Cooling System Build Thread

[Longitudinal]

The water pump bolts to the side of the engine block at the very bottom, and the thermostat (black plastic housing) is bolted to the pump itself.

Why would a controllable electric water pump want or need a thermostat? It seems like unnecessary clutter and complexity to have a thermostat when the pump can be slowed/stopped or sped in order to keep engine temp at target.

 

[Jonohhh]

Why would a controllable electric water pump want or need a thermostat? It seems like unnecessary clutter and complexity to have a thermostat when the pump can be slowed/stopped or sped in order to keep engine temp at target.

It's explained in greater detail above- there is more to engine temperature than the temp at the outlet. The thermostat is useful for its bypass capability, which is also why it is used in BMWs.

 

[lookforjoe]

looks like it (from the tech doc he linked)

 

[Jonohhh]

I believe the CWA200s most common variants do have a removable housing, though the gasket supposedly puts up a fight.

You can also buy the pump itself with a variety of fittings- not just standard clamp types. Quick disconnect style fittings are available too but those tend to become a pain if you have long intervals between service.


The CWA400 comes apart too but it's a bit more complicated, I believe. I haven't been looking into the physical aspect much recently as most of that's as worked out as it needs to be until I get the pump controllable with my controller.

Is the Pierburg pump's compressor housing removable in case I would want to modify it by welding on fittings or elbows?

 

[Jonohhh]

Why would a controllable electric water pump want or need a thermostat? It seems like unnecessary clutter and complexity to have a thermostat when the pump can be slowed/stopped or sped in order to keep engine temp at target.

Sorry for the very short explanation- I wasn't particularly happy with it so here's a follow up.

If the only coolant path is through the engine and then the radiator and back around, regulating engine coolant outlet temperature by slowing coolant flow will cause an excessively cold temperature at the engine inlet (and thus across the engine), since the water flow rate through the radiator is so low, it looses a lot of temperature.

This is very, very bad for composite engines like ours, causing head gasket damage and undue stresses in the engine.

 

[Longitudinal]

Sorry for the very short explanation- I wasn't particularly happy with it so here's a follow up.

If the only coolant path is through the engine and then the radiator and back around, regulating engine coolant outlet temperature by slowing coolant flow will cause an excessively cold temperature at the engine inlet (and thus across the engine), since the water flow rate through the radiator is so low, it looses a lot of temperature.

This is very, very bad for composite engines like ours, causing head gasket damage and undue stresses in the engine.

That may be true, but it's not as though we are talking about pumping ice cold water into the engine. Although I do not know what temperature to expect from the coolant returning from the radiator, I suspect that it is maybe 150F, compared to a 200F target temperature for the engine. It seems that a temperature sensor plumbed in close to the engine downstream of the pump as well as one at the engine outlet, both tied into the control system you will build, would eliminate the need for a thermostat. Maybe there are finer points that I am still overlooking.

In my way of thinking, the advantage to the electric pump is its ability to simplify the cooling system, such as in a K swap where the intake side of the engine is already so crowded. Eliminating the thermostat along with the engine driven WP greatly simplifies the system. I understand that, in such a scenario, coolant flow to the heater core is not always available unless more complexity is added back in, possibly in the form of a parallel system that uses an after-run pump to provide coolant from the engine directly to the core. It seems that power gain from going EWP would be negligible since the engine still has to power the pump, but now through the charging system instead of mechanically.

Your idea is probably less about power gain and simplicity and more about control and precision.

 

[Jonohhh]

That may be true, but it's not as though we are talking about pumping ice cold water into the engine. Although I do not know what temperature to expect from the coolant returning from the radiator, I suspect that it is maybe 150F, compared to a 200F target temperature for the engine. It seems that a temperature sensor plumbed in close to the engine downstream of the pump as well as one at the engine outlet, both tied into the control system you will build, would eliminate the need for a thermostat. Maybe there are finer points that I am still overlooking.

