Announcing a complete solution to mapping engine efficiency
 

(Acknowledgement: This project was inspired and made possible by the MPGuino.)

As part of an extensive research project in optimal control, we have developed a sensor suite that we think can be used to map the engine efficiency of YOUR particular car. The engine efficiency is derived from logging data under normal driving conditions for any individual car. In fact, engine efficiency is probably a misnomer, as we are actually calculating complete system efficiency, including drivetrain losses, etc...

While similar to the MPGuino, this project has some notable differences: 1) no screen, 2) GPS logging, 3) accelerometer, and 4) saving data to an SD card.

The principle and application is described in short here:

http://www.eissq.com/ponthy/preprint.pdf (I had to resort to the obscuring the HTML because of the anti-spam features. I'll delete them as soon as the website says I have enough posts to have the right to post links.[Thanks MetroMPG for fixing it.])

As indicated in the above pdf, the solution is comprised of two parts: a hardware logger (shortly described above, with more details later) and a software package based on an Extended Kalman Filter.

The software package is complete, and is more or less in fire-and-forget state. It’s in alpha because documentation is lacking and it would take a few back-and-forths with me to properly set up your Matlab environment.

The software outputs the engine fuel efficiency, and gives you some other nifty outputs, such as the included Google Earth trip tracking (where color indicates speed).

I have full hardware schematics available, and will distribute the originals on demand. I am providing the jpgs for those curious.

In continuing this research, I would like to extend this project to cars outside the few we are already following. We are looking for people who would like to help us out. We are giving away the Eagle schematic and AVR code base in hopes that some of you will be motivated to build your own, with technical assistance from us. In theory, anyone can build one of these, although in reality the two SMC components (ATMega chip and accelerometer chip) are much easier with an oven. The overall parts cost, not including the GPS and board, can be brought in at well under $20. The GPS can be anything that can output NMEA sentences over a serial port. (I use a FV-M8 at 5Hz.)

As of this moment, I have ~10 circuit boards that are completed and tested, and another two for an upgraded model that awaits soldering. I will give these away for free to anyone interested in installing these units. I will include the AVR, the accelerometer, and the SD Card adaptor, as these might be hard to find. The rest are OTS components: resistors, capacitors, etc… I ask only that people pay shipping.

I also have a new version already written and debugged for the XMEGA, which solves some of the limitations of earlier models. It simply remains to finish the circuit schematic and fabricate the boards. If someone were to complete the circuit schematic for the XMEGA, I would be able to run off 10 or so boards, and then distribute them for free.

This could also be ported to the Primer2 or STM8S-Discovery platforms, if someone were so inclined. Thus you have the convenience of the MPGuino with the power of the Kalman Filter. I can offer a bit of assistance on these projects, but not as much as for the Primer2.

For those of you who have a CAN bus, I also have code for a $75 CAN reader, one where you have nothing to solder but one single 8-pin chip.

If it happens that any of you are in Luxembourg, Luxembourg or Auxerre, France, then I can give you a free working model and help you install it.

If anyone in the community is interested in the programming side of this project, the biggest help would come from converting some of the matlab files into C, Python, or some other language that can run on anyone's computer, without having a multi-thousand dollar software license.

I am interested in any and all comments that can help us improve this. (Notably, I am struggling to explain why I get a fuel efficiency that is globally too high, even if the shape of the curve seems to be spot-on. Which in the end is all that is important, as relative efficiency between engines and engine operating points is far more useful than absolute efficiency.)

As I get an idea of the community and internet’s reaction to this project, I will make a website that explains it a bit better. I admit that this post is a bit disorganized, but alas such is the limitation of forums. Unfortunately, as a result of having spent years on this project, I suffer from being too close to the problem and thus am likely assuming that things are clear when they are anything but. Don't be shy, let us know what you're thinking!

...wow!

http://www.eissq.com/ponthy/On_Mappi...y-preprint.pdf is the link

This is some very impressive work! You've clearly put a lot of effort into this, and it looks very interesting. After reading through the report, I have a few questions about it:

In its' current state, the microprocessor stores the data on an SD card, and then you have to plug the SD card into a computer so Matlab can read it, correct? Would it be possible to output the data directly to a PC/laptop via Matlab rather than storing it on an SD card, in order to have real-time display of the data?

What exactly are the torque and efficiency being mapped against in figures 2 and 3 of the pdf?

Since the data for the different methods is so consistent, why is it still preferred to keep the GPS in the system? Was the data with GPS and DEM considered more accurate?

Do you have a source for the GPS units? I took a quick look for the FV-M8, but only found one for around $100.

Hi Kenn -

Welcome, and thanks for posting. I just had a quick look through your research... fantastic! I'm sure you'll have more than a few members of the Instrumentation forum excited by your work.

I know that the topic of building specific fuel consumption maps has been brought up before in the context of the MPGuino, but this is the first (that I've seen) of it being done.

And verified against a dynamometer ... I'm envious of the tools you have available!

I reformatted your URL to make it active.

(Trust me: with the number of spammers who register here daily simply to hi-jack the site's popularity by posting useless links, the restriction is needed, if annoying to legit new members. Feel free to post & delete 5 total messages to bypass the filter if you want to post more URL's.)

Darin

...anybody got one in USA and playing with results yet?

Absolutely. That's one of the eventual goals in the optimal control problem. With this model, we can estimate parameters online, such as mass changes (passengers getting in/out), air resistance changes (windows open/closed), rolling resistance changes (tire pressure a little low), etc...

What we're missing right now is the time to program the observer. A 17x17 matrix is not chump change for a microcontroller, especially without floating point. Even on my i7 processor, the Matlab calculations only go 10 times faster than real time. There's a lot of optimizing that can be done, though, so I'm guessing that a small DSP (or even FPGA) could do the work. It's certain, though, that the 8MHz AVR can't at that sample rate.

Time. You're the second person to ask that, so clearly I need to do a better job of presenting those graphs. I'll update the caption once I'm back at work.

