Compression analysis for various configurations

[acolunga07]

Has anyone ran any kind of analysis on what the stress distribution is at the first cell vs. innermost cell on various compression designs?

I see two primary challenges when providing compression. The first, is how to provide the correct total amount of compression. There are several posts that have calculated this to be around ~660 lbs for a 280Ah cell as well as a reliable way to get there (compressing a spring a certain distance). The other challenge is how to distribute or spread out that load such that is acts as somewhat uniform pressure on the face of the battery. This can be a lot more challenging and deceptively difficult to do. I have 1/2-1" wood, aluminum or plastic plates used to accomplish this but no real analysis justifying the selection of one thickness/material vs the other.

I'm an engineer and have access to some decent software tools to analyze these kinds of things so I decided to test one option I was considering.
First analyzed design:
EVE304 cells
3/4" thick x 7" x 8" plywood spreader plates
1" x 1" x .125" angle iron beam
165 lbs applied at the washer contact surface area on the angle iron

The two stress plots show stress at first cells surface and at the midplane (between cell 2 and 3). As you can see, there is very unequal stress at the first cell (peak loads exceed 50 psi) but it is very uniform at the middle (right at 12 psi).

Ideas for future iterations include
-using 1/2" aluminum plate instead of plywood
-using c channel instead of angle iron
-insert a shim between the angle iron and plywood at the center

I am most excited about the last one because it seems like a simple/easy fix to shift some of the load to the center of the pack and away from the corners where the angle iron goes over the edge of the plywood. Also, in my application I am limited to 13.5" in depth, so that allows for 13.5" - (4 x 72mm) = 2.16" (for both sides total) of length to figure out a way to spread the load.

 

[Bob B]

I think it's one of those things where perfection is the enemy of good.
The more rigid the compression plate the better.

 

[Lt.Dan]

I did the same kinds of compression analysis as you, trying to design the optimal plate.

I did mine out of 3/8" 5052 aluminum, with 2 bends to rigidize the plate. This kept deformation of the plate to a minimum.

Placement of the rods was critical as well. My first assumption was close to the outer most corners, where in reality, the rods needed to be much closer together than I thought.

 

[SolaroSsaurius]

Has anyone ran any kind of analysis on what the stress distribution is at the first cell vs. innermost cell on various compression designs?

I see two primary challenges when providing compression. The first, is how to provide the correct total amount of compression. There are several posts that have calculated this to be around ~660 lbs for a 280Ah cell as well as a reliable way to get there (compressing a spring a certain distance). The other challenge is how to distribute or spread out that load such that is acts as somewhat uniform pressure on the face of the battery. This can be a lot more challenging and deceptively difficult to do. I have 1/2-1" wood, aluminum or plastic plates used to accomplish this but no real analysis justifying the selection of one thickness/material vs the other.

I'm an engineer and have access to some decent software tools to analyze these kinds of things so I decided to test one option I was considering.
First analyzed design:
EVE304 cells
3/4" thick x 7" x 8" plywood spreader plates
1" x 1" x .125" angle iron beam
165 lbs applied at the washer contact surface area on the angle iron

The two stress plots show stress at first cells surface and at the midplane (between cell 2 and 3). As you can see, there is very unequal stress at the first cell (peak loads exceed 50 psi) but it is very uniform at the middle (right at 12 psi).

Ideas for future iterations include
-using 1/2" aluminum plate instead of plywood
-using c channel instead of angle iron
-insert a shim between the angle iron and plywood at the center

I am most excited about the last one because it seems like a simple/easy fix to shift some of the load to the center of the pack and away from the corners where the angle iron goes over the edge of the plywood. Also, in my application I am limited to 13.5" in depth, so that allows for 13.5" - (4 x 72mm) = 2.16" (for both sides total) of length to figure out a way to spread the load.

Very interesting post!!
Thanks for sharing!

Please remember that what is in the spec sheet is 3 rods per side, which is ignored by most builders.
That will allow a much spread load...

Regards!

 

[acolunga07]

Very interesting post!!
Thanks for sharing!

Please remember that what is in the spec sheet is 3 rods per side, which is ignored by most builders.
That will allow a much spread load...

Regards!

Where is this specification?
And yes I was leaning towards either 3 angle iron members or 2 c chanel (1" x 3") which would create a 6" contact patch between the steel and wood/plastic load distribution plate.

 

[LakeHouse]

Has anyone ran any kind of analysis on what the stress distribution is at the first cell vs. innermost cell on various compression designs?

I see two primary challenges when providing compression. The first, is how to provide the correct total amount of compression. There are several posts that have calculated this to be around ~660 lbs for a 280Ah cell as well as a reliable way to get there (compressing a spring a certain distance). The other challenge is how to distribute or spread out that load such that is acts as somewhat uniform pressure on the face of the battery. This can be a lot more challenging and deceptively difficult to do. I have 1/2-1" wood, aluminum or plastic plates used to accomplish this but no real analysis justifying the selection of one thickness/material vs the other.

I'm an engineer and have access to some decent software tools to analyze these kinds of things so I decided to test one option I was considering.
First analyzed design:
EVE304 cells
3/4" thick x 7" x 8" plywood spreader plates
1" x 1" x .125" angle iron beam
165 lbs applied at the washer contact surface area on the angle iron

The two stress plots show stress at first cells surface and at the midplane (between cell 2 and 3). As you can see, there is very unequal stress at the first cell (peak loads exceed 50 psi) but it is very uniform at the middle (right at 12 psi).

Ideas for future iterations include
-using 1/2" aluminum plate instead of plywood
-using c channel instead of angle iron
-insert a shim between the angle iron and plywood at the center

I am most excited about the last one because it seems like a simple/easy fix to shift some of the load to the center of the pack and away from the corners where the angle iron goes over the edge of the plywood. Also, in my application I am limited to 13.5" in depth, so that allows for 13.5" - (4 x 72mm) = 2.16" (for both sides total) of length to figure out a way to spread the load.

I'm also and engineer with access to fun analytical tools that I use for my day job! I'm on the side of 'compression isn't worth it', but I'll leave any argument in that direction out of this. I just wanted to point out that it looks like you're modeling those cells as perfectly flat, but in reality they likely aren't, or at least, they wouldn't be after a few cycles.
More likely is a slight 'bulge' at each of the electrode locations. Not enough to be a problem or anything, but enough that the nice even stress distribution that you're showing in the center cells is likely not accurate. Even a 1/2mm difference changes all your contact locations enough that the load path should be straight through the electrode locations. This isn't to say that what you've does isn't useful, just that you can probably focus less on what's happening at the corners of your end cells as long as you build in enough stiffness to limit deflection of the end plates.

 

[acolunga07]

I ran a few more cases that are informative in a qualitative sort of way of nothing else.

Stress plot 3 has 1" x 1" .025" thick spacers centered on the angle iron to try and shift some of the compressive load to the center. That seemed to work too well (lost compression on the edges) and as expected, did nothing to center/even the load in the vertical direction. I played around with different thickness of spacers w/out improving the results.

Stress plot 4 has the same spacers but placed at the 1/3 and 2/3 horizontal locations. I also added a third member per the previous suggestions and this combination really helps quite a bit! I think this is definitely in "good enough" territory for me so I will probably build some version of this.

I might run some other analysis just for fun using different thickness aluminum plates (rather than plywood). I'll save the models and if anyone wants me to look at something else, either send me a private message are reply to this thread.

 

[Warpspeed]

The whole aim of the exercise is NOT to statically compress the cells, but to prevent them from expanding !
This small detail is frequently misunderstood by many people.

Under certain unusual conditions such as extremes of voltage, the internal gas pressure within the cell can be sufficient to swell and permanently deform the flat surfaces of the cell.
Its just not economic to make cells sturdy enough to withstand this abuse, so its recommended to constrain the cells within a very rigid box structure of some kind.

Think of it like a deflated balloon within a cardboard box. The balloon can only expand within the limits of the box, and is then constrained.

The box obviously needs to have sufficient structural strength, but is not designed to develop any compressive force upon the cells under normal operating conditions, where the pressure within the cell is, or should be close to atmospheric.
So really all you need are some reasonably stiff metal plates held together with something like threaded rod, with nuts that need only be finger tight, or just nipped up.