In my way of thinking, the advantage to the electric pump is its ability to simplify the cooling system, such as in a K swap where the intake side of the engine is already so crowded. Eliminating the thermostat along with the engine driven WP greatly simplifies the system. I understand that, in such a scenario, coolant flow to the heater core is not always available unless more complexity is added back in, possibly in the form of a parallel system that uses an after-run pump to provide coolant from the engine directly to the core. It seems that power gain from going EWP would be negligible since the engine still has to power the pump, but now through the charging system instead of mechanically.

Your idea is probably less about power gain and simplicity and more about control and precision.

That is exactly true- control and precision are my absolute goal with this project...and having a cooling system that not only *works* but is actually optimal.

Keep in mind the coolant is entering an iron block, heating up, and then going into an Aluminum head. Even if coolant temperature across the engine was zero- there will still be stresses due to the difference in coefficient of thermal expansion between the two. Add in a 50+ degree higher temperature in the part with a higher CoE and you end up with an unhappy head gasket interface. On top of that- the delta is then subject to change based on driving conditions. At high rpm and load, the temp across the engine may not be too bad. Sitting in traffic at idle with the fans on but nearly no coolant flow... you're looking at a high delta. Cycling is just as bad for it, and it's disconcerting to drive around knowing that's happening, so I'm looking to fix it.

A variable speed water pump alone cannot replicate the advantages of a bypass thermostat. Without a thermostat to allow bypass (allows water to circulate from the engine outlet back to the engine inlet without going through the radiator), it is not possible to regulate coolant temperature at the engine inlet and you'll end up with sub optimal temps at the inlet.

There is a reason BMW uses a bypass thermostat in every car they make, regardless of if it has a mechanical pump or an electric water pump that has about as good of a control strategy as you can have.

Mechanical pumps tend to draw an excess of power since at very high rpm, they spin much faster than their efficiency range, but it's a necessary evil in order to have a pump that works properly at idle. Huge gains in efficiency can be found with electric pumps, provided they're controlled properly and the bypass thermostat used is chosen correctly for the target operating temperature. If you calculate the actual amount of horsepower the pump consumes based on its wattage, and assume some number for the efficiency of the charging system, it's actually surprisingly low- much lower than that of the mechanical pump.


If the heater ends up needing more help, I'll run a wire from the blower motor feed (after the resistor) to the Arduino and have it bias pump speed based on blower speed. I considered using an auxiliary pump as a heater booster but it may not be needed- we'll see.

 

[Jonohhh]

So, I got to test fitting today as I got bored of staring at my computer screen.

The bad news is that my nice and shiny stainless expansion tank is probably gonna have to go. The good news is that uh...the end of my coolant feed pipe is rusted right until the point that I'll likely need to chop it at in order to get the pump to fit well.

I may go for a 500 Abarth expansion tank if I need something short and wide, or an E36 M3 one if I need one that's tall and skinny. We'll see. I think using an Abarth one and placing the pump underneath it is my best bet at the moment, as I can then use the lower expansion tank mount for the pump. In that case the hoses will be long enough to allow the pump to be fixed relative to the body, with the hoses absorbing the compliance of the motor mounts.

 

[kmead]

So, I got to test fitting today as I got bored of staring at my computer screen.

The bad news is that my nice and shiny stainless expansion tank is probably gonna have to go. The good news is that uh...the end of my coolant feed pipe is rusted right until the point that I'll likely need to chop it at in order to get the pump to fit well.

I may go for a 500 Abarth expansion tank if I need something short and wide, or an E36 M3 one if I need one that's tall and skinny. We'll see. I think using an Abarth one and placing the tank underneath it is my best bet at the moment, as I can then use the lower expansion tank mount for the pump. In that case the hoses will be long enough to allow the pump to be fixed relative to the body, with the hoses absorbing the compliance of the motor mounts.

The other go to seems to be the spherical VW one which was used on the A4 bodied Golfs/Jettas/New Beetles from 1999.

Volvo and Porsche also have some nice squarish tanks which are flatter rectangles.

Using a bypass thermostat, basically just having the pump installed in effect like the OE pump as just a conveyer of coolant at the end of the ‘coolest’ return to push to supply is the best way to manage the temp in a tight range. Keeping the engine in the zone with minimal cold shocks is desirable and why the progressive thermostat in the X housing will work very well.

Using the temperature of the head or the outlet from the head at the thermostat housing to regulate speed of the pump makes a lot of sense. When cold the pump runs slowly, as the coolant heats up the pump turns faster and the thermostat regulates the temperature by progressively metering in cold return coolant from the radiator as needed.

The BMW electronically regulated thermostat is an interesting solution as it can adjust engine temperature based on the profile of the performance envelope being used. For our rather understressed engines with its crude fuel regulation (carb or OE FI) it would be overkill and would need to be programmed and managed. It would also be another widget to fail. Our wax and spring thermostats can last most of the car’s lifetime.

 

[Jonohhh]

The other go to seems to be the spherical VW one which was used on the A4 bodied Golfs/Jettas/New Beetles from 1999.

Volvo and Porsche also have some nice squarish tanks which are flatter rectangles.

Thanks for the tip- I'll check them out! I have Abarth and M3 ones sitting around here but if I have any fitment issues whatsoever I'll venture into those as possibly better options.

The BMW electronically regulated thermostat is an interesting solution as it can adjust engine temperature based on the profile of the performance envelope being used. For our rather understressed engines with its crude fuel regulation (carb or OE FI) it would be overkill and would need to be programmed and managed. It would also be another widget to fail. Our wax and spring thermostats can last most of the car’s lifetime.

I entirely agree. The complexity isn't worth the gain for our application. They're still regular wax thermostats- just with a heating element in order to trick the thermostat into opening sooner- but coming up with an algorithm to control it is then a bit of a mess because the whole point of it is to maintain a constant cylinder head temperature. Not to mention- they are far too hot for our engines. Without heater intervention, coolant runs at about 112c ( or 104 in some applications ). Perfectly fine for an engine engineered to run at a whopping 234f on purpose...not so fine for ours.

 

[Jonohhh]

(EDIT - wrote this last night & forgot to post it)

Yes, pics of physical installation are more helpful than diagrams to me FYI - mine is K24 drivetrain, not Fiat 1500

Yes, if ambient temps are below around 80ºF, the cooling system temp will typically never reach normal op (190ºF) unless I let it sit idling for long enough for that to happen, and then it will drop to around 170ºF when moving under light load.
Under heavy load & higher RPM (5-8K rpm) temp will typically drop below that.

View attachment 62286

Yesterday, with temps over 90ºF, the operating temp rose promptly to 190 in light load driving & stayed at 190-200ºF in operation, only dropping to 170 range when driven hard.

My rad is custom all aluminum 29.25"x12.75"x approx 2". Stock Honda/Acura water pump, no restrictive components in the whole system (besides t/stat)

I had to use a 160ºF t/stat, as with the coolant temp drop created by the thermostat extension housing (required to maintain use of bypass t/stat setup) off the water pump housing, the engine would run hot (220ºF) with a standard 180ºF t/stat.

View attachment 62290

EDIT 06/01 - 70ºF ambient at 6:20am today - gauge never got over about 150ºF

View attachment 62367

First: I love your car man.

And second:

So, with a 160f stat, it makes sense that the engine coolant temperature drops as soon as flow increases. Since your operating temperature (or rather, where the system reaches equilibrium) is 30+ degrees above your thermostat temp, it's most likely that during idle and low rpm operation where your coolant tends to be hotter, the stat is opening fully and engine outlet temperature is being regulated by the flow rate through the system only. As soon as you get to high rpm and the cooling system can dissipate more heat, temperature falls to the point that the thermostat begins to close and the bypass opens- regulating temperature more closely.

