How Do Compression Forces Affect Battery Electrolyte Distribution?

[kathie747]

Hi, I'm a newbie here.
I'm sorry if my questions are worded improperly or are too simple. I'm a chemist who hasn't had a physics class in over 20 years, so I'm a little rusty.

I have some questions on compression forces (and their propagation). I'm a battery chemist and have an electrolyte (liquid) uptake problem in a battery and have exhausted many chemical ideas, so now I'm looking at the more overall physical system. Before a battery is activated with an electrolyte, the stack gets compressed and put into it's container. You assume everything is even and the electrolyte gets evenly placed, but so much for assumptions sometimes.

To give you a visual of my question, this is kind of what my stack looks like with plates (+ and - ) and separators stack on each other (note, this is not the exact stack, just the first good goggle image I came upon).
http://chem.ch.huji.ac.il/history/kordesch_patent3.jpg

If I apply a force across the top, does that force eventually equilibrate to the applied force? Can there be localized greater forces than the applied force?

Now what if I have a nut and bolt running up through the middle of this cross-section (instead of the 4 at the corners as shown), and was tightening with a torque wrench, and assuming the endplates are strong (er?), how are the forces propagated?

And lastly, what technologies exist to measure these forces? Are there coupled wires I can insert in the stack and measure under, say a hydraulic press?

Thanks so much and let me know if you need more clarification. I'm starting at the basics first, before I might get specific.

Kathie

 

[Q_Goest]

Hi Kathie, welcome to the board.

If I apply a force across the top, does that force eventually equilibrate to the applied force? Can there be localized greater forces than the applied force?

Not sure exactly what you mean, but each bolt acts like an extremely stiff spring. The bolts stretch slightly (get longer) as you tighten them. In turn, the stack also acts as an extremely stiff spring, probably stiffer even. The stack will compress slightly (get shorter) as the bolt is tightened. The compressive stress under the bolt head & nut are highest, with a gradual dissipation of compressive stress (in the axial as well as radial directions) as you look farther away from those points. This compressive stress in the stack is where the stack is most compressed. So if you were to be able to measure the height of the stack before and after tightening the bolts, you’d find the stack is shortest where the bolts penetrate and there’s a bow to the stack where the height is greatest, farthest from the bolts (ie: in the middle of the stack for a square stack with a bolt in each corner).

Now what if I have a nut and bolt running up through the middle of this cross-section (instead of the 4 at the corners as shown), and was tightening with a torque wrench, and assuming the endplates are strong (er?), how are the forces propagated?

Same as above. Unless you say your end plates are ‘infinitely stiff’ (which they aren’t) then the height of the stack is shortest under the bolt head and highest at the point farthest from the bolt.

And lastly, what technologies exist to measure these forces? Are there coupled wires I can insert in the stack and measure under, say a hydraulic press?

Thanks so much and let me know if you need more clarification. I'm starting at the basics first, before I might get specific.

Kathie

Herein lies the problem. The change in length of the bolt is on the order of thousandths of an inch, so it’s extremely difficult to measure accurately. There are some methods, but rather than give you further information, I’d like to better understand what you’re looking for.

 

[kathie747]

Thanks Mr. Goest.
First, I found a picture that shows the exact cell I'm talking about (In more 3D).
http://www.pacificu.edu/as/chemistry/research/images/H2battery.jpg
The cell stack is pressed as a whole before a nut and bolt are even added, therefore, the compression forces would be even across the diameter of the circular plates. Then later, there are specs to which the center nut is torqued.

In turn, the stack also acts as an extremely stiff spring, probably stiffer even.

I think your dead on there, and that is the question (I think) I'm addressing. Each of the components in that stack take an unequal load of the force (when it gets pressed , although I don't know at which step). Not necessarily that each different type of component in and of itself does, although that does happen as each component is a different material.
But I think my question specifically is, can a group of components in the middle of that stack be less compressed than the components at the top or bottom? (the components repeat throughout, being made of positive and negative plates and separators in between)

A quick study was done (not by me) that showed the top and bottom of that stack each "felt" the same applied force (using some dye paper...very qualitative in nature) when the stack was pressed before adding the axial nut and bolt.