This is how I did mine. I fitted thirty 50Ah Winston cells into a standard filing cabinet drawer arranged 6 x 5.
I used 6mm steel checker plate which is far heavier than needed, but I had a lot of this material on hand, so that is what I used.
Three mm steel or aluminium would have been much better.
I used 6mm threaded rod between these plates, and the nuts were finger tight plus maybe one turn.
Its all been in operation now for five years without a single problem.

I am now up sizing my battery to use Chinese 280Ah cells, and fifteen of these also fit wonderfully well into a standard filing cabinet drawer 5 x 3.
Using two drawers that will house the thirty cells I require.

Its common practice to use 16 Lithium cells for a 48v system, but 15 cells work perfectly well where the voltage range of the inverter and solar controller are designed to work with either Lead or Lithium batteries. Likewise, 30 cells are all you need for a 96v system.
Dare to be different !!

 

[acolunga07]

Whether you provide pre-compression or not, the compressive stress distribution will have roughly the same behavior. I ran the following cases last night:

Stress plot 5: qty:2 Al channels (3" x 1" x .125") w/3/4" plywood spreader plate
Stress plot 6: qty:2 Al channels (3" x 1" x .125") w/3/4" Al spreader plate
Stress plot 7: qty:2 Al channels (3" x 1" x .125") w/3/4" Steel (A36) spreader plate
Stress plot 8: qty:2 Al channels (3" x 1" x .125") w/1/4" Steel (A36) spreader plate

General observations:
1) I actually ran .25" thick (same 3" x 1") Al c channels and didn't see any effect by downsizing to only .125" so I just stick with that (stress plot 5).
2) Two wider (3") c/u channels provide the same load distribution as 3 pieces of angle iron (see previous stress plots).
3) Stiffer materials (Al or steel) for the spreader plate does provide better load distribution. It seems that 3/4" plywood is comparable to 1/2" Al or 1/4" steel. You can optimize your spreader plate selection based on availability, ease to work/fabricate, cost, thickness (room to fit battery pack), performance and/or weight. Depending on your constraints, one option might make more sense than the other. In my case I am trying to optimize for thickness (very little room) and weight (don't want to overload trailer/tow vehicle and I already have little margin). If I had my way, the size/weight option would also be the cheapest and easiest to work with but it doesn't appear that way.

 

[HRTKD]

Stress plot 6: qty:2 Al channels (3" x 1" x .125") w/3/4" Al spreader plate

Can you confirm that this was 3/4" Aluminum plate? That seems rather thick for what we're trying to accomplish. I question whether the side of the cell can handle having a plate that is smaller than the side of the battery pressed into it.

I'm no engineer, nor do I have the tools you have, but as an academic exercise, what you've put together is very cool. As @Warrpspeed said, we're looking to keep the cells from expanding. My secondary goal was to ensure that the cells don't move. My cells are deployed in my bumper pull toy hauler. Given the rough roads I take my trailer down, cell movement would be bad and I would likely have connections coming loose.

After putting my DIY batteries into production, I considered redoing the compression fixture to use an aluminum plate (no C channel) instead of the 3/4" plywood. But it's working. I see no stress on the cell edges and my connections are staying tight.

 

[Warpspeed]

Stress on the terminals is a concern here too, and mine is not a mobile application.
I much prefer flexible cable and heavy crimp lugs than the cheap slotted rigid link kits that many of the the battery manufacturers provide.
This also enables having all the cells oriented the same way around.
That becomes an advantage if you are bolting balancing and monitoring boards directly to each cell.

As regards rigidity and the clamping method.
If multiple adjustable threaded rods are used, thinner plate should be well up to the job.
Aluminium needs to be about 50% thicker than steel for similar stiffness.
If every row of cells has a threaded rod at each corner, its going to be immensely stiff and strong.
For safety, some thought needs to go into insulating the top rods. I use neoprene tubing.
The cells can sit on thin wooden battens so the lower rods have a path beneath the cells.

Many ways to skin a cat, but seeing how others have done this may provide some food for thought.
After five years of successful operation, I will do my new battery upgrade in the exact same way, but use 3mm steel plate instead of 6mm steel plate.

 

[Warpspeed]

Whether you provide pre-compression or not, the compressive stress distribution will have roughly the same behavior. I ran the following cases last night:

I am an EE, not an ME, but are you sure that is right ?

