FREE!Ship and Michlet

Is it significant that my hull is pointed the wrong way when I import waves from Michlet into Freeship?

 

Not if you want to sail backwards and have your waves in front of you

 

Well . . . . .Hmmm...

I was hoping for a serious response.

Upon closer examination. The hull form I used, a borrowed file, is built backwards......sigh.

I've been working on crossing the Michlet/Freeship bridge and that was my first result with importing waves. Since I've got you on the line, Martijn, I've got a couple of questons for you. Michlet recommends a 100 x 100 grid for producing waves and documentation recommends a 50 x 50 grid for reasons of computer performance. I noticed a considerable loss of definition in wave form at the lower resolution. But also a considerable loss in computer performance, as predicted, at the higher resolution.

Are the wave forms used for any computations in Freeship? Or is it more along the lines of eye candy?

I've got this 100 X 100 grid imported now and I can't get rid of it.

I've started getting out of memory errors when trying to do certain operations like changing display resolutions and deleting node elements. I tried to save the file and shut down freeship in an attempt to "purge" the memory and restart freeship. I got another memory error and just closed freeship without saving the file. On restart of freeship, there was no more .fbm file. No serious data lost, just a backwards built boat.

I just thought you might be interested in a scenario that eats files.

 

If your hull is defined backwards, it is also send to Michlet backwards and resistance is calculated for a backwards sailing vessel.

Regarding the Michlet grid size: There are two different kinds of grids you can import. I believe they're called sectorial and rectangular grids. I'm not sure which one it is exactly, but one of the two sends twice the number of points in either direction of the grid to the outputfile as specified in Michlet. Leo has been notified about this.

The grid should be disposable like any other layer: Selecting one face while keeping the alt button pressed will select all the connected faces. I think your computer has run out of memory while trying to create the undo data before deleting the waves.

The file-eating scenario sound quite serious. However, whenever a file is being saved the previous one is turned into a back up before the file is overwritten. The backup should still be there (unless you pressed the save button twice) and has the appropriate extension .bak

 

ah, so thats where the un-try'd autocad backup drawings come from

 

Depends on what's in them.....

 

Interior Edges ON. CTRL LEFT CLICK and the whole grid was selected. And disposed of.

The .BAK file was still there AND I did try to save more than once.

I think I'll just put the rudder on the other end.

====================================================

Originally, I was looking for a DELETE LAYER function. Hint, hint,

 

Martijn, some more remarks, comments and wailings.

I reversed my boat by scaling it in the horizontal direction by a factor of -1 to keep the control grid on the port side. What I didn't realize, is this placed the entire craft behind the origin. Obvious now that I think about it. Will have an effect latter.

---------------------------------------------------------------------

I again exported the hull to a Michlet file and created a new wave train that I exported back into Freeship. Just for grins, I set R0 = 0. Zero meaning the middle of the hull by definition in Michlet documentation. In this scenario, the wave actually started ahead of the boat.

I have to assume then that a vessel's aftmost point on the waterline should pass through the origin to get a correct positioning of the wave train.
----------------------------------------------------------------------

In correcting the position error of the hull, I went with a "transform" "move" and inadvertently selected both the hull and the wavetrain. Well this shortly prompted my out of memory error and halted any computations.

My computer didn't hang, but I was unable to do any commands inside of Freeship. I opened up taskmanager and saw that Freeship was utilizing 181,000+ KB of memory. I don't know if it was waiting to utilize more to finish the task or if it just stopped midstream and didn't know where to go with what it had. Again, the program wasn't unresponsive and maybe it was hung-up in a manner of speaking.

I closed the program and reloaded the file. It contained a different 100X100 wave grid and consumed around 121,000 KB of memory. I was successful in deleting the wave train without error. (REF:The basic hull uses around 6,000 KB of memory and a fairly complex hull, interior and sailplan, uses 51,000 KB.)

I don't know if there is code you can incorporate to catch the memory error before it becomes a problem. I guess I just need to buy more memory. I know you're interested in any glitches that come up with your software. Thanks

 

LP,

In CFD the origin almost always is placed at midship location, not the origin of your hull modeler. Just to make things more logical
The R0 setting in Michlet (see manual???) is the distance from the first wave to your origin. So this should be at least half your hull length, preferably more.

Regarding the memory settings: Have you tried lowering the amount of undo memory the program uses in the preferences? Secondly you can also try a lower precision setting, since each incremental step in the precision makes the geometry consume 4 times the memory of the previous setting. In other words, the "high" setting uses 16 times the memory as the "low" setting

Finally, I'll try and implement a memory manager that detects when you're (almost) out of memory. The problem arises as soon as your deleting the grid right?

 

The first time I ran out of memory, I was trying to import a 100X100 wave grid. Another time I was doing a transform move command. Once I get a memory error, I seem to be unable to do anything else without restarting the program.

I've been monitoring memory usage by having task manager running willing working with Freeship. I've seen as much as 300,000 KB in use without a problem. At least on my machine. I've found the undo memory setting and will change that. I think I'll try to get the memory error though and see if purging the undo memory will release things.

I put R0=0 just to see where it would put the origin. No other purpose. I like to try things just to see what will happen. In this case, I learned some details that I wouldn't know otherwise.

 

Full memory

Martijn,

I was able to replicate my out of memory error. I imported a 100X100 wave grid with out any problem. It took a while though. Memory usage went up to over 300,000KB. Then I tried the transform, move function in an attempt to move the hull, one foot forward. It processed the information for a while and finally gave me the out of memory error.

It looks like the transform took place, but there were some oddities. There was a control point left at the origin (no lines to it though) and with hydrostatics turned on the LCF is located at this point. Intersection lines did not function properly. (Did not display)

Left to rignt, most tool bar buttons/functions worked up to the layers button, except for intersection lines as stated above. Purging the undo memory allowed the layers screen to function so I could turn the waves on and off. Transform function did not work. When I say they didn't work, it means I'd get an out of memory error. It took a couple of tries to delete the waves back out. Purging the undo memory a second time did the trick. I didn't try saving. It appears that having a large layer(lots of control points) slows Freeship considerably, whether the layer is turned on or off. The memory usage is the same.

After deleting the waves, the intersection lines came back.

According to task manager, Freeship still was using 300,000 +KB of memory after deleting the waves. Upon minimizing Freeship, memory consumption dropped to 70,000 KB. Restarting Freeship and loading the same file shaow memory usage at 17,000 KB. Not sure if that's helpful knowledge or not.

A further look at the hydrostatics give a 0(zero) value for righting moment and a negativedisplacement.

I'm hoping this is all useful to you.

 

LP,
What precision setting are you using and how many faces does your model contain?

All the problems you've described seem to be caused by an insufficient amount of memory while trying to write the undo data to memory. Obviously some very weird things have happened here. I'm very interested in this model, but also the amount of memory in your computer. Having both I could try to reproduce this myself and try to fix this so other users won't experience the same problems.

Negative displacements can be caused by the fact that surface normals are pointing at the inside of the hull rather than the outside. This in turn can be caused by the weird effects you've just described. As soon as memory problems start to occur computers cease to perform logically

I'm trying to find a solution for this.