But the middle of that stack is showing more signs of degradation with life, and we are trying to figure out why, and intercomponent spacing is being examined (qualitatively at this point as I work in industry that doesn't allow for too much $$ in R & D, economy and all ).

So if you were to be able to measure the height of the stack before and after tightening the bolts, you’d find the stack is shortest where the bolts penetrate and there’s a bow to the stack where the height is greatest, farthest from the bolts (ie: in the middle of the stack for a square stack with a bolt in each corner).

Thanks for that description.
I've measured the thickness of the individual components to be the same as the stack height while bolted (more or less, within statistical error)...so I end up with free space being the porosity of the individual components. That's what has me going crazy. At this point, I can't go back and make measurement while everything was still in one piece.

Herein lies the problem. The change in length of the bolt is on the order of thousandths of an inch, so it’s extremely difficult to measure accurately.

aww, that seem to always be the case.
I had thermocouples in mind when I asked the question. I was thinking along the lines of strain guages (whetstone bridge type) http://signalprocessing.prosig.com/FatigueTesting/strain-gauge-close.jpgand[/URL] placing them throughout a stack in different locations and seeing what happens. But I'm not sure if that'd work and I'd have to buy some more, and I'm trying to make this a small one day project. I also need to look up this dye paper the other guys used.

Kathie

 

[Q_Goest]

Hi Kathie. I think what your saying is you have a problem with a battery stack that’s shown in the picture above. The battery stack is circular with a single bolt through the center. For some reason (I’m no battery expert) the center of this stack seems to degrade faster than the ends and you seem to be suggesting that this could be due to compressive loads on the battery layers. You seem to be implying that the lower the compressive stress, the faster the battery layers will degrade.

Just want to make sure we’re on the same page…

If that’s true, if the layers degrade more rapidly when they are under less of a compressive load than some minimum compressive load necessary for proper life, then you should be able to recognize this very quickly because the very end plates will degrade more on the outer circumference than on the inner circumference around the bolt hole. That’s because the bolt will produce a markedly high compressive stress under the head of the bolt and almost no compressive stress as you move out radially on this particular layer.

Take a look at this picture:
http://www.co-design.co.uk/dpg/bol/figs/bol_fig3_5.gif

It comes from this web page:
http://www.co-design.co.uk/dpg/bol/bol3.html

This picture shows a conical line beneath a bolt head where compressive stress is very close to zero. If you go radially outward from this line, away from the bolt, the compressive stress is essentially zero. Within this conical zone, compressive stress varies slightly as a function of radial distance from the bolt. So the closer to the bolt you are, the higher the compressive stress. (note – compressive stress is the force per unit area tending to compress the material)

You should also note that as the conical section gets larger in diameter, compressive stress goes down. That’s because we have to maintain ‘static equilibrium’ throughout the material. So for static equilibrium to be maintained, the magnitude of the compressive stress has to drop since the area increases. In other words, the total FORCE transmitted through any single layer remains constant, but where that load is distributed across a larger area, the compressive stress (force per unit area) must DECREASE.

So long story short, the layers of your battery might degrade more quickly as you move away from the bolt head, but you should also see this on the layer directly under the bolt head because this particular layer has almost no compressive stress as you move away from the bolt head.

~

Ok, this all gets complicated if you don’t have a conventional bolt head. If the ‘bolt head’ is the entire width of the stack, then it has to also be thick enough to transmit the load through the bolt. But for this to be the case, the bolt head would have to be very thick, and I suspect from looking at your picture, this isn’t the case. So you don’t have an infinitely rigid bolt head.