If the clamping plates are compressed against a static battery case, as you show, I agree that the greatest compressive stress will be near the edges.

But if you use less initial compression during assembly, and then vastly increase the gas pressure within the cell, the greatest force should be at the centre of the flat cell. Cells tend to bulge more in the middle than the corners. That swelling effect is the force we need to counter.

 

[acolunga07]

Can you confirm that this was 3/4" Aluminum plate? That seems rather thick for what we're trying to accomplish. I question whether the side of the cell can handle having a plate that is smaller than the side of the battery pressed into it.

I'm no engineer, nor do I have the tools you have, but as an academic exercise, what you've put together is very cool. As @Warrpspeed said, we're looking to keep the cells from expanding. My secondary goal was to ensure that the cells don't move. My cells are deployed in my bumper pull toy hauler. Given the rough roads I take my trailer down, cell movement would be bad and I would likely have connections coming loose.

After putting my DIY batteries into production, I considered redoing the compression fixture to use an aluminum plate (no C channel) instead of the 3/4" plywood. But it's working. I see no stress on the cell edges and my connections are staying tight.

It was 3/4” Al plate. I subsequently ran 1/2” Al plate and it very much looked like 3/4” plywood or 1/4” steel (hence the comment in the conclusions section).

 

[acolunga07]

Stress on the terminals is a concern here too, and mine is not a mobile application.
I much prefer flexible cable and heavy crimp lugs than the cheap slotted rigid link kits that many of the the battery manufacturers provide.
This also enables having all the cells oriented the same way around.
That becomes an advantage if you are bolting balancing and monitoring boards directly to each cell.

As regards rigidity and the clamping method.
If multiple adjustable threaded rods are used, thinner plate should be well up to the job.
Aluminium needs to be about 50% thicker than steel for similar stiffness.
If every row of cells has a threaded rod at each corner, its going to be immensely stiff and strong.
For safety, some thought needs to go into insulating the top rods. I use neoprene tubing.
The cells can sit on thin wooden battens so the lower rods have a path beneath the cells.

Many ways to skin a cat, but seeing how others have done this may provide some food for thought.
After five years of successful operation, I will do my new battery upgrade in the exact same way, but use 3mm steel plate instead of 6mm steel plate.

Most of what is stated is pretty accurate. I disagree with the statement of simply 4 pieces of all thread at each corner doing the trick. The all thread should not be located at the corners. Instead they should be located such that they are located some distance (d1) from the end so that the max deflection of the plate at the end is equal to the max deflection between the rods (located d2 distance apart). If you work out the math, the ratio is about 9.6:1. So if the height of the cell was 11” (which it clearly isn’t but the math is easy for this example), then the all thread would be 1” from the top and bottom. This is assuming of course that there is no deflection going across the cell which is definitely not true. If you factor that in, it’s closer to 1.75” duch tgst the ratio of d1 (distance from all thread to the end of spreader plate/battery cell) to d2 (distance between all threads) should be close to 4.3:1. Either way, that deflection is proportional to d^4 so using 3 or 4 all threads along the height does wonders in terms of keeping deflection down.

Why is defection so important? Well, in the case of no springs, you’re relaying on the elastic response of steel to maintain compressive pressure while accommodating minuscule movement. Because steel is so stiff, you go from zero to incredibly high numbers and possibly back to zero very quickly. You could try using a torque wrench to relate the torque to some compressive force but that relationship has a lot of uncertainty. This is where springs work great. The bigger/lower the spring constant, the more consistent the force/pressure applied while accommodating small movement.

 

[acolunga07]

I am an EE, not an ME, but are you sure that is right ?

If the clamping plates are compressed against a static battery case, as you show, I agree that the greatest compressive stress will be near the edges.

But if you use less initial compression during assembly, and then vastly increase the gas pressure within the cell, the greatest force should be at the centre of the flat cell. Cells tend to bulge more in the middle than the corners. That swelling effect is the force we need to counter.

This is true. It really depends on how the cell would deform naturally unconstrained as well as the overall stiffness of the cells. If for example the inner two start to expand (but in a uniform way because of how it’s constrained) then it will push the first/last against the spreader plates increasing the existing pressure profile (same behavior but higher magnitude).

