Optimizing performance monitoring system

[Nyman]

First of all, Hello everybody! This is my first thread in this forum.

I am working on optimizing a performance monitoring system. The systems right now is based on daily reporting from ships and is based on; logged speed, draft, wind and sea conditions, fuel consuming, pitch (for ships with CPP) and propeller rpm.

The system calculates a similar condition for a ship with a clean hull and propeller, and the output is a percentage of speed loss due to fouling for hull and propellers.

The problem is that the system is not very precise right now, so i am trying to add some parameters and hopes it will give better results.

Right now i am thinking about adding; water temperature, air temperature and water depth.

Do any of you have any experience in something like this? Any suggestions to other parameters, who haves influence on a ships performance?

Best regards
Mathias

[marshmat]

Hi Mathias,

Firstly, welcome aboard boatdesign.net

What you describe sounds like an interesting idea. No doubt it will not be easy to develop, but the potential payoff (by being able to quantify in economic terms exactly how much money is lost or saved as a result of a change in cleaning intervals, etc.) could be significant.

I suspect you may be confused as to the difference between accuracy and precision. Precision refers to how well a result has been determined, ie. how reproducible it is, without any consideration to its agreement with reality. Accuracy is a measure of how close a result is to the actual value. The distinction is very important in assessing ways of getting a model to better represent reality, even if the distinction is lost on the general public. Building a very precise model is a waste of time if it is not accurate. Likewise, a model may in fact be very accurate, but for various reasons, the estimates of error produced from it may be so large that the lack of precision renders it difficult to use. (Refer to Bevington, P. R. Data Reduction and Error Analysis for further details.)

So let's say you're trying to improve the accuracy of your model. The first thing that come to mind is, how are you calculating the theoretical drag on the ship? Your choice of model here, and what that model includes, could be a big source of error. Your choice of parameters would be dictated by the model chosen, and the amount of effort you put into determining each of those parameters would depend on their relative importance in the model (sensitivity analysis is called for here).

Water temperature, for instance, has little effect on a typical hull drag model. But viscosity- which is highly temperature dependent- has a huge effect. Density is also temperature dependent, but the effect here may be negligible. Water depth may or may not matter, depending on the various Froude numbers for that particular moment in time.

There are people on here who do have expertise in various methods of ship performance simulation. It would be interesting to hear a bit more about what you've done already; there may be someone here who can help.

[Stumble]

I didn't notice these in your list of things you have already considered, so if you haven't corrected for them they might help as well.

Wind direction
Current speed and direction
Wave speed, height and direction
Salinity

Not that I have any great insights, but they could all effect either boat speed directly, or water density.

[Joakim]

Before adding new parameters, you should check how well you are using the current ones. How bad is your accuracy? Is it good even without fouling, waves and wind in deep water? Are you using SOG or speed through water? How do you know the direction and height of the waves? How much actual data do you have for the ship, propeller and motor? Towing tank test based or a universal model for resistance?

Joakim

[Nyman]

Thank for all the replies.

It is an already existing program for performance calculations. The accuracy is not that good at the moment, so I’m thinking about logging the data in a different way, and if it is possible to add some parameters who gives a better accuracy.

Right now it is based on one daily report, where the ship reports;

hours at sea, distance sailed over ground, apparent wind speed & dir, draft fore aft, main engine load, propeller rpm, propeller pitch, and FOC.

The wave height and direction is based on the data from the wind dir and speed. Wave dir = Wind dir, and the height is calculated.

In the program all the necessary hull, propeller, wind load, model test and engine data programmed into a database.


Any of you these kinds of calculations? I think the hull resistance is calculated by holtrop and Mennen and tuned in by sea trials etc.

I'm thinking about making an "automatic" data collector module, that’s log the data by a given interval.

Don’t you think it would be more accurate if you got more incoming data?


/Nyman

[Joakim]

Your data did not report speed through water, which should definitely be used instead of speed over ground, if there is any possibility for currents. Also using daily averages sounds not very accurate.

You can not (accurately) know wave height and direction based on wind measurement, since these depend on wind history, wind far away and fetch.

I think you should start by looking different well defined conditions, like no wind, waves, current and constant speed. If you get that accurately take the next step. If your basic models are not accurate, the extra data just adds confusion not accuracy.

So take e.g. one hour constant data at different speeds and compare to your model.

I have not done exactly these, but I have developed a VPP program for sailing boats and for planning boats (Savitsky) and I have calculated propeller thrusts. My VPP-programs also take into account surface roughnesses.

Joakim