Speaker Enclosure Bracing Simulations - AVS Forum | Home Theater Discussions And Reviews
Forum Jump: 
 4Likes
  • 2 Post By ronny31
  • 1 Post By Augerhandle
  • 1 Post By Bigus
 
Thread Tools
post #1 of 25 Old 02-13-2016, 11:38 AM - Thread Starter
AVS Forum Special Member
 
Join Date: Mar 2015
Location: Norway
Posts: 1,181
Mentioned: 16 Post(s)
Tagged: 0 Thread(s)
Quoted: 676 Post(s)
Liked: 138
Speaker Enclosure Bracing Simulations

I'm simulating bracing alternatives in Autodesk Fusion 360 (its free for 1 year) and will share my results as I get them and edit this post with more results.

Anchoring the parts:
All tests are with artificially anchored side faces, so the simulation is more or less as if each side face is joined to an infinitely strong material instead of just being an edge of an enclosure. Later on I will simulate full enclosures and half enclosures, so we can get more accurate figures as this one-sheet approach will be very optimistic.

Stress amount and type of application of force:
The stress amount is the amount of air pressure produced by a very beefy 18" driver doing 20mm xmax (from neutral) inside a rather small closed enclosure (about 200-300 liters I don't remember, I'll do a more definite number later on but for now I will use this same number so that all the designs are comparable. The software stores that figure so its very easy to use the same one for all the simulations). The pressure amounts to 0.002 MPa (that's 160db, which is what I assume is the limit that the UM18-22 can do internally in a closed enclosure given its motor strength and cone area). The pressure is applied on the outside face of the enclosure, assuming we brace internally. Just because its very very time intensive to apply pressure to a thousand surfaces on the braced side.

Material:
It is mentioned what the material in each simulation is.

Types of simulation:
-Modal simulation, here we find at what frequencies the bracing and the face that is braced, vibrates. And by how much they vibrate.
-Static stress simulation, here we find how much displacement the braced face has, in other words how much it moves inwards when pressurized from the outside, and the reverse (later on we will do sims from both sides I bet, since the bracing probably won't be linear from stress from both directions).

Goal:
The goal is to find out how much to brace before no benefit is to be had. How much benefit there is, is to be decided by us using the total displacement area (Sd sort of) and displacement (xmax, sort of) to then calculate the db output from each surface at modal frequencies. Its not extremely accurate but its a neat yardstick until we have a nine rounder on what yardstick to use.

There's basically two different benefits:
Not subwoofers:
With just tiny differences its possible to remove the need for massive amounts of filler and remove problematic boxiness and internal reflections hitting the inside of the cone of mids and such. The outside enclosure faces themselves can also produce significant output at high frequencies, if the enclosure faces are moving too much as a result of the internal soundwaves.

Subwoofers:
Bracing has more to do with db output than SQ. Because if the cone loads the internal volume air with 150db with perfect solid faces and then only to 149db with real material, then the output from the vent will be less. The movement of the enclosure faces themselves (as well as the cone) moving at sub-bass frequencies produce basically no db at all compared to the vent output because even a badly braced enclosure face isn't going to move 10mm+. Its not a case of "losing some db from the vent from the faces moving, equals the same output in db because the faces produce as much extra db as the vent would have" as one might first assume.

First simulation (not my first simulation, but the first in this thread):
1100x450x9mm sheet with 70mm tall 1mm thick braces across the 450mm axis. PC/ABS plastic is all of the material. It is assumed the sheets are all fastened to each other with a glue stronger or equal in strength to the PC/ABS plastic, and thus simulated as one solid body with this intricate shape.

Modal analysis shows a lot of vibration from the top of the side-ways-unbraced braces.

One modification, eight 5mm diameter poles through the braces along the 1100mm axis, reduces the vibration but the total displacement in static stress test is unchanged.

But what is the amount of material in this compared to a 22mm sheet without bracing?
Well this bracing has the volume of 2 961 000 mm^3, and then there's 9mm of sheet outside that. If we take 13mm of sheet instead of the bracing, it makes 6 435 000 mm^3, with then a 9mm sheet outside that. So the net volume is increased, while keeping the same outside dimensions.
But how much material is lost if we are to make 94 sheets 70mm wide with 1mm thick material, and lose 2mm width next to the 70mm in each cut? We still use less material overall even accounting for losing 2mm on every cut (assuming we don't have to cut the material into 1mm thin sheets first). That is not counting the volume of the eight 5mm diameter solid poles used. They don't have the volume of 2 500 000 mm^3 I assume so I didn't bother.
That's strike 3 as far as I'm aware: (1) more internal volume with same net external volume, (2) less material used and (3) stronger than less complex designs using far more material.
Here's a 22mm sheet of PC/ABS plastic.

