BSFC Numbers for Rotary
 

I have been told for years that rotary engine are extremely inefficient. I am curious how efficient they really are. It is easy to call something inefficient when there are no numbers involved. I have some actual numbers from a dyno run which should be reasonably accurate.

I disregard the higher speed power because I think my ignition is inadequate.

Engine: 4-port (mild street port, factory intake manifold, Racing beat headers)
Air Fuel Ratio:13.1:1
Total Ignition Timing: 38 BTDC
Compression Ratio: 8.2:1 (polished s4 turbo rotors)

Power Output: 247 Hp (198 WHp) = 184.3 kW
Engine Speed: 6500 RPM
Torque:200 Lb-Ft (160 WLb-Ft) = 271.8 Nm
Injector on time: 5.4 ms
Injector size: 460 cc/min

This is all the info I need to do the calculation.
Injector Duty cycle: 5.4ms/(60/6500) = 58.5% (good info to have, not necessary for calculations)
Fuel/time: = 460(cc/min)*4(injectors)/60,000 (ms/min) = 30.6667 uL/ms
Fuel/Revolution = 5.4 ms*30.6667uL/ms = 165.6 uL/rev
Fuel Rate (mL/s) = 17.94 mL/sec
Fuel Rate (g/s) = .72 g/mL* 17.94 mL/sec = 12.92 g/s
BSFC (g/kWh) = 12.92/184.3*3600 = 252.4 g/kWh
BSFC (lb/Hp.h) = 0.415 lb/Hp.h

Please bear in mind that this is a worst-case calculation, as the injectors actually have approximately 200 us of delay before turning on.

If you compare these numbers to those of a typical 8.2:1 CR piston engine at 6500 RPM, it would appear that the rotary holds it's own quite well against a piston engine.

Anyway, for those of you running naturally aspirated rotaries, please post the following information for the calculation (just so everyone knows, please feel free to do the math yourself).
Basic engine info (for information only)
Air Fuel Ratio (for information only)
Total ignition advance (for information only)
Power output (In WHp or Hp)
Compression Ratio (for comparison's sake)
Engine speed
Injector on time
Injector size

This should allow us to develop a good data pool for establishing the relative efficiency of a rotary engine.

We will not include input from engines that ran ridiculously rich (anything rich of 12.5:1 AFR) on the dyno in our average, this will artificially skew the numbers to look very inefficient (as it would for a piston engine, but the numbers out there for piston engines assume decent tuning). Total ignition timing must be at least 25 degrees to be included in the average.

Please feel free to post any data that you have, but if it is rich of 12.5:1 AFR or retarded from 25 BTDC it is not added to the average.

BSFC data is found in this spreadsheet

cool thread!

in for the info

welp, I'm out. :lol:

Actually, a large pool is always better. We should probably submit any numbers that we have and only include optimal ones for the average.

Maybe Turbos get their own category?...

Might be best. Honestly we need a better way to keep track of the numbers. I'll transfer what you've done over to a spreadsheet on googledocs and post the link.

Here you go NoDOHC:

https://spreadsheets.google.com/ccc?...thkey=CKLOhqwE

Fix some of your issues with your math :p

You might be right about a steady state snap shot of the engine, but when most people describe "BSFC" for a particular engine, we're talking about the average BSFC for all ranges.

I've heard 0.60 - 0.65 used in most applications when talking about a 13B.


-Ted

Kenichi Yamamoto's 1971 book, page 97.

Barry

http://i287.photobucket.com/albums/l...mpsumption.jpg

The math is correct as far as I can tell. I used an intermediate density of 0.72g/mL because gasoline covers a broad range of densities. Perhaps you don't like that I used engineering units. (I really didn't want to put 0.0001656 L/rev).

Actually BSFC is typically expressed as a chart with engine load on one axis, speed on the other axis and BSFC indicated by islands (similar to a compressor isentropic efficiency map for a turbo).

Many engines are rated at peak BSFC which is commonly computed at the highest BMEP (highest torque output). This is the point that I picked.

I will develop a full load BMEP curve for my engine at a variety of speeds, but not tonight. I don't have partial load data.

