The Governing Physics of Engine Operation?

[evilhonda]

I'm running Tunerstudio because the software provides the most dynamic adjustability for Megasquirt, but the tune in it is not configured as the creators intended. While it may appear at first look to be set incorrectly, the values are set to constrain the air, ignition, and fuel delivery to the actual physical limitations that govern the engine's operation. In the beginning I did exactly as instructed in order to familiarize myself with the software. Unfortunately, the settings were intended to address issues of common assumption about how an engine executes a combustion event, so the results were exactly as describe to be expected, which was completely unacceptable. Right now and any time, under any condition, I can start my engine and launch the car immediately, with out waiting for a warm up, and it will deliver overwhelming and consistent results efficiently and powerfully. It starts the same every time and this is no accident.

In my tune, the settings windows are utilized independently of their definitions, I only employ their specific functions within the logical physical parameters that govern the operation of the engine. What about fuel condensation, valve temperature, and loss of fuel volatility under varying conditions? Did you know that heat soak problems are simply a matter of adding air during start up? No one can deny that something larger will take longer to burn than something small. So why is there no direct way to advance the timing for larger combustion charges? Conversely, high manifold pressure does not indicate a larger combustion event, it indicates that the engine is pulling air less efficiently per throttle angle and requires a spark retard to place the correct stage of the combustion event at top dead center because at a given throttle angle more pressure in the manifold means less pressure in the cylinder, and consequently, a smaller combustion event per throttle angle. Hence the commonly overlooked reason that timing is retarded when accelerating, but airflow efficiency effects the engines capability to load a combustion charge. When barometric pressure, valve timing, or throttle angle change the air flow efficiency, they need to be accounted for independent of the manifold pressure as an increase or decrease of air mass/volume, and the manifold pressure will indicate how efficiently the engine accepts those changes.

Another point to consider is that the intake manifold never has a consistent pressure throughout it's interior. The point in question is what the pressure at the valve is as opposed to the average pressure throughout the manifold. When air is pushed in by higher barometric pressures, the lowest pressure zone(at the valve opening) pulls away from port fuel injectors and the resulting higher pressures at the injectors and restricts fuel delivery. The net result is that when the throttle is closed higher pressure values require a wider pulse for the same delivery that those same pressures require at open throttle per pressure because the tuning software does not offer a place to discount pulse width for varying valve pressures. This condition is hidden by the fact that the pressures invariably get higher as the throttle opens and the specific variances of pressures during the event are ignored. I'll bet that when you part throttle, the system seems to go lean unless you add a disproportionate amount of pulse width to the calculation. This is caused by the manifold pressure at a specific throttle angle(moving) rising after the opening throttle passes that point(after being the operative word), and the pressure falls after throttle passes the same point when closing the throttle. The tuning software does no specifically allow a calculation under zero to compensate for this, so if you tune doesn't go lean under part throttle you are wasting a lot of gas as the throttle closes. These are real time considerations of conditional change.

I have a lot more if any one is interested. Like what the values from the oxygen sensor really mean. I would like to see this software evolve to perfection, but before you can be so picky about your tune, your Megasquirt needs to be perfectly constructed and configured for your application with no misunderstandings about how the hardware and software work because these considerations will not resolve poor craftsmanship or incorrect diagnosis.

Edit because of distraction during composition. Probably needs more editing.

[Raymond_B]

I'd like to see a few things.

A: Your tune and what vehicle it's for
B: Your credentials
C: A datalog

[evilhonda]

Here is the msq.

2014-09-12_07.26.51evilhonda.msq

2320 lb,1989 Honda Civic, 3.5l, v6, 5 speed. Ms2 and Tunerstudio, EGT at coolant input, narrow band o2, modified Tps, barometric pressure sensor, IAT in the engine bay, manifold air sensor that constrains the TPS value inline, ecu mounted MAP, 26lb injectors 88mm throttle body, v-tec at 2800 rpm(just outside of normal driving range), 1.75" primary x 6 to 4" collector, open 4" pipe to the rear+inverted gun silencer at the tip+debris filter, custom distributor with vr output, Honda igniter by d14 instead of the original coil driver(a bad coil killed the original coil driver), underhood air supply, tires match the speedo.

[evilhonda]

Had to find a smaller data log. I am a twenty five year mechanic, and I work on hundreds of cars per year. The physical data that I used to quantify the physical parameters are searchable, such as atomic weights and the mole count of oxygen and the other components of air per temperature. I'm in college now for a business degree so I understand the need for validity, but the source that I used to verify my theory of engine operation was Megaquirt+Megatune+Tunerstudio as this is the most comprehensive way to study engine physics. My project is a 2320 lb 89 Civic with a 3.5l v6 and 5 speed manual. I've put 60,000 mile on the build along with Megaquirt as a stand alone engine management system. My focus is not with tweaking or modifying Megasquirt, it is to understand the complex dynamics of engine operation as it pertains to component configuration, and the efficiency and performance of engines. The tune that I provided has been that way for several months because I'm watching for global tuning issues that show up with the changing seasons, but I can certainly discuss any questions or concerns about it. I just got the log viewer so if you want specific data I can run a new log for it. Anything that I claim can be proven with Megaquirt+Tunerstudio and scholarly data from reliable sources.

[Raymond_B]

Interesting.

I guess I misread your 1st post, to me it looked like you were calling out the creator of TunerStudio on how it is not correct. But after re-reading it looks you are simply saying that in the context of the person who is building their MS and tuning it with TS.

[evilhonda]

The shape of the maps and enrichment curves have a larger influence on the tune than the actual values do. With the tune shaped that way I can select specific fueling and timing strategies for the best application under varying demands and conditions. The specific shapes of the curves and maps have made the engine and tune very tolerant to a wider range of timing and fueling, for example, the engine will remain stable at a very low rpm in a wide range of fuel mixtures and timing, and deliver consistent high rpm performance. When I used Megasquirt's original tuning strategy the car was dangerously wild, it didn't have nearly enough resolution and a one percent fuel change would make it lunge forward or fall on it's face. One time the car stood on the front tires at around ten miles an hour because of poor fueling resolution, it went lean and the large engine stopped the car. It runs in a manifold pressure range that, for most cars is in the de-acceleration range because there isn't enough load from accessories and the car is very light for the engine. Also, no one has ever presented a faster car to me. The last time I ran it against a street bike(750cc), the guy thought there was something wrong with his bike because he couldn't keep up, I checked it out but nothing was wrong with it. I took a GT 3 as he launched it from a dealer lot with dealer tags on it, he probably didn't buy a Porsche that couldn't keep up with a Civic. Megasquirt made it work.

[evilhonda]

Hey, I just got your post, yes, I'm not hating, I just want to see some progress with the software other than compatibility issues, and usually people are on here just trying to figure out how to use it. I think that there should be a lot more discussion about how engines work since we already have the tools to make them run right.

[kjones6039]

Yikes..............,
I'm just gonna take my 50+ years automotive/electronics and quietly go to my corner to see how this all comes out..........

Ken

[evilhonda]

kjones6039, Try my tune. To make it match you engine, divide your required fuel by 2.55 and enter the number into the required fuel. With a calculator, check to see if the widest possible pulse is not longer than the elapsed time at the rpm where you get peak horsepower by multiplying all enrichments, and make sure that you use the wot value(throttle table) at peak horsepower in the calculation(in cold temps, injector saturation will limit the calculation because the ecu can't calculate a time longer than the event affords so Ms2 just skips events, sync, or whatever it really does), this can be adjusted or scaled by raising or lowering the values in the warm up table and the MAT table. Then set the thermistor values(air is exactly what the real temp is) if you use a coolant sensor run the exact temperature(coolant sensor will cause a rich condition until the engine bay heat saturates(problematic cyclic behavior), if you use egt at coolant then set the thermistor set the temperature range to get -40 at the highest possible egt sensor resistance(cold start on the coldest day) and the lowest resistance(hot active engine) so that it reads in the 200 degree range. You want to stay out of "warm up enrichment". Unless you have a disproportionally large throttle body, the throttle map will work, if it's a normal tb, that drop off will be too sharp and plateau too soon per throttle percentage(a smaller tb will show a more gradual decline that ends closer to wot). The most efficient event is not at the maximum advance point, so don't be too aggressive with the spark advance. Check you port settings and compatibility with component configuration. Check the calibration of the MAP sensor for stable idle.

Edit: Forgot to mention that my msq. has seen a few updates from Tunerstudio since I last went through every setting. Check every setting for spikes or omissions, it's probably okay because the updates have not caused any weird problems in the way my car runs. I'm just really letting it go without adjustment for an extended period of time.

[kjones6039]

kjones6039, Try my tune.

Thanks anyway, but I think I'll pass.

