ComputeForce - Forces on a Floating Wedge

Hi guys,

I am trying to simulate a wedge entry of experiment R. Zhao, O. Faltinsen, J. Aarsnes, Water entry of arbitrary two-dimensional sections with and without flow separation, in: 21st Symposium on Naval Hydrodynamics, 1997.

The wedge is 0.5 m wide and 0.29 m high with 30 degrees dead-rise angle.

The total length of the wedge is 1 m

The total weight of the wedge is 241 kg. And Initial velocity just before contact is - 6.15 m/s

Therefore the Calculated density is Total Mass / (Cross-section Area * length) = 2212 kg/m^3

I got the velocity almost equal to the Zhao results. But the Vertical fluid forces of DualSPHysics is very large.

Can you tell me why I am getting large forces?

Here I am attaching the simulation files


  • Here are some images of comparison with R. Zhao, O. Faltinsen, J. Aarsnes, Water entry of arbitrary two-dimensional sections with and without flow separation, in: 21st Symposium on Naval Hydrodynamics, 1997.

  • And also is vertical slamming force equal to (mass of a wedge particle * summation of acceleration of all wedge particles) ?

    Because in the above force plot, the Zhao results are the vertical slamming force and DualSPHysics results are the mass * acceleration of wedge. The difference is large and I cannot understand why?

    The setup is identical otherwise I wouldn't get the above velocity plot.

  • @C_Kesanapalli Either you can calculate the force as:

    • Total mass of wedge * sqrt(sum(a_x)^2+sum(a_y)^2+sum(a_z)^2)
    • Use DualSPHysics post-processing to find the forces exerted on the wedge object

    The way DualSPHysics calculates forces in post-processing is by analyzing how the water particles "push" on an object, which is the opposite of what you do, ie. calculate how much the wedge "push" on the water particles. Both of these approaches are valid and it HAS to be very close to the same, since IF NOT, there is not a balance of energy in the system.

    My guess is that you have calculated the force wrong, by forgetting to use the magnitude and just using the sums directly.

    Kind regards

  • edited November 2019

    @Asalih3d Thanks a lot for the reply. I really appreciate it.

    I have calculated the force by using ComputeForce and the above and below plotted force is z compound of the force - "ForceFluid.z [N/m]". (wcsph_force)

    I have also calculated the Force using (Total Mass dvz/dt). (wcsph_vel_diff). Here dv/dt = (vj - vi) / (tj - ti)

    But the results are same.

    I don't know what I am missing here.

  • Hmm, I have a difficult time pinpointing what is going wrong then sorry. Maybe something simulationwise is not set properly, densities etc. Because for me, it looks like that both set of data have the same "tendency" but just scaled. Your calculations of force in z-direction seems correct now.

    Kind regards

  • I found the issue.

    In the experiment, the force is calculated for only a part of wedge i.e., mid section of wedge of length 0.2.

    Therefore the net force with (force per unit length ) * (mid section length = 0.2)

    If I scale the results then I will get the proper comparison.

    Can you tell me how to remove the fluctuations in the force ? Should I increase the viscosity/delta-sph or increase number of particles

    Thanks for the help

  • Great to hear!

    Usually they normalize in papers, so good that you spotted that. You cannot avoid the fluctuations when getting data out, but you make it output at more time steps and then apply a smoothing or a low-pass filter (if I recall correctly) - I assume that is the same they have done in the paper.

    Kind regards

  • I was trying to increase the time and space resolution and it was reducing fluctuations magnitude. But smoothing or filtering can be a good idea.

    Thanks for the help..

  • Dear colleagues

    Nice to see that solve the issue.

    About fluctiations, we are working on that... the next beta release v5.0 that we will distribute to delegates of 5th DSPH Users Workshop will include two new functionalites (new deltaSPH and new BCs) that solve these problems to get smoother density/pressure fields so that velocities so that forces

    More info about the event here:


  • Great to hear @Alex !

    I always thought that these spikes were just a consequence of choosing SPH to solve the problem, but glad to hear that we are also able to smooth it out based on physics :-)

    Kind regards

  • Other SPH code include boundary conditions or other formulations more focused on solving accurately academic problems (couette and poiseuille flow, lid driven cavity, etc) are more appropiate for getting nice smooth force results, however those approaches are difficult extended to 3-D complex geometries and engineering problems... and DualSPHysics aims to be an engineering tool as more general as possible,

    That is why the formulation implemented in the online version presents advatanges and limitations, however we are now working to introduce some improvements that can still be applied to real cases such as new deltaSPH formulation and new boundary conditions.


Sign In or Register to comment.