SV Portal Forums banner

Predictive modeling of fuel and timing maps for optimal performance

11K views 44 replies 14 participants last post by  schmidt314 
#1 ·
Have you ever looked at or analyzed some of the maps that either DynoJet gives you as defaults or from a custom tune? I literally have studied hundreds of maps and am confused by them for several reasons. Some maps will go from a -15 to a +40 in just a single RPM interval? I have had a hard time believing this can be optimal settings for a digital fuel injection system. Being a person of numbers and always trying to understand things in a more pure mathematical sense, I figured I would see if I could build a mathematical model that would optimize engine performance with minimal dyno runs. With the help from some sophisticated software I have access to, I have found some really interesting stuff and thought I should share it. This is solid data from my first full tests that were only run at 100% throttle to make sure the model will work. It can then be easily be extrapolated into all throttle positions and RPMs quite easily. This work was done with fuel injection mapping and timing mapping via PCIII usb with a timing box.


The Bike
04 SV1000 (My 32000 mile commuter machine)
Freshly serviced and tuned
Freshly cleaned and oiled BMC air filter
Snorkel in place (Yes, I left it in!)
Fresh oil change and filter with Motorex
M4 full exhaust system
87 octane pump fuel
PCIII usb and timing box

Data collection
The software I am using, with a few input parameters spits out a small series of fuel and timing maps to be run on the dyno from which torque data will collected. These runs include redundancy for prediction of errors including inconsistencies in dyno measurements. The only data needed is the torque values, no AFR requirement because 13.2:1 does not always give best power under certain conditions. If you can tune to give best torque at every interval, who cares what the AFR is. With all the data collected from the dyno runs it is then fitted to a 5th order polynomial equation and can show the relation and interaction between the RPM, throttle position, fuel change and timing. With this data, you can then predict optimal timing and fueling parameters for every position in the fuel and timing maps. Since the software understands the interaction between all the parameters it can often find optimal settings that are far better than the methods of -find best fuel, then tweek timing. The adding of parameters does make it more complicated and time consuming but, it will still cut the total dyno time down and give significantly better results. A full system fuel map (all RPMs and throttle positions) w/o timing can be done in 16-32 dyno runs. You just switch maps quickly, make sure bike is at target temp and make a pull, then repeat. So, after simulating data for quite some time to work out a few kinks, I rolled up the bike on the dyno and had at it. Thanks to my good friend Arlan at LED Performance who lets me use his dyno when needed. Here are my results, I was quite shocked. These numbers do not tell the whole story, but the included graphs do.

-0 Fuel map and 0 timing map
Peak torque 67.5 ft lbs
Average torque 62.2 ft lbs (4500-10000RPM)
Average torque 61.7 ft lbs (4500-7000RPM)
Average torque 63.0 ft lbs (7000-10000RPM)
Peak HP 107.8
Average HP 85.8 (4500-10000RPM)
Average HP 67.6 (4500-7000RPM)
Average HP 101.4 (7000-10000RPM)

-Optimal fuel map and 0 timing map
Peak torque 68.0 ft lbs
Average torque 62.9 ft lbs (4500-10000RPM)
Average torque 62.3 ft lbs (4500-7000RPM)
Average torque 63.8 ft lbs (7000-10000RPM)
Peak HP 109.7
Average HP 86.7 (4500-10000RPM)
Average HP 68.2 (4500-7000RPM)
Average HP 102.6 (7000-10000RPM)

-Optimal fuel map and optimal timing map
Peak torque 68.3 ft lbs
Average torque 63.3 ft lbs (4500-10000RPM)
Average torque 62.9 ft lbs (4500-7000RPM)
Average torque 64.0 ft lbs (7000-10000RPM)
Peak HP 110.0
Average HP 87.3 (4500-10000RPM)
Average HP 68.9 (4500-7000RPM)
Average HP 102.9 (7000-10000RPM)

By displaying the numbers this way I think it displays my philosophy about what makes a bike fast. Peak numbers are often held up as the goal. Pull out the calculus books though and integrate under the torque curves. Two bikes that have the same peak numbers can be wildly different when you get down to real performance. Look for large average numbers, not peak and you will get a winner…
 

Attachments

See less See more
1
#4 ·
I am debating whether to share some map data or not, I might be will to share 100% throttle position only, it is the rest of the map that makes the bike a killer bike to ride every where though. A bit of other work involving this, is also showing me an interesting relation between modifications. The fuel curves required to optimize performance might not be that different from bike to bike even with just the common mods done. My full tilt race bike is undergoing a change in cams, and head configuration so I am interested to see where this work puts my curves for comparison sake.

