Why Apex Seals Fail

[Barry Bordes]

Originally Posted by vex

How did I miss this thread?

Let me see if I can help in adding my schooling into the mix (Hurrah for Fluid Dynamics being applicable in all situations dealing with flow).

I think the ROOT cause of the flow issue is going to take some exploring. First and foremost what do the individual passages look like? I notice from the picture you appear to have a lot of, what appears to be, leakage between the sideplate and the rotor housing. If that's just an illusion then we can forego that issue. However if it's not I think a pressure differential is developing prior to the passage.

It is an illusion, there is no leakage. The #5 passage is just too small.

What I mean by that is that if you have a single source flowing unevenly into the various passages the incompressible nature of the fluid is going to dictate that the area with the higher pressure is going to output the more mass of the fluid (all other things being equal). Consequently ensuring you're getting an equal flow rate prior to the passage entrance would be paramount.

The water is entering through the normal pump passage.

If however there is no leakage and the input is simply just feeding it through the water ports where the waterpump seals, then the issue is indeed hardware related. Consequently since we can assume a uniform pressure distribution (not only for this experiment but for the real world application) we can see that the differences is going to be directly related to the hydraulic diameter and boundary layer conditions of the passages. This provides us with an easy enough solution as well.

First and foremost; enlarging the hydraulic diameter of the passages will allow for a more unified flow condition between all ports in question. There remains however a finite amount of space to achieve that. The other option that can be excersized as well would be to instill a turbulent boundary layer on only some of the passages.

Why only some? Think of it this way: we have a large passage that is putting out a lot of fluid. The issue is attempting to balance the passages to output the same amount of fluid at the end of the day. To do this we can lower the friction a majority of the fluid is exposed to in the other passages, in essence allowing for a faster flow to be achieved on smaller passages, while the undisturbed boundary layer on the larger passage causes the flow to slow down when compared to the tripped boundary layer of the smaller passages.

The solution will consist of finding the appropriate balance of the two.

The experiment is a simple but elegant representation.

Barry

[NoDOHC]

Originally Posted by vex

How did I miss this thread?

Let me see if I can help in adding my schooling into the mix (Hurrah for Fluid Dynamics being applicable in all situations dealing with flow).

Fluid dynamics is applicable, but I think that heat transfer principles and metal expansion rates have more bearing in this situation.

Originally Posted by vex

What I mean by that is that if you have a single source flowing unevenly into the various passages the incompressible nature of the fluid is going to dictate that the area with the higher pressure is going to output the more mass of the fluid (all other things being equal). Consequently ensuring you're getting an equal flow rate prior to the passage entrance would be paramount.

Technically, all passages have the same pressure across them (all tied to the same hose on one end, all rising the same amount, all exiting to atmosphere) The issue is that all passages have different flow rates due to their differing hydraulic diameter. The flow rate does not need to be consistent, as there is very little heat to be rejected at the beginning of the intake stroke as compared to the compression region.

Originally Posted by vex

If however there is no leakage and the input is simply just feeding it through the water ports where the waterpump seals, then the issue is indeed hardware related. Consequently since we can assume a uniform pressure distribution (not only for this experiment but for the real world application) we can see that the differences is going to be directly related to the hydraulic diameter and boundary layer conditions of the passages. This provides us with an easy enough solution as well.

We can solve for the hydraulic diameter given the (measured) flow much more easily than we can compute the other direction. Once again, this is not important, heat transfer is much more important here. Fast moving water is not always better.

Originally Posted by vex

First and foremost; enlarging the hydraulic diameter of the passages will allow for a more unified flow condition between all ports in question. There remains however a finite amount of space to achieve that. The other option that can be excersized as well would be to instill a turbulent boundary layer on only some of the passages.

Enlarging the passages is not always the best solution. What racing beat is doing is much better. They are increasing the surface area that is exposed to the cooling fluid while also inducing turbulence which improves convective heat transfer significantly.

Originally Posted by vex

Why only some? Think of it this way: we have a large passage that is putting out a lot of fluid. The issue is attempting to balance the passages to output the same amount of fluid at the end of the day. To do this we can lower the friction a majority of the fluid is exposed to in the other passages, in essence allowing for a faster flow to be achieved on smaller passages, while the undisturbed boundary layer on the larger passage causes the flow to slow down when compared to the tripped boundary layer of the smaller passages.

Once again, consistent passage-to-passage flow will not make this problem go away. The issue is much more complicated than that. The heat transferred to the coolant in each passage needs to be equal to the heat generated in the nearby chamber and transferred through the aluminum housing. I am guessing that Mazda did some research on this, as the coolant passages on top of the housing (low heat rejection to the coolant) have restrictors in them.

Originally Posted by vex

The solution will consist of finding the appropriate balance of the two.

I admire your attitude and your desire to apply your knowledge to any problem, I just don't want you wasting a lot of time chasing wild geese.

[NoDOHC]

Originally Posted by Barry_Bordes

NoDOHC, check the remaining good seal under a microscope for a crack where the others failed.

Also what water pump are you running?

Did you bevel the spark plug holes? It appears that way from the pictures.

Barry

I looked at it as carefully as I could and didn't see anything (I didn't want to get yelled at for using the magna-flux at work again).

I re-used that seal. I have over 2500 miles on the engine since the failure and have had no issues.

Stock water pump, stock radiator (a lot of fins are rotted out, I am replacing it).

I read this thread earlier and I found the tell-tale pattern on my housings, so I figured that beveling the edges was a good idea that may keep me from losing an engine. It survived 3000+ miles on the street just fine, but didn't survive the dyno run.

I should have backed out of the throttle and had the dyno turn the loading down, My coolant temp was about 125 C when the seal went. I went to the dyno since then and had no issues (with the temp under 115 C) I think that the spark plug boss grows too much relative to the unsupported housing around it. I have considered welding a brace in (parallel with the coolant flow) to keep the surrounding housing growing at a similar rate to the spark plug hole (although the spark plug hole has threads in it).

I like racing beat's idea - if I had this information before I rebuilt the engine, I probably would have done that.

[vex]

I think you're missing what I'm saying.

Originally Posted by NoDOHC

Fluid dynamics is applicable, but I think that heat transfer principles and metal expansion rates have more bearing in this situation.

Quite possibly and is something to consider. However the portion I was dealing with was the difference in the amount of fluid passing through the different passages. Not necessarily the heat transfer as this is two different issues of concern.

Technically, all passages have the same pressure across them (all tied to the same hose on one end, all rising the same amount, all exiting to atmosphere) The issue is that all passages have different flow rates due to their differing hydraulic diameter. The flow rate does not need to be consistent, as there is very little heat to be rejected at the beginning of the intake stroke as compared to the compression region.

Which I already stated.

We can solve for the hydraulic diameter given the (measured) flow much more easily than we can compute the other direction. Once again, this is not important, heat transfer is much more important here. Fast moving water is not always better.

Again, you're trying to tackle a different problem than to what I was referencing (not heat transfer).

Enlarging the passages is not always the best solution. What racing beat is doing is much better. They are increasing the surface area that is exposed to the cooling fluid while also inducing turbulence which improves convective heat transfer significantly.

Again, I wasn't discussing heat transfer specifically but balancing the mass flow between all passages. The heat transfer I did touch upon was coincidental and nothing more.

Once again, consistent passage-to-passage flow will not make this problem go away. The issue is much more complicated than that. The heat transferred to the coolant in each passage needs to be equal to the heat generated in the nearby chamber and transferred through the aluminum housing. I am guessing that Mazda did some research on this, as the coolant passages on top of the housing (low heat rejection to the coolant) have restrictors in them.

Again, I was only referencing the specific post I responded to.

I admire your attitude and your desire to apply your knowledge to any problem, I just don't want you wasting a lot of time chasing wild geese.

Asked and answered a question posted by Barry.

[vex]

Originally Posted by Barry Bordes

Did you guys read the cooling mod thread?

http://www.rotarycarclub.com/rotary_...t=barry+bordes

This thread, (Why Apex Seals Fail) is a peek into one of the conclusions that we came too. We are already running engines with this cooling mod plus some others surmised from this and other tests.

Some think that this may be giving away too much valuable information. My thought is that it helps the whole rotary community.

What I have noticed is that individuals start using the info and forget to give credit to the originators. But that is life... you do what is right anyway.

Along that theme I would like to give credit to Carlos Lopez and Rick Engman for their thoughts on the subject. They are the community's Senior Rotary Advisors.

Barry

I have. I posted in that thread starting on page 2 and page 3. Namely heat transfer and material expansion rates (roughly quoting the math unless you place the free edge of the fin within 0.00076 in then you have nothing really worry about for interference issues).

[Barry Bordes]

Originally Posted by Barry Bordes

If you look closely at this photo you will see the Racing Beat type grooves in the this first iteration cooling mod.

It didn’t seem to help much to alleviate the growth problem as I had hoped.

The next set of mods included severed fins and larger balanced passages.

[project86]

Barry,

You said you were using reliability mods.. What mods exactly were you using? Ported? etc, etc, Just curious

[TitaniumTT]

It seems to me that the stain is in line perfectly with the fin. It appears to me that the fin has kept that area from growing. Effectively retracting it as the rest of the housing grows. Looking at the 16x and the renni, they seem to aleviate some of this material. I'm wondering if this is in an attempt to not contain it, but to let it grow a the same rate. Thoughts?

[vex]

Originally Posted by Barry Bordes

If you look closely at this photo you will see the Racing Beat type grooves in the this first iteration cooling mod.

It didn’t seem to help much to alleviate the growth problem as I had hoped.

The next set of mods included severed fins and larger balanced passages.

Barry, as NoDOHC mentioned I do not think balancing the flow in the passages will help alleviate the problem. The issue I believe may be compounded by too high a temperature difference in the cooling surfaces (IE across the passages). Physically speaking you have a large temperature difference on the lower half of the engine while the upper half is more nominal.
Increasing the engergy transfer around the spark plug hole I believe will yeild more results than not.