This also gets complicated by the modulus of elasticity of the various parts. If the modulus is relatively low compared to your bolt material, that changes things somewhat. You might want to post some drawings or better pictures of what’s going on.

Bottom line - the compressive force through any particular layer is the same. If you integrated the compressive stress across the entire area of any particular layer, the total force should be equal for every layer. Only the distribution of compressive stress changes as you move through the layers.

~

In the end, it seems you’re looking for a way to measure this compressive stress with a bit more accuracy. I can’t imagine how a thermocouple is going to help you, nor a strain gage. Strain gages have to be oriented in the direction you’re trying to measure strain. However, the idea of pressure sensitive film might work. Do a Google search (or ThomasNet) on pressure sensitive film. I suspect the film has to be ‘sized’ for a specific range of compressive load (I’ve never tried the stuff), so if you can provide more detail on bolt diam, threads per inch, torque, lubricants, etc… there are some fairly easy ways of determining what range of compressive stresses you should be looking for.

 

[timmay]

How do you measure degradation of components? From your previous post you mention porosity of the plates, so I'm guessing it's absorption of the electrolyte? And your concern is that absorption is due only to the porosity of the plates, and your central plates are more compressed than the outer ones, leaving less free volume? As the other poster has mentioned, pressure film and strain gauging techniques are both applicable.

For pressure sensitive films (such as Fuji Prescale) you will have to use a film with a specific pressure range. Without making any calculations, you could apply a selection of films with different pressure ranges to the component and see which works best - too low a pressure range, and your film will completely develop, too high a pressure range and your film will not show any development at all.

But once you get the right range you will be able to see the distribution of compressive stress acting on the surface of the component you're testing, as well as the magnitude. This information can either be used qualitatively or you can compare the developed films against a calibrated intensity chart. This could be carried out roughly by eye or by scanning and analysing the films.

A supplier I've had excellent dealings with are Sensor Products who can supply you with a pack of various grades of Fuji Prescale film if you are unwilling to calculate the approximate pressure range.

Strain gauging is the traditional way of measuring stress in a component, although it may prove difficult in your case especially if you've never used them before and do not have the capability of laying them up accurately and measuring their output. Strain gauges change resistance with change in length, and by converting from resistance change to strain change, and using the known elastic modulus for the component, you can measure the stress state at that point. Typically you could calculate axial stress on a component by mounting them axially on the side it, much as in your image.

However, I imagine your plates are about as thin as the strain gauges you would use, so you may have to mount the strain gauges on the normal face of the plate, radially from the centre, measuring radial strain and converting to axial strain. You could mount a series of these radiating from the centre of the plate and get an idea of the stress distribution, as you would see with the pressure film.

As I said though, unless it's something you or one of your colleagues has done before, you will probably find it pretty challenging. For structural and materials engineers it's a very useful skill to have though.

What temperatures are your processes carried out at? One possibility could be that you have different levels of thermal expansion in the middle of your cell to the ends, and that affects absorption. You could use thermocouples in this case to show the difference in temperature between the ends and the middle, and use this to estimate expansion due to temperature. You could even directly measure this if you were to strain gauge your plates in the manner I suggested above.

 

[Dr.D]

Kathie asked, "But I think my question specifically is, can a group of components in the middle of that stack be less compressed than the components at the top or bottom?" The answer is No, the total load on every cross section has to be the same. As Q_Goest has been pointing out, however, the distribution of force across the section can vary radially (assuming perfect symmetry), so that you might have much more load near the center than at the outer edges. This will depend a lot on the stiffness of the end caps.

Kathie, do your observations suggest this, or not?

 

[FredGarvin]

This also gets complicated by the modulus of elasticity of the various parts. If the modulus is relatively low compared to your bolt material, that changes things somewhat. You might want to post some drawings or better pictures of what’s going on.

Bottom line - the compressive force through any particular layer is the same. If you integrated the compressive stress across the entire area of any particular layer, the total force should be equal for every layer. Only the distribution of compressive stress changes as you move through the layers.

