# GM LFX V6

### Help Support HomeBuiltAirplanes.com:

Status
Not open for further replies.

#### wsimpso1

##### Super Moderator
Staff member
Log Member
I'll post my thoughts and see if anyone has a coherent response:

PROS:
Full cooling flow at taxi/climb - most auto conversions struggle with cooling.
Lighter weight - every pound counts.
Reduce belt routing - having an alt and water pump would require an idler.

CONS:
Will not last as long - easily fixed, assign a 1000 or 2000 hour replacement interval. Its only $170 Electric motor will stick out farther - might have a fitment issue, motor is taller than stock pulley. Not done before in AC - extensive ground testing should give an indication of suitability. Swapping back is 30 minutes of work. If I am missing something please let me know.. but I can't see a real downside. The big thing you have been missing is respect for folks who have actually done or know the stuff you have been so blithely dismissing. That we voluntarily try to help with understanding should be worthy of respect as well. An example comes to the fore here. You seem to think that ground cooling is a coolant flow issue. Ground cooling of water cooled airplane engines has been problematic. I have never heard of a case where coolant flow changes mattered one iota. It is all about getting enough delta P across the radiator to drive air through the core with nothing but the prop at idle moving the air. The WWII V-12's all had very large thick radiator cores to get adequate cooling at flight speed and power for the level of waste heat from 1200 to 1700 hp engines at 300+ mph in cruise and at over 100 mph in climb) and then well designed expanding ducts to minimize drag for speed. On the ground, they had engine idle thrust which means they had a gentle breeze wafting - they basically ran with zero air flow through the rads until approaching takeoff speed. We do have the advantage in smaller engines of being able to run thinner radiator cores. Look up Ross' posts on radiators and cooling for how he SUCCESSFULLY solved water cooling for the EJ22 engine in an RV6 as well as how others have SUCCEEDED. With Ross' schemes, the idle level propwash has always given enough rad airflow to maintain temps against idle power heat rejection. With decent airflow, the underdrive pulleys will still move way more coolant than needed in all airplane modes, while greatly reducing accessory power use over stock in flight modes. Could you stick electric cooling fans on the rad for ground cooling? Sure. Your airplane will be lighter and lower drag without it. WEIGHT IS THE ENEMY more than anything else in little airplanes, drag is #2 on the enemy list. Electric cooling fans are both more weight and more drag. A scheme that will ground cool without a fan is better on all counts. I still have strong issues with the idea that an electric coolant pump is both effective and energy efficient in an airplane dutycycle. It may be viable in the airplane and that can have value over routing belts for one more accessory... Once you go with an electric coolant pump, you must consider dual battery or dual alternator or both. But then running EFII, you probably should do at minimum one anyway. Bob Nuckoll's AeroElectric Connection should be required reading. Another guy who has SUCCESSFULLY done this stuff. His perspective on stuff breaking and being able to manage anyway should be mandatory... Billski Last edited: #### spaschke ##### Well-Known Member I have a 333hp engine and a 20x20" triple pass radiator from Griffin Radiator. They recommended NOT using a high flow water pump. The coolant moves too quickly through the radiator to get cooled much. highflow pump >35gph should be good if you have a huge radiator and a large amount of coolant. #### rv6ejguy ##### Well-Known Member I have a 333hp engine and a 20x20" triple pass radiator from Griffin Radiator. They recommended NOT using a high flow water pump. The coolant moves too quickly through the radiator to get cooled much. highflow pump >35gph should be good if you have a huge radiator and a large amount of coolant. Coolant never moves too fast through a HX. The faster you move it, the more heat you can reject (you do reach a point where there is little more to be gained and parasitic losses on the pump increase). Griffin should know better than to repeat this nonsense to customers. This being said, many would be surprised how little water flow you require with a proper rad and duct layout. #### pictsidhe ##### Well-Known Member No, see Kays and London Compact Heat exchangers page 27. It even has the math for it. It isn't intuitive, which is why people struggle to grasp