Resistance prediction tool for arbitrary displacement hulls (modified Michell's theory)

Discussion in 'Software' started by jbasic, Oct 9, 2018.

  1. jbasic
    Joined: May 2010
    Posts: 10
    Likes: 5, Points: 3, Legacy Rep: 10
    Location: Split, Croatia

    jbasic Junior Member

    Hi all!

    Since a lot of things today are web-based, let me present a new online tool/method to predict ship resistance. The resistance prediction tool is intended to be a part of Prelimina - a toolbox for the acceleration of the preliminary design (work in progress). For the moment, this new method can predict resistance of arbitrary displacement monohull forms.

    Short introduction
    During some research, I (accidentally) discovered a physical connection between a modified potential-flow boundary condition and the boundary layer (PDF). It turned out that the modification can make panel methods yield realistic pressure fields, where e.g. even RANS has troubles. I implemented Michell's thin-ship theory along with this modification that accounts for the hull boudary layer (and some fixes for the phase). The results were suspiciously good - aft wavemaking is better and can handle transom sterns. Just recently the calculation method for the viscous pressure resistance was added, so I decided to let you know about the tool. Soon multihulls and dynamic trim/sinkage are coming (so wait a bit if your transom is just touching the WL), and form optimisation... also, the Rhino plug-in if there is interest.

    Validation
    The results for the Kriso containership (KCS) model are corresponding to the experimental data and RANS methods. Holtrop's method is dangerously smoothing out the wave-resistance hump.
    [​IMG]
    Series-60 (with CB = 0.6), where Holtrop's method underpredicts resistance for Fn > 0.3:
    [​IMG]
    And, obligatory, the Wigley hull with a slight resistance bump properly captured (tho it's not really a hull for Holtrop's method):
    [​IMG]
    Using the introduced enhancements that account for viscous and non-linear effects, more accurate results are obtained, as compared to the original thin-ship theory and Holtrop’s method. The trend of the wave resistance curve is properly estimated even for full ship forms. Of course, further comparisons of numerical results to experimental studies are necessary in order to examine the validity of the present approaches and improve their formulations.

    Visit www.prelimina.com and enjoy the calculator! I'd appreciate your feedback.
    - Josip Basic
     
    Last edited: Oct 10, 2018
    pavel915 and Dejay like this.
  2. DCockey
    Joined: Oct 2009
    Posts: 4,371
    Likes: 201, Points: 63, Legacy Rep: 1485
    Location: Midcoast Maine

    DCockey Senior Member

    Very interesting though I don't have access to your papers.

    Have you compared your method to Leo Lazauskas' Michlet code?
     
  3. jbasic
    Joined: May 2010
    Posts: 10
    Likes: 5, Points: 3, Legacy Rep: 10
    Location: Split, Croatia

    jbasic Junior Member

    Dear David, I've modified the post to include the PDF to my paper. I don't think the second one from Noblesse is public. By the way, a new paper about all the details is in the making.
    Yes, Michlet and Maxsurf results were used for the verification of the unmodified thin-ship implementation, which was a starting point.
    By the way, thank you Leo for the advice, fun discussion we had, and for the literature!
     
  4. tspeer
    Joined: Feb 2002
    Posts: 2,228
    Likes: 185, Points: 63, Legacy Rep: 1673
    Location: Port Gamble, Washington, USA

    tspeer Senior Member

    Sorry, I'm not seeing the pdfs - just links to journal sites.

    Can your method handle circulation around the hull, or the wave drag of lifting surfaces?
     
  5. DCockey
    Joined: Oct 2009
    Posts: 4,371
    Likes: 201, Points: 63, Legacy Rep: 1485
    Location: Midcoast Maine

    DCockey Senior Member

    Based on his post, the original poster is using Michell's Thin Ship Theory which only works for symmetric hulls. It distributes sources which satisfy the linearized free surface condition over the centerplane of the hull, with the source strength based on the local slope of the hull surface. I have not seen Michell's Thin Ship Theory applied to lifting hull.

    Tom, what do you mean by "circulation around the hull"?
     
  6. tspeer
    Joined: Feb 2002
    Posts: 2,228
    Likes: 185, Points: 63, Legacy Rep: 1673
    Location: Port Gamble, Washington, USA

    tspeer Senior Member

    I meant a lifting hull.

    I wasn't sure if he was somehow matching the Mitchell's integral with the pressure or potential from the panel code in order to get the wave drag. If that was the case, I was wondering if there was a way of matching the pressures of a hull with side force or the equivalent wave drag of a foil. As I understand it, Mitchell's integral is really a farfield method and doesn't predict the local pressures on the hull. That being the case, it's a little bit like the equivalent body of revolution used in the area rule for aircraft linearized wave drag. A different kind of wave, to be sure, but in both cases it's the farfield that counts.
     