~

In the end, it seems you’re looking for a way to measure this compressive stress with a bit more accuracy. I can’t imagine how a thermocouple is going to help you, nor a strain gage. Strain gages have to be oriented in the direction you’re trying to measure strain. However, the idea of pressure sensitive film might work. Do a Google search (or ThomasNet) on pressure sensitive film. I suspect the film has to be ‘sized’ for a specific range of compressive load (I’ve never tried the stuff), so if you can provide more detail on bolt diam, threads per inch, torque, lubricants, etc… there are some fairly easy ways of determining what range of compressive stresses you should be looking for.

The big thing I see here is that it seems to me that all of the layers are not rigid enough, they need to be thought of as a gasket in stead of as a plain bolted member. If this is the case, the complexety increases dramatically.

The pressure sensitive paper is the way to go. There are a lot of applications where this is used with good results that are easy to see the load distribution.

 

[skeleton]

There is a simple but subtle solution for achieving almost uniform pressure across the sectional area of the battery pack ...

Apply some theoretical analysis in arriving at an appropriately shape bearing plate (I am referring to the plates at opposite ends of the battery pack, which the bolt and nut bear against). This analysis can be simplified by considering the 2-D version, with two bolt sets only.

Imagine a uniformly flat plate and clamped by the bolts. The pressurized battery will act as a uniform group of springs. This is called a "beam on elastic foundation" and its mathematical model is well documented under the subject of structural engineering. The analysis will reveal the corresponding deflection curve - which appears somewhat like a flattened greek omega (w) letter, with bolts at the two local apexes.

Now, fabricate the bearing plates in the same undulating profile as the deflection curve above. Install the bearing plate, but inverted onto the battery pack. Once the clamping force develops across the bolts, the bearing plate will flatten out. The pressure distribution will be nearly uniform.

 

[surinderjitsd]

there are many different forms of compression and i am not sure which one you are talking about to

 

[skeleton]

There is a simple way of determining the pressure distribution across the sectional area ...

Obtain a suitable material that is highly plastic (in the language of structural engineering this means deformable and non-recoverable (inelastic) - like plastercene or wax). It should be appropriately stiff to partially carry the expected pressure level. This material should be shaped as a rectangular plate with uniform thickness. I will call them "squash plates".

It can be inserted at three locations:
1) immediate below the bearing plate.
2) exactly at mid point in the bearing stack.
3) at quarter depth, corresponding to mid-way between positions #1 and #2 above.

Now, finish assembling the battery pack, and torque the bolts to the expected level. Next, disassemble the battery pack and remove the three squash plates. With a micrometer, measure the changed thickness of the squash plates across their 2-D widths. The reduction and variation in thickness is indicative of the variation in pressure profile. If you get that far, there is some mathematics (structural engineering) that relates the thickness with implied pressure.

 

[kathie747]

Dear All,
Just a quick note here to let you know I'm not ignoring you. You all have given some good relevant input and I've been mulling it over for the last days. I want to draw some pictures to see what you think, but it'll take some time to draw them, then get them on the net.

This isn't a new design, just a new problem. The only changes really are the electrical regimes the cell is undergoing and hence why everyone wants to point to electrochemical issues...but I'm thinking the design lends itself towards allowing the issues to happen. (engineers and scientists tend to approach these issues from different directions and hope to end at the same place. The engineers are wanting to solve the symptons, while we scientists want to find the cause. Of course the engineers end up "right" in the end as we can't change the design at this point (these are space batteries with some already in orbit.)

Hopefully I'll be back by tomorrow.

Kathie

 

[kathie747]

Alright, I'm back and I'm trying to address everything everyone has said thus far, thanks mucho everyone!
Q_Goest wrote

For some reason (I’m no battery expert) the center of this stack seems to degrade faster than the ends and you seem to be suggesting that this could be due to compressive loads on the battery layers. You seem to be implying that the lower the compressive stress, the faster the battery layers will degrade.