Something else to consider is that the cooling of the combustion walls is not necessarily the root cause of the problem. I'll have to do some research into the material properties when I'm back home, but could it be the atomic structure of the materials developing fissures over time due to heat cycling? For instance would voids and imperfections in the casting develop the same event?

Originally Posted by TitaniumTT

It seems to me that the stain is in line perfectly with the fin. It appears to me that the fin has kept that area from growing. Effectively retracting it as the rest of the housing grows. Looking at the 16x and the renni, they seem to aleviate some of this material. I'm wondering if this is in an attempt to not contain it, but to let it grow a the same rate. Thoughts?

It's a possibility, but has anyone actually analyzed the stain itself? Do we assume it's a discoloration of the Chrome lining or do we know if it's a Carbon impregnation of the Chrome? Additionally do we know how deep the discoloration is within the Chrome? I ask as that--I believe--will dictate the environment and conditions of the failure. If it's a discoloration of the Chrome caused by too slow an energy transfer through the matrial adjusting the cooling will help, if it is a thermo-mechanical interaction that causes the issue, perhaps modification to the cooling passages to allow easier expansion and contraction would help. From what I see thus far it's a lot of conjecture--can anyone do this testing?

[GoopyPerformance]

Originally Posted by Barry Bordes

But we still haven’t gotten to the root cause. The real problem is the raised spark plug area on the housing. There are at least three approaches to this problem:
.

Our surface re-refreshing has virtually eliminated the spark plug hole raising issue. In fact this high spot is more detrimental than housing edge damage.
The cracks in the OEM Apex Seals are due to their hardness.

I noticed 20-30 compression gains on rotor housings when re-freshed using our Apex Seals and these are housings that would otherwise go to the trash.

A few years ago we posted a 135PSI 12A engine that we built but no one
believed now Judge ito has been building 130 to 140PSI Engines daily with used (refreshed) rotor housings but one thing is for sure the raised spark plug area is a problem if not addressed.
http://rotarycarclub.com/rotary_foru...t=12967&page=5

[Barry Bordes]

Originally Posted by GoopyPerformance

Our surface re-refreshing has virtually eliminated the spark plug hole raising issue. In fact this high spot is more detrimental than housing edge damage.
The cracks in the OEM Apex Seals are due to their hardness.

I noticed 20-30 compression gains on rotor housings when re-freshed using our Apex Seals and these are housings that would otherwise go to the trash.

A few years ago we posted a 135PSI 12A engine that we built but no one
believed now Judge ito has been building 130 to 140PSI Engines daily with used (refreshed) rotor housings but one thing is for sure the raised spark plug area is a problem if not addressed.
http://rotarycarclub.com/rotary_foru...t=12967&page=5

Jonathan, have you thought about heating the plug area while grinding the housing surface?
I believe the factory started grinding either the apex seals or the housing surface that way for better fit while running.
Barry

[j9fd3s]

Originally Posted by Barry Bordes

Jonathan, have you thought about heating the plug area while grinding the housing surface?
I believe the factory started grinding either the apex seals or the housing surface that way for better fit while running.
Barry

i'm not sure where this is mentioned, but a new factory apex seal isn't flat, its bowed slightly, so that when its in a working engine and up to temp its flat.

another problem on used housings, this is why they shrink, and why we measure for width, is that the steel liner actually collapses in the compression area, so the center of the housing bows in. its very bad to have leakage between chambers...


mike

[Barry Bordes]

Originally Posted by j9fd3s

i'm not sure where this is mentioned, but a new factory apex seal isn't flat, its bowed slightly, so that when its in a working engine and up to temp its flat.

another problem on used housings, this is why they shrink, and why we measure for width, is that the steel liner actually collapses in the compression area, so the center of the housing bows in. its very bad to have leakage between chambers...
mike

Mike, a slightly different interpretation of the housing wear could be that the high temps in the spark plug area cause expansion between the torqued plates which in turn compresses the surface, distorts the face, and chafes the faying surfaces (which causes the undersize housing measurements).

Notice the temps where we want to provide extra cooling are doubled that of atmospheric boiling water.
Barry

[j9fd3s]

that very well could be. next time you have some used housings handy, take a straight edge and lay it across the surface, its above the spark plugs that seems to do the worst.

i did build a couple of motors with a stock diameter stud kit, and i was told to tighten it to 50lbs, but i think this is actually bad. it also certainly didn't solve the actual problem which was stock ecu/injectors with a street port and big turbo! or more correctly the owner.

after thinking about it, i think the small diameter studs actually hurt more than they help. the studs DO NOT locate any of the housings any better than the stock bolts, plus #50lbs of torque the rotor housings probably ARE bowing, which is bad!

mike

[vex]

So I did a little more research and I think this modified picture is quite enlightening. Sorry it's hard to make out but it's the best I can do with GIMP at the moment.


We now have flame fronts with angles of eccentric shaft and the temperatures at those flame fronts that match up with the chart. I have a few more pieces of information that I will be posting throughout the week coming from SAE paper 860560:
Material Technology Development Applied to Rotary Engine at Mazda
by
Takumi Muroki and Jun Miyata