Disclaimer

This blog is for entertainment purposes only, and is not meant to teach you how to build anything. The author is not responsible for any accident, injury, or loss that occurs as a result of reading this blog. Read this blog at your own risk.

Friday, April 20, 2018

Ch 23 - Oil cooler fan - part #4

Fan flight testing

I highly encourage you to take a second look at last week’s post, as most of what I will touch upon here is predicated on the ground testing experiment conducted in the shop.

In order to gather the missing data, we will be going flying today. Later we shall analyze the results, and come up with conclusions as to the suitability of this installation to long term flight safety. Then and only then will I feel comfortable leaving the fan permanently attached to the plane as I await the arrival of the first hot summer days. 

Should today's test prove successful, when the warm season eventually arrives, we shall conduct the final functional testing of the fan’s actual effectiveness in reducing oil temperatures on the ground.

But, before we actually get airborne, we need to ready the airplane by temporarily wiring it up, so that our “testing equipment” in the cockpit (basic voltmeter) can connect to the fan in the engine compartment. 

As you might remember, the reason for using a voltmeter in flight is the correlation between volts as read on the meter, with fan rpm, and fan amperage output. We mapped this mostly-linear relationship in the last post, and we are now ready to use it.

10.8v corresponding to safe fan-rpm (see previous post)

Because this is a one time occurrence, there is no need to rip half of the airplane apart just to run a couple of wires, and since the fan is right below the oil filler door on the top cowling, I plan on running a few feet of wire outside the plane from the this door to the aft portion of the canopy, and then inside the plane all the way to the front seat.


Running wires from the oil filler door to the rear cockpit

In the airline industry, maintenance occasionally uses what they call “speed-tape” on the outside the airplane as a temporary fix, you may think of it as airline priced duct tape. In the past I have conducted my own tests on duct tape outside the LongEZ, it is crazy strong (you figure that out at removal time), so I have no doubt our arrangement will work just fine for one flight.


"Speed-tape" over the wiring

I chose to use insulated alligator connectors inside the cockpit to allow for a quick disconnect of at least one of the leads, should it become necessary to isolate the fan electrically, and placed them in a position where I cannot accidentally bump into them, but where I can still reach them with my left hand. Keeping the wires ends somewhat staggered prevents the fan from shorting itself should both alligator clamps become disconnected in flight.


"Sky-Lab"

So, what’s the plan?

I envision the flight testing program as a mostly scientific pursuit, with a bit of art and intuition thrown in, needing to foresee hardware malfunctions, as well as obscure remote failure scenarios, and unpredictable odds. 

Flight testing not only has to establish ways for the plane to collect the sought after data, but it also requires anticipating everything that could possibly go wrong, and design strategies that address those eventualities. From the look of things, we are well on the way to doing just that with our initial voltmeter and wires placement plan. 

Needless to say, the big bugaboo of the flight testing phase is going to be the remote possibility of a self destructing runaway fan on takeoff, or anytime thereafter. Not only would the data collection phase of the flight be compromised, but the safety of flight might be impacted as well.

Aviation safety has come a long way by basically adopting Murphy’s Law that “anything that can go wrong will go wrong” as its mantra, and relies heavily on strategies tailored to minimize and mitigate such eventualities. The point here is that risk cannot be completely avoided but it can and should be managed appropriately.

During our testing I will be using an “action camera” mounted on a chest strap to videotape the flight (airspeed and volts mainly), this way I won’t need a third hand and a spare brain to try to capture the details I need. I did not consider a data-capturing passenger to be an ethical (or legal) option, so this will be a one man flight testing adventure. 

From this point on, all airspeeds will be in knots IAS (indicated airspeed), and any voltmeter values outside of the normal range established during ground testing (0v to -10.8v) will be cause for an expedite return to base, and ending the experiment.

For our purposes, I am going to subdivide testing into three separate phases of flight, each with its own challenges and emergency actions. They will be the takeoff ground roll, the initial climb, and maneuvering flight. No need for additional phases, because if the fan makes it all the way to maneuvering, it will be just fine for approach and landing as well.

The takeoff roll phase will begin at power up, and end at rotation. Because this is the safest time to end testing, and involves the easiest emergency procedure by just closing the throttle and applying the brakes, I decided to extend the ground run to a slightly higher airspeed than normal, up to 80kts before taking “my problems” airborne.


Ready to go fly

Ready for takeoff at the end of runway 5, voltmeter reading 0v as it should, I made one last mental review of what to expect. From this point on I’ll be looking closely at my voltmeter for the first indication of voltage, and by correlation fan spin-up.