As for timings effect on HP, that is included in the graph I attached in the original post.

-ms
 
#5 ·
Sorry I fixed the original post, cut and paste errors...
(4500-10000RPM) full sweep
(4500-7000RPM) bottom end sweep
(7000-1000RPM) top end sweep
I hope that this helps

-ms
 
#6 ·
I also forgot to add just s a seat of the pants ride report. The improvement was drastic when riding it on the street. When you pin the throttle, it has never jumped as hard or been as smooth. It actually had a rush and a very hurried feel to it like it never has had before.

Just remember, this is my data, it might be right or it might be wrong. I am still working on the modeling to refine it a bit more but it appears to have worked in a couple use cases so far.

-MS
 
#9 ·
Is that the zero map modifications schmidt? I'll try it. I liked the Micron Map on the website.

Just like I thought, lean on the bottom and rich up top. I fiddled with my PC3 so much out on the road, it drove me nuts.

I just don't use 100% throttle that much. It would be good for a drag bike though.

This is the Dynojet stocker 100% TP map on the website:
100%

0 - 500 rpm
0
0
0
0
-13 - 1750 rpm
-7
2
5
5
1
10
16
-2
1
6
5
13
4
7
10
2
-5
7
1
-1
2
3
5
1
2
2
-2
3
2
-1
1
-2
-3 - 10000 rpm
0
0
0
0 - 11000 rpm
 
#10 ·
Is that the zero map modifications schmidt? I'll try it. I liked the Micron Map on the website.
The map I showed is the delta from the zero map, so just plug those numbers straight in. I am attaching a couple of other graphs too. One shows the 100TP maps for many of the DJ default maps, and the other is just some straight timing runs. This will show some of the effects of timing, but there is a relation between timing and fuel that this will not show. I have some data, I am still crunching showing the differences between setting the fuel map first, then making timing on top of that. It is showing that they need to be done together to get real optimal tuning. The map I gave you here is the optimal fuel map my program is giving me when running at stock timing figures. The optimal fuel map changes when timing is adjusted. The map I posted was generated in 8 pre defined dyno runs with test maps. The more runs I make does not change the map it spits out for me, it just makes the error bars smaller and smaller which is what I was hoping for.

The map it gave me looks so much different from the other DJ maps I am still a bit skeptical, but the bike runs better and makes more/power and torque in the whole range. The DJ maps are just saw tooth like and that just bothers me, I like smooth curves...

-MS
 

Attachments

#13 ·
:bigclap: Very interesting...

I don't think mapping points that are far apart should have much influence to one another - expect maybe in low gears where these points are still pretty close time-wise and we could see some 'left-over' fuel particles in the intake tract?

I just purchased a PC-III for my track bike (a GSX-R750 K5) and will invest in dyno time, so if you need testers and more data, let me know if I can help... ;)


D.
 
#15 ·
Thanks for the pdf document to go along with this great thread. I get giddy when I think of how this type of fuel and ignition timing mapping can really bring an SV to life. Pretty much convinces me that a PCIII and PC ignition module are worth it.
 
#19 ·
Please don't take this post as an insult ,or let it dissuade you from what you are doing. I wouldn't even know where to begin to design a program like this , and it is awesome that you know how to do something like that AND are willing to share it.

That being said, here are some things that came to my attention when I read this :

1.)Page 7 of the PDF "One point of interest is when the new map reached a point of about 0% fuel in the map (around 8500 RPM) it would seem that the engine outputs should nearly match the 0 map runs at this point. They do not."

While the predicted map is at 0% fuel , the DJ map for the stock bike is at about -7% or so. This is a considerable difference , one that can easily explain the difference in engine output. Am I missing something here ?


2.)On page 6 of the PDF, look at the difference in percentages between the predicted map and the DJ map fueling between the band of 6k and 7k rpm. I am looking at about a 10% difference in fueling percentages.

Looking at the torque graph on page 6 and the HP graph on page 7, the corresponding band of 6k to 7k rpm shows NO discernable difference.


3.)On pages 8 and 9 , the torque graph shows a pretty good increase and the HP graph shows a minor increase in the 6k to 7k band, but this is with the timing map in conjunction with the custom fuel map.

In your post ( not in the PDF) , you showed a graph of JUST the timing with a "0" map , and it had similar (if not the same) increases in the 6k-7k rpm band.

It seems as if the addition of 10% more fuel in this band is not contributing to the power output at all , but that the gains between 6k and 7k rpm are attributable to the timing increase.






I understand that the fuel percentages and the graphs can't tell me what it is like to ride the bike , but it seems like a waste of fueling throughout that band. Perhaps it has better throttle response or something that I can't see in the chart.

I believe the reason the DJ maps are not "smooth" is because the flow of the engine system (airbox, TB's , heads, combustion chambers, exhaust) is NOT linear, meaning it flows differently as it responds to different engine pulses/rpm/frequency. So at some rpms the exhaust might be resonant ( flowing well ), but the airbox might not ( perhaps has reversion ), while at some other rpm, their resonances are similar , so there is better overall flow at that point.


As an example, when you are synching the TB's , you may be able to get them VERY close at a certain rpm, but when you raise the rpm, you will see the balance shift to one cylinder. Raise it some more and it may level out, or might even shift to favour the other cylinder.

Perhaps if the heads have been ported and tuned on a flow bench this would be less of an issue. Then again , equal length exhaust would probably make a difference as well.

I like smooth curves , too. Sometimes they aren't very practical.


I do not mean this as an attack , in any way , shape or form. This is just my point of view based on my limited knowledge. If I am not understanding something here, PLEASE let me know.


..........................Blake
 
#20 ·
<font color="yellow">I am not saying this method is perfect, but I feel it is quite powerful and I am continually trying to collect data to make it better.</font>


1.)Page 7 of the PDF "One point of interest is when the new map reached a point of about 0% fuel in the map (around 8500 RPM) it would seem that the engine outputs should nearly match the 0 map runs at this point. They do not."
While the predicted map is at 0% fuel , the DJ map for the stock bike is at about -7% or so. This is a considerable difference , one that can easily explain the difference in engine output. Am I missing something here ?


<font color="yellow">When the zero fuel map is compared to the new runs you might think that there should be no difference at about 8500rpm because the mapping is the same. This is not a comparison of the DJ map to the predicted model map, it was a comparison of the zero (empty DJ map) and the one the software spit out. This shows that there is another not so obvious relationship in the mapping parameters. I believe it shows that the mapping points are not mutually exclusive, and could have a dependence on the other values surrounding it in the map. This means that changing fueling at 100% throttle position 5000RPM may effect the mapping position at 100% throttle position at 7000rpm.</font>


2.)On page 6 of the PDF, look at the difference in percentages between the predicted map and the DJ map fueling between the band of 6k and 7k rpm. I am looking at about a 10% difference in fueling percentages.

looking at the torque graph on page 6 and the HP graph on page 7, the corresponding band of 6k to 7k rpm shows NO discernable difference.


<font color="yellow">There is a difference but with the scaling of the graph it is difficult to see. I just looked at the raw data. There is actually an average increase over the 6000-7000RPM of just under 0.5ft-lbs</font>



3.)On pages 8 and 9 , the torque graph shows a pretty good increase and the HP graph shows a minor increase in the 6k to 7k band, but this is with the timing map in conjunction with the custom fuel map.

In your post ( not in the PDF) , you showed a graph of JUST the timing with a "0" map , and it had similar (if not the same) increases in the 6k-7k rpm band.

It seems as if the addition of 10% more fuel in this band is not contributing to the power output at all , but that the gains between 6k and 7k rpm are attributable to the timing increase.

I understand that the fuel percentages and the graphs can't tell me what it is like to ride the bike , but it seems like a waste of fueling throughout that band. Perhaps it has better throttle response or something that I can't see in the chart.

<font color="yellow">from the previous answer, there were small gains through this RPM range, and then again a bunch more in timing. I do agree that timing does have more effect here, but the increase in fuel does make a statistical difference in engine output.</font>


I believe the reason the DJ maps are not "smooth" is because the flow of the engine system (airbox, TB's , heads, combustion chambers, exhaust) is NOT linear, meaning it flows differently as it responds to different engine pulses/rpm/frequency. So at some rpms the exhaust might be resonant ( flowing well ), but the airbox might not ( perhaps has reversion ), while at some other rpm, their resonances are similar , so there is better overall flow at that point.

As an example, when you are synching the TB's , you may be able to get them VERY close at a certain rpm, but when you raise the rpm, you will see the balance shift to one cylinder. Raise it some more and it may level out, or might even shift to favour the other cylinder.

Perhaps if the heads have been ported and tuned on a flow bench this would be less of an issue. Then again , equal length exhaust would probably make a difference as well.

I like smooth curves , too. Sometimes they aren't very practical.