it. On a more experimental note, I have personally watched the heat transfer from a radiator increase significantly by lowering the water flow rate. The optimum flow rate varies with air flow rate as well as radiator size, type etc. I used a gate valve with a too large pump, it was far easier to instrument and adjust than try to calculate everything... I have read the book, checked the math and verified it experimentally. There is a very good chance that a radiator manufacturer has done that too. #### poormansairforce ##### Well-Known Member You don't need 125GPM to cool a 250hp engine. Obviously not, if the person designing it knows what they are doing but we have the Raptor thread as testimony that most don't! Even with money to waste, er, spend. If you can raise the delta with a huge rad then great. Auto OEM and planes operate in different environments and under different conditions. See what road/oval racers use as that emulates an airplane better. BMW's pumps are known failure points so no joy with that example. Can we get a 40° delta? 80°? If so and we have the room and can stand the weight gain then we can cut gpm down significantly! But the proposed pump will not pump 35 gpm on an engine! I think many people forget that we are pumping against a lot of restrictions and these pumps are free flow rated! Just another marketing ploy. The only way to save hp is to pump less water! Which requires a bigger rad. Or more airspeed. Oops, here we go again. Realistically, I doubled what the math implied based largely on the OP's comments. I figure he is going to need all the wiggle room he can get. I'm quite sure I would! #### rv6ejguy ##### Well-Known Member Car stuff is very little like aircraft stuff with regards to rads but the same physics applies to all. Cars can have large area/ thin rads and need to be able to dissipate a lot of heat at low speeds (truck towing a trailer up a steep grade on a 120F day. Drag is a minor concern relatively speaking. Aircraft have the advantage of high mass flow available but we have small rad areas to reduce drag which is a major consideration. The information to do aircraft liquid cooling efficiently is readily available here on HBA if you search the threads. The Raptor guys didn't do any calcs that I've seen and break almost every known rule on how to do it with low weight and drag. BMW electric pumps have a known, finite life. Replace them when recommended and they are reliable in my experience. That being said, I don't like the idea. Would rather suffer a .1 mpg hit than have an electric pump that needs to be replaced well before most engine driven ones. You'll never see 80 degree deltas on any aircraft installation unless you were willing to suffer a huge drag penalty. Last edited: #### poormansairforce ##### Well-Known Member The Raptor guys didn't any calcs that I've seen and break almost every known rule on how to do it with low weight and drag. Exactly my point and I don't see it here either! You'll never see 80 degree deltas on any aircraft installation unless you were willing to suffer a huge drag penalty. That was a little bit of frustration over the lack of reality here. (Not you) Time to put the fishing pole away #### rv6ejguy ##### Well-Known Member Search Meredith Effect. #### poormansairforce ##### Well-Known Member I've been there, cool stuff. #### wsimpso1 ##### Super Moderator Staff member Log Member Are you serious? LOL like I would be stupid enough to post anything about myself. For example, I say I have 900 hours and some nutter says something stupid like 'well I have 901 hours so I know more than you'. Funny, the rest of us share some of our credentials on this forum. It is part of having credibility. Careful who you call stupid. Billski #### wsimpso1 ##### Super Moderator Staff member Log Member Yes the alt will need 1-2 more hp to make the additional power, but the removal of a significant load like a water pump GAINS you 8hp overall. https://www.hotrod.com/articles/hrdp-1201-baseline-testing-do-water-pumps-suck-power/ When we swapped to the GMB water pump, using the March underdrive pulleys, we recorded an average of 523.1 hp and 470.9 lb-ft of torque, 8.1 hp less than the Meziere electric pump and 4.4 hp less than the Weiand mechanical pump. It seems that by using an electric water pump and removing the balance shaft you free up almost 20hp and reduce the engine weight 15-25lbs. All for$400.
I have been debating making this post, but in the end decided that I had to expose this particular bit of journalism for lousy science at best. Believe it if you want, but I would not trust what they presented. What is wrong with this test and article?