Multiple reasons: 1) For the highest precision results, we want to use as much data as possible. 2) With the GPS, we can continually update the estimation of road elevation and grade, improving past and future results as we converge toward the true road parameter values. 3) We need the GPS anyway for optimal control. Of course, that being said, it bears repeating that you don't strictly need the GPS and DEM if you want to apply the observer. In fact, the observer becomes much simpler, and might even become small enough to be practical in real-time on a 8MHz AVR. 4) Most importantly, the GPS gives us a very good time fix. Otherwise, you're kind of guessing in the dark when you try to fuse the various asynchronous data, and that is NOT good.

For the FV-M8, I have a source in Germany for less than 50E, and I remember seeing a few in America for under $60. Anyway, I wouldn't use that particular model anymore. The u-blox line of GPSes are far more interesting. For instance, the Falcom FSA03 is cheap, and has a host of features that are good for these kind of embedded applications (For instance, the FSA03 has very advanced firmware support for energy savings, while the FV-M8 has nothing at all.)

The math behind performing all the observability tests is widely understood. It's not complicated once you know what you're getting into, but if you're looking at it for the first time it can be overwhelming. The observer is key to getting good results. Without that, your data will be so noisy as to be totally inconclusive.

Even better, I got to verify against a dynamometer at the Grand Prix racetrack at Spa-Francorchamps. And they said I can come back and put the roadster on the race track once spring comes. "For scientific purposes".

To tell the truth, the dynamometer was kind of scary. It took a long time (around 1 hour) to get a very dense data grid. Since this was my personal car, if something blows up it's not my research budget that pays! But it was a lot of fun, and we'll definitely be going back with some other cars where I've installed the logging unit.

Oh, I believe it! Thanks a lot for fixing it for me. And I'm guessing that I'll get to 5 pretty quickly, anyway.

what exactly would I need to buy in order to use my car as a test car? As a collage student money is tight bit I do drive a consistent route on a regular basis.

Good question. While the exact cost will depend on what you have and what kind of a car you've got, overall it should be quite cheap. I have sitting next to me a number of 1st gen. boards that are completely ready, and tested. Only... they're first generation. The results are as accurate, but there's no automatic shutdown module for the GPS, so either you find an alternate way not losing the GPS fix whenever you turn off the car, or you run the risk of killing the battery if your car is parked for extended periods of time.

I have one additional second gen board, which uses a transistor to turn on the GPS every 30 minutes so that you always have a hot fix whenever you jump in your car. However, for this board it turns out that I don't even have the processors anymore, nor the accelerometer, so you'd have to spend about $20 on chips. Not a lot, but not free either.

So, to answer your question:

1. You should get a GPS. Any kind that outputs NMEA over a serial port AND has a pulse-per-second (PPS) signal will do, but as mentioned in an earlier post, I strongly favor something based on the u-blox chipset. Otherwise, I bet you can find something used on eBay for <$15.
   
2. You'll need to have some way of splicing into the fuel injection signal, so I suggest spending $5 to buy a short extension cord, (just like you would with the MPGuino.)
   
3. Again, just like the MPGuino, you'll need a Vss signal. I've seen cars where this is quick and easy, cars where I found it with a bit of effort online, and, for instance my car, cars where there is no Vss and you have to find some other solution (I used the CAN bus).
   
4. You'll also need an SD card, maximum of 2GB. (>2GB cards won't work.) If you don't have an old one lying around, I just bought a handful for 2€/ea. so they can only be cheaper back in the States.
   
5. If it's a first generation board, depending on the GPS you might need a battery to keep the GPS alive. I used lithium watch batteries, but those are a little pricey (~$5)
   
6. Add in assorted wires and connectors to get everything connected, and you're set for installation.
   
7. You'll also need a computer. If you don't have Matlab at school, you might find a copy that "fell off the back of the internet".

Oh, yeah, and don't forget to factor in a couple bucks in postage. The Luxembourgish post office says 2€ to send it anywhere in the world.

I'd definitely be interested in installing one of these, the only problem for me would be finding the GPS unit. I don't know anything about them, so I'm not really sure what to look for in buying one. I already have an MPGuino set up, so those wires are already tapped, and I have access to Matlab.

Do you know the typical power draw of the GPS units?

It's a pretty large market out there, so it ranges all over the map. The FV-M8 that I have uses about 60mA. The u-blox use about 40mA.

However, one of the reasons I like the u-blox is because it has some nifty shut-down features that let it go to sleep and wake up periodically, all on its own, so power usage becomes much less important. Here's the link: www.lemosint.com/gps/gps_details.php?itemID=571

(Note that if you wanted something like that, you'd have to be able to do some programming in order to support the extended features. It's pretty easy, and I can advise, of course, but without having the GPS unit in front of me I'd be a bit limited. Are you comfortable with that?)

Very impressive work! I haven't read the entire paper you posted yet, but did breeze through it and your dyno map is very close to the one you made yourself.

I also would be interested in doing some testing with you, but would also need a GPS. I also have an mpguino setup in the car already.

How do you feel about programming? Since you've got a Prius, you could actually get all the information I'm gathering directly from the CAN bus (http://www.vassfamily.net/ToyotaPrius/CAN/cindex.html), along with a lot of other goodies. Just connect two simple wires and a ground and hardware-wise you're good to go. There'd be a bit of programming on the software side (not much, I've already done most of it), and that'd be it for the logging. I'd be interested to try to build an efficiency curve from a Prius's drive cycle.

I'm comfortable doing some programming.

I was thinking of putting it in the Paseo, but the Prius would be interesting as well. And, my Prius has a GPS in it. I have no idea what kind of outputs it has though. I'd guess none.

I have a Matrix with a scangauge and a cheep garman GPS unit. would that work or do I need to buy another GPS unit?

That depends on whether the Garmin outputs to a serial port, and whether it outputs a pulse-per-second signal. Do you have the model # handy?