<font color="yellow">I do agree that resonance frequency in the system can play a role, but I do not see any convincing evidence that this is the case for the saw toothed nature of the DJ maps. I have also been comparing some maps from street and race bikes that were set up using the Bazzaz auto mapping. When I graph those values from several bikes I see relatively smooth curves that tend to fit the modeling data. More investigation is needed, but the auto mapping data does appear much smoother. All the bikes these maps were taken from, we tuned either on the street or on the track, and maybe this could be factor too.</font>


I do not mean this as an attack , in any way , shape or form. This is just my point of view based on my limited knowledge. If I am not understanding something here, PLEASE let me know.

<font color="yellow">No prob, I have written many other papers for technical journals (Non motorcycle related stuff) and they tend to go through a pretty rough refereeing process. I really am happy you took the time to read it, think about it, and ask good questions.

-MS
</font>
 
#22 ·
The data I do have does show that fuel and ignition timing are interacting with each other which is what I would expect. I have not run the series you mention.

In the actual DOE space I have run, I always put ignition timing into the mix right away. Leaving it out and doing later might not produce absolute peak results and can actually create even more work. The work here was done serially to prove that I could get a sold response from both variables independently prior to building a more complex data collection scheme.

I have not used this model much recently. It seems to work pretty well in the cases in which I have used it after this write up. The advent of auto tune stuff so easy to use now kind of kills some of this. But what my model will do is ignore AFR and just tune for best power. I don't care if the AFR reads 11.2 or 15.1 If it makes best power at a measured 13.7 then so be it.

I did this to just get my geek on mainly, but it has spurred some great conversations with several folk over the years.

Thanks for reading!
-MS
 
#24 ·
Hi Mike. Just read your awesome writeup on predictive efi mapping. Do you have some timing map numbers you would like to share? Next time I hit the dyno I will do a run with woolich autotune mapping and one with predicted map.

Kristian.


I don’t hand that data out.

Just note that at the higher throttle positions it likes several degrees of advance in the lower rpm, but as it goes up in rpm, you start to pull timing out, and actually subtract out a fair bit up on top.

Very sensitive to engine work though.

-ms
 
#25 ·
Hi Mike.
I am going to hit the dyno tomorrow to see if my latest engine mods yields more bhp.
Mod list:
  • DL 100mm cylinders 1mm squish
  • Light ported heads
  • Pipercross air filter
  • 05 cams, TBs, snorkel, wiring, ECU
  • 106/108 cam timing
  • custom 2-1 exhaust
  • Pair cranckcase vacuum mod
  • STV removed

Bike is AFR autotuned with Woolich software. I use RecoilRobs Ignition map (thanks Rob) where low rpm is advanced and top retarded.
I will do a run with AFR tuned fuelmap and one with one with your predicted map (100% tp).
I am really torn where to leave top end timing. Right now its at 36 degrees (stock is 44) for 100% tp
 
#26 ·
With that engine config, I would not use that map. The engine it was developed on was very different from yours.

36 degrees might not be enough timing, but the dyno will let you know. On these bikes when the timing is set, 100% TP at high rpm, it won’t fall off near as fast post peak power.

Your config should run really hard and should make broad power across the range.

What kind of compression ratio are you running?

-ms
 
#37 ·
I emailed James from JHS racing about cams upgrade. He said their cams typically add 8bhp all the way up rpm scale. Looking at dyno sheets from other users with hot cams, it seems gains are most in the mid-top end. Is there any chance of loosing low end power by going cam upgrade?
I cant make up my mind if I want to sell my SV and keep my newly purchased 01 busa or sell busa and keep SV and beef up the motor. The busa is a nice and fast bike, SV feels ligther and much more easy in corners. They seem to have similar power under 5k rpm where i spend most of my time..
If I keep the SV i would properly go JE 13:1 pistons and cam upgrade and hopefully end in the 130ish bhp with even better low end power.
 
#38 ·
Everyone is a little different but for me the twin just hits a little different from the i4, even if it's not as powerful. It would be neat to have an extra 40 horsepower, but realistically they would be called on very infrequently. I would prefer to have a snappy twin with lots of grunt in the center third of the tach. I'm too addicted to upshifts with the quick shifter to wait for redline.

While I am just starting to ride my lightly built motor I am very happy so far. As in rode it for the first time yesterday after reassembly to bed in the new rings.

Even using the stock exhaust system (so my assertive break in rides drew less attention) it feels much stronger right to the top end than before. Looking forward to swapping back to the 2-1 M4 and getting it dialed on a dyno to finish it off. While I only have half a tank of 91 burned so far, but it does seem to run hard and happy on the 91.

Swapping to the JE pistons in this process was relatively easy. The wrist pin clips are the hardest part, they are very tough compared to the stock ones and like to bounce off of the top of your head and disappear if your hand slips. If you end up doing it let me know and I can mail you a spare or two from the second set I ended up getting.... My recommendation is to use one of the stock ones initially to hold the wrist pin, and then lever the JE clip in with the wrist pin installed for it to back up against. Then pull the stock clip back out, install the piston and do the second JE clip.

