Fuel Octane Testing
A comparison between Sunoco 94 and Torco 104 unleaded
Knock can and will kill engine components (many times silently) if boost, fuel, and spark timing are not kept in check. Pistons, rings, spark plugs, and exhaust valves are placed in great danger if knock persists. While our cars have a knock sensor that attempts to keep things in check, you should never rely on it to bail you out if you have mods. Knock is a function of high pressures and temps that instantaneously combust remaining air and fuel if given enough time.
More detail on what is bad (promotes knock):
Boost set too high - As boost is raised (everything else constant), the cylinder pressure and temperature raise, both of which make knock more favorable.
Fuel lean condition - A fuel lean condition during WOT becomes dangerous as temperatures quickly approach and can exceed safe limits (~1600F at the exhaust manifold). Even if knock isn't occurring, these temps will quickly over-age exhaust valves which can lead to the valve head separating from the valve stem, failing the piston, head, and bore wall as it rattles around.
Timing too far advanced - results in higher peak cylinder pressures which is good for power, just don't go overboard. Spark timing is normally not modified from stock although G-Force can change the ECU and several aftermarket ignition coils allow for timing retard.
So, to keep knock at safe levels, limit boost to a safe level and make sure you have a fuel system that can support the boost you desire. The fuel system must be able to supply enough fuel for combustion and then in many cases some for cooling. Another option, although not popular, is to manually retard spark timing. Some serious owners opt for water (or ethanol) injection for its ability to bring temps down. Sizing your turbos for desired boost (a larger turbo will lower temps at higher boost) coupled with more efficient intercoolers will help matters.
Lastly, a higher octane fuel will reduce knock. Octane can be viewed as a measure of a fuels resistance to knock. So the higher the octane, typically the higher you can set boost without knocking. To be completely safe, you must also monitor exhaust gas temps (EGTs), recall the discussion above about over-aging valves etc.
A word of caution, many high octane fuels (usually anything higher than 104) use lead to raise octane. Lead is dangerous to catalysts and with prolonged use can poison O2 sensors. Also worth mention, a high octane fuel burns slower. For this reason, using high octane fuel can actually hurt performance if your car can safely run on lower octane fuels. So don't go overboard.
As discussed above, a common practice for those of us running high boost on the street or at the track is to protect against knock using high octane fuel.
I'll admit up front that these two runs are far from perfect (as far as driver is concerned). I had a poor launch on the first run, poor shift . . . (read on for more details on my mistakes). But in the name of education, I've chosen to share these two back-to-back 1/4 mile runs.
Sunoco 94 - 1/4 mile run, 15psi boost
(7/24/99 Milan dragway - 94 deg ambient)
For those of you not familiar with or don't know how to read a datalog . . .
TMO Datalogger 101 - Crash Course
First look at the legend at the top of the datalog for a color-keyed listing of all variables logged. I could only display seven graphs at one time, so I opted to not show the first two items (air flow hz & air temperature). The seven graphs from top to bottom are:
Coolant Temperature (degrees F)
Engine Speed (rpm)
Injector Pulse Width (milliseconds) - can calculate Injector Duty Cycle (IDC) once you know engine speed and injector pulse width
Knock Sum (this number is generated by our knock sensor and increases as the magnitude and frequency of knock increases)
Oxygen Sensor (volts) - datalogger only reads rear (left) bank oxygen sensor
Throttle Position (%) - 100% being wide open throttle (WOT)
Timing Advance (degrees Before Top Dead Center (BTDC))
There are many other variables that can be logged, but these are the ones I tend to monitor on my 1/4 mile runs.
Back to the 94 octane datalog. Looking at the datalog, you can see I launched at an aggressive rpm (~6,000) and slipped the clutch (another bad practice). I messed up the 1 - 2 shift, and finished with a slow 2 - 3 shift netting a 106mph trap. Note, rpm jumps a certain amount (more prevalent at higher engine speeds) due to extremely small time the ECU has to calculate rpm based on cam position sensor readings. This was further compounded by a slipping/grabbing clutch that was heat soaked from my poor launch (worst near peak torque) causing a bouncy looking rpm signal.
The main variable to look at here is the green "Knock Sum" line. With 94 octane the knock sum reached 20 or more in all gears. This knock correlates with retarded timing (thus less cylinder pressure and less performance). TMO list-server members have discussed the relationship between knock sum and timing retard as follows:
degrees spark retard = 90*knocksum/256
The ECU wants to deliver peak power under WOT and will tolerate some knock before timing is retarded. Many TMO users believe knock isn't retarded until a knock sum of ~15 is reached after which the equation above may be used. The knock sum will also decay (decrease over time) until additional knock is sensed. This helps to minimize timing retard. The ECU will also reset the knock sum to zero as soon as you let off the throttle.
Side note: The timing advance value recorded by the datalogger is in error. TMO members believe it may read 50% high on our V6 engines. The datalogger was first developed for our four cylinder, 2.0L, cousins (Eclipse / Talon / Laser) and carried over to our V6s. As a result, a 50% error (1 - (6 cylinders / 4 cylinders)) may be present. This error is further compounded by not including base timing. I plan to degree my damper and determine the relationship between actual timing and TMO recorded values with a timing light.
104 Torco Unleaded - 1/4 mile run, 15psi boost
(7/24/99 Milan dragway - 94 deg ambient)
Minutes after completing the Sunoco 94 run (~2 gallons remaining in tank), I topped off with Torco 104 unleaded (actually has a trace amount of lead) and got back in line for another run. Note the much improved launch and 1 - 2 shift which resulted in minimal clutch slippage and a better 109mph trap.
Like magic, the knock nearly disappears with only a trace knock found at the end of first gear and at the end of third. For a more direct comparison of the two runs, continue to the next graph.
Another nice feature of the TMO datalogger is its ability to export raw data to a spreadsheet. This allows the user to manipulate the data to meet their needs. Above, I took a 3.5 second sample of both runs (~5,000 rpm to ~6,400 rpm) comparing several variables.
In the spreadsheet, I calculated Injector Duty Cycle (IDC) as follows (simplified form):
IDC = 100% x (injector pulse width x Engine Speed) / 120,000
injector pulse width (milliseconds)
Engine Speed (crank rpm)
As you can see, the 94 Octane run (red lines) differs from the 104 Octane run (blue lines). The higher knock sum of the 94 octane results in more spark retard. Note that the piggy-back fuel computers (VPC & GCC) were adjusted to fairly rich settings. Because of this, the oxygen sensor measured at least 0.92V at all times on both runs (fuel rich). The 94 octane run also resulted in a higher injector duty cycle. This may be part of the ECU's strategy (i.e. high knock results in longer injector pulse width in an attempt to cool the combustion chamber and reduce knock).
The only confusing result of the datalog is the leaner O2 reading (lower voltage) on 94 octane despite the higher IDC. The higher oxygen concentration in the exhaust gas may be explained by the poor combustion due to too much fuel (partial combustion and perhaps misfiring).
So in review, the lower octane fuel resulted in high knock which retarded timing (loss of performance), dumped more fuel for cooling (loss of performance), and slightly worse combustion. Was the run with 94 octane dangerous to my engine? Probably not, but now that I have my datalogger, I refuse to run anything higher than a knock sum of 15 (and when I get my EGT gages repositioned to the exhaust manifold, no higher than 1600F).
since August 7th, 2002
Last Updated: 08/08/02 07:50 PM