The test is hugely flawed several ways;

Statistical quality of the data is concealed from us;

The data that is shared with us indicates testing validity errors;

Let’s get into each one in turn, and then talk about what could have been done better to give us confidence in what we find…

The test is hugely flawed several ways:

Basic Methodology Part 1 - If you want to know what torque and power a pump uses, test the pump. Get the conditions and requirements of the pump, manage the operating conditions, run the pump at the requirements, and measure the torque required to do that over the range of operating speeds. In this way, you can know that the pump is doing the thing that you need it to do while you can accurately measure the work going to the pump to do it.

Measurement Error - When you try to measure something small with a gadget designed to measure something big, you run into the very common problem of measurement variation being about as big as the thing you are trying to measure. Let’s say this dyno has a measurement error of 0.25% (and that would be a very good dyno) and its torque capacity is 1500 ft pounds (that is about right for dynos that are used for both engines and engines with transmissions). The standard error for that gadget would be 3.75 ft/lbs.

Loading Error - The tool applying load has errors that you want to be both known and controlled. Here we had known engine, but did it make the same power on each run? The shop running it felt that they needed to make three pulls on each run and average the data… We do not know anything about the standard variation in this engine, nor do we even know if torque and power were corrected for local atmospheric pressure and density, which can easily change 1% per day all by itself. It does take time to cool down the engine enough to work on it, drain coolant, disconnect plumbing and everything else in the way, remove one pump, install another pump, reinstall pulleys, belts, wiring, plumbing, get it all running again, warmed up and stabilized and then make the next three runs. I bet there were an overnight or two in there… A 1% shift on 475 ft-lb is 4.75 ft-lbs and I expect reality to be somewhere around that number.

Now I am not saying that the standard run to run error of this engine on this dyno is 8.5 ft-lbs, but I am saying that this amount is typical. Even if it were halved again, you would be trying to make a measurement on differences between things that use 4 ft-lb with a measurement tool that has errors of 4 ft-lb.

All of this applies to both the water pumps and to the alternator…

Test Methodology Part 2 – Understand that you can cool an engine on a dyno with no water pump at all… Yep, easily done too. Set the input water pressure 20” H2O higher than the outlet. This is what the pump normally does. Knowing this, if the cooling system on the dyno does not have control and data taking of the water in and out pressures, we do not know if they were loading the pumps or even windmilling the pumps sometimes. Engine dyno usually do not worry over this, they just want to make sure that the engine does not blow on their stand… They usually make sure that plenty of cool water is running through the engine. Knowing that dynos can be anywhere on water through your engine, how do we know if the water pumps on the engines were being loaded the same or not?

Test Methodology Part 3 – Same problem for the alternator as for the water pump. Was it loaded? How was it loaded? We should know that the alternator was making a certain electrical power on each pull, and in that way, we can know how much work it was doing vs how much work was being extracted from the engine to do it…

Statistical quality of the data is concealed from us:

The test ran not five sets of samples, but they ran three pulls on each data point. We have not seen the individual pulls only the averaged data, nor has it been offered. This would allow calculation of the standard error of this engine on this dyno. If the data was extremely noisy we would know it and be able to determine p values etc to tell if the differences we saw were real or buried in noise. As it is now, we have added to our unknowns.

The data that is shared with us indicates testing validity errors:

Normally aspirated engine torque output varies linearly with atmospheric density that it is ingesting. Thinner air, less power, thicker air, more power.

Normally aspirated engines must have their mixtures adjusted to the atmospheric density or more power is lost as pressures change. This effect is small with the dyno in one place, but it is real and present anyway.

We can check some of the data for validity:

The water pump could have been assisted or hindered in the pumping it does by how the dyno is run. A centrifugal pump impeller runs just like a propeller – torque to turn it goes with the square of speed and power with the cube of speed. Very little power should be consumed by bearing and seals, so we should see water pump torque at 5300 rpm be about (5300/6600)^2 = 64.5% of what it was at 6600 rpm. Power should be 51.8% at 5300 of what it is at 6600 rpm. Similar checks can be made for slowing the accessories with lowered pulley ratios.