And just to compare, here's 3 times the material, 66mm PC/ABS plastic.

And here's 22mm Steel ASTM A36:


But now is the question, how much worse is it if we half the height from 70mm to 35mm and use the 35mm extra material instead to brace in the other axis. PC/ABS plastic again. Technically slightly more internal volume if you used this instead of the one-axis 70mm bracing because this is equally much material minus the 8 cylindrical braces. But much worse strength.

Strangely, this 44mm material not filled in, is apparently stronger than the 66mm sheet itself. But I'm guessing that's down to the finer mesh used in the braced simulations. So I'm going to have to be better at making sure results aren't because of changed mesh settings. As we wait I'm doing even higher polygon-count simulations on smaller sections to see if the fidelity of the mesh may make more or less optimistic results on complex bracing simulation.

The goal is to get most strength from least material and least assembly complexity. 1mm plastic is well within realms of practicality, even if there are many of them, as long as there's not too much CNC cutout lines (they get paid by each meter of cutting, pretty much). 22mm titanium however.. Well you get the idea.
If the bracing has a lot of volume loss from the enclosure, resulting in a bigger enclosure net external volume, its not ideal, but sometimes it can be a worthwhile design choice if the strength and practicality of the bracing scheme is good enough.

Feel free to install Fusion 360 yourself and get cracking making better bracing schemes. If you're familiar with google sketchup you should get to grips with Fusion 360 after a few furious hours fusing the keyboard to your forehead while watching youtube tutorials.

EDIT/Update1: I have completed a very high resolution simulation of a 200mm by 450mm portion of the 70mm tall braces with 8 cylindrical braces. But I anchored it wrong (on the 450mm sides in addition to 200mm sides), so the results are not comparable to the 1100x450mm simulation with lower resolution. Another hour+ long simulation it is then.

EDIT/Update2: Here's a new simulation between braced 9mm sheet and 79mm solid sheet. Since we had that funny one above where the braced thing with less overall material was stronger than the entire untreated sheet.


Now it seems they are correct, they have the same dimensions, only one lacks tons of material and the other is solid. Now the correct one is strongest by a huge margin. No idea what happened in the comparison between the square pattern 35mm tall bracing scheme.

EDIT/Update3: Lets test how anchoring affects the outcome:
braced edgewise by anchor:

Not braced on the immediate edge but another edge, so the corner is free-standing, with one pressure face so it can be compared to the last:


Then also the comparison between internal pressure and external pressure.
Two faces under pressure inwards:

Two faces under pressure outwards:

Note how the displacement changes. Especially between force applied into the middle or out from the middle, since there's more surface area on the outside than the inside and the force applied is in pascal (newtons per square meter).

EDIT/Update4: Lets use Pythagorean theorem to cut the use of bracing material while keeping the same enclosure strength. Question: What is the most effective brace on the bottom side of this profile of an enclosure? The answer is the nearby faces to the sides. We use the Pythagorean theorem to calculate what distance from the wall this is true. As long as sqrt(4x^2) with x being the distance to the nearest wall, is shorter than the distance to the opposite wall, this is more efficient.

The amount of material used in bracing: 37 000 mm^3

The amount of material used in bracing: 13 143 mm^3.
Both sims are anchored on the left wall face.
That's a 74.5% reduction in material used. This extra saved material can easily be used to brace the span between the corner braces enough to get more strength than the alternative braces. But now if you do the math and compare the bracing locations in the span in the middle, you get the figures that you are better off bracing against the braces instead of running material all the way across the enclosure.
How many here are going to consider using bracing like the first simulation now?

EDIT/Update5:
Then I tried to find (or rather just confirm) the optimal placement of the corner to corner braces.

The below one has the center-line of the brace 65mm from the corner, and the total length from corner to corner is 190mm internally and 200mm externally. So we can see that bracing with 2 braces is best by splitting up the span into three equal pieces (obvious really). So when you make your braces you should measure 1/3 out on the face if you are using 2 braces like this.


The volume of bracing used in the best placement is 18 799 mm^3, compare that to the 37 000 volume of bracing to the opposite face (the picture below in update 6). So bracing corner to corner requires half as much material as the alternative below in update 6, and translates to less external enclosure size if you want the same performance tune.

EDIT/Update6: Here's the optimal 1/3 placement of braces across to the opposite face as well, just to compare to the optimal placed corner braces:

These are quite small models, btw, hence the low displacement figures. Not like the entire sheets I simulated first.
LTD02 and fill35U like this.

Last edited by ronny31; 02-14-2016 at 03:30 PM.
ronny31 is offline  
Sponsored Links
Advertisement
 
post #2 of 25 Old 02-13-2016, 01:12 PM
AVS Forum Special Member
 
Bigus's Avatar
 
Join Date: Feb 2003
Location: The South
Posts: 5,701
Mentioned: 13 Post(s)
Tagged: 0 Thread(s)
Quoted: 641 Post(s)
Liked: 585
I kind of stopped reading about halfway through.

Fully constrained edges are appropriate as modeling a real enclosure where the corner joints are adequately strong and stiff (bending moments are opposed on adjoining panels).

I think what you are doing is good as a learning exercise, and could be useful to illustrate relative performance of different bracing schemes. But it won't match reality well with the simplistic input modeling.

Dynamic vibration analysis is a monster (see a previous post of mine on submarine isolation rafting). Amplitude of panel vibration could be more or less than calculated static deflection given a periodic or nonperiodic input. You just can't say without complex modeling, or better yet real world testing.

But relative is good enough for your purpose. One suggestion, spend time on bracing schemes people might actually consider building.
Bigus is offline  
post #3 of 25 Old 02-13-2016, 03:03 PM - Thread Starter
AVS Forum Special Member
 
Join Date: Mar 2015
Location: Norway
Posts: 1,181
Mentioned: 16 Post(s)
Tagged: 0 Thread(s)
Quoted: 676 Post(s)
Liked: 138
Quote:
Originally Posted by Bigus View Post
I kind of stopped reading about halfway through.

Fully constrained edges are appropriate as modeling a real enclosure where the corner joints are adequately strong and stiff (bending moments are opposed on adjoining panels).

I think what you are doing is good as a learning exercise, and could be useful to illustrate relative performance of different bracing schemes. But it won't match reality well with the simplistic input modeling.

Dynamic vibration analysis is a monster (see a previous post of mine on submarine isolation rafting). Amplitude of panel vibration could be more or less than calculated static deflection given a periodic or nonperiodic input. You just can't say without complex modeling, or better yet real world testing.

But relative is good enough for your purpose. One suggestion, spend time on bracing schemes people might actually consider building.
I'm out to find the first principals to which gets efficient bracing. Then devise bracing that takes advantage of those concepts.
ronny31 is offline  
Sponsored Links
Advertisement
 
post #4 of 25 Old 02-14-2016, 03:17 AM - Thread Starter
AVS Forum Special Member
 
Join Date: Mar 2015
Location: Norway
Posts: 1,181
Mentioned: 16 Post(s)
Tagged: 0 Thread(s)
Quoted: 676 Post(s)
Liked: 138
Read Update 4 if you want to see how you can cut material used in bracing by over 2/3.
ronny31 is offline  
post #5 of 25 Old 02-14-2016, 04:15 AM
Advanced Member
 
chadamir's Avatar
 
Join Date: Jul 2003
Posts: 513
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 177 Post(s)
Liked: 60
Quote:
Originally Posted by ronny31 View Post
Read Update 4 if you want to see how you can cut material used in bracing by over 2/3.
What happens when you make the corner braces longer so that they almost meet in the middle of each panel?

Also, on the one with cross bracing, why not center those more?
chadamir is offline  
post #6 of 25 Old 02-14-2016, 04:29 AM - Thread Starter
AVS Forum Special Member
 
Join Date: Mar 2015
Location: Norway
Posts: 1,181
Mentioned: 16 Post(s)
Tagged: 0 Thread(s)
Quoted: 676 Post(s)
Liked: 138
Quote:
Originally Posted by chadamir View Post
What happens when you make the corner braces longer so that they almost meet in the middle of each panel?

Also, on the one with cross bracing, why not center those more?
See the next update that I just did.
ronny31 is offline  
post #7 of 25 Old 02-14-2016, 04:35 AM
Advanced Member
 
chadamir's Avatar
 
Join Date: Jul 2003
Posts: 513
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 177 Post(s)
Liked: 60
What about if the braces are at 1/3? From the wall that is. Also it would be rather difficult to do diagonal bracing when you don't have a wall to brace the other side against.
chadamir is offline  
post #8 of 25 Old 02-14-2016, 04:48 AM - Thread Starter
AVS Forum Special Member
 
Join Date: Mar 2015
Location: Norway
Posts: 1,181
Mentioned: 16 Post(s)
Tagged: 0 Thread(s)
Quoted: 676 Post(s)
Liked: 138
Just did the face to opposite face simulation. See update 6. Its slightly better than the optimal middle picture in the series of three corner braces. But it uses twice as much material (37 000mm^3 no matter where you place them on the face).
If I fiddled the placement slightly towards the corners I could probably improve upon it a little bit more because its not a perfect 1/3 between the braces. But its a lot of work to do it.