Ken

[Matt Cramer]

It sounds like you're calling for a model based ECU. The current MS2 and MS3 code incorporates a few model based calculations, such as Enhanced Acceleration Enrichment modeling the amount of fuel stuck to the intake ports. However, more complicated forms of model based processing are likely to require more processing power.

[evilhonda]

You may be right Matt, I had to look up "model based calculation", it seems to describe the process that I used to determine the necessary values for the tune, only I did it in my head with standards taken from physics. The model part of it produces a breakdown into the application of the components, including the Ms2 ecu. The model based approach may produce a set of applicable variables that can be used to accurately tune an engine. If you guys produce an ecu specific to that model(engine dynamics) I would certainly buy in.

Thank you for the input.

[cygnus x-1]

You may be right Matt, I had to look up "model based calculation", it seems to describe the process that I used to determine the necessary values for the tune, only I did it in my head with standards taken from physics. The model part of it produces a breakdown into the application of the components, including the Ms2 ecu. The model based approach may produce a set of applicable variables that can be used to accurately tune an engine. If you guys produce an ecu specific to that model(engine dynamics) I would certainly buy in.

Thank you for the input.

I believe one of the original stated intents (by Bowling and Grippo) of the MS3 ECU was to explore model based engine control. This was back several years when the MS3 was still being designed and I haven't heard anything more of it since, but I found the idea intriguing. So I'm happy to see that someone is thinking about this still.

I've read read your posts a few times now, and looked at your .msq and log. I must say your tune looks completely unworkable and I'm not completely following your description of what you did, but I am interested in hearing more about it.

It sounds to me like your goal is to find a more accurate way of determining the amount (mass) of air that gets into the cylinders so that the correct amount of fuel can be injected, correct? You also appear to be using the same physical measurements (manifold pressure, baro pressure, throttle angle, etc.) but in a somewhat different way. Could you elaborate on this a bit more?

I'm also specifically interested in the idea of using the exhaust gas temperature instead of the coolant temperature. Could you explain this as well?

I know this is probably a lot to explain at once, so if you want to answer one topic at a time that might be better.

I look forward to hearing what you have to say, so thank you in advance!

C|

[evilhonda]

I've been thinking about how my tune could apply to any engine for a long time, but the tune is based completely on concept instead of actually dialing in the tune in a dyno or street setting, and the concept has been developed by a model of the engine's function that I created in my head. I recently saw a brief report about scientists using a program similar to a video game to examine possible genetic structures of dna. They had individuals attempt to assemble different molecular components with possibilities derived from their human brains, with some success in discovering structures that served their purposes. They were doing this because a model based calculation to determine the possibilities was far too complicated to create. If Megasquirt were to employ a virtual model of the engine operation(a collection of Identified relevant values in numerical form working together as a functional equation), it would be the product of extensive research and the size of it couldn't fit into the final product(ecu). I think that they could create a functional model(set of identified,constraining parameters, in a real time calculation), and use it as a tool to establish a set of variables that could be entered into a processor(end user ecu) to evaluate and operate relevant system components.

[evilhonda]

I appreciate your request to extend this dialogue.
The exhaust gas temperature reflects the amount of work the engine has recently done, when placed close enough to the cylinder(s). I use it to get a reaction of the tune from the added heat present after the engine has done work, for example when the car stops, the engine idles high on the fixed idle setting(same air volume all of the time), partly because of the added heat's contribution to the combustion event, so because the events are completing faster, the timing needs to be retarded, the fuel needs to be reduced, and the air volume needs to be reduced, not just the air volume alone. when the actual heat in the cylinders was ignored, only the air volume was reduced to control the idle speed and the calculations supporting the fuel and timing would not accurately compensate for real time changes in the events' potential to complete thoroughly and in time. My tune attempts to conform the engine to fueling and timing requirements in real time because the engine and user's demand of the engine is never constant. The load continually changes and the throttle rarely remains still under operation. When the throttle is moving(not fixed) or the engine temperature is changing(the most influential heat source is the cylinder heat), the rate of change and intensity of the changing values needs to be calculated. The exhaust gas sensor does this by changing in real time, relevant to the real conditions in the shortest period of time for the shortest and closest reaction to engine conditions. I use the egt as part of how the afterstart fuels and times the engine, after I discovered the cold starts were mostly a condition of the cylinder temperature, so my afterstart is really two parts, one in the pulse and cranking for fuel condensation an evaporation at the ports, valves and cylinders, and the value and reaction of the egt(at the coolant sensor input) for the short term fueling and timing required until the cylinder heat reaches the optimal temperature range. In my car, when the engine starts cold, the fueling and timing quickly reduce as the cylinders heat up and the calculation that causes that also applies accurately to the real engine requirements all of the time instead of needing to change the calculation after the engine has come to operating temperature for air density and temperature changes that are not reflected by the sensory input(problematic cyclic behavior).

[evilhonda]

In response to the question of the relevance of the values at the throttle table, barometric pressure, and manifold pressure, I have to point out that much of the influence to the tune for these values are hidden in the way my particular system operates. They are the real time affects of the behavior demonstrated by those sensors. The manifold pressure enrichment values are based on the physical limitations of the engine(pressure at the injector, throttle position, and barometric pressure), and what proportion of the total fuel/timing value is required during a given change of the throttle and barometric pressure, so the numbers on the manifold pressure map represent the scale of contribution from manifold pressure, more so than any absolute value(the absolute, in this tune, is the shape of the map).
The throttle map reduces that way because I had a persistent problem with the fueling being too lean during an opening part throttle change where the calculation was otherwise accurate. I tried every possibility before discovering that the calculation that I was looking for would need to discount the multiplication of the parameters(a negative calculation occurring in real time along with positively multiplying values). I theorized deeply about the cause for this and determined that there was a problem with the pressure at the fuel injectors causing the system to go lean at pulse-widths that were expected to deliver a given volume of fuel. When I created this real time static at the changing position of the throttle that caused the calculation to go positive or negative in multiplicity based on how much fuel could be delivered at given pulse width and pressure, the problem with lean part throttle completely disappeared. That is why the throttle map reduces fuel in the higher throttle angle. The throttle timing map looks nothing like that because there is no influence of air volume because of this, so it just reflects the volume and of air entering the engine as it pertains to the length of time required to complete the event. Still confusing, I know.

[evilhonda]

I also mentioned (priming)pulse and cranking, so I think that these settings windows need to be addressed. The car wouldn't start consistently, partly because of the temperature of the internal engine parts and the air in the engine at start up. To resolve this, the priming and cranking pulse window are set to only allow enough fuel to maintain the volume of fuel required to overcome condensation or evaporation depending on the temperature(a trace amount of fuel and not enough to start the engine). The priming pulse delivers a volume of fuel to saturate the metal parts at cool temperatures and to cool the metal parts during a hot start and the cranking enrichments are to maintain the fueling necessary to overcome this until the engine starts. The cranking rpm setting is intended to limit how fast (takes a period of time to reach and how many cylinders pass) the engine turns before beginning to calculate based on the load where it then fuels enough to start the engine. In my tune the setting is ninety rpms so that during that time all of the stale air and fuel that may be present is passed through before the engine starts. I calculated the volume of the manifold area outside of the valve ports and determined that when the cranking engine reaches ninety rpms the manifold and cylinders have fresh air and fuel(because of the priming and cranking pulse values) and will start predictably. The need to supply clean air and fuel before start up excludes the skip pulse setting as a solution for start up concerns, I have it set to one pulse and when I set it higher(three pulses to clear the manifold and get a consistent tach pulse) the rpms are too high and there is no time left for a quick start that is consistent and accurate(cranks too long before starting), so this process of priming and cranking pulse width allows the engine to get to the starting rpm without prematurely initiating a combustion event and provides accurate fueling at startup from the fuel maps and enrichment tables.

[cygnus x-1]

I've been thinking about how my tune could apply to any engine for a long time, but the tune is based completely on concept instead of actually dialing in the tune in a dyno or street setting, and the concept has been developed by a model of the engine's function that I created in my head. I recently saw a brief report about scientists using a program similar to a video game to examine possible genetic structures of dna. They had individuals attempt to assemble different molecular components with possibilities derived from their human brains, with some success in discovering structures that served their purposes. They were doing this because a model based calculation to determine the possibilities was far too complicated to create. If Megasquirt were to employ a virtual model of the engine operation(a collection of Identified relevant values in numerical form working together as a functional equation), it would be the product of extensive research and the size of it couldn't fit into the final product(ecu). I think that they could create a functional model(set of identified,constraining parameters, in a real time calculation), and use it as a tool to establish a set of variables that could be entered into a processor(end user ecu) to evaluate and operate relevant system components.

That's an interesting idea, and I can see how it could *potentially* lead to more accurate engine control with limited ECU resources. However given the significant complexity of engine operation and the large variability of possible configurations, it would take a very long time and an incredible amount of effort to produce a functional model accurate enough to represent the physical engine. It would also mean that the end user would have to supply (to the functional model) a lot of data about the physical engine configuration than wouldn't otherwise be needed with the current approach. I suspect this would be a rather daunting task to many end users.

C|

[cygnus x-1]

I appreciate your request to extend this dialogue.
The exhaust gas temperature reflects the amount of work the engine has recently done, when placed close enough to the cylinder(s). I use it to get a reaction of the tune from the added heat present after the engine has done work, for example when the car stops, the engine idles high on the fixed idle setting(same air volume all of the time), partly because of the added heat's contribution to the combustion event, so because the events are completing faster, the timing needs to be retarded, the fuel needs to be reduced, and the air volume needs to be reduced, not just the air volume alone. when the actual heat in the cylinders was ignored, only the air volume was reduced to control the idle speed and the calculations supporting the fuel and timing would not accurately compensate for real time changes in the events' potential to complete thoroughly and in time. My tune attempts to conform the engine to fueling and timing requirements in real time because the engine and user's demand of the engine is never constant. The load continually changes and the throttle rarely remains still under operation. When the throttle is moving(not fixed) or the engine temperature is changing(the most influential heat source is the cylinder heat), the rate of change and intensity of the changing values needs to be calculated. The exhaust gas sensor does this by changing in real time, relevant to the real conditions in the shortest period of time for the shortest and closest reaction to engine conditions. I use the egt as part of how the afterstart fuels and times the engine, after I discovered the cold starts were mostly a condition of the cylinder temperature, so my afterstart is really two parts, one in the pulse and cranking for fuel condensation an evaporation at the ports, valves and cylinders, and the value and reaction of the egt(at the coolant sensor input) for the short term fueling and timing required until the cylinder heat reaches the optimal temperature range. In my car, when the engine starts cold, the fueling and timing quickly reduce as the cylinders heat up and the calculation that causes that also applies accurately to the real engine requirements all of the time instead of needing to change the calculation after the engine has come to operating temperature for air density and temperature changes that are not reflected by the sensory input(problematic cyclic behavior).

Ok, I see what your saying. Basically the EGT reacts faster and is a better predictor (than coolant temperature) of the changes in combustion based on temperature.

Regarding temperature based timing and fuel compensation, I believe the MS3 has this capability whereas the MS2 doesn't.


C|

[cygnus x-1]

In response to the question of the relevance of the values at the throttle table, barometric pressure, and manifold pressure, I have to point out that much of the influence to the tune for these values are hidden in the way my particular system operates. They are the real time affects of the behavior demonstrated by those sensors. The manifold pressure enrichment values are based on the physical limitations of the engine(pressure at the injector, throttle position, and barometric pressure), and what proportion of the total fuel/timing value is required during a given change of the throttle and barometric pressure, so the numbers on the manifold pressure map represent the scale of contribution from manifold pressure, more so than any absolute value(the absolute, in this tune, is the shape of the map).
The throttle map reduces that way because I had a persistent problem with the fueling being too lean during an opening part throttle change where the calculation was otherwise accurate. I tried every possibility before discovering that the calculation that I was looking for would need to discount the multiplication of the parameters(a negative calculation occurring in real time along with positively multiplying values). I theorized deeply about the cause for this and determined that there was a problem with the pressure at the fuel injectors causing the system to go lean at pulse-widths that were expected to deliver a given volume of fuel. When I created this real time static at the changing position of the throttle that caused the calculation to go positive or negative in multiplicity based on how much fuel could be delivered at given pulse width and pressure, the problem with lean part throttle completely disappeared. That is why the throttle map reduces fuel in the higher throttle angle. The throttle timing map looks nothing like that because there is no influence of air volume because of this, so it just reflects the volume and of air entering the engine as it pertains to the length of time required to complete the event. Still confusing, I know.

Regarding the lean condition during opening part throttle, I assume you tried the standard acceleration enrichment methods (MAP/TPS based and EAE)? With some tuning these methods are usually sufficient to compensate for all but the most dynamic conditions or unusual engine configurations.

Also, does your fueling system not have a fuel pressure regulator that is compensated by manifold pressure? This is essential to ensure that the amount of fuel delivered by the injectors stays consistent as manifold pressure changes.

In one of your other posts I saw where you describe your engine setup. It's not exactly stock but it didn't sound like something that would be out of the range of the *typical* tuning methods. So I'm a little surprised you had to come up with a radically different tuning method. Of course without first hand experience with your engine I can only speculate.


C|