It is obvious that by
repeating the measurements the random errors can be reduced. One single test
point will contain a larger random error than the average of a number of test
results. The aspect of removing random errors is the very reason to have a minimum
length range in the towing tank to measure the parameters and to utilize this experiment.
Therefore, it is worthwhile to select a reasonable number of data points per experiment
and to incorporate some checks to ensure that those data points are consistent and
reliable.
It is much more difficult to dispose of
systematic errors than to get rid of the random errors. Within the same test
setup or with the same computational procedures they do not vanish. The
repeatability of an experiment does not indicate that systematic errors are
absent. It just tells that random errors are small. The only manner to dispose of
the systematic errors is to carry out the experiments at a different
establishment with same models at the same conditions. In practice, this is
never or only seldom done.
Usually the level of
some of the parameters measured is verified by checking against calculations
and statistical standards. The standard deviation of the power correlation
coefficient is composed of 3 independent contributions:
• Inaccuracy of the
model experiment;
• Inaccuracy of the
extrapolation process, and
• Inaccuracy of the
full-scale experiment.
The last contribution,
which is in fact the uncertainty in the full-scale verification, gives the
largest contribution to the natural variation of the power correlation
coefficient (Holtrop, 2001).
The 3 sources of uncertainty in the
correlation allowance factors can be further subdivided. The uncertainty in
model experiments, the inaccuracy of the extrapolation and the uncertainty of
the full-scale measurements all contribute to the variation in the correlation
coefficients.
-
Ref.: Muslim, Mazlan, Powering and
Resistance of Fishing Vessel, Lambert Academic Publishing, Germany, 2014.
Available online from www.amazon.com and www.morebooks.de,
also from www.lap-publishing.com.
Comments