What dioagonally unbraceable shape are you referring to specifically?
ronny31 is offline  
post #9 of 25 Old 02-14-2016, 04:50 AM
Advanced Member
 
chadamir's Avatar
 
Join Date: Jul 2003
Posts: 513
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 177 Post(s)
Liked: 60
Quote:
Originally Posted by ronny31 View Post
Just did the face to opposite face simulation. See update 6. Its slightly better than the optimal middle picture in the series of three corner braces. But it uses twice as much material (37 000mm^3 no matter where you place them on the face).
If I fiddled the placement slightly towards the corners I could probably improve upon it a little bit more because its not a perfect 1/3 between the braces. But its a lot of work to do it.

What dioagonally unbraceable shape are you referring to specifically?
The corner bracing method.
chadamir is offline  
post #10 of 25 Old 02-14-2016, 05:07 AM - Thread Starter
AVS Forum Special Member
 
Join Date: Mar 2015
Location: Norway
Posts: 1,181
Mentioned: 16 Post(s)
Tagged: 0 Thread(s)
Quoted: 676 Post(s)
Liked: 138
Quote:
Originally Posted by chadamir View Post
The corner bracing method.


But there's always a nearby corner to brace to. Even if the A side of the corner don't have a corner to brace to, there's still the other side of the corner. Brace the red line when the green isn't viable.

Last edited by ronny31; 02-14-2016 at 05:36 AM.
ronny31 is offline  
post #11 of 25 Old 02-14-2016, 11:15 AM - Thread Starter
AVS Forum Special Member
 
Join Date: Mar 2015
Location: Norway
Posts: 1,181
Mentioned: 16 Post(s)
Tagged: 0 Thread(s)
Quoted: 676 Post(s)
Liked: 138
So with the diagonal bracing I overturned a century of AVS-forum accepted-and-recommended bracing culture. And I got 1 like for sharing it with a hundred viewers. That's nice.
ronny31 is offline  
post #12 of 25 Old 02-14-2016, 10:44 PM
AVS Forum Special Member
 
Augerhandle's Avatar
 
Join Date: Jan 2009
Location: About 25" away from my computer screen
Posts: 5,338
Mentioned: 32 Post(s)
Tagged: 0 Thread(s)
Quoted: 1254 Post(s)
Liked: 1256
I really doubt that this forum has been around for a century, considering that the internet has only existed for around 25 years.

What about your GFR ribs that you swore by just a few days ago?

Quote:
Originally Posted by ronny31 View Post
Here's my preferred method to stiffen the cabinet.

Grey is the wood material, black is polyester/epoxy resin, orange is sticks. First the grey surface is coated with resin, then the sticks are applied, then ideally I also use some chopped mats glassfiber and add it on top of the whole thing over the sticks and between them when I apply the resin on top of the sticks. The sticks go the long axis of the surface they reinforce. For db drag I would use a lot of glassfiber and the distance between the sticks would be smaller than a home-stereo speaker. Home speakers may do just fine with only one layer of glassfiber, and on the smallest surfaces no glassfiber at all.
This works even on LCR speakers, because you don't need tons of stuffing to avoid high frequency sounds coming out of the ports, just a 1 inch or 5/6 inch layer of stuffing on all the internal surfaces.
I agree that diagonal bracing is more efficient. I learned that in my 9th grade math class. The reason window or dowel bracing is used is for ease of building. It's easier to cut and install a perpendicular edge than an angled edge. Additionally, there is less shear force on the glue joint. Have you calculated that shear force? Which glue would be optimal for that shear?


Why are you simulating in ABS plastic? The MoE of ABS is about 2.50 GPa compared to almost 3.6 GPa for MDF. Just because the simulation shows flex doesn't mean a MDF subwoofer box his excessive flex or that it needs more/better bracing. Monster speaker wire has all kinds of unnecessary "improvements" as well.


Keep working with the program though, you're on the right track.

"The wise understand by themselves; fools follow the reports of others"-Tibetan Proverb
_____________________ http://www.scientificamerican.com/article/auger-handle/ ________________________

Last edited by Augerhandle; 02-15-2016 at 12:57 AM.
Augerhandle is offline  
post #13 of 25 Old 02-15-2016, 04:30 AM - Thread Starter
AVS Forum Special Member
 
Join Date: Mar 2015
Location: Norway
Posts: 1,181
Mentioned: 16 Post(s)
Tagged: 0 Thread(s)
Quoted: 676 Post(s)
Liked: 138
Quote:
Originally Posted by Augerhandle View Post
What about your GFR ribs that you swore by just a few days ago?
That was a-few-days-ago-me, I don't have to have his opinions. Now I'm leaning towards using 1mm braces like this:

It uses less material than a bare 22mm sheet because it has a 9mm sheet with thin bracing spaced far enough apart to get out even on material used.
Then I would brace this diagonally to the neighboring walls.
The main problem with this is that I haven't worked out how to assemble such a design without it taking as long as building a stadium. And also how to cut the material myself from the actual sheet dimensions the material comes in, without hiring expensive CNC processes.