If it does output to a serial port, and not USB, then you'll still need one additional chip. It's called a MAX233, and you can get them for free as a sample from Maxim. Very easy to wire up.

it is a Nuvi 200, the only port is 1 USB. Does my Matrix have a CAN bus and if so would that save some work?

Ah, that won't do it then. My design eliminates the crystal, but depends on the GPS's PPS to keep the RC oscillator in line.

According to Does an '06 Toyota Rav4 have an accessible CAN-bus? - MP3Car.com it does not. Moreover, it would actually take more time, believe it or not, because the CANbus protocol would have to be reverse engineered. I've already done it for the Smart Roadster, and it's a PITA. Someone else has already done it for the 2006 Prius, so I could base things off that information. But for another car...? There's no way of telling what we'd find in there.

You see, there's no standard for what information a manufacturer puts on its CANbus, and there's no (legal) requirement for them to divulge this information, so it's sort of the wild west in there. That might explain why it's said that there are more lines of code in your navigation system than there are in the whole Airbus A380.

Anyway, the CANbus reader is something you would have to buy for around $75. It's not something that I have here in my lab, so if you're on a student budget the original approach is cheaper.

The biggest problem I see is with the inputs for the road load equation: M, Ma, Cd*A, Crr. M varies quite a bit. For example, my car sheds 27kg as the fuel tank runs from full to empty. I've removed some parts and added others, so if I wanted to know my car's mass, I'd have to have it weighed.

My Cd*A changes with every aeromod I do, and Crr depends on tire temperature, pavement type, and too many other factors. I think the best approach here is to have the user do a high-speed neutral coast to discern Cd*A and a low-speed coast to determine Crr. I think lots of EcoModders would like to have a precise datalogger to help calculate these numbers.

I would love to study the BSFC of the first-gen Honda Insight. I'd love to quantify the effect of lean burn, find the optimal rate of acceleration, and study exhaust and intake mods.

How long does it take to fill the 2GB SD card?

Programming the stuff is no problem, as long as it's nothing too difficult, I think I'll be able to handle it.

GPS things are something that I've just never really looked into, I didn't even know you could buy separate chips like this before reading your report, so that's really my only concern - I don't know any of the features, benefits of one vs. another, etc. But recommend me a decent, cheap one, and I'll be all over it.

Very true, although mass loss through fuel burn is something that is directly measurable (since we're measuring fuel consumption after all). Harder is mass change due to passengers, but if the model is good enough we can identify all these things.

I might add that the 27kg mass change is perhaps negligible when considering that the true mass of the vehicle is probably unknown, and other (important) parameters such as air density, wind speed and direction, air temperature, engine temperature, oil temperature, tire pressure, etc... are not included in the model. It's not really that important though, as all these factors tend to average out, leaving you with good results even if you have more uncertainty than you would like.

Some of these above parameters we can easily include in the model once we have the measurements. I'd be especially interested to know if we can get an idea of absolute air pressure changes based on the MAF.

One of the beautiful things about observers is that we can estimate all those parameters. High-speed/low-speed tests are a great way to have a first order approximation, but they miss some essential information that we can reconstruct with a parameter estimating observer.

That being said, an online observer and a parameter estimation observer are slightly different, and I haven't yet written the code to do that with this model. The first order approximation is good enough for my purposes right now.

You can lean burn on the Insight? Cool. I had no idea. How do you do this? What are you using to measure the effects of lean burning? Do you have a CHT or EGT gauges?

Quite a while, although it's speed dependent. Figure roughly 1MB/min, and you've got an idea of how long you can go before emptying the card.

The car leaves the factory with a stratified charge lean burn scheme, that allows AFR as low as 22:1. It happens automatically when the catalyst is warm and the engine is lightly loaded, and the driver notices, because less power is produced at the same throttle position. Also, the fuel economy gauge shoots upward when you enter lean burn mode.

I'm sure Honda used EGT and NOx sensors while developing the engine, but the car only needs a wideband oxygen sensor and lots of emissions equipment as a result of lean burn.


Having a fuel economy gauge is very helpful, but a fuel economy gauge will tell you heading down a hill at very high speeds is good, braking is not bad, and heading up a hill at full throttle is bad. We know all these things are untrue.

I'd like cars to come with a BSFC gauge, and a road load gauge. The BSFC gauge would tell you how efficiently you're producing power, and the road load gauge would tell you how much power you're using. Although your project doesn't (yet?) give the output in real-time, it would still improve my driving.


I would like to install one of your boards in my car. Perhaps I can help find where your road load model deviates from reality.

The goal of our research, of which this efficiency estimation is only a small part, is to be able to tell you what you should do going up a hill, in real-time.

As you say, a simple fuel economy gauge is difficult to understand in an optimal way. I'd like to see an energy gauge, one that calculates how much energy is being expended instead of just saying average economy. This would take into account potential energy and kinetic energy, for instance.

I admit it sounds interesting, but you're driving an Insight, and I have no way of measuring power flow in the electric circuit. To make matters more interesting, you have a CVT, which again makes life a little harder. If there's a way to log all that data while driving, then I can work with it. Otherwise, I'm afraid that the only way this would work on the hybrid would be if you could completely disable the electric motors for your tests.

Actually, I have a five-speed manual transmission. Also, I drive around with the electric motor disabled all the time, because I save gas by avoiding the constant motor->battery-> motor energy conversion losses.

The battery pack has a current sensor that outputs a voltage proportionate to current. The efficiency of the inverter and motor, once warmed up, vary with RPM and load in a knowable manner. So it would be possible to account for the torque from the electric motor, but you'd probably need a lot of dyno time to come up with the efficiency map of the electric motor for each model of hybrid you want to support. I've come across maps for the Prius, first-gen Insight, and Honda Accord V6 hybrid.

...when the "integrated" version becomes available, I'll get one; but, I'll just be a curious spectator during its' "component" development stage.