I agree that you have heard those numbers, so have I, but I also heard that I should never exceed 26 degrees of ignition advance and to always run rich of 12:1 AFR. If you do that, you will have a BSFC of 0.6 lb/HPh

Vex, I like your spreadsheet idea. I updated it to be more linear (for the easy addition of more data). You should be able to copy the formulas down to each new row.

Nah, engineering units aren't the issue. I just didn't know where you were pulling some numbers so I left it blank. Besides engineering numbers are master units; L, in, lb, kg, not slugs, g, etc (though I suppose in can easily be substituted for ft):p
That looks a lot better than what I originally had in there. May I suggest you put the link in the first post so others can add their numbers if they desire (I'm planning on doing something similar with the db profiles when I get around to getting to it).

I realize that engineer-types would insist on a graph version, but most laymen would just see a single number.

I have to disagree that this single number is highest BMEP...

Taken from: http://en.wikipedia.org/wiki/Brake_s...el_consumption


-Ted

I apologize, I was unaware of this technique for determining BSFC (fuel economy cycle average). The BSFC numbers that I use to compare engines for work are all taken at a certain speed and WOT, this is what I am attempting to document.

If we could find a way to simulate this driving cycle repeatably on different dynos with different cars, we could use that instead. Especially if there is a large amount of piston engine data available for this cycle. The closest thing I can find to data on BSFC during this driving cycle is largely computed from fuel economy numbers on the EPA cycle. We already have this data for RX7s, the EPA fuel mileage rating of the RX7 is awful.

The main reason that I am attempting to produce this chart is because I think rotaries are very inefficient at high manifold vacuums, but not bad at all under full throttle.

I am aware that many online calculators use BSFC for determining injector sizing. This is considering the engine at maximum output. Most users out there are interested in the number for this reason, not to determine average fuel economy.

Engine BSFC is well-documented at WOT on a typical (piston) gasoline engine. I can get curves when I get back home (traveling right now).

Basically Ted, I see your point, but I am mostly looking for a benchmark that we can use to compare. The only benchmark that we have right now is using a terribly tuned stock ECU with retarded ignition timing and non-optimal gearing for fuel economy to compare against the fuel economy of competing piston engines.

Let's not turn this into an arguing match over whether BSFC is an EPA driving cycle or determined at peak torque at WOT....

Because peak torque is what we can measure easily on a dyno and because the EPA driving cycle is not something that we can measure easily on a dyno. I recommend that we take our benchmark at WOT and peak torque.

This information is well documented for piston engines and will give us a good opportunity for comparison without having to account for gear ratios, engine load, poor tuning on the factory ECU, etc.

Hey man, don't sweat it...

We both know that "numbers" can come in laymans or expert terms.
A BSFC graph plot is the ideal.
I was just puzzled when you said a BSFC "number" was best peak.
Most piston engines average around 0.50, as a consensus.
Some of those Honda VTEC's can hit in the 0.4x range under best conditions, I think.

As another example...
If you know some sound engineering stuffs, drivers / speakers can be given frequency range numbers - i.e. "30hz - 20kHz"
Anyone who is seriously into this stuff knows that those numbers are useless - at the very least, you need a deviation % spec for it to be useful - i.e. "30Hz - 20kHz, 0.1%"
Take it one step further, and you'd insist on an actual graph plot of the frequency curve!


-Ted

If I can afford the dyno time, I would love to document a full BSFC curve. This issue with this is that drivetrain loss is a little funny at lower throttles (manifold pressures), as some of it drops with load (bearing and gear resistance) and some is fixed (windage and oil displacement). Still, this curve would be more useful than an actual engine BSFC curve to develop optimal mileage gearing....


More data would be awesome here, as I would not consider the number from a single engine a representative sample.

The data is easy to get, simply datalog injector on-time and engine speed during a dyno run and compare the graphs. You should have fuel pressure and manifold pressure to be really accurate.

With enough data, we could develop compression ratio comparisons (which should show lower BSFC with higher compression). Porting comparisons, aspiration comparisions, AFR comparisons (mostly done already by Kenichi Yamamoto) and ignition timing comparisons.