Quote:
Originally Posted by Augerhandle View Post
I really doubt that this forum has been around for a century, considering that the internet has only existed for around 25 years.
Comedian Omid Djalili: "I'm the only iranian comedian, which is three more than germany".

Quote:
Originally Posted by Augerhandle View Post
I agree that diagonal bracing is more efficient. I learned that in my 9th grade math class. The reason window or dowel bracing is used is for ease of building.
Cut a sheet with 45 degree ends and cut holes in it and install it like you would a normal brace we use everywhere.

Or the more efficient designs of this corner-brace. The right one below, or even more efficient ones:


Quote:
Originally Posted by Augerhandle View Post
It's easier to cut and install a perpendicular edge than an angled edge. Additionally, there is less shear force on the glue joint. Have you calculated that shear force? Which glue would be optimal for that shear?
Most glue is stronger than the material we are gluing with it. Or at least it is if you pick the good glue. If its the absolute cheapest wood glue, not a guarantee that it will hold out longer than the material. But if its the absolute cheapest polyester resin, its going to be stronger than the wood.

Quote:
Originally Posted by Augerhandle View Post
Why are you simulating in ABS plastic?
Because MDF doesn't come in 1mm thick sheets, so I can test designs that wouldn't be possible with MDF.
And its not important what material it is as long as I use the same material for all comparative tests to make them comparable. Plastic also isn't so strong that the 0.002 MPa makes too microscopic amounts of displacement results.

Quote:
Originally Posted by Augerhandle View Post
Just because the simulation shows flex doesn't mean a MDF subwoofer box his excessive flex or that it needs more/better bracing. Monster speaker wire has all kinds of unnecessary "improvements" as well.
We can calculate how much flex is the point of no extra worthwhile improvement for subwoofers and speakers. Feel free to have a gander at defining that point.
ronny31 is offline  
post #14 of 25 Old 02-15-2016, 07:47 AM
AVS Forum Special Member
 
Augerhandle's Avatar
 
Join Date: Jan 2009
Location: About 25" away from my computer screen
Posts: 5,338
Mentioned: 32 Post(s)
Tagged: 0 Thread(s)
Quoted: 1254 Post(s)
Liked: 1256
Quote:
Originally Posted by ronny31 View Post
That was a-few-days-ago-me, I don't have to have his opinions.
Which means that you admit that you were wrong in the Bracing again thread
Quote:
...The main problem with this is that I haven't worked out how to assemble such a design without it taking as long as building a stadium. And also how to cut the material myself from the actual sheet dimensions the material comes in, without hiring expensive CNC processes.
So you say it is cost and time prohibitive (as you were also politely informed in the Bracing again thread)
Quote:





Cut a sheet with 45 degree ends and cut holes in it and install it like you would a normal brace we use everywhere.


Or the more efficient designs of this corner-brace. The right one below


...
Wow, you picked the same brace suggested to the OP (and that you argued against) in the Bracing again thread
Quote:

Most glue is stronger than the material we are gluing with it. Or at least it is if you pick the good glue. If its the absolute cheapest wood glue, not a guarantee that it will hold out longer than the material. But if its the absolute cheapest polyester resin, its going to be stronger than the wood.
You never answered the questions. Have you calculated the shear with an angled brace? What is the best glue to use in shear?
Quote:



Because MDF doesn't come in 1mm thick sheets, so I can test designs that wouldn't be possible with MDF.
And its not important what material it is as long as I use the same material for all comparative tests to make them comparable. Plastic also isn't so strong that the 0.002 MPa makes too microscopic amounts of displacement results.
Might as well model it in copier paper then
corradizo likes this.

"The wise understand by themselves; fools follow the reports of others"-Tibetan Proverb
_____________________ http://www.scientificamerican.com/article/auger-handle/ ________________________
Augerhandle is offline  
post #15 of 25 Old 02-15-2016, 08:38 AM - Thread Starter
AVS Forum Special Member
 
Join Date: Mar 2015
Location: Norway
Posts: 1,181
Mentioned: 16 Post(s)
Tagged: 0 Thread(s)
Quoted: 676 Post(s)
Liked: 138
Quote:
Originally Posted by Augerhandle View Post
So you say it is cost and time prohibitive (as you were also politely informed in the Bracing again thread)
A stereo is a high cost item that provides absolutely zero fulfillment of our biological needs. You can not take the "that is a pointless expenditure of time and money" argument. Others have completely different ideas about how much money and hard work they are willing to put into getting a better stereo, than you.

Quote:
Originally Posted by Augerhandle View Post
Wow, you picked the same brace suggested to the OP (and that you argued against) in the Bracing again thread
That brace works well in some situations are crap in others. I argued against bracing from side 1 to 6 with such bracing (if you picture a dice). Here I am suggesting bracing from side 1 to the other sides, not side 6, and from side 6 to the other sides, not to side 1, etc. With this brace.
If I said to not fill your diesel car with gasoline would you jump down my throat if I said to fill gasoline on the next guy's gasoline car?