Since the stat is essentially doing nothing to regulate temperature at your 190-200f equilibrium point, its going to be heavily dependent on airflow through the radiator (when in motion) and ambient temperature (when sitting still)- potentially explaining why ambient temperature has such a strong impact. I wouldn't be surprised if, under the same operating conditions, coolant temperature remains a fairly steady amount above ambient temperature- tracking closely with it on colder and warmer days. If this is not the case, then something strange may indeed be going on.

Ideally you could work out a way to go back to using a warmer thermostat- or one that the fully open temperature is closer to the crack temperature, which should offer similar characteristics while hot, without letting temperature fall as low during more ideal cooling conditions.


Now, it's strange that it was operating below 160f under some conditions... that should not be the case if your thermostat begins to open at 160f.
Does the thermostat you chose have a fail safe latch? It's possible that being such a low temperature stat, the safety mechanism is also set to trip at a lower temperature, has tripped after hitting 200f+, and is no longer closing. Typically these failsafe thermostats are very simple- there's a little spring tab inside the stat that, if the piston hits the end of its stroke, gets jammed on the cylinder, preventing it from closing. They do not automatically reset- it becomes mechanically jammed open at that point. May be something to look into.

 

[Jonohhh]

Well, I had to make a tough decision the other night, but I think it's wiser in the long run.

While I was making progress on learning C++, and the code was coming together okay-ish, I think I have decided that for the time being, I need to suck it up, get the TinyCWA controller from Techomotive, and revisit doing my own controller solution at a later date.

Between my Summer Calculus course, replacing the Abarth's engine, doing maintenance on the families cars, and trying to make some money this Summer, I simply don't think there is enough time for me to not only learn a new language (coding), but also come up with a robust and resilient Arduino implementation.

On top of the Arduino itself, I would also need a logic level shifter to shift the PWM signal to 12v instead of the Arduinos 5v, and a voltage regulator and filtering circuit for the Arduino itself to prevent the harsh automotive environment from murdering it at the first voltage or EMF spike. All of this has to be connected, likely via a perf board. That's lots of solder joints (done by an amateur) to fail- and just a single one could disable the car. The only thing I don't like about the controller is that it cannot PWM control the fans based on a radiator temperature sensor- it only uses engine temp, and that's something I'll look to improve later. I'll likely leave the tinyCWA controlling the pump in the long run, but do my own solution for fan control as the two systems can be completely disconnected from one another.


Beyond that though, I have finalized how I am going to handle the pump and expansion tank mounting. By going to a more compact expansion tank (of a better design, too), I will be able to fit the pump and the expansion tank into the space previously occupied by the factory expansion tank only. Since they're so close, the pump and expansion tank will share one 3D printed bracket, and sit together in a "stack" containing the two, that mounts using the factory mounting solution for the original expansion tank. I'm pretty confident this should work well and be easy to service if I ever need to, so I'm pleased with this solution so far.

I hope to have a design for the lower mount done tonight- though I can't go much further than that as I am still waiting for the new water pump to come in.

I have been making update videos but not posting them, since I haven't been able to make one I am happy with, so sorry for the lack of updates.

 

[Jonohhh]

Tecomotive kindly provides CAD of the CWA200, which makes my life much easier. So far, I have managed to incorporate features for the pump and expansion tank mounting relative to each other- but I have not yet worked on the features that actually mount it to the car. I'll need to get my printer working before moving ahead with fine adjustments to the design, as at this is very much a rough draft, and is in no way strength optimized yet.

I wish I had access to a 3D scanner because the dimensions of this expansion tank aren't particularly friendly to someone who only has access to a set of calipers and engineering paper.

 

[lookforjoe]

First: I love your car man.

And second:

So, with a 160f stat, it makes sense that the engine coolant temperature drops as soon as flow increases. Since your operating temperature (or rather, where the system reaches equilibrium) is 30+ degrees above your thermostat temp, it's most likely that during idle and low rpm operation where your coolant tends to be hotter, the stat is opening fully and engine outlet temperature is being regulated by the flow rate through the system only. As soon as you get to high rpm and the cooling system can dissipate more heat, temperature falls to the point that the thermostat begins to close and the bypass opens- regulating temperature more closely.

