sometimes closer to mode 3, sometimes closer to mode 2 and the difference in covered distance will diminish.

It seems useless to coach on a very pronounced mode of seat motion.

For details of the simulation, see: Recover, Details and Model.