Quote:
Originally Posted by Augerhandle View Post
You never answered the questions. Have you calculated the shear with an angled brace?
No, I haven't calculated the shear force. Well, Fusion 360 does calculate something akin to it to let you know if parts of the model are under danger of breaking apart with crack formation, but I haven't focused on that. Feel free to do something yourself. The more calories you spend on this the more cake can you eat.

Btw, 0.002 MPa is 2000 Newtons per square meter. That's 449 pounds on a 39 by 39 inch area. Its about where I estimate the absolute upper limit of pressure inside a powerful subwoofer is. Without going into 10 grand DB drag subwoofers.
Atmospheric pressure on both sides is an even 101325 newtons per square meter, or about 10 newton per square centimeter. Its like having 22 770 pounds on a 39 by 39 inch area. On both sides of the enclosure sheets.

Last edited by ronny31; 02-15-2016 at 08:48 AM.
ronny31 is offline  
post #16 of 25 Old 02-15-2016, 12:48 PM
Member
 
s_yang's Avatar
 
Join Date: Jan 2010
Location: Indiana
Posts: 34
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 15 Post(s)
Liked: 10
Quote:
Originally Posted by ronny31 View Post
Strangely, this 44mm material not filled in, is apparently stronger than the 66mm sheet itself. But I'm guessing that's down to the finer mesh used in the braced simulations. So I'm going to have to be better at making sure results aren't because of changed mesh settings. As we wait I'm doing even higher polygon-count simulations on smaller sections to see if the fidelity of the mesh may make more or less optimistic results on complex bracing simulation.
I noticed that you only used one layer of elements through the thickness in those solid sheets. It wouldn't give you accurate results in bending. Try use 3 or 5 layers through the thickness and compare the results.
s_yang is offline  
post #17 of 25 Old 02-15-2016, 01:53 PM - Thread Starter
AVS Forum Special Member
 
Join Date: Mar 2015
Location: Norway
Posts: 1,181
Mentioned: 16 Post(s)
Tagged: 0 Thread(s)
Quoted: 676 Post(s)
Liked: 138
Quote:
Originally Posted by s_yang View Post
I noticed that you only used one layer of elements through the thickness in those solid sheets. It wouldn't give you accurate results in bending. Try use 3 or 5 layers through the thickness and compare the results.
Hm, fine thinking, I will try that. (doing it as you read this)
ronny31 is offline  
post #18 of 25 Old 02-15-2016, 02:12 PM
Member
 
s_yang's Avatar
 
Join Date: Jan 2010
Location: Indiana
Posts: 34
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 15 Post(s)
Liked: 10
Quote:
Originally Posted by ronny31 View Post
Hm, fine thinking, I will try that. (doing it as you read this)
Another way is to use second order element if Autodesk supports. It also would give you better results with less elements through thickness.
s_yang is offline  
post #19 of 25 Old 02-15-2016, 02:21 PM - Thread Starter
AVS Forum Special Member
 
Join Date: Mar 2015
Location: Norway
Posts: 1,181
Mentioned: 16 Post(s)
Tagged: 0 Thread(s)
Quoted: 676 Post(s)
Liked: 138
Hm, not sure what to make of this. Top is 11x600x2400mm times two, the second is 22x600x2400mm times one. ABS plastic, under 0.002 MPa.


When I adjusted the mesh size on the second one, the slightly bigger mesh than this had exactly the same displacement (0.001 difference). So I think there's some pinches of salt to take with these simulations on different mesh settings. On future more advanced simulations I will curve the edges and apply much more force I think, so that we can have significant differences and more similar mesh densities between the different models. Because I had to go from 5% setting to 3% setting to get the second picture to have closer to the same mesh as the first (and probably could have gone to 2%).
ronny31 is offline  
post #20 of 25 Old 02-15-2016, 06:05 PM - Thread Starter
AVS Forum Special Member
 
Join Date: Mar 2015
Location: Norway
Posts: 1,181
Mentioned: 16 Post(s)
Tagged: 0 Thread(s)
Quoted: 676 Post(s)
Liked: 138
There's a mesh refinement feature, which refines the mesh after simulating stress, then it runs the simulation again, then you can put in how many times and some other settings for this. I'm going to try that tomorrow and see if that yields closer results between models that should be equal.

EDIT: It went a lot quicker than I thought.

Adaptive mesh refinement settings:
maximum number of mesh refinements: 6
Minimum refinement step difference (%): 1
Portion of elements to refine (%): 50
Results for baseline accuracy: Displacement, total.