...talking on a cell-phone is already dangerous; don't need something *interesting* to distract me even more (ha,ha).

I commend you on the effort. It is an interesting area of study and I hope you continue. However, I do have some comments/concerns about the paper.

First, and minor, reference [3]. It is very doubtful that you are achieving +/- .1% flow measurements from the electrical pulse width controlling the injectors. When commercial automotive fuel systems are tested gravimetrically, the variation is already larger.

Also, it is a mistake to assume that this method is readily applicable to modern fuel efficient designs. Consider, some systems intentionally alter injectory opening and closing times conditionally, others will leave the injectors on some cyls closed while still cycling others.

Second, I have some real concerns about your basic mathmatical model. My prediction would be that it would generate a fairly innacurate instrument of engine efficiency, and this seems to actually match your own data.

Consider figure 7, efficiency calculated from dyno readings. It does rougly match the expected V/P curve for a conventional combustion engine. And also, as expected, you have an 'efficiency island' near peak torque. However, your efficiency island is elongated from typical engine efficiency models.

Mathmatically, this deviation in the curve should be exaggerated by under sampling and an EKF techique, and that is what we see in your data. Consider figure 6. You point out that peak matches, but that is the lowest possible hanging fruit, like finding lambda 1.0 on a UEGO sensor. The question is, what is the typical error as we move away from the easily identifiable set point?

Based on your included data, pretty high. Notice how much larger and elongated the efficiency island is on figure 6. The expected V/P slope is not even really visible on the chart. If we did a scatter plot of point to point deviation between the two it would appear that, aside from peak, the instrument would almost always over estimate efficiency, and in a non-trival operational envelope, extremely so.

My third concern is that the paper never really establishes a link between the acheived measurement and the stated goal. Look at part of your first paragraph:

"optimal control to make great reductions in emissions and fuel consumption"

Now, we could debate rather or not your measurement of engine efficiency is accurate enough to be of value. But the more fundemental question is, is the data even really connected to either fuel consumption or emissions?

Virtually everyone in the field knows that every conventional combustion engine has a fuel economy sweet spot - wide open throttle (so there is no induced vaccuum reducing VE) at peak torque.

But this is not an available decision for routine operation of a vehicle. In fact, most people here (myself included) go out of our way to avoid this point of operation, because it is terrible from both the perspective of emissions and *operational* fuel economy.

At peak torque wide open throttle, the mixture is generally quite rich, to keep CHTs and pressure down so that destructive abnormal combustion does not occur. But this also means that the vehicle is operating way out side of the cat efficiency envelope, making emissions soar.

I'm not saying this to discourage you. Just the opposite. You are attempting to inexpensively measure a very narrow definition of effiency. What I am suggesting is that you try an experiment to demonstrate rather or not your target metric is of any practical use in either the area of fuel economy or emissions.

Good Luck!
-jjf

...my thoughts were: "...too many 'user-supplied' inputs for the average-Joe Blow-driver to have reasonable knowledge of..."

...and, unfortunately, I don't believe the automotive companies would help by making ANY of their "proprietary Intellectural Properties" available...to anyone.

Sorry, I missed this earlier. I would contend that that is almost wholly false. The factors tend to work additively, and many act exponentially, which is why you over estimate effiency over so much of the operational curve.

-jjf

Very interesting comments. Here are my responses:

For instance in http://ecomodder.com/forum/showthrea...tml#post152293, it seems that that is the accuracy that people on this forum are getting.

Furthermore, Luxembourg is home to Delphi’s fuel injection research center (although they might have others). I spoke with one of my colleagues who researches direct gasoline injections, and he felt that this level of repeatability sounds reasonable.

That being said, I don’t believe everything I read on the internet, and my colleague couldn’t give me another citation, so if you have some additional information on this, could you send me the reference?

Agreed. For instance, some cars can switch between sequential and grouped firing patterns depending on efficiency vs. power requirements. However, this is a trivial problem, as by simply adding a voltage sensor to each injector, instead of only one, the problem is solved.

A bigger problem is what to do with diesels. Most cars sold over here are diesels, and this technique does NOT work for them.

Note, Fig. 7 is a 4th order curve fitting, not calculated efficiency. The curve fitting will have a tendency to reinforce this island effect. It’s somewhat worse on the 2nd order model, which is what we’re actually using now because it seems to provide acceptable results for the optimal control problem.

Hmm… my initial thought is that the fact that a 17-dimension non-linear observer with asynchronous outputs (data measurements) converges to a correct value is not low-hanging fruit. In future work, it can be refined, augmented, etc… in order to get fruit higher up the tree, but, of course, on principle you are right, what good is this model if it’s telling us what we already know? I’ll try to answer that at the end.

I’m going to show my ignorance here. The people I work with do observer and optimal control research, and I am a controls engineer (See e. busvelle - Google Scholar , jp gauthier - Google Scholar , u boscain - Google Scholar), so we’re not as strong as we'd like to be in car industry experience. What is the expected V/P slope, and keeping in mind that we’re looking at total system efficiency, where should we expect to see it?


Someone once told me, “When you can say it better than anyone else, write it. When you can’t, cite it.” I think citations [6,11,14] answer your first question better than I could.

As to the second, I’m not sure I follow. Are you asking if fuel consumption data is “connected to either fuel consumption or emissions?” I think I might be missing the scope of your question.


Interesting point about the catalytic converter. Can you give me some references, and perhaps a simplified model? This is exactly the kind of thing that we can/should include in the cost function. (More on cost functions below.)


I think we’re talking about different things here. You’re interested in absolute efficiency, whereas I’m interested in the shape of the efficiency curve. I going to say we’re both right, as it pertains to our particular problems.

One of the most nagging questions for me during this project is why the efficiency is overestimated. Looking at my equations, an overestimated efficiency implies either an underestimated fuel flow (going back to the +-0.1%, I do not know if this is the case for my car or not, due to a forgetful fiancée and the fact that we refill the car every two months) or an overestimated force. Judging from the dynamometer results, we are overestimating force. Still, it’s not important if the hypothesis that the overall shape is correct and that we have a bias error holds.

I’m not taking it at all as discouraging. Quite the contrary! I appreciate the fresh voice and the experience.

We have real-world trials in the pipeline, hopefully in the coming weeks, but it will be months before we can present the data. Taking your comments into account, I think a new visit to the dynamometer test bench is in order, this time to look at emissions as a function of optimal control.

A sincere thanks for your comments, and I hope you will find the time to continue the conversation. In summation, I would like to reiterate the goal of our project, and how this fits in.

The problem comes down to this: people want to drive optimally, but do not know how. Optimal control studies are performed, but they are always constrained to working hand-in-hand with manufacturers, making publishing data difficult at best. Furthermore, the data they publish is for an ideal engine in ideal test conditions. The real-world is far different, most of all because we’re not looking at the engine as a single unit, but as a part of the whole. The transmission, differential, etc… must be included, too. And that’s even before we start talking about differences in manufacturing and wear. Our observer is a method of overcoming these obstacles on a case-by-case basis without investing in hundreds of thousands or millions of dollars of equipment.

So how does it all fit together?

One of the great weakness of optimal control is defining the cost. At this moment, our cost is only expressed in terms of gasoline used. Of course we can use far less gasoline if we run in extremely lean conditions, but then there’s too much NOx production. So it would be interesting to know what else could be included in the cost. I welcome any and all suggestions, keeping in mind that for the purposes of making a complete synthesis we need to be able to idealize.

Optimal control results are heavily dependent on the solution method. One of the most common, dynamic programming, is one of the worst when it comes to accuracy. Dynamic programming is easy, you hardly need to think about the math at all. However, it takes tremendous computing costs and is completely impractical for a small microcontroller implanted in a car. NLP (non-linear programming) comes out a little better, but still needs a fair amount of processing power.

For optimal control to work in cars, we need a way to get an answer without throwing a supercomputer at the problem. By fitting the observed efficiency map to a polynomial, we are capable of finding analytic solutions, instead of being forced to find only numeric solutions.

Our goal is not to provide a bullet-proof efficiency solution, that will have to be left to people who have better access to the sensors, and better models. We hope that by publishing this observer, researchers will be able to focus on the more interesting parts of optimal control by saying, “the solution to the problem of mapping efficiency has already been demonstrated, so now let’s get on to the fun stuff.”

So in the end, once we find the analytic solution for a given polynomial, finding the best-fit constants can become more important. For the moment, it suffices to show that our approach finds a polynomial that is sufficiently good. After that, if it is really worth it, we, or others, can spend time refining the model. After all, that’s what happened with the MPGuino: it launched a method that was lacking in certain refinement, and my project was to take it a step further. I’m certain there’s far more to be done, and I thank you for the encouragement.

Sorry, I'm going to be terse and may miss some issues, I'm trying to help a plant get to rich burning LP engines the size of houses within the upcoming emissions compliance today because we're anticipating a lot of rain.

Think about that. 1. They calibrate measured to actual with a fudge factor. This covers a number of variables, including driver behavior. 2. They are measuring their results with fuel fillups. Are you certain that they are filling their stock tank systems at existing gas stations to the level of accuracy you are asserting?

Then he probably does not work on the fluid dynamics side. I'd strongly recommend that you read some of the SAE research papers in this area. Or, just try an external, and more precisely controllable, tank feeding the conventional one and measure for yourself.

Remember, if you can't measure it, it isn't science!

No. Look at the trend in efficiency over the last 3 decades. A big key is higher compression. This means that peak fuel delivery has to high, which makes operating injectors fast enough at idle speeds difficult. There are several technologies in use, but look at peak-hold injectors, which are expressly meant to address this problem.

Now look at modern control methods of peak-hold injectors, which include techniques like PWM. In PWM control, open and close times are conditionally varied - that is, they cannot be simply averaged out, they need to be measured and modeled continuously. This makes for a considerably more demanding measurement system and computations. More sophisticated device, multipled by every injector... It seems contrary to your basic assertion. Add the current trend for muliple injectors and even duel fuels, and it sees dead end.

But there are relatively low cost methods that do. For example, wideband UEGO measurement is becoming mroe common in modern vehicles. If you combine good lambda measurement with, say, MAF, you have volume of air and combustion ratio, so you know the volume of fuel burned.

Then you may be thinking about the wrong end of the problem. Look at your simplified model and consider the influence of torque. Now look at your instrumentation, it is all, generally speaking, at best accuracy at peak torque.

Now look at the torque curve and the weighting of your data predictor. You have the least error in your sample prediction and your calculation at peak torque.

On the flip side, forget the physics, math, and sensing and look at the data! Ultimately, you have to explain why your chart has the lowest error, essentially zero, at peak torque, but huge amounts of error elsewhere!

If the cause of error is not understood and established, then there is no reason to presume that incrememtal improvement can occur. In other words, we can't assume it is a matter of better math. It could easily be the precision and limits of your underlying measurements, or even a flaw in your foundational models and premises. In science, everything is on the table until the data is explained and the results replicated.

Actually, you are not looking at total system efficiency, you are looking at a fairly narrow area of system efficiency, post drive train (but while ignoring many real world events)

As far as the V/P curve and slope, that is too big a question to properly answer here. I'd recommend starting with basic texts in engine design. But, in super brief, think of a single cyl in an engine. Mechancially, we have a constantly changing volume (pistone up, piston down). In combustion, we are creating additional pressure which also follows a curve (picture the flame front radiating from the ignition source while generating an envelope of gases).

If we could somehow create a perfect engine, plotting these two on a two axis graph, we would basically have a repeating rectangle. But chemistry and simle mechanics do not allow this, so we get a deflated and slighly twisted football (American). We do different things to try to draw portions closer to an ideal rectangle, like turbulence in the mix giving faster burns at higher RPMs, but, ultimately, the optimum point of peak pressure to occur is fixed, literally built into the cyl. So all engines have a fairly small peak efficiency island.

No, I'm saying that your model appears to be too simple to be of much use in actually predicting operational economy. Consider a seemingly simple question, why do people here hate driving in winter?

Simple, fuel economy is worse. But why?

There are actually multiple reasons, even the fuel composition is different, but let's look at just two, emissions and aircharge.

The effiency of a cat is very narrow, both in terms of gas composition and in terms of required exhaust temp. So a modern automobile runs, as much as possible, at lambda 1.0 (actually, the vehicles are closed loop to equivelency ratio, but we often talk about the reciprocal, lambda). This gives both peak EGT, and a cat friendly gas composition.

Now, it is winter, and the air charge entering the engine is denser, so it takes more fuel to reach the same stoichiometric ratio. Simply by virtue of air being dense, you have to burn more fuel at even the lightest loads to keep emissions systems operating.

If you are a small plane pilot, you love cooler denser air, because you are taking off and generally climbing at wide open throttle. That denser air means you climb faster (for multiple reasons, some more important than others, but a big one is the plane is probably normally aspirated and the density altitude is lower).

If you are piloting an big car, it is another story. Like the plane, your peak performance is improved, but you aren't using peak performance, you are driving the same speed as always. How do you do this? You throttle back.

What does a throttle create behind it? Vacuum. So the VE of the engine is lower. That is, to obey the rules of the road, the engine is operated farther from peak efficiency...

And this is just the tip of the iceberg. The point of all this is that I doubt that a typical user here would be able to, say, save any fuel using your efficiency measurement, because your model simplifications mean that you inherently are least accurate in the parts of the opeartional envelope where they have no choice but to operate.

I also doubt that it is much use on the vehicle development side. If you make a criteria that a car only has to drive on surface streets, you can make it more fuel effient because you can put a smaller engine operating at higher effiency at relatively low speeds. Once the vehicle goes on a highway, you have to carry more weight to protect the occupants, and you need a bigger engine to propel the increased mass at higher speeds. Inherently, when that vehicle is operated at lower speeds, you will, by nec. be farther from the efficiency island of the engine.

The directions we have been going are smaller, turbo charged, higher compression engines, and you are not really compatible with those fuel delivery technologies - and making certain forms of very ineffient operation more efficient, and you are, again, least accurate in these parts of the operational envelope with your measurements and calcuations.

I'd be very happy to proved wrong, but you will need to do some experimentation and collect some data to do so!

Good Luck
-jjf

P.S. I mean it, I short changed a lot of important subjects above, but I didn't want to let it sit for days, since your paper is still relatively fresh in my mind today.

Repeatability is repeatability. I’m not a fluids expert, but I do expect that the gas station pump is accurate to beyond 0.1%. Otherwise, they’re making 0.1% on every liter they sell (after all, they’d be stupid to do anything other than have their pumps be at the absolute limit of the law). So if people see +-0.1% on multiple fuel-ups, that’s a strong indicator that they’re on to something.


He does CFD models of gasoline fuel injector asymmetrical dispersion patterns at high pressure. (Good lord, the processing power he’s got to throw at this kind of thing. He’s the kind of guy who asks for a 64-core computer, and then runs a process on it for 2 weeks at a time.) But this is besides the point. I’ll explain why just after.

By measuring current, instead of voltage, we can do the same on peak-hold as we do on saturation injectors. This complicates the sensor package a tiny bit, as now hall sensors or current shunts have to be used instead of simple voltage probes, but does not radically change the problem, nor the solution.

But again, this is getting lost in the details. The internal management system has far better ways of calculating fuel flow than I do. Our method demonstrates that with a certain number of inputs you can reconstruct the efficiency of the system. Whether or not this particular sensoring approach works in the future is not important. In the future we will have newer, better ways. You list one below.


No! That’s not measuring, that’s inferring. And if you’re not using a model based stochastic filter to do the inferring, you’re not getting the most out your data. Pick your poison-- EKF, UKF, sliding point observers, particle filters, information filters, etc…-- but definitely don’t go without.

In an abstract sense, the UEGO/MAF approach and the injector patency approach are identical, only the physical parameters change. My goal is to observe power_in, and both are valid approaches.

Even better, by the sound of it they are complimentary approaches.That is a perfect proof to my above point that “we’ll find a better way”. So this sounds like a promising line of attack for better models, and more importantly for diesels. Do you have a good reference?

I can’t explain that, in part because I don’t believe that. My observations show a peak of 35% holistic system efficiency. That’s too big, by a heftier margin than I’m comfortable with. But I’m expecting that that is in part because I’m using manufacturer’s data for things such as air resistance and am overestimating tire rolling resistance by a substantial margin (by >25% according to what I saw today.)

Could you specify what it is exactly that is bothering you? For the life of me I can’t see the problem you describe.

That is a question of philosophy. Anyone who’s ever studied observers knows that EKF (Extended Kalman Filters for everyone else out there) cannot be shown to globally converge. There is no mathematical proof for it. Yet, in no physical case have they ever been show not to globally converge. So even at the upper echelons of the mathematical world, there is cutting edge research being done, based on something that can’t be proven in a global sense. Even mathematicians are willing to fudge on things and not explain details that aren’t important when the system “just works”.

Now, I’m not saying that this is an excuse nor a reason for every Tom, Dick, and Harry who’s too lazy to cross his t’s, dot his i’s, and expects his new-fangled system to run on imaginium, but I do think it’s a mistake not to go further in this case because we have some nagging unknowns.

Ah, I see. You’re talking about the volume vs. pressure thermodynamic graphs. (Personally, I always preferred entropy vs. temperature, but I guess that just depends on who and what you work with.) Now that I know what you’re talking about in thermodynamic terms, I’m doubly unclear about why I should see the curve, nor which part of it I should expect to see in an efficiency map, esp. a curve-fitted one.

I am a small plane pilot, and I do love cold days, and cold, clear nights even better. (Just so long as there’s no ice accumulation due to radiative heat loss). But you comparison is unfair. You’re assuming that we claim that the summer efficiency map is the same as the winter efficiency map. Not so.

While we don’t even have enough data yet to show how the holistic efficiency map changes across the seasons (hurry up and get here, summer!), it’s a reasonable guess that it will only change the magnitudes of the efficiency curve, and not its general shape.

Well, look at it this way. We have three possibilities:

1) Drivers follow the optimum trajectory through R_17 space and see a net fuel economy increase.
2) Drivers follow the optimum trajectory through R_17 space and see a net fuel economy decrease.
3) Drivers follow the optimum trajectory through R_17 space and see no net change in fuel economy.

Many experiments with manufacturer provided data have shown that #1 is the outcome. See for example the excellent results from E. Hellstrom in “Look-ahead control for heavy trucks to minimize trip time and fuel consumption”. The only question left is, then, “Can we provide a similar increase in system performance with an efficiency map observed from real-world driving?”

Dollars to donuts gets you that the answer is “yes”.

Note that the typical user will NOT be able to guess the optimal control trajectory. If there’s one thing I’ve learned about optimal control as a mechanical engineer, it’s to shut up when the math speaks. Optimal control defies all physical intuition, and it’s terribly frustrating because it is optimal. There is no room for debate with optimal results, there is only room for debate with the model compromises that led to that result. The scary thing is how terribly “wrong” most of what we do is. Well, not wrong, but sub-optimal.

***

Exercise for the reader: you are the driver of a car on a road that is horizontally straight, but that has exactly the shape of a sine curve in the vertical dimension. (Imagine a roller coaster) The car starts of on the righthand side of one of the waves, pointing uphill. You have one single gear, and an ideal clutch (i.e., a clutch that instantly engages without slipping). Your engine efficiency curve looks like a bell curve (i.e. there’s a certain speed toward the middle of the curve at which you have maximum efficiency). Whenever you disengage the clutch, the engine turns off. You have no air or rolling losses whatsoever. (In fact, no losses at all aside from those dictated by the efficiency curve.)

Your goal is to get to the top of the hill three hills to the right, all while using the least energy possible. What is your optimal strategy?

I’ll send a free logging board and an AVRDragon to the first person who gets it right before next week Friday.

The correct approach is to engage the clutch until you have just enough kinetic energy to crest the first hill, then disengage the clutch and coast the rest of the way. Also, disengage the clutch if engine speed exceeds the peak of the efficiency curve, and re-engage if that is necessary to climb the first hill. The result will be that once the vehicle gets up to the speed corresponding to peak efficiency, it will hold that speed by running the engine at whatever duty cycle is required (this is familiar to ecomodders as pulse and glide).

Vehicle speed at the crest of each hill will approach zero, and the driver will die of old age before he finishes the course.

I could write a long post about this paragraph alone, but you do see the problem, repeatability, you just missed the application. People here are testing their accuracy against *fillups*. We do not see inside our tanks, so we are relying on a pressure system to kick off before we pour gasoline on the ground.

Do you really think that point is being hit with a high degree of precision? ;)

However, you are actually wrong even in your disbelief. A 15 degF change in fuel temperature makes about a 1% difference in volume with gasoline. The fuel at a station is generally stored in an underground tank, so it's temp doesn't change all that much, but the fuel lines between the tank and the nozzle do. When consumer reports tested morning fillups vs. afternoon fillups, they saw about a .1% change in fuel economy. Edmunds found higher, but over a hotter period of days.

That, by the way, is a clue for one of the reasons that I find your flow accuracy claim dubious when matched to other literature.

Not if they are under modern PWM control. I wish I had a graphic of a typical schematic handy, it would be interesting just where you expect to put your current measurment.

I think that this one paragraph is a perfect metaphore for ths conversation. There is no assumption in my comment on what you can and cannot measure in winter. There is only a verifiable example to back up my point - that there is a difference between engine efficiency and operational efficiency, and the difference varies depending upon application.

The only assumption that I am making is that you are young, and you are trying too hard.

I am old, so I have little patience for this sort of game. Look at your comments about flying. Private Pilot or not, you are still trying too hard to impress.

As it happens, I am also a pilot (ATP rated and CFI/CFII ticket), there are two essential ingredients for icing with a small aircraft, visible moisture and below freezing temperatures. The temps do not have to be around the plane, water droplets can get supercooled high in a thunderstorm and fall on you. But both elements have to be present (it's actually a question in both the private and instrument written test question blocks)

The suggestion that you can be out in clear, below freezing air and experience a dangerous accumulation of icing in flight is patently false. Cold clear air contains very little moisture and sublimation cannot occur under a boundary layer of 100 kt air.

NOTE for any lurking pilots, this does not mean you should ignore frost in your preflight, accumulated frost can have a fairly significant impact on lift. No, you almost certaintly wouldn't fall from the sky, but you may find your deptarture shorter on runway or a lot closer to obstacles than you wish to be - so take a few minutes and properly deice it. AOPA's Air Safety Foundation has a good online course on this.

Now, aside from establishing that you don't seem to know what the FAA would like you to know about icing, what has this accomplished? Nothing.

Likewise, what possible reason could I care about the job title of a nameless person that annecotally supports a claim? Does that meaningfully compete against accurate measurement? If so, then what does it mean that I am (truthfully) part of the team that was recently awarded the SAE/MIT Best Innovate award for green/efficiency automotive technology? Does that suddenly make your friend wrong, or do we have to go into a runoff where we measure lab coat stains or count pens in our pocket protectors?

All this sort of noise does is distract from meaningful dialog. Aside from the surprisingly fierce battle between annecdotal opinion and the existing published literature on precisely measured performance of fuel systems, I'm really making only two very simple, (and I would think) easily understood points.

1. If you print your figure 6 and figure 7 on transparency in matching scales and lay one on the other, redish color on one will fall on blueish on the other.