  7. jbasic
    Joined: May 2010
    Posts: 10
    Likes: 5, Points: 3, Legacy Rep: 10
    Location: Split, Croatia

    jbasic Junior Member

    It is possible to find the hull pressure distribution from the far-field. The tool already calculates the local-flow pressure distribution on the hull. The one thing missing is a quasi-static solver to actually lift and trim the hull by those obtained pressures, i.e. the hull is fixed at the moment. It must be noted that truly planing boats should use "different kind" of Michell's theory (see Tuck & Lazauskas', or Yeung's work). I believe that actual horizontal foils near free surface could employ this so-called "pressure-patch" thing to obtain the wave drag, but I don't know if there is global interest for that.
    Secondly, I was intrigued to try the Michell's theory on a heeled yacht (SYRF Wide-Light design) simply by examining separately two sides of the heeled yacht - and it actually gives adequate results, at least for this design. So I guess it could be possible to predict forces and torques of slightly heeled/yawed hulls approximately with this theory. I'm adding this to my to-do list. Kutta condition is a lie anyway :)
    So in the next weeks you can expect: demihulls; deep transom sterns; lift and trim solver; heeled/yawed hulls; hull form optimisation (with a free-form deformation box).
     
  8. Dolfiman
    Joined: Aug 2017
    Posts: 1,021
    Likes: 433, Points: 83
    Location: NICE (France)

    Dolfiman Senior Member

    I read the Wide-Light final report, interesting but a bit frustrating that they don’t deconstruct the tank total drag into viscous and residuary as usually done. In particular for the heeled results of such beamy hulls, it is evident (and easy to compute from just inclined hydrostatics) that the wetted surface decrease and mostly influence the heeled total drag, leading to less drag that the upright one. A deconstruction approach would bring more focused info on the wave (residuary) drag of the heeled hull, which is the interesting point theoretically as practically. But from this report, no info on that unfortunately.
     
    jbasic and Mikko Brummer like this.
  9. Dejay
    Joined: Mar 2018
    Posts: 350
    Likes: 48, Points: 28
    Location: Europe

    Dejay Senior Newbie

    Hello Josip, thank you for making this tool and making it available! It's easy to use and to copy values into a spreadsheet. And looks very sleek :)
    I'm an amateur and interested in building a very low resistance / low power cruising catamaran powered by solar power.

    Does prelimina predict the drag from wave interference between the hulls?

    Here a little feedback from my "playing around":

    The expected axis alignment is a bit weird when exporting from polycad or Fusion360. It would be great of course if you could switch orientations easily and scale the hull.

    At first I got a mirror of the hull when it was misaligned. Am I correct that the mirroring is only done if there are no vertices on the Y axis? (X is forward, Y is port, and Z is up, correct?)
    Or do I have double faces if I don't import a half-hull?

    The value for the density seems to change somewhat unexpected, depending on what validation test you clicked first.

    Values for speed min and max get reset when clicking prepare for calc.

    Attached are the values that I got. For my weird 12m LWL 5m beam 4t displacement test it gave me 472.9N total resistance at 3m/s. So about 2.84kW electrical power assuming 50% efficiency for a speed about 6 knots. Going to 15m doesn't improve the efficiency at that speed.
    Is this a realistic value to expect?

    Thanks again!
     

    Attached Files:

  10. jbasic
    Joined: May 2010
    Posts: 10
    Likes: 5, Points: 3, Legacy Rep: 10
    Location: Split, Croatia

    jbasic Junior Member

    Thanks for the feedback! Since I don't want to spam, all updates were going silently. After testing cats, I'll release the dynamic trim/sinkage solver, transom effects, and some other stuff.

    Yes, the wave interference is included in the prediction. The local pressure interaction when two hulls are too close, not yet.

    I've tried to let user import whole hull or only one (arbitrary) side of the hull. The mirroring is done if the user imports only one side ofc. If you find a bug, please send the problematic hull. For now, X is forward, Y is whichever, and Z is up. The code has an option to choose arbitrary axes orientation, but this is not yet on the to-do list for the interface.

    Thanks, I'll look into it. Validation tests were done in fresh water hmm.

    It is often expected once you reach large values of slenderness. So for your cat sailing at Fn = 0.3, try setting a smaller value of hull separation, and I bet you'll get to around 420 N, which is still very nice for 4 tonnes sailing on 6 knots! The side effect is that above that speed, the drag will abruptly rise (but that's the idea of the design speed).
     
  11. Dejay
    Joined: Mar 2018
    Posts: 350
    Likes: 48, Points: 28
    Location: Europe

    Dejay Senior Newbie

    Please feel free to spam :) I think you're doing poseidon's work!