Adaptive mesh refinement settings:
maximum number of mesh refinements: 6
Minimum refinement step difference (%): 5
Portion of elements to refine (%): 40
Results for baseline accuracy: Displacement, total.

Note to self. Mesh refinement.

Last edited by ronny31; 02-15-2016 at 06:26 PM.
ronny31 is offline  
post #21 of 25 Old 02-15-2016, 08:05 PM
AVS Forum Special Member
 
Augerhandle's Avatar
 
Join Date: Jan 2009
Location: About 25" away from my computer screen
Posts: 5,338
Mentioned: 32 Post(s)
Tagged: 0 Thread(s)
Quoted: 1254 Post(s)
Liked: 1256
Quote:
Originally Posted by ronny31 View Post

Btw, 0.002 MPa is 2000 Newtons per square meter. That's 449 pounds on a 39 by 39 inch area. Its about where I estimate the absolute upper limit of pressure inside a powerful subwoofer is. Without going into 10 grand DB drag subwoofers.
Atmospheric pressure on both sides is an even 101325 newtons per square meter, or about 10 newton per square centimeter. Its like having 22 770 pounds on a 39 by 39 inch area. On both sides of the enclosure sheets.

449 lb. on a 39 x 39 inch area is equal to only 0.2952 PSI. It doesn't take much to restrain that.

"The wise understand by themselves; fools follow the reports of others"-Tibetan Proverb
_____________________ http://www.scientificamerican.com/article/auger-handle/ ________________________
Augerhandle is offline  
post #22 of 25 Old 02-15-2016, 09:31 PM
AVS Forum Special Member
 
Bigus's Avatar
 
Join Date: Feb 2003
Location: The South
Posts: 5,701
Mentioned: 13 Post(s)
Tagged: 0 Thread(s)
Quoted: 641 Post(s)
Liked: 585
Solid elements just aren't optimal, bordering on inappropriate, for this modeling exercise. Plate elements are for just this purpose.

Also, try to keep in mind that the design of a part in a fighter jet has a somewhat different criteria list than the design of a part in a steamroller. A sub box is much more like a steamroller. We don't need a solution for the optimum efficiency weight to stiffness bracing solution... we need the one that is the easiest and quickest to build perhaps even using scrap material from the enclosure and performs "plenty good enough."

The reason your grp bracing idea was cricized earlier is that the end result probably wouldn't even be as stiff, but much more importantly was a vastly more labor intensive and costly solution.

So are these 1mm fin braces (and your simulation isn't evaluating plate buckling failure in those elements I would bet, something I suspect could actually be important with the dimensions you have used). No one wants to build a MN enclosure like that, regardless of the results, so your time is good only for educational purposes. Which is fine, but moving to more practical solutions would be nice.

Like the corner braces. Plenty of people have used that in the past, especially near the driver opening to navigate bracing around the basket and motor. But it's nice to see the analysis showing that it isn't a compromised method. Its a bit trickier to assemble in a box, so many will just stick to perpendicular bracing, but those are the types of options people would be interested in.
Augerhandle likes this.
Bigus is offline  
post #23 of 25 Old 02-16-2016, 04:56 AM - Thread Starter
AVS Forum Special Member
 
Join Date: Mar 2015
Location: Norway
Posts: 1,181
Mentioned: 16 Post(s)
Tagged: 0 Thread(s)
Quoted: 676 Post(s)
Liked: 138
Quote:
Originally Posted by Augerhandle View Post
449 lb. on a 39 x 39 inch area is equal to only 0.2952 PSI. It doesn't take much to restrain that.
"much" isn't a unit of measure I recognize. It must be imperial. This is just opinion, and we can go nine rounds about whether or not our opinion is that a surface under 0.2952 Psi requires "much" strength to not flex, but it would be pointless. Because there is no right answer. Its just individual opinion of what defines as "much".

There is however an answer to how much strength is actually required to get x micrometers of movement from 0.2952 Psi. If we then just agreed on what amount of movement is the absolute max we can allow with subwoofers and a separate figure for what movement is allowed for full-range speakers, then we would be able to adjust the design to get the precise amount of strength required for Z Psi. I say separate figure for full range speakers, but really its a separate number for each frequency because as you double frequency the amount of xmax you need to get the same db goes down by 75%, so eventually the tiny movement may make audible sounds as long as the frequency is high enough. And if not audible, it may at least be noticeable on the output of the speaker, lost energy moving elements instead of pressurizing the air as intended. And hard-earned moneys spent on insanely low distortion drivers and amps and audio files may be p****d into the wind because the enclosure makes ten times that distortion amount due to poor bracing.

Quote:
Originally Posted by Bigus View Post
Solid elements just aren't optimal, bordering on inappropriate, for this modeling exercise. Plate elements are for just this purpose.