2. It is not really practical to maximize operation of a car at the center of the red zone - since it essentially identifies peak torque.

That's it, take it, leave it, deny it, circular file it, whatever...

I sincerely hope you continue with your studies, but if you are this fierce about annecdote vs. prior art, research is going to be a very stressful field for you!

Good Luck
-jjf

Very well explained.

:) I needed that. It is hard for me to not write as a grumpy old man.

Thx

In the interests of discussion, I can't comment yes or no until the week is up.

Again, repeatability is repeatability. We cannot comment on how members who report their fuel economy do their fill ups. It is as reasonable to suppose that they stop the first time the handle kick off as they they fill until they see the gasoline reach the filler inlet. As long as people accept 1 part-per-thousand, that’s what I’ll use. And when they no longer do, I’ll change it. You tell me about “the literature”, but for heaven’s sake, my good fellow, give me a specific citation! If the citation says +-1%, I will put that in the study instead, it makes no difference to me.

But we’re getting lost in the details. Let the experts figure out the best way to measure fuel flow, the truly interesting problems for us are in the overall structure.

I don’t know, but I’m sure someone out there will figure it out and then I’ll have my answer. As problems go, there are far bigger fish to fry than this one. What about electric cars? What about hybrids?

If there’s a modeling or measuring problem, I’m certain we can overcome that when we get to it. The whole reason I’m seeking help on ecomodders is because I only have one car, so I can’t possibly test all the various possibilities.


This analogy works both ways. The world is as full of stories of old men who claimed it couldn’t be done as of young men who failed trying it.

Woah, slow down pardner! You’re dangerously close to the line of “assume makes an ass…” The only thing I’m saying is that I don’t like having to miss my weather window because of ice accumulation (specifically frost, but laymen reading this conversation won’t appreciate the difference as it pertains to aircraft) due to radiative heat losses on clear nights as the airplane is sitting on the tarmac. I also don’t like when condensation on control cables ices up. Kind of gets your sphincter in a tight knot when you’re at 11,000’ and you realize you don’t have ailerons anymore.

(That being said, to all those reading this, what jfitzpat said is correct down to the letter. But in the interests of scientific conversation it is not exactly relevant to the topic at hand, and I kind of wish we could delete this whole aeronautic detour through the woods. Clouds. Whatever.)

First, look at it from my perspective. You are a voice on the internet. You could be a crank, or a genius. I have no way of telling. I can only judge based on data. Until you support your arguments with data and/or references, I’m stuck. It is entirely within the realm of reason that you are correct, but you’re asking me to take that on faith over what a fuel injector engineer tells me. A nebulous voice on the internet over the professional down the hall. What would you do?

But the problem is that all this is missing the point. If the error is higher, that’s fine, I just have to detune the observer a tad bit. It doesn’t really change whether I can show gains through optimal control.

Second, getting on to optimal control, who ever said that we would be driving in the area of maximum operation? That’s a big… assumption. Honestly, I can’t tell you what the optimal trajectory for your car will be. I’ve said it before, and I’ll say it again: intuition fails when it comes to optimal control. Your intuition, that you’ve gained after so many years, is standing in your way. It’s an obstruction to you seeing the overall picture.

You seem to know what you’re talking about, but I think you could contribute far more to the conversation if you responded to my questions about sources and references. I would love nothing more than to incorporate your experience into my observer and make for a better model. But what I’m getting out of you is that “I’m trying to hard to impress” and “research is going to be a stressful field”. You seem like a valuable resource, but not like this. A change of direction is in order.

Because the fact is, right now I’m engaging in a dialogue that I don’t have time for either, and that isn’t helping me. My whole theory, rather debate, with my colleagues is that it is far more interesting to engage the internet and get the experience of people who don’t necessarily read journal articles, than to go to conferences, publish in unread reviews, etc… I argue that there is a whole world of untapped cross-disciplinary experience, but we have to go to it instead of sitting on our duffs and expecting it to come to us. Right now I’m losing the battle, severely.

Mr. Fitzpatrick,

I take it from your posts here and in the "reliable PCB" thread that your specific field of expertise is Antagonism Engineering. Consider also the fact that you are a self-described grumpy old man. Your posts have been longer than necessary, drifting off-topic, and full of sublte personal insults and irrelevant questioning of the qualifications of our guest experts.

Please change. You appear to have considerable technical knowledge and experience. Can you apply it primarily to improving EcoModder projects, instead of simply pointing out their shortcomings?

All you have to do is to generate the energy needed to get to the top of the first hill in the most efficient way possible.

Engage the clutch, start the engine, set engine speed (and load obviously) at max efficiency while climbing till you have just enough kinetic energy to make it over the top of the hill at which point you disengage the clutch while you're still climbing, shutting the engine off automatically, you'll just barely make it over the top and you then coast this way going up and down hill for a while with the engine off and just brake (or barely engage the clutch, the resistance of the engine should stop the car) when you want to stop at the very top of the third hill.

re Exercise for the reader:

In the real world there are soo many variables that I would probably give a mpguino like program/device an accelerometer (because we are on a changing slope) and throttle control, and maybe a hint that the load would be decreasing and have it seek/adjust the most efficient throttle position, then try to make sense out of it :)

course I don't know what exactly an "ideal clutch" means.

I believe there is more than a 0.1% error in fill to fill variations. Especially if you fill at different pumps. Although I believe they measure fuel accurately, I do not think they will all stop at the same level. Also, since a fuel tank is an irregular conainer, if your car is tilted in different ways, the error can easilly go past 0.1%.

You should try to do coastdown tests to validate the manufacturers value you input in your model (Cd and Crr). Maybe revalidate the frontal area too. As already suggested, you could validate you fuel measurements from MAF and O2 sensor output.

Also, Crr gets lower as the distance driven increase (as the tire gets hotter, hysteresis losses are reduced). There are a lot of other variables that will change like that depending on the trips you log.