estimate BSFC sweetspot for partial load 1

[pickler]

I have a naturally aspirated 2.5L boxer 4-cylinder engine. I would like to estimate brake specific fuel consumption sweetspot for my engine at partial load (say +50% based on MAP readings). I do have a chassis dyno curve which i thought some one could use here to help estimate BSFC from. Basically this is a 16 valve boxer 4 cylinder oversquare engine with variable valve lift (Not timing). I have been told valve lift occurs at high loads and/or higher engine speeds and that one intake valve is always open.

I have attached the dyno graph. Looking at the dyno, the engine has a relatively flat torque curve for a 4 cylinder. Peak torque occurs at 4400rpm and peak horsepower at near 6000rpm. AFR stays stoich until just under 4000rpm. My estimates for BSFC sweet spot are 1500-2000rpm at 80% and 2000-2500 at 60% load, but i'm not sure how valve lift at higher loads affects BSFC.

 

[Tmoose]

Didn't Subaru start that option ~ 2005?
Does your car have an instantaneous MPG display option?
If not, there are some phone apps that interface with the OBD port and can display that info.
Or there are more expensive separate meter options.
Then drive the speeds you want and test.

A friend with a new normally aspirated Subaru insists his instantaneous mpg is better in 5th than 6th around 65 mph on the highway in our slightly hilly section of Eastern Massachusetts.

 

[pickler]

it does have a MPG monitor but it's inaccurate since it works based on air intake volume from MAF sensor. sometimes up to 12% off. So it's not a good indicator for me. i'm not worried about my instant MPG but rather BSFC during acceleration. So i want to burn least amount of fuel for most power when leaving a red light.

 

[Lou Scannon]

If you're that obsessed about it you'll have to set yourself up to datalog a few parameters from the engine control. Namely, injector pulse width (or calculated fuel), time, and vehicle speed. From these you can integrate the fuel (or pseudo parameter) consumed over a given acceleration event. Once you have this capability I recommend you do a series of repeated accelerations as identical to each other as possible (e.g. same grade if not exact same stretch of road), and evaluate your experimental repeatability. If you're satisfied with it, go ahead and perform different acceleration scenarios and see if you can resolve an optimum method out of the data.
You also have to decide what criterion to use as a yardstick. Myself, I would use amount of fuel consumed to go an absolute distance, with the proviso that this distance must be sufficient for the slowest accelerating scenario to reach a given cruising speed, which must be the same for all tests.

"Schiefgehen will, was schiefgehen kann" - das Murphygesetz

 

[GregLocock]

If you look at a few BSFC plots for NA SI automotive engines as a rule of thumb the optimum BSFC point is at max torque rpm, typically at about 80% throttle.

Quite why this is so is an interesting question to which I have no pat answer.

However that assumes the torque curve is at optimum af ratio, yours is not. If you fully fueled it lower down you'd get more low down torque



Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[pickler]

hey hemi i tried several runs couple months ago on a straight relatively flat road that's 400 meters long. It starts by rolling from a stop at different engine loads to reach 32mph.

observations :
Run....Load(%).......Shifts(RPM)....Consumption(L/100km)....Acceleration Time(s)...
A.........60%.............1800rpm.............11.5...........................17...
B.........80%.............2300rpm.............12.0...........................12...
C.........90%.............3000rpm.............11.9...........................9...

consumption is liters per 100km, so lower the better. Load is based on Manifold absolute pressure and NOT throttle.

 

[tbuelna]

I would agree with Greg Locock. With N/A SI engines, best BSFC at any given engine speed will normally be at MBT or peak BMEP. MBT implies 100% load and WOT. Just to be clear, in engine terms I believe "load" refers to torque, and not power. And the reason best BSFC at any given speed tends to occur at WOT is due to the significant effect on efficiency from pumping losses produced by any intake throttling. Reducing the need for intake throttling over a wide range of engine speeds is one of the main reason variable valve control strategies are now widely used on auto engines.

 

[pickler]

yes i used to believe that efficiency would be best by doing WOT operations at peak torque. But after some tests i found that my consumption was increased by about %15-20 when accelerating at WOT to 4500rpm. After some investigation i found that the fuel mixture was enriched just before peak BMEP.

 

[pickler]

PS. here is a BSFC chart for an N/A SI engine:

you can see that BSFC sweet spot is no where near the peak torque

 

[GregLocock]

hah, good one. OK, now explain how the max torque line doesn't define the max torque? That is, how can you have a bsfc number for values above the max torque line? I suppose it could be artifically throttled back, but that ain't a true max torque line in that case.



Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[GregLocock]

Here's a page full of bsfc charts, seems like my rule of thumb is busted when it comes to more modern engines

http://ecomodder.com/wiki/index.php/Brake_Specific_Fuel_Consumption_(BSFC)_Maps#Saturn_1.9L_DOHC

Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[pickler]

yea i've seen those which makes sense considering that most modern engines run stoich until higher engine speeds and some like mine have valve lift technology which further increases burn rate. seems like 2000-2500rpm is the magic number.

 

[Lou Scannon]

hey hemi i tried several runs couple months ago on a straight relatively flat road that's 400 meters long. It starts by rolling from a stop at different engine loads to reach 32mph.

I can't make anything out of those data. If you follow my suggested method and understand what you're doing, you'll be able do the analysis yourself.

"Schiefgehen will, was schiefgehen kann" - das Murphygesetz

 

[Tmoose]

How are you defining 60/80/90 % load?

I wonder about the impact of the fuel use during 20 seconds of a 30 minute real life road trip. I'm guessing throttle control and dRiViNg during the other 29.6 minutes might swamp the results.

I was expecting others to comment on the quoted 12% innaccuracy of a MAF sensor. I'd think that would clobber the A/F ratio and emissions.

I'd repeat the test 30 times using ~ 1/2 throttle (effectively full throttle at low revs and shifting at 2500 or so.

 

[pickler]

i didn't say my maf was 12% off i said my trip computer/MPG gauge was 12% off. that's a combination of fuel cutoff calibration, winter blend gas (evaporates in tank) and the fact that it doesn't measure how much fuel is being injected but how much air is being drawn in. Also as i said above i measure load based on MAP reading so when my map is showing 11psi i divide that by 14.5psi (atmospheric pressure) and i get 75% load. I don't have to go full throttle to fully load out the engine at 4th gear or 2000rpm, just 1/3 throttle will do that.

 

[pickler]

hey guys i had some time this weekend to test for various loads with the impreza today. I drove around in all sorts of roads and conditions and tried to keep it as accurate as possible. I had a friend who would log time of acceleration, engine load, RPM and the resulting consumption for every run. Then I looked at the log and created charts on fuel used vs load and rpm. I then plotted these efficiencies as dots on the torque chart created on excel and ended up creating this graph:

basically higher the efficiency colour the better gas mileage. What i found was that at lower RPMS higher loads are beneficial, but as RPMs climb load must fall too. It might have some thing to do with the variable valve LIFT system (i-AVLS). Anyway during one run we managed to shift at low enough RPMS without logging or kicking in the valve lift and that's when we scored one of our best runs. each run was approximately 800 meters long. Beyond 3500rpm we found the efficiency to drop significantly no matter what the engine load was. Mixture was consistently enriched above 3500rpm. Sweet spot seemed to be 2100rpm with 6 in.hg of vacuum or approximately 30kw/45hp held steady at 1500-2500rpm.

thanks for the help.

 

[Lou Scannon]

That's a very pretty chart but how did you calculate efficiency? And does your plotting method allow you to mark, on the same chart, the data points from which the contours are determined?

"Schiefgehen will, was schiefgehen kann" - das Murphygesetz

 

[Tmoose]

Was the acceleration during those trials something you can 'live' with ?

 

[Tmoose]

where does 60/65/70/75 mph cruise end up?

 

[pickler]

hemi, Efficiency was figured out using a OBD2 reader and logger. The lower the consumption for the run the higher the efficiency, the highest efficiency is considered 100%. The chart used to be covered in data points which i then joined together to form those islands using photoshop, as excel is very poor in creating areas. It also makes it simpler to understand but it is less accurate. Some small efficiency islands where omitted. Anyways i found that chasing small islands is a waste.

Tmoose, the acceleration rates are fine...35kW and shifting at 2500rpm allows me to flow with the traffic if not faster. For highway cruising i don't use BSFC data unless i'm pulse and gliding. Cruising at say 65mph requires 13kW which is no where near the 35kw sweetspot. But BSFC is not a helpful tool for cruising on a relatively flat surface. It does help in a sense that lower cruising RPM will benefit fuel economy due to lower friction but higher load in general means higher throttle which is not exactly what you want when cruising. For cruising I try to minimize speed and load to reduce drag on the freeway. Updated chart: