Michlet

Discussion in 'Software' started by pavel915, Dec 8, 2009.

  1. patc
    Joined: Jan 2010
    Posts: 20
    Likes: 2, Points: 0, Legacy Rep: 36
    Location: Area65

    patc Junior Member

    PS I added 10% error bars -- since apparently experiments are so unreliable and the data is not to be trusted (despite the fact that the Wolfson Unit had been using the same tank, dynamometer and data acquisition system for may years with considerable success -- see all the AC and Whitbread work they did, amongst other stuff) and theoretical models are 100% reliable, even if contradicted by what we observe in the real world ;-)
     
  2. Guest625101138

    Guest625101138 Previous Member


    To produce the error band I used the measured drag, added 0.2N to it for the upper range and subtracted 0.2N from it for the lower range. This provides the error band for the measured data.

    I subtracted the viscous drag as a force from the experimental range and the Michlet prediction. I then reduced the resulting wave resistance component of the force to coefficients.

    The attached tables has the two sets of data being the raw forces in N and the wave resistance coefficient so determined with the experimental data as an upper and lower band.

    The main point to note from the table is how a very small error in force measurement will result in an extremely wide error band at the low speed end.

    At higher speed any error in force measurement becomes less significant and the experimental error range narrows into the Michlet prediction.

    Rick W
     

    Attached Files:

    1 person likes this.
  3. patc
    Joined: Jan 2010
    Posts: 20
    Likes: 2, Points: 0, Legacy Rep: 36
    Location: Area65

    patc Junior Member

    but what is "Michlet less viscous" -- it doesnt look like (Rt-Rv) / (0.5 rho S v^2)
    given that Rv = Rf since you not seem to be applying a form factor.
    If you sent over the spreadsheet it would be much easier to see how you are getting the numbers
     
  4. Guest625101138

    Guest625101138 Previous Member

    The data in the table is what the Michlet in_npl.mlt file you sent me produced. I ran it without any changes.

    File with force error band on experimental data attached.

    Rick W
     

    Attached Files:

  5. Leo Lazauskas
    Joined: Jan 2002
    Posts: 2,696
    Likes: 150, Points: 63, Legacy Rep: 2229
    Location: Adelaide, South Australia

    Leo Lazauskas Senior Member

    Rt-Rv = Rwd + Rwt + Rh

    where
    Rwd = diverging wave resistance
    Rwt = transverse ditto
    Rh = hydrostatic drag.

    I reproduced Rick's Michlet results for the total resistance and its components he showed in a previous post, so at least he and I now agree on that much. :)

    I think the comparisons of the components of the total resistance will depend on whether sinkage and trim are included in the calculations, your form factor, and how hydrostatic resistance is handled by our two codes.

    Your Rw depends on the value of the form factor, k=0.3 for the NPL4a.
    However, this factor was estimated by putting the hull in a bow-down position so the transom stern was not immersed.

    Your Report 71 (Fig. 72c) shows that for the beamier NPL4b hull, 1+k=1.55 if the transom is immersed, and 1+k=1.37 if the transom is not immersed. That's quite a big difference. When the hull squats at higher Froude numbers, the transom is immersed even more than for the Froude numbers used to estimate the form factor, so there is more than a little doubt over the exact value that should be used. Or if one should be used at all, of course.

    In the Couser et al paper, p. 5, you state that "the hydrostatic correction has been linearly reduced from its full value at Fn=0.5 to zero at Fn=0.1."
    Is this component used in Hullspeed when estimating the total resistance?
    I don't like it, but Michlet uses a partly wet, partly wet transom at low Froude numbers. The actual extent of the wettedness depends on a transom depth Froude number. To be honest, I have literally dozens of different ways this component can be estimated. All are hacks, even if they depend on some theoretical results due to Tuck and Vandenbroeck.

    I'll try to knock out some comparisons in the next couple of days including some predictions of the sinkage and trim.

    Of course there are also a whole raft of issues to do with Froude's hypothesis. Including viscous damping in Michell's integral (the "Michell-Lamb integral") opens up another can of worms. It does a pretty good job of smoothing out the humps and hollows in the Cw curve at low Fn, but choosing an appropriate eddy viscosity is tricky. But then so is choosing a form factor, picking the length of the virtual appendage, and coming up with a way to reduce the hydrostatic resistance from zero to full. :)

    All the best,
    Leo.
     