The alternator could be run at high or low load, depending upon the set point of the regulator and the battery state. Alternator efficiency could be another good topic for investigation. Alternator rotor torque to turn is a function of how much current is applied to the field windings. Once you have torque to turn set, then power to turn the rotor goes with rpm. But there is another element – the fan on alternators consumes power too, and at the same speed cubed relationship as props and water pumps. If the battery is fully charged, alternator power is small and the fan may draw far more power at high rpm.

Now let’s see if the data fits anything we know…

Electric pump only is our baseline. With no alternator on this run, we can assert is this is the engine with zero accessory loads.

I calculated hp at the peak torque point by TQ*5300/5252, then subtracted run HP from the baseline HP to get accessory HP at 5300 as well as at 6600 rpm.

Weiland pump with 1.01 ratios. The data shows 3.7 hp drop at 6600, which should mean 1.9 hp drop at 5300 rpm. We actually got 1.4 hp drop. Might be OK…

Weiland pump with 0.87 ratios. The data shows 0.4 hp drop at 6600, which sounds too good to be true. I calculated that it should be 2.4 hp based upon the 1.01 ratio run and the lowered accessory speed. At 5300 rpm, I expected 1.3 hp, but got 1.8 hp. At 6600 rpm, we got too small a power drop, and at 5300 we got a power increase over the both expected and over the previous run with higher accessory speed. Something somewhere is way off about this data.

GMB pump at 0.87 pulley ratios. 8.1 hp loss at 6600 rpm and that should translate to 4.2 hp at 5300 rpm, but it gave 5.4 hp drop.

Last was to add in the alternator to the Weiland and 1.01 ratio setup. We know that the Weiland and 1.01 is 3.7 hp and 1.4 hp, so we can get the alternator hp at 8.1 and 4.0, with 6.5 expected at 5300 based upon the 6600 rpm number and constant torque from the alternator – that does not fit… If the alternator was running at constant power (14 V and some constant amp output), it would have the same power as at 6600 rpm, but it is half that – efficiency differences with rpm are never close to accounting for that. I tried third power of rpm, thinking that maybe alternator electrical power was close to zero and fan power on the alternator was big. 4.2 hp lines up nicely with the 4.0 hp the test shows. Maybe the fan really is causing most of the losses at this alternator or maybe the data has a LOT Of noise. Or maybe – as suggested by the large errors between anticipated loss changes due to accessory speeds – the data in this test is so noisy and unreliable as to leave me with no confidence in any of the conclusions.

Improvements - What could we do to know more about the test and extract results or finish dismissing the data as meaningless?

Provide the atmospheric pressure and temperature for each pull and then correct torque and hp for those figures OR provide us with data stating that all engine outputs were corrected for density. Maybe that was done, but nobody said so.

Provide data from all individual pulls. Anyone trained in statistics can extract the standard error within samples and then check to see if our data points are buried in our standard error or if the data points stand out from the standard error.

Even better would be to run one standard condition an extra time as an indicator of how much the engine changed with time. My preference would be with the electric pump and no other accessories once at the beginning and once at the end.

Let’s give some idea what the terms I am about to use mean.

Statistical significance is important. It lets us know if the differences we just saw are likely real or more likely imagined. There are lots of “old wives tales” out there competing with real changes, and statistical significance or the lack thereof help us to know which we have.

ANOVA is a set of statistical tests to let us tell if some variable’s effect on values is statistically significant in the face of other variables and of the noise in the data.

Given all of this, we run ANOVA to look for statistical significance on pump type, ratio, and alternator presence.

All of this stuff is done so that we can have a chance at knowing if the data we want is mingled in garbage from the test, or if it stands out above the garbage that comes from running the test.

OR, you can run tests of just the accessories. You want to know what the coolant pump and alternator are taking from you? You want to know where a water pump can not cool the engine or the alternator ceases to charge the battery so you can choose belt ratios intelligently? Run the accessories on test stands and document the torques and horsepowers used, the flows at pressures obtained, etc. I bet that some of the accessory builders have data for you on a lot of this stuff. A great series of articles could be written collecting the data, running some of it independently, and publishing the summary of which look best for the various communities.

