Michael Schmiechen, Berlin:
Towards a rational evaluation of ship speed trials
Abstract
The present paper is a summary of recent work undertaken by the author to
promote a more rational evaluation of ship speed trials. In a series of
sample evaluations it has been shown that the consistent application of
systems identification techniques provides trustworthy results with a
minimum of plausible and acceptable conventions and without reference to
model test results, as it should be.
The paper contains direct links to all relevant material being published
on the website of the author and to be found as well in his
Bibliography on Propulsion
and under his Recent papers
in the section 'On the evaluation of ship speed trials'.
An earlier version of this summary, already including the reduction to the
no wind condition was the note on
Further evaluation.
The data and the evaluation of example 21010,
the latter including the the first attempt at a reduction
to the no wind condition, have been removed as obsolete
from the website at a rather early stage of the development,
but have of course been archived for purposes of research.
A new example has been added 1999.06.24:
05010_data_orig,
05010_eval_rat.
The examples may still contain errors due to misinterpretations of
traditional sign conventions!
Introduction
The traditional way of conducting and evaluating ship speed trials is very
costly and involved and at the same time the results are not very
trustworthy. The reason for this situation is that the logics behind the
whole procedure is not very clear to say the least.
Consequently an attempt has been made to promote the necessary
clarification and perform a systematic reanalysis of trials data made
available to the author up to date. The ideas developed although in
principle simple and conventional, but not traditional are certainly not
readily acceptable for the community concerned, but maybe they can form the
nucleus of a discussion, and the economic advantages will finally lead to
their adoption.
In a proposal for a clear-cut procedure, prepared as a contribution to the
current ISO/WD 15016, it as been shown how the power functions of ship
propellers in the behind conditions can be identified together with the
current velocities with systems identification techniques from a minimum
of test runs, and with a minimum of conventions and without reference to
model test results, as it should be.
The approach is based on the theory of conflict resolution, on simple
models of hull-propeller interaction and current velocity as well as the
basic facts of systems identification in noisy feed back loops. This is
the approach systems engineers, knowing little or nothing about ship
theory, would take. It is not only more adequate, but much more convincing
and trustworthy than the traditional approach, avoiding unnecessary and
maybe even irrelevant diversions. The sketchy style in the format of the
layout of the standard serves to stress the point and needs of course to
be changed, if the ideas are followed up.
The work is related to experience gained in applications of the rational
theory of ship hull-propeller interaction during full scale tests with
the German research vessel METEOR and the Blohm + Voss experimental SES
CORSAIR. Material concerning these tests and their evaluation has been
published on the website of the author as well and is to be found in his
Bibliography on Propulsion
and under the Recent papers
in the sections 'On the propulsion tests with METEOR, Proceedings 2nd
INTERACTION Berlin '91' and 'On the propulsion tests with CORSAIR'.
Motivation
The stimulus for the recent activities of the author has been the Japanese
ISO Committee Draft (CD) 15016. The first reaction of the author was a
different Draft proposal together with
Cover letter and two examples,
Data of example 1,
Evaluation of example 1,
Data of example 2,
Evaluation of example 2.
The reservations concerning ISO/CD 15016 were, and are even stronger now,
that an ISO standard should not just continue to refine past practice,
but should meet the highest 'standards' and take advantage of the latest
state of the art and technology of systems identification, not only in view
of the legal implications, but the requirements of ISO 9001 as well.
And the standard must be the result of a joint effort of the whole
community concerned. In the interest of the profession, science and
technology, and the costumers, yards and owners, a serious discussion not
only of the details, but of the fundamentals in the first place, is
strongly suggested. Naval architects need to take the discomfort of the
industry, they are serving, very serious and come themselves up with
adequate solutions before outsiders or industry tell them what they better
should do or should do better.
Principles
The principles proposed are to keep the models as simple and the method as
transparent as possible in order to make the results as truthworthy as
possible. Consequently the rule adhered to is to keep separate problems
separate as far as possible and keep the exposition as simple as possible.
The resulting procedure is:
- 1. identification of the current and the powering performance in the
behind condition at the given load condition from measured data only,
- 2. comparison with the predicted performance at that condition,
- 3. comparison with the contracted performance at that condition,
and, for the purposes of the study started to convince the community
concerned,
- 4. comparison with the final results of traditional evaluations,
whichever has been used by the yard or institution supplying the data.
The first task has been solved in terms of a powering model and a current
model as shown already in the draft proposal and in the accompanying
examples and in the many new examples, last update and additions 1999.02.09:
- 01010_data_orig,
01010_data_corr,
01010_eval_rat
- 01020_data_orig,
01020_data_corr,
01020_eval_rat
- 03010_data_orig,
03010_data_corr,
03010_eval_rat
- 21020_data_orig,
21020_data_corr,
21020_eval_rat
- 21030_data_orig,
21030_eval_rat
- 21031_data_orig,
21031_eval_rat
- 22010_data_orig,
22010_eval_rat
under the
Recent papers in the section 'On the evaluation
of ship speed trials'.
As mentioned in other places the data and the evaluation of example 21010,
the latter including the the first attempt at a reduction
to the no wind condition, have been removed as obsolete
from the website at a rather early stage of the development,
but have of course been archived for purposes of research.
The usual, rather involved iterative solution of a problem with at least
five unknows is replaced by the straightforward solution of a system of
linear equations. The second and third tasks could not be performed in the
examples due to lack of data. Task 4 has been performed as far as data
have been made available.
One general observation concerns the choice of the power to be used in the
evaluation. In proving the conformance with contract conditions, the shaft
power measured, the brake power, appears to be the most adequate reference,
as it does not require further assumptions on top of those necessary
anyway, i. e. for the strain gauge measurements in the usual absence of
calibrations proper.
Evaluation
The advantage of the procedure advocated is that a minimum of assumptions,
i. e. conventions to be agreed upon, are necessary. No references to
model or resistance data are necessary. Further, even with moderate
deviations from the contract conditions the powering performance does
not change, provided the submergence of the propeller does not change.
Changes of salinity can be accounted for computationally.
If the submergence of the propeller changes with load conditions,
especially from fully to partially submerged, trials have to be conducted
at all relevant conditions anyway. Subsequently the powering performance
can to identified separately at all these conditions.
The objection to this procedure was and is that the traditional procedure
goes further and establishes powering performance at certain service
conditions, e. g. at the no wind condition. By doing so, a Pandora's box
of problems is being opened. But if one wants to do this, for whatever
reason, one should look for a solution following the principles stated.
After having reached this point it may be rightly suspected that any
educated guess, i. e. systems identification of the wind effects and
reduction to the no wind condition, might be more convincing and consistent
than the traditional procedures followed so far. Consequently, as a first
attempt, a very simple procedure has been developed, the first results
published in the example 21010, later considered as obsolete for various reasons
and consequently removed from the website, perfectly agreeing with those
obtained by the yard following the traditional procedure!
The model used in the program is a linear to cubic interpolation of the
power required due to water resistance and a cubic interpolation of the
power required due to wind resistance. This model is of course open for
discussion as are the extensions to account for other deviations from
contract conditions. In order to identify these effects correctly the
conduct of the trials will have to be changed as has been proposed earlier
by the present author since 1980; see his
Bibliography on Propulsion.
The idea can be extended to phenomena as e. g. changes of trim etc, as soon
as the corresponding changes are being performed during the trials and the
parameters are available for purposes of correlation. And in due course the
effects of waves, shallow water etc will have to be considered consistently
with the procedure developed sofar.
Of course this will lead in many cases to the use of data, which have been
used up to now as well. But the use may be different. In that sense the
whole exercise is to be considered as a necessary rationalisation of the
traditional procedure. The former appeal is repeated: naval architects
should do this better themselves before other people tell them what to do
better.
Request for data
In the meantime, on occasion of the ONR Symposium in Washington 1998,
the fundamentals and the first two examples published earlier have been
explained to a group of colleagues. In a paper
On the Logics with
presentation, prepared for that purpose, a more
detailed exposition is to be found.
One problem in carrying out the systematic reanalysis of trials data,
which aims at the comparison of the results of traditional evaluations
with those of the rational evaluation proposed, is that many of the data
sets provided are incomplete.