    Another small "gotcha" I've noticed:
    When I change the hull separation and then directly click on "calculate resistance", I do get different values. If I click on "prepare for calculations" first, then I get yet other values. Might trip someone up.
    I think ideally you could remove the "prepare" button and/or just automatically invalidate / update on changing input fields.

    I did get slightly lower values of 459.5 at 3m separation. But I figure there is a lot to tweak and optimize there. I don't really know what I'm doing, just wanted to get some preliminary values for power needed and what range could be possible at what length. So thanks again!
     
  12. jbasic
    Joined: May 2010
    Posts: 10
    Likes: 5, Points: 3, Legacy Rep: 10
    Location: Split, Croatia

    jbasic Junior Member

    Major update:
    • Trim and sinkage solver
    • Transom effect on a) wave-making b) viscous pressure resistance
    The trim and sinkage solver separately assumes the pressure distribution from wave-making and local flow. At the moment, local flow is approximated based on this, which is very crude, but still okay for an initial approximate guess for (semi)displacement ships. The method that includes transom effects on wave-making is a novelty, more info soon.

    This is a development version, please keep that in mind, and expect more updates soon! Visit prelimina.com to test it out. At the moment there are > 80 active users that occasionally do calculations. Thanks to those that gave me feedback!

    Validation (SYRF yacht):
    Let's look at the results below for a modern wide and light yacht. The trendline of trim and sinkage is nicely predicted. So reaching Froude numbers 1.0 is now not a problem, along with an immersed transom on higher speeds, which significantly affects wave-making. Furthermore, this hull form is far from slender forms, it's a huge improvement that the modified thin-ship theory can handle it.

    upload_2019-10-31_22-46-58.png

    Validation (NPL 4a):
    Very nice prediction of the trim on the image below, but not so much the sinkage (due to too approximate local-flow solving). Larger trim angle (2 deg) modifies the wave-making significantly, which is captured by the solver. NPL has deep immersed transom, so these results are encouraging.

    upload_2019-10-31_22-50-43.png


    In any case, for the moment the results are not bad for a: free, web browser, 2 seconds-per-speed, dynamic solver. Have fun and let me know if you need some features!

    Future features:
    • Rhino plug-in.
    • Improve the local flow solver.
    • Validate and improve transom effects on the wave-making.
    • Transition to planing regime (with boundary layer theory + Morabito pressure distribution)
    • Appendages (both lifting and non-lifting)
     
    Dejay likes this.
  13. Dejay
    Joined: Mar 2018
    Posts: 350
    Likes: 48, Points: 28
    Location: Europe

    Dejay Senior Newbie

    Thanks for the update! I'll give it a try soon.

    This might be a stupid question but does this method work for a power trimaran if I export this as a single obj file? (ILAN type, long slender with smaller outriggers)
     
  14. jbasic
    Joined: May 2010
    Posts: 10
    Likes: 5, Points: 3, Legacy Rep: 10
    Location: Split, Croatia

    jbasic Junior Member

    It's a good question, and the answer is no, but I might implement it in future. Timarans and other multi-hulls are low priority for me since I never had chance to design one.
     
    Dejay likes this.

  15. jbasic
    Joined: May 2010
    Posts: 10
    Likes: 5, Points: 3, Legacy Rep: 10
    Location: Split, Croatia

    jbasic Junior Member

    Dear all, here's a quick visual update to the discussion above. The solver separates the wave-making flow and local (double-body) flow, and it is possible to get the pressure distribution from both steps [ref]. The pressure field (magnitude in Pascals) around the KCS hull, for Fn = 0.26, is rendered in the image below. The pressure is obtained from both sub-steps separately, and the later it's superposed. The pressure field can be obtained in milliseconds on a multi-threaded CPU, which means it's gonna be fun using the solver for the hull optimisation in the preliminary design stage. (Sorry for the bad image quality, this is a quick Paint overlay of contours over the image.)
    pressure_field_test.png
    The top (wave-making) pressure field is obtained by original Michell's theory - notice the too large stern wave. The beauty of the novel method is the inclusion of the displacement thickness of the boundary layer when calculating the pressure field, and resistance from the far-field. The comparison of the original and enhanced theory is given in the image below. The real-deal would be to get the streamlines and calculate the displacement thickness along them, since this is currently approximated from experimental data, using geometrical similarity and interpolation. For now I haven't noticed any large discrepancies.
    upload_2019-11-6_12-57-49.png
    Anyway, this pressure field information is used for the trim & sinkage equilibrium solver, and the solver can handle fuller hull forms. Hope you like this.
     
    Dejay likes this.
Loading...
Forum posts represent the experience, opinion, and view of individual users. Boat Design Net does not necessarily endorse nor share the view of each individual post.
When making potentially dangerous or financial decisions, always employ and consult appropriate professionals. Your circumstances or experience may be different.