Also, try to keep in mind that the design of a part in a fighter jet has a somewhat different criteria list than the design of a part in a steamroller. A sub box is much more like a steamroller. We don't need a solution for the optimum efficiency weight to stiffness bracing solution... we need the one that is the easiest and quickest to build perhaps even using scrap material from the enclosure and performs "plenty good enough."

The reason your grp bracing idea was cricized earlier is that the end result probably wouldn't even be as stiff, but much more importantly was a vastly more labor intensive and costly solution.

So are these 1mm fin braces (and your simulation isn't evaluating plate buckling failure in those elements I would bet, something I suspect could actually be important with the dimensions you have used). No one wants to build a MN enclosure like that, regardless of the results, so your time is good only for educational purposes. Which is fine, but moving to more practical solutions would be nice.

Like the corner braces. Plenty of people have used that in the past, especially near the driver opening to navigate bracing around the basket and motor. But it's nice to see the analysis showing that it isn't a compromised method. Its a bit trickier to assemble in a box, so many will just stick to perpendicular bracing, but those are the types of options people would be interested in.
You can't get the meal before its cooked. One thing at a time, and eventually I'll boil it down to a series of bracing schemes that anyone with a pulse can copy. Then the person with pulse can just pick whatever level of bracing is the preferred level of calorie-expenditure.

I am however up for constructive criticism about the simulation method, what do you mean by solid elements being wrong? Explain.

You may need "plenty good enough", but I want exceptional. And I also want to gain net internal volume if possible instead of losing a few precious liters to bracing.

Here's the cross-bracing illustration:


Of course the corners should be redesigned.

I'm essentially figuring out how they strengthen bits and pieces in the real world and then applying it to enclosures. After that we can figure out how to make it with bits and pieces of scrap material like Angus MacGyver.
ronny31 is offline  
post #24 of 25 Old 02-16-2016, 06:25 AM
AVS Forum Special Member
 
Bigus's Avatar
 
Join Date: Feb 2003
Location: The South
Posts: 5,701
Mentioned: 13 Post(s)
Tagged: 0 Thread(s)
Quoted: 641 Post(s)
Liked: 585
FEA applied to solid material mechanical analysis comes in three major flavors, 1D, 2D and 3D elements. 1D elements are represented by lines in the software, and are beam elements. You supply the program with their properties such as modulus, mass, cross section, and MOI in each axis. 2D elements are plates, and are perfect for modeling, well, plates. Flat structures like enclosure panels. You give the software properties like thickness, modulus, mass, and they are represented as polygons with no thickness.

You can combine beam and plate elements quite easily. Along a row dividing plate elements, you can superimpose a beam element, stipulate its properties, as well as offset if any from the plate surface (ie, is the beam centered on the plate, or flush with one side).

While using 1D and 2D elements may seem crude, it really isn't. It often lets you get a much finer mesh still with significantly reduced computational demands. 1D and 2D models can be quite complex. A 2D element in a good solver (probably not Autodesk I would guess) will let you input a table of orthotropic layer properties specifying thickness of plies, orientation, weave properties (modulus, strength) in a multilayered composite sheet and produce pretty good results including interply shear stress, delamination, etc.

3D elements are solid elements. They work great when used properly, but you have to be really careful. It may seem that going to a 3D element means you have more fidelity, but that isn't necessarily the case. Because the computational demands skyrocket, you may actually have significantly fewer total elements in a solid model. You need to pay attention to how many elements are used through the thickness of a material (this was mentioned already in this thread). With only one element you don't get a good representation of reaction through the plane. This applies to bracing structures as well. 3D elements tend to force the software to approximate stress risers in sharp corners and joints, and if fidelity of the model is sufficient there you get garbage. Modal analysis in particular seems to suffer from low fidelity 3D models. And 3D models really should have a lot of thought put into the arrangement/layout of elements, automesh typically produces subpar results, even if the mesh looks pretty.

You're letting it automesh, and assuming the results are good because the pictures are pretty. FEA may be easy to use, but not always easy or obvious to use properly.

Bigus is offline  
post #25 of 25 Old 02-16-2016, 08:26 AM
Member
 
s_yang's Avatar
 
Join Date: Jan 2010
Location: Indiana
Posts: 34
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 15 Post(s)
Liked: 10
Ronny,


In your plate simulation, I would suggest you to remove the restraint on three edges and only keep on one edge. So it becomes a simple cantilever plate with uniform pressure loading. You can compare your simulation results with the analytical results, which is much simpler. You can also try shell elements as Bigus suggested and compare the results.
s_yang is offline  
Sponsored Links
Advertisement
 
Reply DIY Speakers and Subs

Thread Tools
Show Printable Version Show Printable Version
Email this Page Email this Page


Forum Jump: 

Posting Rules  
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off