Consequently more and complete data sets in a format discussed below
have been asked for and as well as the permit to publish the results on
the website of the author in the format of the examples;
see the request for data
to colleagues and institutions.
The data made available so far are very different not only in extent, but
in layout, including the proper identification of the data. Evidently
there is a very strong demand for a standard concerning this format,
long before any standards concerning the execution and evaluation of full
scale trials and model tests. Such rigorous standards, i. e. conventions,
will be necessary as well due to the fact that in the presence of noise,
the results strongly depend on the procedure.
The differences in the presentation of model test results has in fact been
the reason for the formation of the former, famous ITTC Presentation
Committee, the name of which has been changing through Information
Committee to, presently, Symbols and Terminology Group.
'Some Fundamental Considerations Concerning the History and Recent
Development of the ITTC SaT List, the International Towing Tank
Conference Symbols and Terminology List' can be found in a
paper, which has been
presented
at the First International Conference on Maritime Terminology.
The current ITTC SaT List can be downloaded from the
ITTC Website.
The uniformity of presentation is required more than ever before for
the purposes of quality assurance according to ISO 9001 and for the
purposes of product data model technology (PDT) aiming at the exchange
of data in neutral formats, e. g. the ISO/STEP format. Evidently this
format is not concerned with neither the generation nor the use of the
data.
A a later stage the format will have to be linked with the ISO/STEP
Shipbuilding activities and the emerging Application Protocol Ship
Hydromechanics. But before this the fundamental questions associated with
the change in paradigm promoted have to be solved with a minimum of
overhead and as convincingly as possible. The environment of Mathcad is
considered to provide a particularly intuitive access to problems and
solutions.
Proposed format
From the exercises in conjunction with the creation of the rational theory
of ship hull-propeller interaction and for the purposes of the present
study the following appear to be the minimum data necessary:
- Identification
The code presently used for the data consists of five digits: two digits
for a given yard or other institution, two digits for a given ship, one
digit for a given load condition or other modifications concerning the
same ship.
Each yard or institution, which has provided data, has obtained the
identifications of its data. New data can then be coded by the yard or
institution itself in order to secure the privacy of the data.
- Measured values
Displacement, density of water, propeller diameter, time, modulus and
direction of the speed over ground or equivalent, course of the ship,
rate of revolution, shaft power, modulus and direction of the relative
wind speed or equivalent.
- Reduced values
Current speed over ground in ship direction, ship speed in ship direction
relative to the water, shaft rate of revolution, shaft power at no wind and
no wave condition.
According to one or various methods the results of each to be listed
separately and identified. The traditional methods of analysis are
evidently different from institution to institution.
Very desirable data are:
- Contracted values
Displacement, density of water, propeller diameter, ship speed relative
to the water, shaft rate of revolution, shaft power at no wind and no
wave condition.
- Predicted values
Displacement, density of water, ship speed relative to the water, shaft
rate of revolution, shaft power at no wind and no wave condition.
According to one or various methods: based on past experience, on model
test results and/or on computational results the results of each method
to be listed and identified separately. The traditional methods of
prediction are different from institution to institution, and, accordingly,
will provide different results based on the same measured input.
In the present study predicted values have not been made available and
contracted values only on a very limited scale. One observation in
analysing the data is that very often the quality of the data is very
'poor', i. e. that they exhibit systematic errors. This would not matter
if this would happen randomly, i. e. if the samples would be large enough
for a statistical treatment proper. In case of the small samples available
other methods have to be relied upon. In fact some yards and institutions
appear to use normalized data in the way proposed to scrutinize the data.
Conclusions
According to the comparisons available so far the current velocities and
the powering characteristics in the behind condition can be identified
without any reference to resistance and model data. The results are in all
cases very close to the values obtained by traditional methods. And the
technique proposed permits to establish transparently the conformance with
conditions predicted and contracted on the basis of model tests or
determined by other procedures as e. g. the traditional method and detect
inconsistencies in the latter.
Looking forward to the serious consideration of these remarks as well as
the elaborations and the numerous examples provided, and maybe some
response the author remains with best regards,
Michael Schmiechen.
© 1999 Michael Schmiechen, Berlin
Last update and additions: 1999.06.28