 

[Skypower]

I have found by practical experience that a “pusher block” with a centralized die spring and the fixed end of a single row box must be made of two pieces of 3/4” thick Radiata pine ply glued together. 1/2”and 3/4” glued together will eventually bow approximately .020”.

 

[Warpspeed]

This is true. It really depends on how the cell would deform naturally unconstrained as well as the overall stiffness of the cells. If for example the inner two start to expand (but in a uniform way because of how it’s constrained) then it will push the first/last against the spreader plates increasing the existing pressure profile (same behavior but higher magnitude).

This is perfectly logical thinking. If only two cells near the middle of a stack have been abused, and suffer from the swelling effect.

Normally we will be considering a series connected string of cells, where all the cells receive pretty much the same charge and discharge current, and in a perfect world, also have very closely matched terminal voltages and internal gas pressures.

Swelling is only caused by abuse, its not something that happens if you keep within the normal recommended operating voltage envelope.
Its much more likely that all the cells in a clamped row might have similar internal gas pressures at any time, either normal, or very high.

If one cell fails or goes rogue for some reason, all bets are off.

I would not worry about twenty thou of bulge.
If swelling occurs, its probably more like half an inch or an inch per cell in a whole row, cumulative !
That can only happen if no attempt at all has been made at clamping.
Even a half arsed attempt is better than nothing.

Here is a picture I found on the internet, they are not my cells, but fairly typical of the problem.

 

[acolunga07]

This is perfectly logical thinking. If only two cells near the middle of a stack have been abused, and suffer from the swelling effect.

Normally we will be considering a series connected string of cells, where all the cells receive pretty much the same charge and discharge current, and in a perfect world, also have very closely matched terminal voltages and internal gas pressures.

Swelling is only caused by abuse, its not something that happens if you keep within the normal recommended operating voltage envelope.
Its much more likely that all the cells in a clamped row might have similar internal gas pressures at any time, either normal, or very high.

If one cell fails or goes rogue for some reason, all bets are off.

I would not worry about twenty thou of bulge.
If swelling occurs, its probably more like half an inch or an inch per cell in a whole row, cumulative !
That can only happen if no attempt at all has been made at clamping.
Even a half arsed attempt is better than nothing.

Here is a picture I found on the internet, they are not my cells, but fairly typical of the problem.

Yes, it does seem that “something is better than nothing,” but I have read somewhere that we can expect better longevity (the difference between 4K and 6k cycles) by more optimal clamping arrangements.

And, in the event you’re receiving, having a stiffer arrangement (that nominally applies uniform pressure) will also increase pressure locally wherever bulging would want to occur. It the seems the same goal of trying to create uniform pressure will also combat local bulging more effectively. The main issue I see are designs are simple/cheap/less effective solutions that apply most of the compressive load at the edge or corners where it is least needed. It is my understanding that the bulging tends to occur in the middle which is where you need/want the compression. I can think of two ways to do that:

1) Get stiffer plates and/or structural members that go across the batter, or

2) insert shim/spacer between structural member and spreader plate to counteract the bending the bending of the structural member and intentionally determine where the force/pressure should be applied.

I explored both options in a analytical way for the benefit of the community (or at least those that were wondering about this).

As previously acknowledged, I recognize there is a very good argument that what I am doing is impractical but the same could be said about a lot of the things done/discussed around here. I am doing this because I find it interesting/fun and posting about it to share the information with those that might be interested.

 

[acolunga07]

I decided to run a case with stiff (aluminum) plates and no structural member. The design comes from boywonder here.

Stress plot 9: 1/2" thick Al spreader plate

This is my favorite so far. It seems that boywonder and I have similar space constraints so a stiffer and unfortunately more expensive material is the way to go.

 

[Warpspeed]

Stiffness is only required over the distance between tension members, or whatever holds it all together.

As with boiler stays in the fire box of a steam locomotive, lots of small closely spaced stays beat a few wide spaced larger ones to minimise bulging deformity under boiler steam pressure.
Weight is not really a constraint for us, because the cells themselves are going to be really heavy anyway.
Available space may be limited though.
I think steel might be the preferred material if it all has to be fairly compact with least cost.