Ricardo Wave Experience

[Chuck Dean]

I think it would be very worthwhile for people to share their experiences in working with Ricardo Wave. It would be beneficial to all of us in helping us use this incredible tool more effectively.

This isn't a post asking for engine specs, flowbench data, etc....but those are certainly welcome http://fsae.com/groupee_common/emoticons/icon_biggrin.gif

I think the important thing to discuss here is our lessons learned. For example:

-What areas were determined to be the causes of an inaccurate model?
-How important are accurate friction correlations? (sensitivity)
-IRIS cylinder vs. regular cylinder?
-Helpful tools within WAVE
-3D vs 1D...discharge coefficients...DELX...DIAB
-Restrictor Modeling in Wave
-What NOT to do
-Combustion Model insights

...the list goes on.

While troubleshooting a WAVE model can be very "educational", I think it would be best to post suggestions of what works and what doesn't work.

With such a powerful program available to FSAE teams, it would be a shame for us to spend all of our time trying to get it to work properly instead of using it to its full potential.

[Chuck Dean]

I'll start it off.

Today, for example: I used a function from the top menu bar called "veiw parameters" (or something like that). Anyways, this feature lets you look at the diagram of your complete engine system in respect to a certain parameter. So, if you select initial temperature, it will provide a color-coded scale for temperatures from low to high. Your exhaust ducts and junctions should all show up red (hot) and your intake ducts/junctions should be closer to the blue (cold) spectrum. So it shows the properties of each element in a visual way.

You can also use this function for initial pressure, diameter, bend angle, etc.

It's very useful in checking your model for consitancy - rather than clicking on one element at a time and viewing properties. If you forgot to specify a temperature in a header duct, this tool will show that.

[Jersey Tom]

This is our first year using WAVE, was part of one kid's independent study in intake design. When our intake is done we'll try validating it.

[Belo]

I'm using it for the first time this year and I can tel you it's been a pain!!!

Thanks to Marc from Polytechnic of Montreal I've been able to finally get it to run!

First of all I would really recommand you use the 4 cyl model to start and modify it with your engine specs... Otherwise you'll spend hours building a non functionnal model! (I did it...)

Many options are hard to understand and the help file has rarely been usefull in my case...

If you use the 4 cyl model, make sure you set the output data to be mean values, not instantaneous! I had a 600cc engine pulling 200hp at first... took me about 2 weeks to find the problem! I don t have access to the program right now so I can t say exactly in what menu you can set that but its somewhere we you can had a filter and set the frequencies of it. It might be in Wavepost.

All my friction factors are set to 1.0, the final results are an F4i pulling 56Nm of torque at 8000RPM and about 53KW at 10 000RPM. My intake runners are straight and the exhaust bend angles are all set to zero. That for sure will be an advantage but friction factor set to 1.0 might compensate. I didn't really mind about the results, I was mostly interested by the effects of runner lenght and plenum volume change. The program as been a great tool for that purpose. I'll see on the dyno if the curves are close to reality.

Ricardo should do some models of the most common FSAE engine, that would be great!

Anyone knows if there is a tech support?

The controls for moving and editing the model are not right

The best way to work with the program is to run input checks everytime you make a change, if you wait at the end, you ll end up with problems...

Make sure to use octane http://fsae.com/groupee_common/emoticons/icon_wink.gif

It would be interesting to validate our model by putting our values into a proven model... I think Poly of mtl as a pretty accurate F4i model, maybe marc could help us! If mine looks good, I could for sure share it in exchange of a rewarding thank you for the time you have spend swearing at the computer!!!