But please do not just take somebody's test and assume that they know what they are doing without checking it out first...

Oh, my data is included. The equations are available. Yeah, there are no statistical calcs. Can not do it with sample sizes of one on each sample...

Billski

#### Attachments

• 23 KB Views: 1
Last edited:

#### pfarber

##### Well-Known Member
HBA Supporter
Is the engine for a BD-4 or something else?

If a BD, did you build the wing with the bonded skin?

BJC

BD-4B

I am rebuilding the wings, the currently fiberglass. I will be going all metal on rebuild. There was a BD flown with a custom race v8 and a belly radiator. No issues with it. Not sure if still flying.

#### Winginitt

##### Well-Known Member
What a lengthy piece of Billsh*t ! OK, now that I have everyones attention, I was "JUST KIDDING". I personally recognize and appreciate the amount of time invested here and in most of Billski's posts to present in depth explanations and sources to support his points of view. I don't always agree with him, sometimes don't understand,and sometimes vehemently disagree....but I always appreciate the fact that he is one of the most prolific posters on HBA and spends his time writing in depth enough to make reading worthwhile. Folks, we should all appreciate the amount of time and effort invested in these type responses.

Now as far as the transferrance of heat goes, I know one thing for sure. If I swipe my finger against a hot surface that I just finished welding, the longer my finger is in contact, the more heat that transfers to my finger. I have personally tested this on numerous occasions, sometimes even utilizing my whole hand ! By my personal observation size of contact patch and length of time make big differences.

Let's all lighten up a little......

#### pfarber

##### Well-Known Member
HBA Supporter
I have been debating making this post, but in the end decided that I had to expose this particular bit of journalism for lousy science at best. Believe it if you want, but I would not trust what they presented. What is wrong with this test and article?

The test is hugely flawed several ways;

Statistical quality of the data is concealed from us;

The data that is shared with us indicates testing validity errors;

Let’s get into each one in turn, and then talk about what could have been done better to give us confidence in what we find…

The test is hugely flawed several ways:

Basic Methodology Part 1 - If you want to know what torque and power a pump uses, test the pump. Get the conditions and requirements of the pump, manage the operating conditions, run the pump at the requirements, and measure the torque required to do that over the range of operating speeds. In this way, you can know that the pump is doing the thing that you need it to do while you can accurately measure the work going to the pump to do it.

Measurement Error - When you try to measure something small with a gadget designed to measure something big, you run into the very common problem of measurement variation being about as big as the thing you are trying to measure. Let’s say this dyno has a measurement error of 0.25% (and that would be a very good dyno) and its torque capacity is 1500 ft pounds (that is about right for dynos that are used for both engines and engines with transmissions). The standard error for that gadget would be 3.75 ft/lbs.

Loading Error - The tool applying load has errors that you want to be both known and controlled. Here we had known engine, but did it make the same power on each run? The shop running it felt that they needed to make three pulls on each run and average the data… We do not know anything about the standard variation in this engine, nor do we even know if torque and power were corrected for local atmospheric pressure and density, which can make easily change 1% per day all by itself. It does take time to cool down the engine enough to work on it, drain coolant, disconnect plumbing and everything else in the way, remove one pump, install another pump, reinstall pulleys, belts, wiring, plumbing, get it all running again, warmed up and stabilized and then make three runs. I bet there were an overnight or two in there… A 1% shift on 475 ft-lb is 4.75 ft-lbs and I expect reality to be somewhere around that number.

Now I am not saying that the standard run to run error of this engine on this dyno is 8.5 ft-lbs, but I am saying that this amount is typical. Even if it were halved again, you would be trying to make a measurement on differences between things that use 4 ft-lb with a measurement tool that has errors of 4 ft-lb.

All of this applies to both the water pumps and to the alternator…

Test Methodology Part 2 – Understand that you can cool an engine on a dyno with no water pump at all… Yep, easily done too. Set the input water pressure 20” H2O higher than the outlet. This is what the pump normally does. Knowing this, if the cooling system on the dyno does not have control and data taking of the water in and out pressures, we do not know if they were loading the pumps or even windmilling the pumps sometimes. Engine dyno usually do not worry over this, they just want to make sure that the engine does not blow on their stand… They usually make sure that plenty of cool water is running through the engine. Knowing that dynos can be anywhere on water through your engine, how do we know if the water pumps on the engines were being loaded the same or not?

