tank capacity

[widelto]

If your plc is a compact/control logix you can use a FXgen instruction with all the points you mention, no equations.

 

[balash]

balash it is the most used system on ships as the tanks are very difficult positioned and HFO does not like to have a capacity or ultrasound in it.

hmm... interesting

but if there is a tank table book then i assume there is no certain formula V=f(h) i.e. V=f(p) approx. excel method could work with enough points...

 

[jvdcande]

Because all available space is used as much as possible the tanks have the weirdest shapes. Furthermore you have to take into account that the ship is almost never levelled out, not even if the sea itself is. The calculations to obtain the correct values for different angles of pitch and roll are done once and placed in the tables (yaw doesn't play that much a role). When sounding a tank, the hight of the tank level and pitch and roll at the moment of sounding are taken into account when applicable and/or available and the tank volume is read.

When the ship builder does the calculations for the tables he can use the drawing information and the finite element set they used to also calculate e.g. the various stressing forces on the ship. Normally this is done on a dedicated computersystem which they don't have aboard the ship itself.

Kind regards,

 

[shooter]

yaw does not play a role unless the ship is turning rapid like a centrifuge hihi.
and yes the tables are made with the autocad files and then fill it virtually. this results in a book or yes an excel sheets.

and balash if you send the tank data i can make your book, or the excel sheets. their are also programs for this.

 

[balash]

and balash if you send the tank data i can make your book, or the excel sheets. their are also programs for this.

you meant siggibald as he need it now...
but thanks, you never know where i may end up yet... (now when we are in EU )

 

[Lancie1]

PS: excel gives: y = 0,523x4 + 0,9221x3 + 4,5827x2 + 9,0937x + 0,1318 or less accurate but linear y = 14,286x + 1,8047
where y is volume, and x is height

Balash, I calculated the Y points using your Excel results, and was amazed at how far off they are, with errors ranging from -3% to -2203%.

I tried entering the points into an Excel trendline and got this 3rd-order equation:
y = -0.151665x^3 + 1.123842x^2 + 13.680117x

R2 = 0.998979 which indicates an overall accuracy of .998
The largest point error was -22% for point 2, followed by -4% for Point 3, with the remaining points +/- 2% accurate.

The accuracy can be improved slightly from R^2 = .998 to 0.999 by going to a 4th order equation: y = 0.051298x^4 - 0.905031x^3 + 4.488813x^2 + 9.307259x
R2 = 0.999849

 

[balash]

Balash, I calculated the Y points using your Excel results, and was amazed at how far off they are, with errors ranging from -3% to -2203%.

ahahahaha, of course you are, (i was too when checking now) when i made a sign error writing it here, there should be minus before 3rd potention:

y = 0,0523x^4 - 0,9221x^3 + 4,5872x^2 + 9,0937x + 0,1318

with R=0,9999

 

[Peter Nachtwey]

You guys aren't approaching this right. Has anybody plotted the data? If so you would see that there are two linear sections to the data. It looks like the tank has one surface area from 0 to 4.5m and a smaller surface area from 4.5 to 7.5m.

It look like the volume as a function of level can be approximated best by two linear equations. There is no way one polynomial is going to fit the data well.

 

[balash]

well guessing is just that - guessing. your way is equally good, i consider it too.
best way would be to do what jvdcande said as it is done by manufacturer who knows the ship best.

 

[Lancie1]

You guys aren't approaching this right. Has anybody plotted the data? If so you would see that there are two linear sections to the data.

Yes I plotted the data and saw that the shape is irregular. I suspect it is shaped like a pottery jug, with a smaller bottom, flaring out to a larger diameter, than tapering off toward the neck.

My only point was that he should not use Balash's first equation, because it was 2200% off at the high end, and some ship might get in the middle of the Atlantic ocean and run out of water.

 

[Peter Nachtwey]

Yes I plotted the data and saw that the shape is irregular. I suspect it is shaped like a pottery jug, with a smaller bottom, flaring out to a larger diameter, than tapering off toward the neck.

You got that backwards. The tank is narrower at the top. If you divide the change in volume by the change in level you get the average surface area of the tank between those two levels.

My only point was that he should not use Balash's first equation, because it was 2200% off at the high end, and some ship might get in the middle of the Atlantic ocean and run out of water.

He would be better off using two second order equations or simply do linear interpolations between the points.

The point I have tried to get across is that it is rarely best to simply fit a fourth order polynomial to data. Get information about the object being modeled or plot the data. One can usually tell by inflection points what order of polynomial will work.

 

[Lancie1]

You got that backwards. The tank is narrower at the top.

Sorry, I meant that "tapering off toward the neck" in hillbilly language means it gets narrow at the top just like those old jugs we used for moonshine! Sorry for my slang!

 

[NetNathan]

Isn't it just a log scale conversion?
The pressure signal output appears to be "raw" log...
I have run not this a lot with vacuum sensors and gauges.
In graph below:
X= ma signal output (min - max of meter output)
Log(x) = pressure (min - max of sensor range)

 

[balash]

what it definitely is not - is not a log function. in it's basics: p=ro*g*h. if ro and g are constants it will depend on height but because of weird shape of the tank that height is not proportional with pressure.
as mA signal output - it is proportional to pressure as pressure sensors are built that way or at least i never heard for logarithmic one

PS: and log(20) = 1,3 and max pressure is 0.6bar

 

[shooter]

i do NOT see any answer from siggibald