80 knots, 0 volts

With rotation speed easily made with no incidents, I entered right into phase two of flight testing, the climb. 

The expectations for the climb, as for the rest of the flight, are the same as for the ground roll, less than 10.8v (absolute value) reading on the meter. 

The emergency action are quite different however, should that not prove to be the case, and they include wrapping the pattern immediately to the left for a close downwind, slowing down to the lowest voltmeter reading, and landing immediately. I could be back on the ground in about one minute at this point, so the possible danger from the over-speeding fan would be kept to the very minimum.

Of course, by now I have already figured out that 80kts is a safe speed for the fan, and because it is also a flyable speed, my stress level just took a major dump. I now know that at this airspeed, there is not enough airflow through the oil cooler to spin the fan, and should anything bad happen, I can always slow down to 80kts to stop the fan. 

Granted, it is possible it might take more energy to start the fan than to keep it running, and perhaps slowing to 80kts might not stop the fan, however we just proved we aren’t remotely close to fan overspeed territory here, and that’s what matters.

After a few seconds climbing at 80kts with 0.000v showing on the meter, I decided to lower the nose, letting the airspeed build up and seeing at what point I’d get initial fan rotation. 


100 knots, 0 volts

I am sure by now you are probably wondering whether the voltmeter is even working, or perhaps it might have gotten disconnected. I had the same exact thought, and at around 500’ above the ground took a quick peak at the leads…

Nope! Still attached.

Well, that is just weird! Where’s the heck is the oil cooler airflow? 

Obviously the fan is not spinning, since even turning it with one finger produces enough tension to be read on the instrumentation, but does that necessarily imply that there is no airflow?

Let’s open a quick parentheses here… 

I later tested the fan holding it out of the window of my truck with another aircraft builder driving it (thanks Eddie), and quickly discovered it takes around 40kts (wind adjusted with opposite direction runs) to turn the blades over. 

The obvious implication is that just because the fan is not spinning, it doesn’t mean there is no airflow through the fan (read oil cooler). There could actually be up to 40kts of airflow through the fan before I can be alerted to it by the voltmeter. 

Close parentheses.


110 knots, -1.3 volts

At 110kts, I knew we had this b#*ch nailed!

Finally some voltage started reading on my meter, confirming not only that the fan had started windmilling, but that the fan-rpm were well within safe parameters at this airspeed.

Much relieved by how things had gone in this one minute flight so far, it was time to enter phase three… maneuvering flight.

The only emergency action at this stage would be to slow the airplane down to a previously identified safe airspeed where the fan would be controllable once again, then test some more, or call it done and go home.

Personally, I needed to see that I had full control of fan-rpm via changes in airspeed. I spent a some time going back and forth from 110kts to160kts with complete predictability, though the voltages were slightly different at individual airspeeds depending on whether I was accelerating or decelerating toward them, but all in the safe sub 10.8v (absolute value) range.


120 knots, -1.7 volts

130 knots, -1.8 volts

140 knots, -2.7 volts

150 knots, -3.2 volts

160 knots, -3.6 volts

What a great day of testing! 

So far I had managed to safely takeoff, climb, and maneuver to a pretty high cruising airspeed without any incidents or issues.

Because I rarely fly faster that 160kts IAS, I could have just packed it in for the day, and be satisfied with the results. But with an airplane capable of so much more, it wouldn’t have been a complete vetting of the installation had I not pushed it all the way to redline.


170 knots, -4.3 volts

180 knots, -4.9 volts

190 knots, -5.5 volts

195 knots, -5.8 volts

Ok, so technically I was a couple of knots short there, nevertheless I think we succeeded in proving a major point today…

The oil cooler fan installation is safe!

Whether it works as expected on the ground in reducing summer's high engine oil temperatures remains to be proven, but we can now take flight safety concerns off the “most wanted” list.

Unfortunately the camera was bumped into at some point before takeoff, and not all footage is usable as the speed-tape got cut out of the shot quite a few the times, plenty of it remained however to validate our results.

For those interested in the data, I compiled a graph based on some of the speeds and voltages taken from the inflight chest-cam video footage.  

Voltage output from the fan at various airspeeds

From the experimental knowledge we gained during ground testing, it is now possible to correlate actual indicated airspeed to fan-rpm.

Correlated fan-rpm at various airspeeds

As you can see, the highest fan-rpm reached at the fastest airspeed this LongEZ is capable of, is merely 2200 rpm, roughly half of what it normally spins when powered by the battery (4300 rpm), and from the table at the beginning of this post, we know that the maximum induced current would only be -15.5a, which is easily stopped by an appropriately sized diode.

I feel comfortable closing the inflight portion of the oil cooler fan testing now, and while the summer round of ground tests might still disprove our main assumption, we will still have had fun, safely learning and growing in the process. 

Isn’t that exactly what the FAA had in mind when they created the amateur-built certification category after all?


Tuesday, April 10, 2018

Ch 23 - Oil cooler fan - part #3

Ground testing

The principal goal of this post will be establishing clear safe operating parameters to be used in testing the oil cooling fan installation. Minimizing and managing the risks involved in flight testing is going to be priority number one.

Let me just be upfront by saying that I’m not an expert in fan performance, and what follows might prove to be mildly upsetting to a lot of fan engineers out there. 

So... Now that both fan and shroud are mounted to the plane, we could just jump in and go fly. 


Completed oil cooler fan installation (for ground use only)

But should we? Let’s think about it for a minute…

For one thing this is an automotive type fan, and I bet it was never intended to see speeds of 200 kts (230 mph or 370 kph) across its blades. Still, while that’s an achievable airspeed in a Long EZ (shallow dive with some added power), it doesn’t necessarily means that the air flowing through the oil cooler will move that fast. 

“How fast does the air go through the oil cooler then?” you might ask.

The answer is… “I have no idea!”  Nor does anyone else I have talked to, for that matter.

The main reason why we don’t know is that, due to its own compressibility, air changes speed and pressure as it moves through variable size passages. 

Clear as mud so far? I thought so.

Case in point, the Long EZ’s main air intake, which is a NACA shaped trough carved into the belly of the plane (actually grafted onto the plane, but it  does look carved).


NACA inlet duct (bottom engine cowling removed)

The peculiar shape of this passage is such that it allows the high speed air to slow down as it enters the engine compartment, while at the same time increasing its static pressure. 

The NACA inlet duct thus pressurizes the bottom half of the engine cowl above the outside static air pressure, and does so with very low drag, as opposed to a less efficient scoop type inlet. This is why NACA ducts are still the preferred inlet method for bottom/up engine cooling schemes.

This high pressure/low speed air at the bottom of the engine compartment is in an ideal state for being channelled through the contorted engine baffles and cylinders cooling fins. 

The differential pressure that forms between below and above the engine keeps air flowing in the right direction, and its slower speed allows tight bends in the airflow to be achieved with less separation, turbulence, loss of efficiency, and internal drag.

This air flow eventually exits the rear of the plane from the top cowling, above the engine


Air flow through the cylinders cooling fins

Of course, some of the incoming air is also diverted and used up by other items, such as carburetor, oil cooler, heater muff, and possibly a few more pieces of equipment depending on the installation. 


Air flow from the NACA inlet

The slower air also spends more time in contact with the hot items in the engine compartment, and expands as it heats up. Then, it picks up speed and expands even further as a result of the converging aft cowling geometry (Bernoulli's principle). This shape promotes acceleration of the airmass back to its original speed (theoretically), before it is ejected out of the engine compartment, smoothly and efficiently rejoining the fast air above and below the cowling (again…theoretically)

That’s roughly the big idea.


Top cowling hot air exhaust

Wow! I think I lost track of this post's subject for a minute. Sorry about that! Let’s get back to testing.

Because we don’t know for sure how fast the air goes through the oil cooler, we also do not know how fast the fan will be windmilling, so we cannot be sure whether the fan bearings will be able to hold up to constant inflight spinning .

Unfortunately for us, the fan does not have a fan-rpm indicator, so if we just went up and flew the plane right now we wouldn’t know this very important detail to the life expectancy of the fan. 

If the fan spun too fast for too long the bearing could seize (most benign failure mode), or disintegrate altogether (worst case scenario) rattling inside the bottom cowling until something got damaged, eventually making it past the engine, and ejecting onto the wooden propeller. 

Neither option is very appealing, but the second one could certainly be disastrous.

The fan manufacturer publishes only limited engineering data, however what data we do have is at 13.5 voltage, representative of engine running and alternator charging the battery. 

Almost by definition then, whatever rpm the fan is able to reach at 13.5v (normal bus voltage in a 12v system), must represent the designed normal rpm, and will constitute our "Never Exceed" rpm value for flight testing. Notwithstanding the safety margin possibly engineered into this fan, I will consider safe an rpm envelope that is at or below this value.


Maradyne's fan data


The static pressure Maradyne charts here represents the resistance to airflow (friction) caused by the air moving through pipes, ducts, hoses, filters, hood slots, air control dampers or louvers. 

