Hull Resistance

For those of you that don't know, Markla and I work in the marine industry as engineers. Naval architects to be precise. We have at our disposal some pretty neat tools and it is always cool to put them to the test.

For this reason we decided to compare the full scale tow tank resistance data of Bill Beaver and John (better copy and paste this) Zseleczky with results from one of the tools in our kit. It is a program that predicts resistance of ships using a number of methods, one of which is called the Slender Body Method. Basically it uses an analytical method to calculate the resistance of narrow hulls rather than using traditional statistical/regression methods. blah, blah, blah, that's enough background.

Firstly I produced a hull model traced off the Hungry Beaver linesplan in the report (Bill has since sent me a file of his hull which I haven't had a chance to run with yet.). Static drafts for each of the towed displacements were calculated and then run through the program. Neither the dynamic trim nor squat that was present in the tank tests was modelled, for better or worse.

Ok, raw data comparison graphs attached...


So far so good, despite not accounting for trim or squat with the computer prediction, there seems to be some good (but not perfect) correlation. The biggest problems are underprediction of resistance at higher speeds and higher displacements/drafts and the existence of a hollow in the computer prediction at about 4 knots. There isn't any tank data points around there though so it could be a resolution issue.

Well I'm interested in the correlation at the faster speeds (not really that fast) so here's a graph at 8 knots for varying drafts...


This shows a clear discrepancy at the deeper draft for this speed. The difference is nudging 10% of the towed result. Well there's a number of reasons this could be:

• Incorrect/lack of calculation of transom drag at greater drafts.
• Effects of static trim
• Effects of dynamic trim
• Effects of dynamic squat/heave

I have no knowledge of transom drag calculation so I'll forget about that for now; static trim may be the cause but apart from the general hydrostatics 0 deg trim I'm not sure what the trims were for each displacement; we have nice dynamic trim/pitch data so this is a possibility but requires many more computer runs to get more resistance data (cbf right now and isn't as straight forward as my lazy arse would like). So this leaves us with dynamic squat (or just plain squat because it only occurs dynamically and really refers to shallow water behaviour of moving ships).

When referring back to the beaver hungry paper, squat increases with increasing displacement/draft (as well as speed, I know, but one thing at a time) so let's interpolate the computer resistance at the squatted drafts obtained in the tank tests.


mmm, that didn't work so well. Maybe not, while the average error seems to have increased (from -1N to +4N), the standard deviation of the errors has decreased (from 3N to 1.5N). Some gains, some losses.

So next thing for me is to get some dinner. On the resistance topic though, I'm happy enough to move forward with assessing hull designs based on the raw computer data. The main reasons for this are that I am happy with the correlation we've seen so far and predicting squat and trim is too hard/impossible with the tools and knowledge I have available so they can't be accounted for in the prediction. I'll probably run the actual hull model through the program to iron out any errors in the results from that area, but I don't expect major changes.

I'd like to thank Bill for his correspondence and congratulate Bill and John on releasing some useful info to the mothing, and sailing, world.

ps - all results are in fresh water, made that mistake too many times with this validation to ever forget again.

Wave picture