Well more or less, just speculating on what's cause and what's effect.
To get a little more specific, this is a fuel cell and during certain parts of cycling we have H2 and O2 recombining. With proper gas channelling, that meeting is friendly, but if we have gas pockets, that meeting can be less than friendly.
Q_Goest wrote

If that’s true, if the layers degrade more rapidly when they are under less of a compressive load than some minimum compressive load necessary for proper life, then you should be able to recognize this very quickly because the very end plates will degrade more on the outer circumference than on the inner circumference around the bolt hole.

and Dr. D wrote

do your observations suggest this, or not?

Well, you could say we have different things going. but I'm going to go and do a complete quantitative burn mark count, and I'll see if it follows a pattern. What complicates matters is we can only simulate so much here on Earth that won't be significant up in space, more specifically, gravity likely gives us different results here on earth.
Q_Goest wrote

Bottom line - the compressive force through any particular layer is the same. If you integrated the compressive stress across the entire area of any particular layer, the total force should be equal for every layer. Only the distribution of compressive stress changes as you move through the layers.

Thanks..got it and thanks for the link. So here are a couple of situations...let me know if I'm thinking in the right direction. And please pardon the inaccurate drawings, I did it on an excel spreadsheet.

Figure 1 is where the core/nut exert stress equally in both directions and meet in the middle as shown by the red lines. Now, if I had end plates that had a higher tensile strength (and sorry if that is not the proper term) than the nut/core system, could I get the distribution of stress as shown in figure 2.
And could you say it is likely that a system could actually have a mixture of the two figures? (as in, are forces always 45 degrees? and don't meet in the stack middle)

Now what might happen if I add a bunch of washers to one end as in figure 3? I have shown it with unequal distribution (and therefore I change the 45 degree angle.) Does one side show increased stiffness?

And lastly, say I insert an infinitely stiff (for the purposes of discussion) plate in the center as in Figs. 4 and 5. Do I get a distribution more like figure 4 or 5...or do the stack ends adjacent to the core/nut ends show the 45-degree distribution, and the stack center fans the entire cross section (like Fig 2).

I hope my questions make sense.
Timmay wrote

A supplier I've had excellent dealings with are Sensor Products who can supply you with a pack of various grades of Fuji Prescale film if you are unwilling to calculate the approximate pressure range.

Fred Garvin wrote

The pressure sensitive paper is the way to go. There are a lot of applications where this is used with good results that are easy to see the load distribution.

Thanks so much, I'll look into them. This likely what the other lab did a quick test with, but if they had it equally spaced from both ends, it isn't showing me a whole stack distribution.

Timmay wrote

As I said though, unless it's something you or one of your colleagues has done before, you will probably find it pretty challenging. For structural and materials engineers it's a very useful skill to have though.

I've mounted them on the outside of canisters and know how challenging they can be on smooth surfaces. I hadn't thought about the complications with the porous plates (which are about 1 mm thick). I think I ought to pass on that idea for the time being. Thanks for the heads up.

Timmay wrote

What temperatures are your processes carried out at? One possibility could be that you have different levels of thermal expansion in the middle of your cell to the ends, and that affects absorption.

Actually, it is temperature that effects many situations as the batteries are kept at about 0 C...and if we have water pooling, we likely have water freezing. Our fix will likely include heating the cells. I think thermal expansion is not as significant as the fact that the positive plates can swell as they change morphologically during cycling from chemical species that are incorporated into the structure during the electrochemical reaction (s). The plates though, are made to porosity specifications to allow for this chemical change(although with long life it becomes more significant.)

skeleton wrote

Now, fabricate the bearing plates in the same undulating profile as the deflection curve above. Install the bearing plate, but inverted onto the battery pack. Once the clamping force develops across the bolts, the bearing plate will flatten out. The pressure distribution will be nearly uniform.

