Accuracy of Holtrop's method.

Hi,

Since Sunday I've been wondering how accurate are power prediction methods (Holtrop's method for example) in practise. Any ideas? Is it more like 1%, 5% or 10% accuracy?

Best regards,
Marius.

 

Any power prediction method is really what its name suggests, a prediction, not a calculation which purports to be exact or very rough.
In even the predictions made in tank test should be contrasted with the "test of the mile." Made this test and extrapolations of various parameters, that almost no one understands, it is said that the ship, which in the conditions of the test has achieved the speed she has achieved, in the theoretical conditions of the prediction would reach the intended speed.
What I mean by this?, Which, in my opinion, no one would venture to say what degree of accuracy is a prediction because, among other things, you do not know what will be the actual conditions at the time of testing.

 

But then again, one should be able to validate the prediction after tests on the mile right? That actually confirms that the prediction method can be trusted. The basin model tests and resistance prediction methods are usually done in a way that enables their results to be used!

I didn't ask for exact values, but the observations of designers that did the prediction and then checked it with for example basin model tests.

Best regards,
Marius.

 

I would say not higher than 10% for regression methods such as Holtrop and 5% for systematic series, if hull shape of design is close to parent hull shape of series.

Of course this works if the parameters of craft are within parameters of method/series.

 

As I was told once by a wise old NA who wrote a book on power predictions:

"Always overpredict (power required for contract speed) by 10-15%. If you are over speed by 0.5 knots they are buying you drinks, if you are under by 0.01 knots you are in court."

FWIW, correleation between power prediction and and at sea testing is never better than 5% and often worse than 10% even in the best conditions. So no matter what the prediction method, if you are looking for a given speed, you need add something for what is called "sea margin". This takes into account water temperture, winds, sea, fuel heating value, stern tube alignment, how the Chief is feeling that day, etc....

 

What hullform ? That's important. Holltrop is not very good at all for some types and as Alik said series are a better indicator if they match your parameters.

But even then I've had significant differences 10% between tank testing and Holtrop and I always do what JEH suggests above

 

Power margins is know-how of the designer and is something layman never considers. Moreover, I often see the situation where our prediction give higher power demand just because we have put all necessary power margins in our calculations, those are compared with someone else's who is just crunching the numbers in the computer resulting lower and more attractive figures for customer ("oops, I forgot to add air drag!"). And they guy might not even have realistic weight predictions as well!

Power margins we usually take are:
- resistance on waves, aero drag, appendages - calculated directly;
- margin for inaccuracy of prediction method - usually 5% for conventional hull shapes, and higher for uncommon shapes;
- engine derating factor for higher air temperatures, humidity, fuel quality, etc. We have often have our prediction compared vs. performance of similar boat somewhere in Europe, that does not consider the fact that de-rating factor in tropics is 10% for some engine models.

And of course correct prediction of performance needs fair comparison. We often see advertised numbers unrealistically high and probably 'jumps' of speed on hand GPS are reported, specially for pleasure craft and by amateur designers. In reality such craft does 3kts less if speed is measured in common way during sea trails.

 

This is an extract from a short article I have on my FAQs page regarding performance predictions (written some time ago - as you can tell)

"Purely theoretical /computed tests are even more of a guess. There are far too many assumptions that have to be made. The monohull race boat rule, the IMS is a good example of such a "theoretical" rule.

I'm sure I'm not the only one who has a background of dinghy sailing. Thousands of dinghies race thousands of handicap races every year. So we "know" a Laser 1 rates 1077 and a Laser 2 1035 (using the UK's RYA data from 2000). That's what, 4% faster? Even my mother could look at the two boats sailing and say the Laser 2 is the faster boat. After all it is longer, and has a trapeze and spinnaker.

I believe the IMS is supposed to be accurate to +/-2% and people are happy with the results because the formulae look complex and results can only be calculated by computer so therefore must be right! Yet if a Laser 1 speed was calculable under IMS it could be 1050, say, ie within 2% of the real value. Similarly a Laser 2 could work out at 1050, also within 2%. So the super accurate, everyone is happy with the results, IMS would make the Laser 1 and Laser 2 the same speed!! As I said a second ago, my mother is a better handicapper than that.

Compare the complex measurements needed, the small speed range that monohulls use and the type forming of the IMS rule and I think you can see that no completely theoretical calculation can be much better than a guess dressed up by computers to convince the uninitiated."

Apart from that, I agree with Alik. Do you quote a performance when the engines are new, or one that reflects an aging engine. Friends with a 28ft powercat (not my design) reported that when first launched the boat would do 22 knots, 5 years later it would only do 16, even when empty and a clean bottom

Do you assume a sailor has laminate sails, or dacron ones? We know there is at least a 2% difference in speed between the two sail cloths. Never mind making allowances for humidity and air temperature (again as Alik indicates). I recall that back in the IOR days the American SORC boats found they had very different speeds when they sailed in the wet and windy UK

Richard Woods of Woods Designs

www.sailingcatamarans.com

 

One can never have too many margins.... I still take off 1 knot...even with tank test results.

 

Yep, and if you deal with "go fast" boats for people in "the small package trade" missing speed can be deadly.

 

The accuracy of Holtrop statistical analysis method is known in MODEL test to be +-4% in Ctl, equivalent to +- 1% in speed.

Calculations of inputs are very complex and errors will creep in if not too careful. The method uses three dimensional resistance analysis, statistical wetted surface area calculations, and three range of speed for wavemaking resistance.

If you want more accuracy, Lin, Day & Lin new measures of prediction error are given in comparison with methods of Holtrop.

Leo probably has this in his library but as AH says, he is "deep in thoughts" at the moment. http://www.boatdesign.net/forums/design-software/michlyt-demo-46909-2.html#post627950

DMacPherson also has a new study on the series 64 hull but have not finished it yet. Same link.

 

Depending on type of hull; it can be lower at humps speeds specially.

 

Yes it seems. Looking at the second hump at Fn 0.298, it seems a steep curve for its cycle.

 

So long as the hull falls within:
Cp .55 to .85
L/B 3.9 to 15
B/T 2.1 to 4.0
FnL 0 to .8
accuracy is good.