Since the stat is essentially doing nothing to regulate temperature at your 190-200f equilibrium point, its going to be heavily dependent on airflow through the radiator (when in motion) and ambient temperature (when sitting still)- potentially explaining why ambient temperature has such a strong impact. I wouldn't be surprised if, under the same operating conditions, coolant temperature remains a fairly steady amount above ambient temperature- tracking closely with it on colder and warmer days. If this is not the case, then something strange may indeed be going on.

Ideally you could work out a way to go back to using a warmer thermostat- or one that the fully open temperature is closer to the crack temperature, which should offer similar characteristics while hot, without letting temperature fall as low during more ideal cooling conditions.


Now, it's strange that it was operating below 160f under some conditions... that should not be the case if your thermostat begins to open at 160f.
Does the thermostat you chose have a fail safe latch? It's possible that being such a low temperature stat, the safety mechanism is also set to trip at a lower temperature, has tripped after hitting 200f+, and is no longer closing. Typically these failsafe thermostats are very simple- there's a little spring tab inside the stat that, if the piston hits the end of its stroke, gets jammed on the cylinder, preventing it from closing. They do not automatically reset- it becomes mechanically jammed open at that point. May be something to look into.

Thank you.

So, the fact that it is a 160ºF t/stat, does not really effect the equilibrium point as you have described, due to the offset/extension housing that locates the T/stat housing. There is a 20º drop between the water pump housing on the engine block, and the t/stat (with bypass) housing. With a 180º t/stat, the engine always ran hot, because the t/stat wasn't opening before the coolant temp arrived at <200ºF in the block. With the 160ºF t/stat, coolant temps sit at 190º in normal operation (when ambient temps are 80º+, or if I idle long enough for it to get to normal op before driving.

As an aside, what is the projected HP gain from using electric vs. belt driven water pump?

 

[Jonohhh]

Thank you.

So, the fact that it is a 160ºF t/stat, does not really effect the equilibrium point as you have described, due to the offset/extension housing that locates the T/stat housing. There is a 20º drop between the water pump housing on the engine block, and the t/stat (with bypass) housing. With a 180º t/stat, the engine always ran hot, because the t/stat wasn't opening before the coolant temp arrived at <200ºF in the block. With the 160ºF t/stat, coolant temps sit at 190º in normal operation (when ambient temps are 80º+, or if I idle long enough for it to get to normal op before driving.

As an aside, what is the projected HP gain from using electric vs. belt driven water pump?

Ahh, I see what you're saying. In that case, the hard to manage variable is likely the fact that the temperature drop between the engine outlet and thermostat is a function of flow- the higher you rev, the faster water flows, and the less the temperature drops before it gets to the stat, so at high rpm it behaves like it "should" for a 160f stat. (Minus the sub 160f occourances?- may still want to make sure that it's closing entirely)

I'm not familiar with this extension housing- I couldn't make it out in the picture. Is it that black housing next to the stainless pipe? If so, there's something going on there: a 20f drop is very, very significant- I could only imagine that happening at extremely low flow rates. Are the dimensions of the extension housing identical to the OE setup (specifically the distance between the thermostat mounting surface and the sealing surface for the bypass valve)?

Is it possible to insulate the housing to help prevent a temperature drop? The block it's connected to should also be fairly close to coolant temp, so some insulation may just help it.



And the power gain- I have not calculated it out yet as it's not a main focus, but doing so should technically be possible. The difficult part is determining the power consumption of the stock pump as "rules of thumb" can be quite erroneous, and it varies not only by engine but the style of pump installed.

Assuming a 60% charging system efficiency, and the pumps 200w power draw, that would equate to about 0.45HP for the electric pump when operating at full speed- which it should rarely have to do anyway. The mechanical pump is sure to be more than that but by how much I don't know.

