Dedicated Michlet computer

[johnhazel]

I have a spare computer to play with and want to configure it as a dedicated Michlet machine. It is a 2.8GHz Celeron D 336 with 512Mb ram.

I am wondering if I can set it up to run deep and wide Godzilla evolutions at much higher speeds than in it's current Windows XP configuration.

Would stripping the computer down to a DOS only system and running Michlet in the text mode give me a huge speed increase? (Eg load up my copy of Win 98 and start in dos mode then run Michlet from command line)

What I want to end up with is a data set of hull parameters optimized for a range of length, displacement, speed, and water depth. Draft, beam, and shape functions would be unrestricted. This data set would be used to choose hulls for outrigger stabilized human power boats with the outriggers pared down to bare minimum.

I'd like to repeat the process using multipoint optimizations. One example would be to look at what hull optimim is reached using a 3 speed approach with speeds U, 0.9U, and 1.1U. It would be intresting to have this set for comparison to the single point optimum data

The span of the input space I'm looking at would be:
Length 4-10m
Displacement 0.05-.25m^3
Speed 2-5m/s
depth 0.2-2m

How many lengths between 4 and 10 would be needed to capture the parameter trends? Likewise how many Displacements, Speeds, and depths?

How should the steps be distributed? for example :
linear depths 0.2, 0.5, .08, 1.1, 1.4, 1.7, 2.0
exponential depths 0.2, 0.29, 0.43, 0.63, 0.93, 1.36, 2.0

[Leo Lazauskas]

Originally Posted by johnhazel

I have a spare computer to play with and want to configure it as a dedicated Michlet machine. It is a 2.8GHz Celeron D 336 with 512Mb ram.

I am wondering if I can set it up to run deep and wide Godzilla evolutions at much higher speeds than in it's current Windows XP configuration.

Would stripping the computer down to a DOS only system and running Michlet in the text mode give me a huge speed increase? (Eg load up my copy of Win 98 and start in dos mode then run Michlet from command line)

What I want to end up with is a data set of hull parameters optimized for a range of length, displacement, speed, and water depth. Draft, beam, and shape functions would be unrestricted. This data set would be used to choose hulls for outrigger stabilized human power boats with the outriggers pared down to bare minimum.

I'd like to repeat the process using multipoint optimizations. One example would be to look at what hull optimim is reached using a 3 speed approach with speeds U, 0.9U, and 1.1U. It would be intresting to have this set for comparison to the single point optimum data

The span of the input space I'm looking at would be:
Length 4-10m
Displacement 0.05-.25m^3
Speed 2-5m/s
depth 0.2-2m

How many lengths between 4 and 10 would be needed to capture the parameter trends? Likewise how many Displacements, Speeds, and depths?

How should the steps be distributed? for example :
linear depths 0.2, 0.5, .08, 1.1, 1.4, 1.7, 2.0
exponential depths 0.2, 0.29, 0.43, 0.63, 0.93, 1.36, 2.0

I doubt that you will get much extra speed from a DOS only text version. I haven't released a text version for several years, so the old one might not give results that are the same as with the new (graphic) version.

The number of lengths to use is tricky because it depends on the number of speeds as well. This page shows how the optimal length changes for a hull which has a fixed shape:

http://www.cyberiad.net/library/rowi...ond/misres.htm

Good luck!
Leo.

[johnhazel]

Originally Posted by Leo Lazauskas

The number of lengths to use is tricky because it depends on the number of speeds as well. This page shows how the optimal length changes for a hull which has a fixed shape:

http://www.cyberiad.net/library/rowi...ond/misres.htm

So it seems that I should keep Fnv>1.2 if I want smooth variations. That means for displacement = 0.1m^3 a minimum speed
Umin = 1.2*SQRT(9.8*0.1^(1/3)) = 2.44M/s.
I wonder if this will be true for shallow water also.

On the page you reference, under figure 3(b) is:
"Of course the actual total drag Rt is much larger for larger shells, once we multiply Ct by L*2=D2/3. "

But rearranging from the top of the page:

" Ct=Rt/(1/2rhoU2 L*2) "

doesn't Rt=1/2*rho*U^2*L*^2*Ct ?
I'm missing where the dynamic pressure term went.

[Leo Lazauskas]

Originally Posted by johnhazel

So it seems that I should keep Fnv>1.2 if I want smooth variations. That means for displacement = 0.1m^3 a minimum speed
Umin = 1.2*SQRT(9.8*0.1^(1/3)) = 2.44M/s.
I wonder if this will be true for shallow water also.

Of course that depends on the water depth. Finite depth effects are only significant for depth-based Froude numbers greater than about 0.7. It wouldn't surprise me if there is a jump (or a cusp) in the curve of optimal length near the critical depth-based Froude number.

Optimal hull length might not vary quite as quickly for super-critical Froude numbers because there are no "pure" transverse waves.
On the other hand, hull shape effects might then be more important because they affect the behaviour of the diverging wave pattern.

If you also allow hull shape to vary, you could end up with all sorts of funny jumps in the optimal parameters. There are some examples in this report on "optimal" racing kayaks:

http://www.cyberiad.net/library/kaya...ing/racing.htm

Originally Posted by johnhazel

On the page you reference, under figure 3(b) is:
"Of course the actual total drag Rt is much larger for larger shells, once we multiply Ct by L*2=D2/3. "

That's just a scaling factor: it's not an exact equation for Ct.

Leo.