Development of a reliable performance prediction tool for semi-planing hull forms

Hi ,

I am a naval architect on the quest to develop a reliable performance prediction tool! Along the years I have used the many naval architecture programs commercially available but have never found a satisfying method for a reliable performance prediction of semi-planing hull forms.

As a result of that we end up having to test all our designs in towing tank which is rather expensive or more recently using CFD with Open Foam which is promising but still very slow.

I want to use the mass of data that we have accumulated from towing tank tests, sea trials and CFD calculations in order to correlate the hydrodynamic resistance of semi displacement hulls with their form parameters.

My initial approach was to follow the method described by the Delft University
Here is one of the papers they published in 1996:
http://www.hiswasymposium.com/assets/files/pdf/previous/14th - 1996/Keuning@hiswasymposium-1996.pdf

The obvious difference is that this is applied to sailing yachts resistance which are not so much affected by the transition stage from displacement to planning regime (apart from the mighty open 60 and Volvo 70 of course!) but also this was a systematic series.

Although I have a lot of data available, unlike a systematic series it doesn’t explore the design space homogeneously and makes a good correlation slightly harder to find.

However I am starting to get good results using the following model:
Rr=Δ*(a0+(V^(1/3)/LWL)*(a1*(BWL/LWL)+a2*(V^(2/3)/WPA)+a3*(Vtun/V)+a4*(LCB/LWL)+a5*tan(α))
Where Rr is the Residual resistance, Δ displacement, V volume, Vtun volume of the tunnels and α angle of entrance
And the coefficients a0, a1, …. vary with the Froude number following a polynomial equation of the 5th order.

Does anyone know of a better approach than the one I described here? Maybe something like a response surface method? Could a software like Matlab offer the tools I am after to derive the best statistical law? I am currently using the good old excel coupled with the premium solver.

Also, it became obvious to me that these types of tools would gain a lot of accuracy if we could gather all of our information together on something like an open naval architecture database where everyone could input their data and different regressions would automatically be generated. This would also be a way to check your towing tank results in that if it really doesn’t match the existing regression it probably means that something is wrong with your results.

I know I am probably being overly optimistic and this will probably not get done since we all want to keep our data for ourseleves. However this approach may have more potential than most 2DOF CFD work that everyone seems to be focusing on. I am not saying CFD doesn’t work but it is just not suited yet to quick performance prediction of planing and semi planing hull forms.

I would very much like to have your thoughts on these matters.

Many thanks

 

There is a long and lively discussion about the definition of planing (http://www.boatdesign.net/forums/hydrodynamics-aerodynamics/definition-planing-45248.html) that may show you how difficult it is to find it. We haven't yet been able to reach any satisfying definition. I think you may be more accurate and successful by limiting the operational speed brackets for the vessel. That would remove all the confusion about displacement, semi-planing and planing modes and focus on the characteristics of the particular vessel.

 

Thanks for your reply.
It has become increasingly important to predict accurately (and minimise) the resistance at cruise speed as well as top speed. That is why we cannot focus on one specific Fn range.

I have in the past used “composite” resistance curves where I used different methods for different Fn ranges but it is far from being ideal.

I found the discussion on what defines planing very interesting indeed but it seems that after this long debate no consensus could be found. The concept of displacement/semi-planing/planing is evidently very important but they are just words trying to define a hydrodynamic behaviour.
At the end of the day a resistance curve is not stepped nor are the trim or heave curves.

What I am currently working on is a global model that could quantify the resistance for a wide range of Fn for a given set of form parameters. I am not saying it is the way forward but I feel that if enough data is available it should be achievable. I have attached a snap shot of a validation I did using a hull that wasn’t used in the regression data. It shows good correlation but I would like to extend the area of validity of the model in the future.

 

The results look well within acceptable variance. Congratulations.