Test Methodology Part 3 – Same problem for the alternator as for the water pump. Was it loaded? How was it loaded? We should know that the alternator was making a certain electrical power on each pull, and in that way, we can know how much work it was doing vs how much work was being extracted from the engine to do it…

Statistical quality of the data is concealed from us:

The test ran not five sets of samples, but they ran three pulls on each data point. We have not seen the individual pulls only the averaged data, nor has it been offered. This would allow calculation of the standard error of this engine on this dyno. If the data was extremely noisy we would know it be able to determine p values etc to tell if the differences we saw were real or buried in noise. As it is now, we have added to our unknowns.

The data that is shared with us indicates testing validity errors:

Normally aspirated engine torque output varies linearly with atmospheric density that it is ingesting. Thinner air, less power, thicker air, more power.

Normally aspirated engines must have their mixtures adjusted to the atmospheric density or more power is lost as pressures change. This effect is small with the dyno in one place, but it is real and present anyway.

We can check some of the data for validity:

The water pump could have been assisted or hindered in the pumping it does by how the dyno is run. A centrifugal pump impeller runs just like a propeller – torque to turn it goes with the square of speed and power with the cube of speed. Very little power should be consumed by bearing and seals, so we should see water pump torque at 5300 rpm be about (5300/6600)^2 = 64.5% of what it was at 6600 rpm. Power should be 51.8% at 5300 of what it is at 6600 rpm. Similar checks can be made for slowing the accessories with lowered pulley ratios.

The alternator could be run at high or low load, depending upon the set point of the regulator and the battery state. Alternator efficiency could be another good topic for investigation. Alternator rotor torque to turn is a function of how much current is applied to the field windings. Once you have torque to turn set, then power to turn the rotor goes with rpm. But there is another element – the fan on alternators consumes power too, and at the same speed cubed relationship as props and water pumps. If the battery is fully charged, alternator power is small and the fan may draw far more power at high rpm.

Now let’s see if the data fits anything we know…

Electric pump only is our baseline. With no alternator on this run, we can assert is this is the engine with zero accessory loads.

I calculated hp at the peak torque point by TQ*5300/5252, then subtracted run HP from the baseline HP to get accessory HP.

Weiland pump with 1.01 ratios. The data shows 3.7 hp drop at 6600, which should mean 1.9 hp drop at 5300 rpm. We actually got 1.4 hp drop. Might be OK…

Weiland pump with 0.87 ratios. The data shows 0.4 hp drop at 6600, which sounds too good to be true. I calculated that it should be 2.4 hp based upon the 1.01 ratio run and the lowered accessory speed. At 5300 rpm, I expected 1.3 hp, but got 1.8 hp. At 6600 rpm, we got too small a power drop, and at 5300 we got a power increase over the both expected and over the previous run with higher accessory speed. Something somewhere is way off about this data.

GMB pump at 0.87 pulley ratios. 8.1 hp loss at 6600 rpm and that should translate to 4.2 hp at 5300 rpm, but it gave 5.4 hp drop.

Last was to add in the alternator to the Weiland and 1.01 ratio setup. We know that the Weiland and 1.01 is 3.7 hp and 1.4 hp, so we can get the alternator hp at 8.1 and 4.0, with 6.5 expected at 5300 based upon the 6600 rpm number and constant torque from the alternator – that does not fit… If the alternator was running at constant power (14 V and some constant amp output), it would have the same power as at 6600 rpm, but it is half that – efficiency differences with rpm are never close to accounting for that. I tried third power of rpm, thinking that maybe alternator electrical power was close to zero and fan power on the alternator was big. 4.2 hp lines up nicely with the 4.0 hp the test shows. Maybe the fan really is causing most of the losses at this alternator or maybe the data has a LOT Of noise. Or maybe – as suggested by the large errors between anticipated loss changes due to accessory speeds – the data in this test is so noisy and unreliable as to leave me with no confidence in any of the conclusions.