Clearly from the chart above, increasing the static pressure a fan must operate against reduces the amount of airflow delivered by fan, and looking at the data you can see that the fan blades are in a stalled condition at 1.2” H2O, and that the current jumps to 6.3a. When a fan stalls like this it becomes incapable of moving air, all it does is make noise, produce vibrations, and drain the battery. 

In our situation we will be operating in synch with the direction the air naturally wants to move because of the already higher pressure inside the cowl, so we’ll only have the oil cooler resistance to contend with, so fan stalling should not become a factor.

So, how do we find out the normal range fan's rpm? This value is easily measured using a hand held tachometer, and a variable power supply.


Gathering data for table #1 (fan at 4635 rpm at 14.4v and 5.9a)

It turns out that at 13.5v this fan likes to spin at 4300 rpm, so we will use this number as our maximum safe value. 

Next, I moved on to testing the fan in a windmilling scenario, but before doing that I tried pushing the fan to the maximum voltage my bench power supply could muster (20v) in order to highlight any obvious flaws, like weird sounds or vibrations at higher rpm, all indicative of substandard or deteriorating bearings.

Since I'm not sure yet what would happen to the fan in flight, I wanted to ascertain its behavior in a worst case scenario, way above the designed rpm range. 


6796 rpm at 21.8v and 10a 

I found that the fan spun flawlessly at almost 7000 rpm, no weird sounds, no rattling, no unusual vibrations, though it was a bit scary to be so close to.


Table #1 - fan being driven by the variable voltage DC power supply

Obviously an electric motor turns when supplied with electricity, but you might or might not know that it can also produce electricity if it were to be mechanically turned, as in an air driven turbine.






Fan acting as an air driven generator




This creates a new set of concerns, like how to handle the reverse current flow in case the Ground Cooling Fan switch is inadvertently left ON during takeoff.

There are a few ways of doing this, but in the case of low currents, the easiest way to prevent reverse flow is to insert a diode in line with the load. 


Ground Cooling Fan wiring diagram

At higher currents different strategies might be necessary, so it is imperative that the circuit be properly sized to handle the actual current (forward and reverse) that will flow through it. 



important M073K fan specs

The reverse current will depend on the yet to be known windmilling fan-rpm, which in turn depends on the aircraft speed, and the particulars of the individual installation. 

This data can only be gathered during flight testing.

As for ground testing, I knew I’d be limited in scope by my equipment since I don't own anything that can spin super fast, but I could get up to 5000 rpm using my converted belt driven Mini-Mill. So, I attached the fan (blades removed) to the mill, ran it at different rpm, and charted the results.


Fan motor being spun by the mill (multimeter off camera)


Ungracefully, amperage testing abruptly ended when I blew the internal fuse in my 10A multimeter at about 22 amps (“Doh!”), hence the lack of windmilling data (red NA) above 4200 rpm.



Table #2 - fan generating electricity while being spun by the mill

No direct fan-rpm reading is possible once airborne, but windmilling voltage could be made available to the pilot during flight testing, and a fan-rpm inference may be drawn from this voltage using the experimental data of table #2. 

Depending on what rpm the fan ends up being driven to in flight, this installation might or might not show to be acceptable, as far as fan bearing integrity is concerned. 

With our self-imposed structural fan-rpm limit set at 4300 rpm, the induced voltage limit is -10.8v (from the chart above). Note though that at this fan-rpm the motor is producing over 22.2 amps. Ideally, I'd like to keep this maximum back current to about -15a or less, which means about -5.5v, at a paltry 2000 fan-rpm.

If flight testing showed high windmilling rpm (i.e. high reverse flow current), this would make the fan switch even more of a critical item, to be absolutely checked in the OFF position before every takeoff. This might pose an unacceptable hazard to the electrical system.

Flight testing the windmilling fan at all usable airspeeds is a definite must, and it's going to be the key to unlocking the suitability of this installation to the safety of flight.

From our newly acquired experimental knowledge (recorded in table #2) we now know that 10.8v corresponds to an acceptable inflight fan-rpm regime, at least as far as its physical integrity is concerned. 

Given the unknowable nature of the actual fan-rpm, a small risk will have be taken in the quest to gather this data. However, using table #2 we can now design a flight-test schedule that addresses our concerns, and minimizes the dangers. 

Next time I will talk about flight profiles, as well as emergency scenarios appropriate to different phases of flight. Then, I will assume responsibility for the risk I have been trying to minimize and manage all along, and go flying. 

All that could have been done has been done at this point, and there is only one thing left to do to put this case to rest  once and for all, one way or the other...

"Cash me outside... howbow dah!"