Alex

[LCP_KL]

-What areas were determined to be the causes of an inaccurate model?
-How important are accurate friction correlations? (sensitivity)
-IRIS cylinder vs. regular cylinder?
-Helpful tools within WAVE
-3D vs 1D...discharge coefficients...DELX...DIAB
-Restrictor Modeling in Wave
-What NOT to do
-Combustion Model insights

this is our second year using WAVE and first year using F4i. due to resource constraints we did not measure the head and piston/con rod dimensions, things like con rod offset, vol at TDC, etc and only use the value that we can find in the manual. valve profile was simply expanded from the example given in the tutorial, since we could not afford to open up the engine. lastly all the coefficients were approximated. in short, our model is very crude.

we mainly use WAVE to see the effects of intake and exhaust lengths. beats dynoing all the different iterations. that was several months ago. recently, we dynoed our actual engine and found the WAVE results to be relatively accurate. local peaks and troughs and shape of graph were roughly the same. The WAVE results look like the acutal graph only shifted about 1000 rpm up. actual torque and hp were about 10 too low.

as for the restrictor, heard that it is much more accurate to model it as a orifice with a certain Cd. since WAVE cannot discretize a tapered pipe well.

my 2 cents

[Chuck Dean]

I have seen some problems with the simulation not liking too much valve overlap.

I have been working on a seperate engine model (Honda B18C1) aside of our FSAE engine model to gain additional experience working with the program.

The power and torque values were extremely low on the low end even though I had complete flow data and accurate cam profiles for the head. After playing with the model a little bit I found that if I decreased valve duration by 20% the model's power/torque curve was almost identical to the stock configuration.

I think if you input cam profiles with lots of overlap, the simulation assumes TOO much reversion and overscavenging. Just a guess of course.

I'd be curious to see if this would differ if the geometry of the combustion chamber was COMPLETELY modeled via an IRIS cylinder.

Has anyone else seen accurate cam profiles produce the wrong shape of a power curve in their simulations?

[BeaverGuy]

I haven't used WAVE, but with Virtual 4-Stroke when I put in the actual lift profile, as opposed to just the lift and duration, for the intake cam I got overly generous numbers. I'm not really sure what the problem is though. Hopefully, they will get the new cams in the motor soon and I can compare those to the simulation predictions and that can shed some light on the issue.

[Kamil]

We've used Ricardo Wave for a couple of years now at UWA Motorsport. We've found that geometric parameters such as runner length, diameter, valve lift profiles etc are extremely important, as well as modelling of y juctions. I'd always use complex y junctions because Wave actually models them as 3D volumes. For restrictor modelling, we're sticking restrictors on the flow bench and then using Wave as a flow bench and correlating mass flow vs static pressure difference to calibrate our model. To properly tune our model, we compare things like pressure traces at different locations. Also, one we're going to be doing this year is vol efficiency. As far as i'm aware, your friction parameters do affect output torque higher (~9000rpm+) in the rev range.

Kamil
UWA Motorsport

[nathan s]

That said, can anyone tell me how to get the effective volume of a cylinder(make it spherical) to put into the y-junctions of a plenum?

That said, now for my opinion of the program...

Ricardo is the most touchy program I have ever used. I love it, when it wants to work. If it doesn't work, I hate it.

Also, a word of advice...To make entering specifications easier, make variables in the constants table for all of your lengths and diameters. That way, you can change specifications for all tubes at the same time. http://fsae.com/groupee_common/emoticons/icon_biggrin.gif

[Kamil]

Check out phase 4 of the SI tutorial in the Wave help file. It explains how to use complex y junctions for plenum volumes and how to set up DIAB and DELX values correctly.

Kamil
UWA Motorsport

[Chuck Dean]

I'm going to keep this thread rolling. This is exactly what I had in mind. This could end up being a knowledge-base for all of us Ricardo users.

Some more tips:

Full RPM sweeps: Doing complete runs with over 100 rpm points can be beneficial in more accurately displaying the ouput. However this should probably only be done once. Doing multiple runs like this creates a LOT of data. Not only does the simulation have to analyze over 100 cases, but Wavepost takes FOREVER to load this data up. So keep your number of RPM points low until you feel your model is where it needs to be.

Using Excel:
Excel can be used to input the various tables into Ricardo. It can be used to input your 'Constants Table'. So you can setup your constants table in excel and do interpolations between values to avoid using the tedious interface in WAVE. You will have to save it as a *.txt (tab delimited). This can be done for other table such as cam spec and flow coefficients.

**Question:
Has anyone ran into problems importing your cam specs. I had the cams defined based on opening/closing points and max lift and it worked fine. When I imported the actual cam specs (per every 1 degree), I started getting negative torque values!! The cam profiles show up just fine within the valve profile editor. Opening and closing points are just fine. Very wierd.

[Kamil]

To get a nice, continuous torque curve output, try using a transient sim. This is covered in the intermediate level tutorials. Its definitely quicker than running 100 cases, probably just as quick as 10 cases. The ouput file is small as well. The only drawback is that you don't have pressure, velocity data etc for steady state analysis.

Kamil
UWA Motorsport

[Garlic]

Originally posted by Chuck Dean:
**Question:
Has anyone ran into problems importing your cam specs. I had the cams defined based on opening/closing points and max lift and it worked fine. When I imported the actual cam specs (per every 1 degree), I started getting negative torque values!! The cam profiles show up just fine within the valve profile editor. Opening and closing points are just fine. Very wierd.

You have a timing problem. It's easy to do. It can be confusing to get the opening and closing points correct. Triple check them, and do some hand calcs to prove to yourself that everything is happening at the correct time.

It's true that an engine has negative torque when the valves are closed when they should be open, etc.

[Chuck Dean]

Trust me, it's been trippled check and then some. WAVE shows the correct valve profiles and valve phasing - but something screwy is going on within the program.

[Garlic]

Problems like that, are never the program. Believe me, 99% of WAVE users input cam profiles, and it works.

Post a screenshot of your cam profiles. You can blank out the values, if you like. I''m sure there's an issue in there.

[Charlie]

What version of WAVE are you running?

[ARCTurbo]

Has anyone modeled turbo engines? I am trying to play around with this now and am having some trouble with the input of compressor speed and control of the wastegate.

Any hints?

[jonnycowboy]

http://www.ricardo.com/download/pdf/wave_engine_simulation_wave.pdf

[jon109]

hi, first of all, i would like to congratualte the person that started this forum... it's really good to discuss things you are working on.

i am using ricardo wave for my thesis project. In a few words i am simulting an FSAE kawasaki engine so as to finally try to tune the valve timing etc... i am having problems to model the restrictor.

I've already done experiments to get the characteristics of the restrictor (i.e mass flow versus downstream pressure). but now i would like to use the data i have in my model. I also found the respective orifce area to allow the same flow at resopective downstream and upstream pressures. so here i derived a 2D interpolation map of orifcie areas for the respective upstream and downstream pressure. but unfortunately when i tried to implement it on wave, it didnt work.
can anyone help me to find a way to model the restrictor well.

note: i also tried to model it in wave by a nozzle and diffuser and did a flowbench test on it. apart of taking long times to solve, it is not as accurate since the diffuser modelling is quite hard to get in wave. i also tried to increase some pressure loss and friction coefficients so as to get the right flow characteristicts. Hopefully there should be something more professional to do this...

tnx in advance for any helphttp://fsae.com/groupee_common/emoticons/icon_smile.gif

[adam shortall]

Originally posted by jonnycowboy:
http://www.ricardo.com/download/pdf/wave_engine_simulation_wave.pdf

This is a great document, however I can't see the screen shots, can you post a higher resolution version, or send it to me at adam.shortall@student.manchester.ac.uk

[Bazanaius]

Hi guys,

I wonder if anyone can help, I'm having a bit of trouble setting up WAVE.

Essentially, there are only two people likely to use it this year - me and a housemate, and so I am setting it up with my PC as the license server, also running a copy of WAVE on there. I'll worry about my housemate's copy when we've got one working, but I'm having trouble getting it to see the license.

I followed the instructions that came with the license server and got no errors. The Ricardo Licenses manager service is running ok, and I have installed WAVE. In the license server manual it says that I might have to change the ricardo_license_file environment variable to get it to work, however I can't find how to do this, and the manual also says that if I've followed the instructions exactly it should find the license automatically!

When I start WAVE, it tells me it can't contact the license server, and shuts the program.

I wonder if anyone can talk me through the stages from the end of license manager install to how to install WAVE correctly, on the license manager machine. Or if anyone has any ideas on where I might be going wrong?

thanks in advance,

Baz

[Charlie]

Your Ricardo rep is your best bet. Contact Ricardo directly- they should be able to talk you through it.

I can't say I've ever had any issues with the Ricardo license.

[Nishant Jain]

Hey guys,

We're just about to get Ricardo WAVE installed on our systems in a few days.

For preparation before using the program, can someone refer me to the kind of books I should read to help me in specific aspects of engine modeling or simulation on Ricardo?

I mean, aside from the usual engine textbooks (I'm reading Stone) and the manual of the F4i (which we're using), should I look out for anything in particular?

[Mikey Antonakakis]

I'm not sure how helpful it would be, but Design and Simulation of Four Stroke Engines (forget the author at the moment) is one book on the subject.

[Exhausted]

The author of that book is Gordon P. Blair and it is an excellent resource for engine simulation no matter what code you're using. He wrote the code that Optimum-Power's Virtual Engines/Automated Design is based on.

I also know of a user forum based on Virtual Engines/Automated Design -- FourStrokeDesign.com (http://fourstrokedesign.com/forums/index.php)

[Nishant Jain]

Thanks guys, I've heard of that book too. Actually seen it mentioned in Stone. Was thinking about purchasing it, and now I'll definitely go ahead and do it.

Thanks a lot.

Aren't there any specific publications or tutorials on Ricardo anywhere? I mean, most softwares come with tutorial/introduction sort of books too, dont they?

[Bazanaius]

Ricardo does have tutorials, which will at least get you a 4 cylinder engine running very quickly. You then have the job of making it actually represent your system accurately.

The tutorials can be found in help --> tutorials.

[Mikey Antonakakis]

More specifically, open any WAVE program, click help>WAVE Knowledge Center>Tutorials
They're pretty good.

[Nishant Jain]

Haven't gotten access to Ricardo just yet. Waiting for our legal documents to arrive.

I have run into a query while (theoretically) planning our engine modeling approach in Ricardo. No doubt it would be cleared as soon as the software and help files arrive, but I'm curious to know asap.

So, can anyone tell me how map values of the ECU are input to the Wave program for engine model and simulation? And if these numbers are input somehow, can't they be also easily altered and the simulation of the consequences seen? That would make Ricardo useful for ECU modeling too, in a way?

This is mostly just stuff in my head, so please excuse any silly lapses.

[samphlett]

WAVE has a wide variety of actuators and sensors that can be used to measure/modify most things as the model runs. There are some fairly simple control system modelling elements in WAVE itself. It can also be hooked up to Simulink (and someother programs), where the other program gets to play with the actuators and sensors.

In this way, ECU modelling (control strategy development to be more precise) is a common use, with WAVE as the high fidelity plant model.

[Garlic]

http://www.ricardo.com/enginee...wave_availablepapers (http://www.ricardo.com/engineeringservices/software.aspx?page=wave_availablepapers)

[Nishant Jain]

@samphlett: Thanks. Our electrical guys will be glad to hear of that.

@Garlic: I had downloaded a few of those papers before, but not with reference to ECU modeling. Thanks, that was a useful reminder!

[Tincho]

Hello, i was wondering if any of you could explain me a couple of things.
i'm trying to make the SI Wiebe Combustion Model to work, the thing is i need the 50% Burn Point and the combustion duration to be dependable on the ignition advance. I read that that cold be achieved with an actuator, but i don't know how to use those.

i would apreciate the help.

[BB]

hi guys.
i have use WAVE with a license serve at a adress.
but yesterday i can't use it any more.may be the license serve was changed.
if any one have the new adress,please sent it to me .
my email is lb15926@126.com.
thanks

[Hector]

You need to contact Ricardo.

[atul.gupta]

Hi everyone,

This happens to be my first post on this forum,
I have previously used AVL Boost, and now using Ricardo Wave. Over the course of using Wave, I have done the troubleshooting of a lot of things, and now can do most of it within no time.

Recently I have developed a 1.4 liter Gasoline direct Injection with Turbocharger (with PID) and VVT.

If you have any questions or queries to get sorted, I might be able to help if I have come across the thing.

Regards,
Atul

[Jon Burford]

Recently I have developed a 1.4 liter Gasoline direct Injection with Turbocharger (with PID) and VVT.



Hi Atul, I am interested in hearing about your engine, do you have any pictures of it, how is it running?

[atul.gupta]

Not physical. I meant on Ricardo Wave. I probably should have rephrased my term "developed". Sorry about that.

I am MSc. Automotive Engineering student at Coventry University. I also carry 2 years professional experience in Engine R&D before that.

Right now I am writing my dissertation, so using Ricardo Wave.

Regards,
Atul Gupta

[Jon Burford]

I see,
I would be interested in reading your dissertation when it's complete. Are you looking to produce an engine from your work?

[TerpsWP]

Atul,
I'm new to using Wave and us at terps racing have been trying to more acurately model our engine to produce more reliable data for proposed changes in design. Until now we have been using the SI_Wiebe combustion profile that we used during the tutorials for the program, but we are trying to understand how this profile works and how we would go about creating a new combustion profile for our engine. what are the purpose of the 50% burn point and the combustion duration, and how would one come across these figures for any given engine? from what I've found another way to get a combustion profile is to enter a Pressure Profile, and indicate at what carnk position the ingniton starts, and Wave will calculate this profile for you. I'm trying to figure out if it would be possible to obtain this data, and what the benefits of doing so would be, any information would be helpful


Walter

[atul.gupta]

Hi Walter,

That is a brilliant post, will help a lot of users to know about it. Fortunately I have some experience on it.

The SI_Wiebe is a very simplified combustion model, nonetheless works very well for what most users require out of WAVE.

The 50% burn point specifies the crank angle position when 50% of the fuel mass has burned. When you provide this data to WAVE, it calculates the combustion start point.
Now, as per my knowledge, 50% burn point cannot be measured on an actual engine.

In actual engine, P-Theta (Pressure curve) can be measured rather easily by a pressure sensor in the cylinder head. This P-theta curve can be fed to WAVE, wherein you provide the Combustion start point and the duration of combustion.
Feeding the P-theta curve or Rate of Heat Release curve to WAVE is much more accurate than the 50% burn point method as approximations are eliminated.

I have tried this in AVL Boost as well, and I find Pressure curve or Rate of heat release method more accurate.

Hope it helps!

Atul

[TerpsWP]

Atul,
thank you so much for your input, do you have any idea where in the head we should be putting a pressure sensor? or possibly more than one? we have been doing alot of instrumentation this year and might be able to create a pressure profile for the engine and create a better combustion profile for our engine, again, any input is very helpful,
Walter

[atul.gupta]

Hi Walter,

Well you will have to drill a hole right in the cylinder head so that the pressure sensor can be placed in it snugly. There would not be a much space to put the sensor if its a 4 Valve system, and/or multiple spark plug system and/ or Direct Injection system.

Here are the sensors you can use:
https://www.avl.com/pressure-s...-combustion-analysis (https://www.avl.com/pressure-sensors-for-combustion-analysis)

and this is a good (although length, but good) documentation on in cylinder pressure measurement.
http://www.sussex.ac.uk/Users/...inderMeasurement.pdf (http://www.sussex.ac.uk/Users/tafb8/eti/eti_17_InCylinderMeasurement.pdf)

Regards,
Atul

[TerpsWP]

Atul,
I was under the impression that this was a pressure curve as a function of crank angle w/o a running engine, as in I would simply put a pressure sensor in place of the spark plug and then turn the motor over by hand, from what you've said this seems like I would need to run the motor with a pressure sensor in the head to make a profile of a running motor as combustion occurs, is that so?

[Kirk Feldkamp]

IIRC, there are a number of types of in-cylinder combustion analysis pressure sensors. Kistler (http://www.kistler.com/mediaaccess/100-460e-12.10.pdf) sells some too. Guaranteed not to be inexpensive, but I'm sure worth every penny.

I was under the impression you use these on an operational engine. Then you get to see all sorts of fun stuff happening as you change ignition timing and such. I remember a few years back there were also some aftermarket/racing companies selling plug gap ionization sensing equipment. I don't know if those are still around, but from what I was reading they sounded pretty interesting, and possibly more practical from an in-car tuning standpoint. I guess it all depends on how much of a science project you're looking for versus a practical/fast development approach. Usually, tuning to minimum best torque (MBT) across the load/RPM range is sufficient for a racing application. *MOST* "professional" race cars get much less attention than that! Then again, if you guys are using this for sims, then you probably already know what you're looking for.

-Kirk

[Simon Dingle]

Hi Terps,

Yes, you would need to get the in-cylinder pressure curve while the engine is running.

In order to get your real in-cylinder pressure curve, you will need to use an in-cylinder pressure sensor as advised in the previous posts. Kistler and AVL both sell spark plugs with pressure sensors built in to them and this is by far the best option for a production engine as you don't need to modify the head. This will set you back around $5000 for the sensor and charge amplifier. On top of that you will need a digital shaft encoder (at around $400) and some high speed data logging equipment (probably at least $1000). If you are looking into doing this, it will not be cheap!

However, if you just need to know the 50% mass fraction burned point, MBT is at around 8-10* aTDC for every engine I've seen. And then rettardss roughly linearly with spark retard (although it's not exact). I should probably know why 8-10 is the 'magic' point, but I don't. Maybe that'll be my task for today.

EDITED because I didn't read things properly the first time.

[atul.gupta]

Perfect answer by Simon!

Walter, this is why engineers use simplified combustion models on simulation softwares. Although an approximation, they give out fairly accurate results (reasonable limits).

What else you might consider doing is that if you have the actual engine performance dataset from the test bed, you can compare it with the simulation and adjust the input parameters. Start by matching the air flow, then the fuelling, and then knock intensity.

Simon's task for today, became my task too now. Can't stop myself thinking about it!

Kirk: MBT (Maximum Braking Torque Point), although means the same what you intended to, which is fine as well.

Regards
Atul

[Owen Thomas]

However, if you just need to know the 50% mass fraction burned point, MBT is at around 8-10* aTDC for every engine I've seen.
Simon,

Are you saying that the 50% burn point occurs at 8-10* after TDC while the engine is operating at MBT timing? I'm just a little confused by the wording... If this is the case, I am both surprised and astounded; especially since MBT seems to range 20* or more through an RPM sweep. I'm guessing it has more to do with piston velocity vs flame propagation..?

Regardless, it never occurred to me to think of engine performance in this way (which probably makes me a n00b). If you don't mind sharing, I think lots of us would like to see the results of your "magic number" investigation.

[atul.gupta]

Owen,

There seems to be a confusion of information.
I guess what Simon intended to say was 50% burn point occurs around 8-10 degrees ATDC, which leads to Maximum Brake Torque, which as you correctly said is around 20 degrees ATDC.

It wasn't meant to be said that 50% Burn point = Maximum Braking Torque point.

Both of you are correct.

Regards,
Atul

[Kirk Feldkamp]

Originally posted by Simon Dingle:
...MBT is at around 8-10* aTDC for every engine I've seen. And then rettardss roughly linearly with spark retard (although it's not exact).
Originally posted by atul.gupta:
Kirk: MBT (Maximum Braking Torque Point), although means the same what you intended to, which is fine as well.

Yeah, I guess I'm using "tuner" parlance rather than research/simulation parlance. Sorry. I haven't ever had the opportunity to play around with serious simulation or research equipment, so all of my experience is in the practical tuning realm. Minimum Best Timing (oops, said Torque before), as I intended to use it, simply means the minimum ignition advance required to make maximum torque at a given operating point. This will almost certainly correspond to Simon's 8-10deg aTDC ideal *maximum cylinder pressure* point. It's approaching the same thing from two different sides. You guys are talking about the result, I was just talking about how to actually get there. As you said Owen, the actual advance required to do this can vary a large amount across the operating range of the engine. I also assume this is due to piston velocity and flame propagation speed.

-Kirk

[Simon Dingle]

Firstly, I'm looking more into what I said as it does now seem odd to me, but I've never questioned it until now. I'll reply (to myself i suppose!) when I've got everything clearer in my own head.

Secondly maybe I should clarify on what my abreviations meant.

CA50 = 50% mass fraction burned point = the crank ancgle at which 50% of the fuel mass is burned and 50% is unburned.

MBT = Maximum Brake Torque. This is probably the most loosely and poorly used accronym in engine development and it seems that everyone has their own definition of it, this results in no-one really knowing what anyone else is saying. I meant it to be an engine operating condition. MBT being when the engine is producing the maximum torque for a given fuel and air flow rate (fixed throttle and fixed fuel timing).

Atul, what do you mean when you say MBT?

---------------------------------------------

Owen,

To answer your question, yes I meant that the 50% burn point occurs at 8-10° after TDC while the engine is operating at MBT timing. (If by "MBT timing", you mean the spark timing that leads to MBT). I will say that I have to widen that range to at least 7-12°aTDC after searching for this in several papers today.

[Simon Dingle]

Hi All,

So I've found a relevant paper:

http://www.vehicular.isy.liu.s...hD/99_PhD_580_LE.pdf (http://www.vehicular.isy.liu.se/en/Publications/PhD/99_PhD_580_LE.pdf)

Don't worry about the length, the pdf is a collection of papers and the one we're interested in is Publication 7. More specifically, the part that relates to what we're discussing is section 6.1, on page 156 (that's page 180 in the pdf).


Figure 19 shows many simulated heat release curves (these are converted to mass fraction burned curves by normalising all curves against the maximum heat recorded release value). They are all optimally phased, that is, they have all had their spark timing set to MBT,but they all have very different combustion development parameters. It can be seen from this graph that despite the various different burn rates, all of the lines pass through a common point. Figure 22 (page 160) shows the same graph but with the curves converted to Mass Fraction Burned and zoomed in around MFB=50. This clearly shows that for a wide variety of flame developments and burn rates, the 50% Mass Fraction Burned point is very similar and ranges from 9.5 to 12.

Owen, Figure 19 sheds a bit of light on why a large range of spark timings (the start of combustion) might be needed to maintain optimum combustion phasing, while the 50% Mass Fraction Burned point remains relatively unchanged.

Reading though some of the earlier section in the paper reference some work by Bargende, who apparently supports this theory empirically. But this is one for the German speakers, unless I can find an English translation somewhere.

Michael Bargende. Schwerpunkt-kriterium und automatische klingelerken-
nung. Motor Technische Zeitschrift, Vol. 56(10):632–638, 1995.

Section 6.3 might shed a bit of light on engine to engine variation. Figure 27 (page 165) shows how the position of 45% Mass Fraction Burned point changes with the heat transfer coefficient in his simulation model. It shows that as heat transfer increases, the 45% (and therefore also the 50%) Mass Fraction burned point retards away from TDC. So engine to engine variation in the optimum 50% Mass Fraction Burned seems most likely to be due to the heat transfer between the combustion and the block/head/piston. Which is primarily governed by parameters such as the coolant jacket design, engine material and cylinder flow regime (swirl, tumble, reverse tumble etc.) While the author says that the heat transfer coefficient in the model is typically around 1, he doesn't comment on how much it ranges by.

The paper doesn't discuss the reasons for while the narrow range of 50% MFB points for MBT timed combustion is where it is (around 10* aTDC), but the more I think about it, the more I think that it might just be a product of the crank/con-rod geometry.

Anyway, I hope that helps. Feel free to chime in with comments etc.

Cheers,

[Owen Thomas]

Thanks Simon! I think we all appreciate the effort you put in to that post.

In retrospect, it seems obvious now that the "optimal spark advance" (which I think I'll use instead of MBT from now on :P) is the means to maintain the correct 50% MBF timing.

I suspect that going even further into detail, the 50% MBF benchmark is simply a way of getting close to another, more relevant target. Taking a wild guess, I'm thinking it has to do with getting the maximum burn rate to happen at or near peak pressure in the cylinder.

There always seems to be more detail, no matter how close you look... Sure can make you feel like a dummy!

[atul.gupta]

Brilliant posts!
The document is really informative.

Simon, yes I meant Maximum Brake Torque (point) by MBT.

Now guys, lets come back to Ricardo WAVE. Walter, I hope all the information helps programming the WAVE code.

[madrules]

i wanna use ricardo wave for manifold designs, where do i get ricardo from and is it good?

[Jay Lawrence]

Space.

Maybe.