Improvements - What could we do to know more about the test and extract results or finish dismissing the data as meaningless?

Provide the atmospheric pressure and temperature for each pull and then correct torque and hp for those figures OR provide us with data stating that all engine outputs were corrected for density. Maybe that was done, but nobody said so.

Provide data from all individual pulls. Anyone trained in statistics can extract the standard error within samples and then check to see if our data points are buried in our standard error or if the data points stand out from the standard error.

Even better would be to run one standard condition an extra time as an indicator of how much the engine changed with time. My preference would be with the electric pump and no other accessories once at the beginning and once at the end.

Let’s give some idea what the terms I am about to use mean.

Statistical significance is important. It lets us know if the differences we just saw are likely real or more likely imagined. There are lots of “old wives tales” out there, and statistical significance or the lack thereof help us to know which we have.

ANOVA is a set of statistical tests to let us tell if some variable’s effect on values is statistically significant in the face of other variables and of the noise in the data.

Given all of this, we run ANOVA to look for statistical significance on pump type, ratio, and alternator presence.

All of this stuff is done so that we can have a chance at knowing if the data we want is mingled in garbage from the test, or if it stands out above the garbage that comes from running the test.

OR, you can run tests of just the accessories. You want to know what the coolant pump and alternator are taking from you? You want to know where a water pump can not cool the engine or the alternator ceases to charge the battery so you can choose belt ratios intelligently? Run the accessories on test stands and document the torques and horsepowers used, the flows at pressures obtained, etc. I bet that some of the accessory builders have data for you on a lot of this stuff. A great series of articles could be written collecting the data, running some of it independently, and publishing the summary of which look best for the various communities.

But please do not just take somebody's test and assume that they know what they are doing without checking it out first...

Oh, my data is included. The equations are available. Yeah, there are no statistical calcs. Can not do it with sample sizes of one on each sample...

Billski

So what did you discover that the article didn't already discover? The best performance was obtained with the electric pump. All of your test cases are a waste of time. Basic rules off mathematics are all we need. Calibrate for air density? LOL SURE. While it may affect the HP generated it would not affect the performance of the pumps. Not one bit.

Going off on a tangent regarding underdrive pulleys..... ARE YOU KIDDING ME? On a water cooled aircraft, you're going to further reduce pump GPM for cooling for exactly what reason? The Laws of Thermodynamics would like a word with you.

I guess for the uninformed, your prose regarding measurement bias, scale errors and other such minor bits of trivia are useful to them. But in the end if you are going to make an assumption that a professionally run dyno shop isn't going to know what calibration is or say 'all the tests are invalid because they are within the machines tolerance' then I'd like to see you find a dyno shop that will have a tolerance of +/- .5hp to ensure accurate measurements. But since the tests were all done multiple times, and the average was consistent, you must accept the data at face value.

All of this proves one thing: The lightest, least power robbing, highest GPM pump on an otherwise stock engine is going to be the electric water pump. Can I get an A-men!

#### pfarber

##### Well-Known Member
HBA Supporter
Now as far as the transferrance of heat goes, I know one thing for sure. If I swipe my finger against a hot surface that I just finished welding, the longer my finger is in contact, the more heat that transfers to my finger. I have personally tested this on numerous occasions, sometimes even utilizing my whole hand ! By my personal observation size of contact patch and length of time make big differences.

Let's all lighten up a little......
So you are saying that lower water flow is preferred for liquid cooling? The Laws of Thermodynamics would like a word with you.

Removing BTUs from the block to the radiator is the job of the coolant. The coolant can only absorb so much heat. The faster you circulate the water to the radiator the more BTUs you can remove from the engine.

This may be informative
https://www.stewartcomponents.com/index.php?route=information/information&information_id=14

#### BJC

##### Well-Known Member
HBA Supporter
BD-4B

I am rebuilding the wings, the currently fiberglass. I will be going all metal on rebuild. There was a BD flown with a custom race v8 and a belly radiator. No issues with it. Not sure if still flying.
Thanks.

BJC

#### bmcj

##### Well-Known Member
HBA Supporter
Now as far as the transferrance of heat goes, I know one thing for sure. If I swipe my finger against a hot surface that I just finished welding, the longer my finger is in contact, the more heat that transfers to my finger. I have personally tested this on numerous occasions, sometimes even utilizing my whole hand ! By my personal observation size of contact patch and length of time make big differences.

Let's all lighten up a little......
There you go... the size and time made a difference in the amount of heat transferred to your hand. But the speed it moved probably did not affect it. As your hand warmed, closer to the temp of the metal, it (you) probably lost some of its ability to transfer more heat at the same rate (that makes sense, because once your hand reaches the same temp as the metal, heat transfer stops... and you probably use some words not appropriate for general audiences). Now, if you swipe your hand across fast enough that you can follow it with your other (cooler) hand, you would probably absorb more heat than you could by leaving one hand on for twice the time. This would require fewer inappropriate words.

#### Winginitt

##### Well-Known Member
So you are saying that lower water flow is preferred for liquid cooling? The Laws of Thermodynamics would like a word with you.

Removing BTUs from the block to the radiator is the job of the coolant. The coolant can only absorb so much heat. The faster you circulate the water to the radiator the more BTUs you can remove from the engine.

This may be informative
https://www.stewartcomponents.com/index.php?route=information/information&information_id=14

Did you ever see a radiator with a blockage welded into it to cause the water to make two or even 3 passes across the radiator? A racers trick. The flow wasn't increased but the time that the coolant was in contact with the aluminum surface was increased. The metal inside the engine or the inside of the radiator only has a certain area of contact to transfer heat. The radiator didn't change size, but the coolant was "essentially" slowed to provide more contact time. You will reach a point of no return where coolant at a normal temperature will not give up additional heat because of reduced contact time. If this were not true, a radiator could be a simple straight Tube instead of a series of tubes and fins. If we needed more cooling we would then simply increase the rate of flow with a bigger water pump. In my "finger" explanation the object of my welding mistakes could not transfer sufficient heat to my finger to burn me if I was quick enough. Sadly, sometimes I wasn't....Since there are variations in all things that are designed and manufactured, as well as variations to the daily conditions these things operate , I'm reasonably sure that sometimes a better result can be obtained by increasing flow in an inefficient system....but I don't believe that simply increasing flow always results in improvement.

Many years ago I worked at an amunition manufacturing plant on assembly line where we had to place different size charges in dummy shells as they passed by. When they turned the speed of the conveyor belt up, we would be unable to keep up sometimes, and empty shells would go by.

Now I think what I said sounds perfectly logical, but if others see it differently, I'm OK with that.

#### wsimpso1

##### Super Moderator
Staff member
Log Member
So what did you discover that the article didn't already discover?
Three big takeaways for everybody interested in learning what works and what does not:

#3 - It is MUCH better to figure out what small things do by testing them by themselves on small test equipment than to run them on big things on big test equipment;
#2 - Standard error on the test was as big or bigger than any signal available from the small things, making the test results close to impossible to learn anything from it;
Drumroll...#1 We still do not have any data (at any quality level) indicating that an electric coolant pump driven from the alternator (as an airplane must use it) is more efficient than a belt driven coolant pump. That case was not run on the test nor was any data that could be combined run. Hot Rod would have had to run the alternator with and without the added draw on the alternator needed to run the pump and published the data on both cases.

Even if the case of note (alternator driven electric pump vs engine driven pump) was run, there was so much error in the test as to make that result difficult or impossible to resolve.

If you really want to know, pursue the manufacturer for their load vs speed and output vs speed on accessories, then pick pulley ratios that allow you to run them only barely fast enough at min operating rpm to keep the engine cool and the battery charging, while minimizing the power sucked off at takeoff and climb power. If you really think that the electric pump is an improvement, have at it and tell us how it works out.

Billski

Status
Not open for further replies.