Wow, that is so beyond my comprehension at the moment, but I'm assuming this is the essence of figure 2. But your discussion made me think of the fact that we have belleville washers on one end also. They are actually to allow for that plate expansion I mentioned above.

skeleton wrote

Now, finish assembling the battery pack, and torque the bolts to the expected level. Next, disassemble the battery pack and remove the three squash plates. With a micrometer, measure the changed thickness of the squash plates across their 2-D widths. The reduction and variation in thickness is indicative of the variation in pressure profile. If you get that far, there is some mathematics (structural engineering) that relates the thickness with implied pressure.

That is indeed a simple enough test that I hadn't thought of and will give a good measure. I am hoping not to have to get so complicated in the math, but that's what co-ops are for (haha).

Thanks

Kathie

 

[Q_Goest]

Hi Kathie,
It’s nice to see someone respond to their own posts with so much dedication so thanks for that. Let’s see if any of this helps . . .

Well more or less, just speculating on what's cause and what's effect. To get a little more specific, this is a fuel cell and during certain parts of cycling we have H2 and O2 recombining. With proper gas channelling, that meeting is friendly, but if we have gas pockets, that meeting can be less than friendly.

Ok, so maybe if the stack isn’t compressed, you get bigger gas pockets and that’s allowing larger quantities of H2 and O2 to pool up before combining? If that’s the case, then perhaps you’re thinking that if you can eliminate or minimize these pockets, there won’t be large buildups of H2 & O2. Hence the interest in a uniform compressive load?

How large are these poockets?

Figure 1 is where the core/nut exert stress equally in both directions and meet in the middle as shown by the red lines. Now, if I had end plates that had a higher tensile strength (and sorry if that is not the proper term) than the nut/core system, could I get the distribution of stress as shown in figure 2.
And could you say it is likely that a system could actually have a mixture of the two figures? (as in, are forces always 45 degrees? and don't meet in the stack middle)

The picture I’d posted a link to previously just shows a generic bolt and a generic compression distribution under the head. Really isn’t the best picture. Attached is another picture taken from a textbook I have. Here you can see the lines of constant compressive stress for a ‘typical’ bolted joint. It looks similar to the previous one but you can see the lines aren’t 45 degrees. Also note that directly beneath the bolt head is where the compressive stress is highest. This stress distribution will vary depending on the modulus of elasticity of the material (not tensile strength) and also depend on overall geometry.

In your case, the plates between the two bolt heads are probably not very ‘stiff’ (ie: they probably have a relatively low modulus of elasticity) so the stress distribution would look different because the material is different. However, the general distribution would be similar.

Now what might happen if I add a bunch of washers to one end as in figure 3? I have shown it with unequal distribution (and therefore I change the 45 degree angle.) Does one side show increased stiffness?

The only reason adding ‘washers’ (preferably a single ‘washer’ of the same thickness) would make any difference at all is because this would make the bolt head stiffer, but a typical bolt head is about as stiff as you’ll get so adding anything of the same diameter isn’t likely to change anything.

If your intent is to produce a uniform compression throughout the stack as shown in your figure 2, I can think of 2 ways that might be accomplished. The first way is as skeleton mentions. Have you ever seen an 18 wheeler pulling a flatbed trailer going down the highway, but the flatbed trailer is bowed up in the middle and there’s no load on it? The beams that make the trailer are actually made bent like that – they’re bent slightly so that when the trailer is fully loaded, they straighten out.

Now imagine a circular plate that’s made to do the same thing - flatten out when a load is produced in the center. It would look a bit like a contact lens. If you took 2 of these slightly bent circular plates and placed one on each end, such a plate could theoretically produce a uniform load over your entire stack of plates as you show in figure 2. However, to design such a plate would require a large team of coops, an expensive FEA program, a few dozen pizzas and a large box full of cash.

The second option may be patentable, so I’ll send you a PM.