I wish I owned a dyno sometimes, I would get a lot more data if it didn't cost me $60 per hour to collect it.

I have all these numbers for the 13B-REW & Cosmo in std form, street ported and various version of race porting, across all types of fuel mixture and ign timing values, (compression ratio's from 7.8:1 to 9.0:1) apex seals from 2mm to 3mm single piece and multi piece........... but it took me 15 years of work/learning to accumulate the data and not about to share it on the internet... sorry :seeya:

One thing I will say! you must have the ability to measure the air mass flow rate, cause you will be surprised as to the ACTUAL volumetric efficiency of even a standard ported engine when free from OEM constraints (bolt on's)...... Wanted to add, be very careful making ASSumptions on fuel flow! (you actually need to test this). Also be even more careful relying on compressor maps (if you are doing turbo work!) you will see why when you start measuring turbine speed and trying to reconcile PR ratio's and calculated airflow rates! the more you get into this the more you see how many big fundamental mistakes you can make if using other peoples data/tables/figures.

I have never ever come across anyone (in performance game) who has accurate BSFC data that is believable (including some Racing Beat published articles on 12A Fuel injection V's Dellorto V's Holley set ups) and can be cross checked to other measures to be varified from modern rotary engines. There is plenty of very old texts that cover up to 1982/90 developments and some of those figures for BSFC on old 3mm two piece apex seals at heavy fuel mixtures in turbocharged application are flat out scary :o11: but again TOTALLY USELESS if you want to draw conclusions for your own stuff based on say a 13B-REW standard engine with associated bolt on's let alone tuning set up you have calibrated the engine too.

In short if you are doing it part time it will take you years to collect the data, you will need lots of equipment, and money and the qualifications in mechanical engineering (along with the interest to follow it through).

What you really need is an engine dyno and a super accurate fast acting lab (not Leb!) spec scales) I did BSFC testing on our FSAE car engine and simple effective way to measure fuel flow was to have the whole fuel system sitting on the scales! this negated the need to have $20,000 worth of calibrated fuel flow sensors in both supply and return line to determine the exact fuel flow into the engine. these sensors are still only as good as the data they are calibrated too and only work on certain fuel types!

With 3 reference meters we could see air fuel ratio and thus determine air flow rate in the said engine, from this data we could then along with power on the brake (on equilibrium!) establish that correct output less intertial variances! (very important). The output fuel usage (by mass) was feed into a data logger and all the collections automated.

Seen all kinds of set ups, there are many ways to do it, but as anything in life lots of ways to do it incorrectly! Dealing in an Aeronautical and Mechanical Engineering lab even 4th year Thesis students tended to make some very fundamental mistakes and ASSumptions that they think would cut it in doing up a proper report.

Hey, that would be awesome! I have a set of 6kg scales that are accurate to 0.1g. All I need is a fuel pump in a jug and set it on the scales. I could only have about 2 gallons in the jug, but that is enough for a pretty long dyno run.

Actually, I have a gram scale that goes to 50kg +- 5g, maybe that would be better for this application. I could use a ten gallon fuel cell on that and make the dyno guy a little less nervous.

I have a second fuel pump that I could plumb in and it would be easy to install a couple of ball valves. This would not only provide accurate BSFC numbers, but also allow calculation of injector energizing time. It would also indicate the accuracy of the previous data collection technique.

The Mustang dyno I use is capable of maintaining steady-state load on the engine. There will still be a ?? loss in the drivetrain.

I really like this scale idea.

You saved me from purchasing a set of fuel flow meters (not cheap).

Thanks!

A note on accuracy: I know that this data is not exact, but I don't think that the error exceeds 10%. Even at that, the BSFC does not match the commonly posted numbers. As I pointed out earlier, the calculations that I made are worst-case - they do not account for the inertia of the fuel (assuming instantaneous acceleration of fuel into injectors). I did get flow test results on the injectors (they are rebuilt) and they were all within 3% of 460cc/min (whether that is true now is questionable).

The goal in using this method is to get an easy data source so that those who do not have precise measurement abilities can post data too.

I like it! keep me updated :001_005: