After fixing the boundary condition and drag issue, the simulation is done and the drag history is shown in Fig. 1. The mean drag is very close. The y-axis is at the scale of \(10^5\), so there must be a unit conversion I was missing in the calculation.

convertToMeters 1; should be used as convertToMeters 0.001; to do the unit conversion because the model is in a unit of a millimetre. So, the drag coefficient should be 0.19874? The next question is how can I verify this.

Figure 1: Using two ways yeild similar mean drag

Figure 1: Using two ways yeild similar mean drag

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
>> mean(Cd_good)

ans =

   1.9876e+05

>> mean(Cd_bad)

ans =

   1.9874e+05