Not sure how the difference of an aftermarket cam balances with other head upgrades, like 40mm valves. Down the road I would like to work these heads for the 40mm valves, or fill and re-port a set of TL heads, which is part of why I wanted the pistons with larger valve relief already installed. Also I wanted to refresh the rings and Suzuki wants enough for stock rings to pay for basically half of the JE piston set.

Looking forward to seeing and comparing what the curve looks like for this motor after it is tuned, since I think other than the compression bump it is similar to yours.
 
#39 ·
Jumjum I don't know how the speed limits are in your country but in mine they are becoming more restrictive, so I would choose the lightest bike with the best low range, when you can take advantage of the high power you will already be very out of the law. I envy your work, I have very few hours to work on my motorcycle and I am very slow, but I progress little by little. Now I am focused on the adaptation of the TB Sv in my DL 06, and I have already started and tested it, it works!
 
#41 ·
I was in the same quandary a while back once the first round of engine mods were finished....what to do next?(if there is a 'next') I came up with some basic rules: Increase power and increase mileage...OK. Increase mileage without hurting power...OK. Increase power while decreasing mileage...no good. Increase power or mileage without decreasing durability...OK. No decrease in durability is permissible so we've got to be careful here. This put some real limits on what can be done and the two things that are available that can obey my rules are cams and larger intake valves.

The SV heads are superior to TL everywhere except up near redline where they're about equal...SV ports are much better and even though the intake valves are smaller they flow enough for the job at hand most of the time. We have a saying "not quite enough is always better than way too much" and the 36mm intakes don't hurt us badly until near redline and adding the larger 40's would free up some midrange to top end power even with the stock cams. As long as the ports aren't modded and just the bowls blended for the larger valves this shouldn't badly affect the low and midrange and is where I'm going if the heads ever come off again. Larger valves will extend the power band and to take advantage of them or hotter cams would also require an increase in rpm allowance over the stock 10,600 rev limit. Revving higher while holding torque better will make lots more horsies available in your corral. I don't exercise mine fully very often and not often enough to warrant a tear-down to make the changes....but if it comes apart for something else THEN we'll do it. :)

Bigger cams can get tricky....increased lift without altering durations will make more power without hurting the bottom end much at all...but will increase the stresses on the valve train which might affect the durability. Stock springs and retainers likely wouldn't take kindly to lots more lift and the farther you move away from stock the more engineering you're going to be needing to do. Increasing duration will help the top end power...but will need increases in compression or the bottom end and midrange will suffer and for a street engine this isn't good for me. I DO use the torque we have on every ride...just don't wind it to the redline very often so midrange punch is what I'm after mostly and if we can do that and gain some top end too...all the better.

Don't overlook the Wiseco TLR slugs as they're also cut for 40's and give about 12.5:1 with 1mm squish and relatively stock head CC's and .030 head gasket. They work fine with stock cams though I always run our best 93 octane (R+M/2) while the JE's will have more compression I don't know where the limit of pump gas suitability lies. If you go beyond pistons and retimed stock cams the Woolich Racing tuning kit and its' offering of higher redline makes more sense. Suzuki was playing some games with the tach while hedging their bets on the motor durability, but if it's 100K mile durable at 10,600 then with a lighter piston it 'should' be good at an honest 11,000 or slightly higher but going way higher is likely to let you see the insides of your motor from the outside.

I've been looking for another bike for a few years now and haven't found one that offers as nice a package as the SV1K...especially at an affordable price point. There are many bikes that are faster, but in the Real World this additional speed is nearly worthless. I ask a question to those that say 'it doesn't have enough power': 'How often is the amount of power you have determining how fast you are riding'? Unless you are on a very high speed road circuit, an honest answer would be 'almost never'. For the street...responsiveness, torque production at any revs and good handling are far more important to me than having 200 HP I can't use. And truth be told....those people with 200 HP pretty much never get it delivered anyhow because they have wheelie and traction control acting on a fly-by-wire throttle that barely opens most of the time regardless of how aggressively you crank on the handgrip. Again...not quite enough is better than way too much for me. :)
 
#42 ·
How is this for a crazy build!

Machine out sleeve in stock cylinders.
Put in this liner 104mm
Use this piston

Problems: Seal between sleeve and cylinder and piston pin is 23mm not 22m on CP carrillo
 
This is an older thread, you may not receive a response, and could be reviving an old thread. Please consider creating a new thread.
Top