 

[lookforjoe]

Ahh, I see what you're saying. In that case, the hard to manage variable is likely the fact that the temperature drop between the engine outlet and thermostat is a function of flow- the higher you rev, the faster water flows, and the less the temperature drops before it gets to the stat, so at high rpm it behaves like it "should" for a 160f stat. (Minus the sub 160f occourances?- may still want to make sure that it's closing entirely)

I'm not familiar with this extension housing- I couldn't make it out in the picture. Is it that black housing next to the stainless pipe? If so, there's something going on there: a 20f drop is very, very significant- I could only imagine that happening at extremely low flow rates. Are the dimensions of the extension housing identical to the OE setup (specifically the distance between the thermostat mounting surface and the sealing surface for the bypass valve)?

Is it possible to insulate the housing to help prevent a temperature drop? The block it's connected to should also be fairly close to coolant temp, so some insulation may just help it.



And the power gain- I have not calculated it out yet as it's not a main focus, but doing so should technically be possible. The difficult part is determining the power consumption of the stock pump as "rules of thumb" can be quite erroneous, and it varies not only by engine but the style of pump installed.

Assuming a 60% charging system efficiency, and the pumps 200w power draw, that would equate to about 0.45HP for the electric pump when operating at full speed- which it should rarely have to do anyway. The mechanical pump is sure to be more than that but by how much I don't know.

Sorry - I'm diverting your thread. The extension housing is an aluminum piece I had to make in order to get a bypass t/stat setup, that also aligns with the Fiat water pipes. The black housing is the Volvo t/stat assembly. The Honda t/stat assembly (same design as the Volvo unit) normally bolts directly to the water housing & therefore sees full engine coolant temp.

 

[Jonohhh]

Sorry - I'm diverting your thread.

No need to apologize, no harm done whatsoever.

The extension housing is an aluminum piece I had to make in order to get a bypass t/stat setup, that also aligns with the Fiat water pipes. The black housing is the Volvo t/stat assembly. The Honda t/stat assembly (same design as the Volvo unit) normally bolts directly to the water housing & therefore sees full engine coolant temp.

Ah, I see what you're saying.

I'm a bit intrigued because 20F is an absolutely huge temperature drop to have across a housing like that- which is surely not sustainable at any meaningful flow rate. I know this is a little repetitive of me, but do keep in mind that the temperature drop is going to inversely proportional to flow rate- the higher the rpm and flow, the less it'll drop, and the less the thermostat will be biased.

Are all of the inlets and outlets flowing the same direction and serving the same purpose as they did in the Tstats original application?

 

[lookforjoe]

No need to apologize, no harm done whatsoever.

Ah, I see what you're saying.

I'm a bit intrigued because 20F is an absolutely huge temperature drop to have across a housing like that- which is surely not sustainable at any meaningful flow rate. I know this is a little repetitive of me, but do keep in mind that the temperature drop is going to inversely proportional to flow rate- the higher the rpm and flow, the less it'll drop, and the less the thermostat will be biased.

Are all of the inlets and outlets flowing the same direction and serving the same purpose as they did in the Tstats original application?

Yes, all functionality matches original layout. I was very surprised that there was much of a temp drop. There's no question that the engine ran 20ºF too hot under all normal driving conditions in 80º+ weather, with the 180ºF t/stat though. With K-Tuner, I logged the temps back when I was trying to figure out why it ran so hot originally.

 

[Jonohhh]

Yes, all functionality matches original layout. I was very surprised that there was much of a temp drop. There's no question that the engine ran 20ºF too hot under all normal driving conditions in 80º+ weather, with the 180ºF t/stat though. With K-Tuner, I logged the temps back when I was trying to figure out why it ran so hot originally.

I'm also surprised how biased the system gets by ambient temperature (from the sounds of it). If the "degrees above ambient" suddenly falls off at a certain temperature (as ambient gets colder), something very strange is going on indeed. Sorry I can't really be of any real help, I've never had behavior like that before to know where exactly to start!

 

[Jonohhh]

Also, the goods came in today! The pump, along with the controller!