Problem Michlet import from freeship

Discussion in 'Software' started by killwater, Apr 26, 2009.

  1. killwater
    Joined: Oct 2007
    Posts: 13
    Likes: 0, Points: 0, Legacy Rep: 10
    Location: Poland

    killwater Junior Member

    Hello
    I have a problem with importing hull shape from freeship. My freeship hull have flat transom but unfortunately after importing to Michlet it is slightly pointed. If I cut and copy the last line composed of zero offsets and paste it on begining transom is ok but i have bow as a straight line (for one station lengt) I can not delete this line of offsets becouse Michlet need odd number as number of stations and waterlines.
    On attached image red lines are with pointy transom, other are with deformed bow.
    Has anyone found a solution to this problem?

    Michal Klimek
     

    Attached Files:

    • 1.JPG
      1.JPG
      File size:
      14.2 KB
      Views:
      655
  2. Guest625101138

    Guest625101138 Previous Member

    Can you do a screen dump of your design waterplane from Freeship so it can be compared with what is being sent to Michlet?

    How many waterlines are you sending to Michlet?

    Rick W
     
  3. Martijn_vE
    Joined: Apr 2005
    Posts: 254
    Likes: 24, Points: 0, Legacy Rep: 401
    Location: Netherlands

    Martijn_vE Marine software developer

    Note that only the underwater part of your hull is exported to Michlet.
    Also, if your model has a sloped transom the waterline start might fall in between two stations
    In that case increasing the number of stations might help in your case. Just set it to the max and try.
     
  4. killwater
    Joined: Oct 2007
    Posts: 13
    Likes: 0, Points: 0, Legacy Rep: 10
    Location: Poland

    killwater Junior Member

    Thank you for fast reply.
    I'm using 41 waterlines and 41 stations. I'm also aware that only underwater part is send to michlet and the part above water is assumed to be straight vertical. Transom is not sloped, it is straight vertical and horizontal. I've always had problem with exporting from freeship to michlet boats with transoms and result always looked like this one above no matter what hull I was sending. Below are screenshots from freeship.

    Michal Klimek
     

    Attached Files:

    • 2.JPG
      2.JPG
      File size:
      133.2 KB
      Views:
      573
  5. Guest625101138

    Guest625101138 Previous Member

    I rarely use bluff transoms so have not taken particular notice before. All the Michlet hull tables have zeros on the first and last station rows and first water line column. I expect it is to avoid singularities with infinite slopes in the analysis. Doing what you are doing to get the flat transom introduces a large error.

    You could better approximate the flat transom by increasing the number of stations and waterlines but I doubt it will change the result much. I rarely use more than 31 stations. You could try a few examples say 11, 21, 31, 41 and 51 stations to see what the result approaches at the speed of interest. I expect you will find little change.

    Even the mlt offset tables generated by Godzilla have the same first and last row of zeros and the first column of zeros so it is a required format.

    The Michlet offset table introduces some odd situations when the number of waterlines and stations are not the same but the results still work out well.

    Rick W
     
  6. killwater
    Joined: Oct 2007
    Posts: 13
    Likes: 0, Points: 0, Legacy Rep: 10
    Location: Poland

    killwater Junior Member

    I'm certain that you can get flat transom in michlet. Godzila can give you flat transom if you are using 8 series hull functions. It also takes in account pressure drag on transom. In manual you can find that only first row of zeros is required and if you want to make transom like mine you need to have non zero last row. Problem lies particularly in exporting from freeship to michlet: it exports like there is no transom. I hoped that someone know how to fix it either in freeship or in mlt file. I also know that inserting zero row will give me fake results so I'm asking if anyone know the solution :)

    Michal Klimek
     
  7. Guest625101138

    Guest625101138 Previous Member

    The attached shows a sample of a hull with a transom. Like yours the waterplane shows a slight pointy shape. BUT take a look at the bodyplan in the upper left. The transom is clearly present. I can assure you that in the calculation the transom drag is taken into account.

    You should be able to check the aft end of your bodyplan to observe the same thing. If it shows in this view then it will be accounted for in the drag.

    One test would be to trim the bow down and see how the drag reduces at low speed by lifting the transom above the water level and thereby removing the transom drag.

    Rick W
     

    Attached Files:

  8. marshmat
    Joined: Apr 2005
    Posts: 4,127
    Likes: 146, Points: 0, Legacy Rep: 2043
    Location: Ontario

    marshmat Senior Member

    Killwater,

    This problem is quite common in Freeship -> Michlet translation. Hopefully Leo will chime in here with the authoritative explanation.

    The vast majority of my designs have transoms. Many of these begin with a Godzilla "series 8" optimization. In my own runs, and in Leo's examples, there is no requirement for the final row to be zeros. (In fact, if the final row is zeros and the hull actually is supposed to have a transom, the predicted drag will appear substantially higher than it should be.) It took me about 40 tries to catch on to this, wondering why I was getting such enormous variations in drag from miniscule changes to the hull- but in the screenshot Rick posted, what appears to be the transom is actually the second-last station, and what Michlet is seeing as the "true" transom is revealed only by two pixels just off the centreline. I suspect it thinks the water is staying in contact with the hull, ie. making a near 90-degree turn at the second-last station, not breaking away cleanly as it would from a transom. It is my understanding that if the final row is non-zero, Michlet recognizes this as a transom and adapts its calculation accordingly.

    The solution I use is to:
    1) Use lots of stations and waterlines (I generally use 81 of each, which on a Pentium 4 Northwood space-heater is still very quick to compute)
    2) If the final (transom) row of offsets appears as zeros: Copy the second-last row of offsets onto the last row. The error introduced by doing this is very small, as long as your hull sides are not sloping sharply inwards at this point.

    This effect can be demonstrated with the ddg51.mlt example input file. As included in the package, it has a transom, and the transom drag term Rh appears in the resistance plot. If you change the final row to zeros, the Rh term becomes zero, and the total resistance at 20 m/s increases by 398 kN- an error of more than 9%.
     
  9. Guest625101138

    Guest625101138 Previous Member

    Mat
    The attached shows the comparison for only the transom drag with the modification you suggest along with the standard export file. They are vastly different. As you point out the std export produces a much higher value. On the other hand the shape of the std looks more like I would expect.

    I have never bothered to do the integration to see if the transom drag produced with the zeros is close because I mostly work in regimes when no transom is best. My aversion to transoms was supported by the data from Michlet so I may need to rethink. The 18 function version of Michlet/Godzilla will actually optimise to a transom with high speed hulls.

    The particular data I have shown has a beam of 4m and a draft of 0.8m. Transom area is roughly 2sq.m. So the worst case drag with no wave trough would roughly average to 3kPa over 2sq.m given the shape. This is 6kN so a peak value of 4kN would seem in the ball park. A peak value of 360N seems to be way too small for such a large transom. I know pushing a bluff stern of a small dinghy through shallow water (literally) is tough work until you can get it to plane.

    Clearly there is a large difference so it needs checking.

    Rick W
     

    Attached Files:

  10. marshmat
    Joined: Apr 2005
    Posts: 4,127
    Likes: 146, Points: 0, Legacy Rep: 2043
    Location: Ontario

    marshmat Senior Member

    Rick,
    Just checking, are we talking about the same value here?

    In the version of Michlet I use (8.07), the manual gives:
    * Rh is the added resistance caused by there being a transom, with air behind it instead of water
    * Rwtrans is the resistance caused by the transverse wave system

    In the plot you posted, the curves are listed as Rwtrans(Mod) and Rwtrans(Std). Their behaviour is as I would expect for the transverse wave system: A clean breakaway at the transom ("Mod") flattens out the transverse system, while curving the hull sharply in and up between the final station and the stern ("Std") draws the flow up into a large transverse wave.

    (edit) The attached images are what happens when the Michlet offsets are read back into a graphical form (using the DDG51 destroyer example file in this case). Zeros in the final row (left) produce a sternpost; a non-zero final row (right, from Leo's example file) produces a transom. Michell's equations assume that the flow follows the hull; thus, the former case would cause a large transverse wave to be drawn up behind the ship, while in the latter case, Michlet sees the transom and corrects for it. The case with a sternpost has Rh = 0 at all speeds, ie. no transom drag, but its actual Rwtrans will be hard to predict as the real ship would have separation effects that Michell's theory doesn't account for. My understanding is that Michlet just solves Michell's equations for the given surface in this case, even though they may not be applicable for this type of stern. In the case with a transom, Michlet gives a negligible Rh at low speed, increasing with speed until (in this case) 12 m/s at which point the flow is breaking cleanly away from the transom, and Rh is 42.7 kN. This transom is something like 9.8 square metres, so -4.3 kPa on average.

    At lower speeds, hauling a transom through the water will usually give higher drag than a streamlined, pointy stern will. The transom's advantage only comes when you're going fast enough for the transverse wave system to be a major drag component, and the transom's ability to flatten out the transverse wave outweighs the added hydrostatic drag.

    I would love to play with that 18-parameter version of Michlet. (I have the executable for 8.09b, but the Allegro graphics dll is corrupted somehow and it's not on Leo's site anymore.)
     

    Attached Files:

  11. Guest625101138

    Guest625101138 Previous Member

    Mat
    You are right. The graph is of the transverse wave system not the transom drag.

    The transom drag produced by the standard export file is zero.

    Using the modified input file the transom drag rises to 1.5kN at 6m/s so looks reasonable. Obviously this is offset by a much smaller transverse wave system.

    Letting the water leave the hull reduces the energy in the transverse wave system no doubt!!!!

    The only way to fix this is by overwriting the last row with the second last as you previously pointed out. Just making more stations with the last row zero will not help.

    The Series 20 version of Godzilla actually produces transoms in some optimisations and the last row is not zeros. I even discussed the hull shape I got with Leo and his response is post #65 on this thread:
    http://www.boatdesign.net/forums/design-software/michell-v-cfd-efd-round-2007-a-20497-5.html

    I will now see transoms more favourably and if I export one from Delftship make sure I adjust the input file. It is certainly something that needs correcting. I appreciate your input.

    Rick W
     
  12. Guest625101138

    Guest625101138 Previous Member

    Actually it is 20 functions not 18. It takes a lot of fiddling with constraints to arrive at something that can reasonably build. The attached shows optimisation for a stable form using the Series 20. After Leo's feedback on things like sinkage it made me think it is not worth the trouble unless the drag calculations takes this into account.

    Rick W
     

    Attached Files:

  13. marshmat
    Joined: Apr 2005
    Posts: 4,127
    Likes: 146, Points: 0, Legacy Rep: 2043
    Location: Ontario

    marshmat Senior Member

    For interest's sake:
    In case anyone following the discussion doesn't already have a copy, here's J.H. Michell's 1898 paper describing how the equations for wave resistance are derived. The math may be a bit nasty, but for anyone using Michlet, it's interesting to note the assumptions made in deriving the theory.
     

    Attached Files:

  14. killwater
    Joined: Oct 2007
    Posts: 13
    Likes: 0, Points: 0, Legacy Rep: 10
    Location: Poland

    killwater Junior Member

    Sorry for not responding for such a long time :)
    My problem is exactly the same as marshmat wrote. I've been thinking about copying the row before last row to last row position but i was afraid that it will introduce a big error. Unfortunately i cannot see any other solution to this problem so i have to use this one.
    Does anyone know how much (in percents ) it can be?
    You don't need to use 20 series to make optimisation with transom: the 8 series gives a possibility to cut some part of afterbody (in percents). It does not give a great control over shape but after some trials You can get reasonable effects.

    Michal Klimek
     

  15. killwater
    Joined: Oct 2007
    Posts: 13
    Likes: 0, Points: 0, Legacy Rep: 10
    Location: Poland

    killwater Junior Member

    Sorry for not responding for such a long time :)
    My problem is exactly the same as marshmat wrote. I've been thinking about copying the row before last row to last row position but i was afraid that it will introduce a big error. Unfortunately i cannot see any other solution to this problem so i have to use this one.
    Does anyone know how much (in percents ) it can be?
    You don't need to use 20 series to make optimisation with transom: the 8 series gives a possibility to cut some part of afterbody (in percents). It does not give a great control over shape but after some trials You can get reasonable effects.

    Michal Klimek
     
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.