Based on my reading, the octane rating for a given fuel is calculated by finding the compression ratio at which it knocks and finding a mixture of $X\%$ isooctane and $(100 - X)\%$ n-heptane which knocks at the same compression ratio. The $X\%$ isooctane is then that fuel's octane rating.

But how does this work with negative numbers or those over 100?


2 Answers 2


I recommend the @Stanley Yu own answer, as I admit I was not particular familiar with the shape of the ON number curve.

It is done by an extrapolation of the critical compression ratios of mixtures 0-100% of isooctane and 100-0% of n-heptane for ON values <0 and >100.

If, just illustratively, 100% isooctane had the critical compression ratio 1:20, 90% isooctane 1:19 ,... 90% n-heptane 1:11, 100% n-heptane 1:10, then the fuel with the critical compression ratio 1:21 would have the octane number 110.

The octane number/rating differs for different procedures. See Research octane number RON, Motor octane number MON, Anti-knock index AKI=(RON + MON )/2 and Observed Road octane number RdON.

As seen in the OP own answer, the extrapolation of the octane number (ON) scale above 100 is done by linear extrapolation by the empirical aviation formula for performance number $$PN=100 + 3 * ( ON - 100 ) where on the graph is coordinate ON substituted by tetraethyllead addition.

  • $\begingroup$ Thanks. While that certainly makes sense in "completing the line", is it not actually defined? At that point these numbers bear no relation to the amount of n-heptane / isooctane. Seems dangerous to be comparing these numbers outside their domain. $\endgroup$
    – Stanley Yu
    May 11, 2020 at 23:12
  • $\begingroup$ What is the danger ? $\endgroup$
    – Poutnik
    May 11, 2020 at 23:22
  • 1
    $\begingroup$ My understanding is that motor vehicle fuel is not made with pure ratios of those chemicals anyway, so "octane rating" is just a proxy indicator or critical compression ratio. Through that lens, it seems benign having fuels that are not possible with reference reagents. $\endgroup$ May 12, 2020 at 1:23
  • $\begingroup$ Maybe "dangerous" was poor word choice. I mean it is difficult to interpret its meaning with regards to its original definition. If we detach the octane number from its definition, and apply it to its intent, then I don't see any issue. It makes sense. $\endgroup$
    – Stanley Yu
    May 12, 2020 at 2:54
  • $\begingroup$ @Stanley Yu Why difficult ? Consider ON = f(x(IO)) as the calibration curve, not as the direct comparison of reference and test fuel with the same behaviour. It is common practice to extrapolate calibration curve if it behaves well and statistical errors are low. BTW, even in IO+PbEt4 case, the performance number for the mixture is primarily defined by the above extrapolated function for n-heptane + isooctane. That mixture is just a practical, secondary standard to avoid testing the full calibration scale on an eventual non standard engine in ON testing context. $\endgroup$
    – Poutnik
    May 12, 2020 at 10:54

According to the standards group ASTM International, octane numbers above 100 are defined using the formula:

$$ON=100+\frac{28.28T}{1.0 + 0.736T + (1.0 + 1.472T - 0.035216T^2)^{0.5}}$$

where $T$ is "mL tetraethyllead (TEL) per U.S. gal in isooctane" [1]. This is the standard used by the FTC to regulate automotive fuel ratings in the US [2]. It should be noted that this formula is not an extrapolation of the octane curve from 0 to 100 as many seem to think. In fact, if we were to continue the octane scale, "[the] increase in the size of the octane unit [would be] so rapid as to make the unit become infinitely large if the scale were extended to about 116." [3] We can see this graphically:


So how was this formula determined? I could not find this directly specified anywhere, but ASTM gives us a clue in another publication relating to another number usually reserved for aviation fuel, the Performance Number, via the following formula [4]:

$$ PN = 100 + 3(ON - 100) $$

Performance number on the other hand is very well defined for numbers above 100: "This scale related percentage-wise the knock-limited imep (indicated mean effective pressure) obtainable from the test fuel to that obtainable from isooctane, assigning a value of 100 to the imep obtainable from isooctane." So for example, a performance number of 130 would mean that fuel had 130% the in-cylinder pressure of a fuel containing 100% isooctane. We can see a graph relating ON, PN, and TEL here [5]:

enter image description here

So what does all this mean? It means that an octane number above 100 is simply 100 plus the knock-limited indicated mean effective pressure over that of 100% isooctane divided by 3. Why divide by 3 and not use PN / KLIMEP instead? @Poutnik makes the observation that the factor of 3 ensures the PN-ON curve remains a linear extrapolation past 100.

I could not find anything regarding negative octane numbers. I can only assume people are extrapolating from the existing octane curve, but as seen above, the nonlinear, asympototic nature of octane numbers certainly would give me pause about the information conveyed by such a number. Again, if anyone has more info, please post!

[1] ASTM D2699-04, Standard Test Method for Research Octane Number of Spark-Ignition Engine Fuel, ASTM International, West Conshohocken, PA, 2004, www.astm.org

[2] 16 CFR § 306.5

[3] BROOKS, DONALD B. “Development of REFERENCE FUEL SCALES for KNOCK RATING.” SAE Transactions, vol. 54, 1946, pp. 394–403. JSTOR, www.jstor.org/stable/44548282.

[4] “Standard Method of Test for Knock Characteristics of Motor and Aviation Fuels by the Motor Method.” ASTM Manual for Rating Motor, Diesel, and Aviation Fuels: 1973-74, American Society for Testing and Materials, 1973.

[5] MacAuley, J. B. “Advances in the Utilization of Tetraethyllead.” Third World Petroleum Congress, The Hague, 1951: Proceedings, E.J. Brill, 1951, p. 157.

  • $\begingroup$ So at least the aviation PN is an extrapolation of the ON curve. $\endgroup$
    – Poutnik
    May 20, 2020 at 9:32
  • $\begingroup$ Not sure I'd call it extrapolation, @Poutnik. In the second picture, the right side uses a completely different x-axis (mL TEL) and we can see the curve, instead of continuing upwards to infinity, curves "down" instead. But it certainly continues off from where 100% isooctane ends ($100_{ON}=100_{PN}$) $\endgroup$
    – Stanley Yu
    May 22, 2020 at 3:47
  • $\begingroup$ But the picture on the right is just the extrapolation aviation formula PN = 100 + 3*(ON-100), for ON=100..120 with the coordinate X substitution ON=f(T) by the formula on the top, getting PN=100 + 3 * ( f(T) - 100). The factor 3 is then the empirical slope of the tangent line dPN/dON. $\endgroup$
    – Poutnik
    May 22, 2020 at 7:49
  • $\begingroup$ So they used the aviation linear extrapolation formula PN=g(ON), determined dependence of PN=h(T) and from that they derived ON=f(T). $\endgroup$
    – Poutnik
    May 22, 2020 at 8:34
  • $\begingroup$ Hmm... I see what you're saying. I couldn't find an explicit statement of motivation, but the desire to keep the same tangent line would explain the factor 3 and makes sense. $\endgroup$
    – Stanley Yu
    May 28, 2020 at 22:44

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.