    Last edited: Jan 14, 2010
  6. Leo Lazauskas
    Joined: Jan 2002
    Posts: 2,696
    Likes: 150, Points: 63, Legacy Rep: 2229
    Location: Adelaide, South Australia

    Leo Lazauskas Senior Member

    See, we're in total agreement.
    Theoretical models are indeed 100% reliable when augmented by fudge factors that are 100% reliable!

    Have a good le weekend!
    Leo.
     
  7. patc
    Joined: Jan 2010
    Posts: 20
    Likes: 2, Points: 0, Legacy Rep: 36
    Location: Area65

    patc Junior Member

    Oh I now see where the differences lie -- you are dividing by running WSA -- all the coefficients Ive published are based on static WSA, which for this model is: 0.338m^2 and the Fn is based on a fixed length of 1.6m
    BTW how do you calculate running WSA are you calculating a near-field wave pattern right up against the hull? -- I would suggest that thats pushing the limits of a linear thin-ship threory? If its a quasi-static WSA based on trim and sinkage, isnt that a bit pointless -- as it makes the results even more sensitive to having the trim and sinkage already calculated (from experiment for example)?

    I think all the discussion is pretty pointless -- the only difference between Hullspeed and Michlet is the way the transom is dealt with, either hydrostatic correction or some fictitious closure by extending the hull in some way. Both moth methods have their merits, neither represent whats actually happening -- they are both approximations, maybe with different assumptions, or at least different ways of treating the problem. As I already said my work is focused on high speed -- where transoms are beneficial; Rick's work is low speed so he doesnt have a transom (but I cannot see how you can justify a hydrostatic-based transom correction when the transom is wet?)
     
  8. patc
    Joined: Jan 2010
    Posts: 20
    Likes: 2, Points: 0, Legacy Rep: 36
    Location: Area65

    patc Junior Member

    At the end of the day we need to provide practical solutions to ship designers otherwise we wouldnt get funded. The form factor is a nothing more than a bodge factor, but at the moment its the best weve got and with experience you can pick one that allows you to correlate model to ship.
    TTFN,
    pat
     
  9. Guest625101138

    Guest625101138 Previous Member

    Pat
    Simply putting a 10% error band on the derived value of Cw from the experimental data does not make much sense. You have even stated that if you were interested in low Froude drag you would have used larger models.

    Preferably you should consider the instrument and reading error for each individual component that is used to derive the coefficient and stack them so they all add for the upper bound and all subtract for the lower bound. This will give a more realistic range of error in the experimental value.

    In the error range I charted, I only considered force measurement but the coefficient also has velocity included so you should factor this error range as well. I am assuming other variables such as density and temperature were controlled to a range where their variation would not be significant.

    In manufacturing I have seen some very expensive warranty claims where machine tolerances resulted in loose fits that should have been interference fits. An improbable outcome but it is possible to get all the ducks in a row.

    I have learnt from my experience in a few industries that analytical models are powerful tools as the basis for development. They do not need to be perfect but they do need to be based on basic physics to have broad application. Measured performance is used to validate the model. You can then play with the model to gain better understanding and try things out. Sometimes this leads to new insights. Michlet has proven valuable in my understanding of hull shapes.

    I have also learnt that getting reliable field data is a challenging task. There are so many variables that can come into play that useful information is buried. I have good knowledge of measurement error as an electrical engineer in a range of industries. My worst experience was with an air flow meter instrument error costing AUD60M (1999 terms) over a period of 6 months. I knew something was amiss when the field data was not matching my plant model. So you might appreciate my skepticism about measurement accuracy.

    When we were testing the V11 hull in Calgary we zeroed the SRM power meter for every test and the span was recalibrated every week or so. Some of the velocity measurements proved to be flawed because we assumed a small lake would have no current - bad assumption.

    Rick W
     
  10. Leo Lazauskas
    Joined: Jan 2002
    Posts: 2,696
    Likes: 150, Points: 63, Legacy Rep: 2229
    Location: Adelaide, South Australia

    Leo Lazauskas Senior Member

    The wetted area is used to estimate Rf. People can choose to use the static or the "running" static WSA to non-dimensionalise. It's not of much ccncern to me. I am not really interested in extrapolating up from model-size.

    The static WSA, if it is indeed 0.338m^2 for the NPL4a is wrong in Couser et al and Report 71 where it is given as 0.348 m^2. I don't think that either value was used in any actual calculations in your spreadsheet.

    I agree completely that each approach has its merits and that they essentially differ in their approach to approximations of the transom stern flow, none of which are correct. We are also both trying to estimate skin-friction at Rn where the ITTC line (and my preferences) are not very reliable. Hence my suspicion of the use of small models.

    The original point of all this is to explain to Daiquiri what is happening on the Hullspeed page. The answer to one question is that the result for Michlet is not really comparable to the Hullspeed definitions of Cwp and Cw in the graph for the NPL model. Michlet values in that graph include the hydrostatic resistance.

    In Michlet there is no hydrostatic resistance correction for the wet part, only for the dry portion.
    Indeed it can be sensitive to the hull attitude and it can be worse if experimental values of squat are used as input. Hence my reference to the issue that was raised at the IWWWFB.

    Yes it does push linear theory, and nothing works consistently. Lawry Doctors, Bob Beck and Kevin Maki conducted yet another extensive research program into transom stern flows a couple of years ago. Interesting as it was, I don't think that anything earth-shattering came out of it. Similarly, earlier work by Doctors which looked at measuring the length of the hollow behind the dry transom was not all that successful.

    In Michlet the user can place the hull in a sunk and trimmed attitude and the WSA is used to estimate total skin-friction. No wave effects. Or the user can specify that the hull remains in the static attitude and use the static WSA.

    In other codes I have I do calculate the near-field. Again, users have a variety of choices when it comes to wetted surface area, how to model the transom flow, hydrostatic resistance, squat, friction line, viscous damping etc.

    I appreciate that you are interested in the high-speed region. Me too, hence my interest in estimating dynamic sinkage and trim. But for some applications I am interested in, such as SES that have to travel off-cushion in harbors and some other areas, the wet regime is also important.

    Thanks for your input. I also appreciate that you have other work to do :)

    Cheers,
    Leo.
     
  11. Leo Lazauskas
    Joined: Jan 2002
    Posts: 2,696
    Likes: 150, Points: 63, Legacy Rep: 2229
    Location: Adelaide, South Australia

    Leo Lazauskas Senior Member

    It's the best we will ever have if nobody tries other methods and approximations :)
     
  12. Guest625101138

    Guest625101138 Previous Member

    Pat
    I disagree that it is pointless.

    It has highlighted to me how small measurement errors can be dramatically amplified when reduced to coefficients - I have learnt something here and it differs to my initial thoughts on the curves. This is a good lesson for anyone looking at experimental data believing it to be the holy grail.

    If the measurement errors dwarf the measurement then it is not much value. You are trying to decipher something useful from noise - it is something I have done but not recommended. The experiment needs to be set up in such a way that the measurement error is not significant to the comparison being made. Given the error band in your data from force alone, having curves plotted below Fn of 0.4 is pointless. I have not considered what a 1% error of range in velocity measurement would do to the coefficient.

    Rick W
     
  13. patc
    Joined: Jan 2010
    Posts: 20
    Likes: 2, Points: 0, Legacy Rep: 36
    Location: Area65

    patc Junior Member

    hi Leo -- just wondering, when you are dealing with calculation of the wave pattern, how do you deal with transom closure? Or do you leave it open to infinity behind the transom?
     

  14. Leo Lazauskas
    Joined: Jan 2002
    Posts: 2,696
    Likes: 150, Points: 63, Legacy Rep: 2229
    Location: Adelaide, South Australia

    Leo Lazauskas Senior Member

    For large transoms, I use a series involving terms arising from the integration by parts of the P,Q functions.

    I use an open wake model for the portion of the boundary layer trailing behind the vessel - the "leak" through the boundary at infinity is too small to worry about.

    Leo.
     
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.