This blog entry is an update to my first report on my car’s fuel consumption (FC) for various petrol brands.
Background of study
About one and a half years ago, I started to measure the fuel consumption (FC) for my new car (Nissan Grand Livina). Like any new car owner, I was curious about my car’s FC. I then decided to determine if there was any difference in FC between petrol brands in Malaysia. After six months of measurements, I reported my results here in my blog (Read here).
I reported then that there was a difference between petrol brands, notably between Shell and Petronas. I discovered that Shell gave the lowest FC for city driving but Petronas the lowest for highway driving. My results then was surprising because I had always presumed that even if there was any difference between petrol brands, their differences would be too small to be appreciated or noticeable in normal driving conditions.
I continued with my measurements because I wondered if my initial findings would remain stable over prolonged time (in other words, was my initial results due to chance?).
This blog entry is to update my results after more than one and a half years of FC measurements.
As before, I like to stress emphatically that my study is in no way of a scientific study setup. There are many weaknesses in my study that can distort (and even invalidate) my findings. For one, to be scientifically rigorous, my study would need more than one car and one driver (at least 12 cars and 12 drivers, to be exact). Second, each car must be filled with only one petrol brand (no mixing of petrol brands in one car). Third, all cars must be driven out for FC testing at the same time, and fourth, all cars must be driven along the same route – akin a caravan of test cars (and the order of the cars randomized for every trip). Many more conditions are required to make my study much more scientifically valid.
So, with that understanding, readers should kindly take the following results as only a rough or speculative indication of the FC differences between some petrol brands in Malaysia.
After a total of more than 100 petrol fill ups, my car’s average FC is determined to be 7.85 L per 100 km (which is equivalent to 12.81 km per L or 30.12 miles per gallon). At the current price of petrol, my car’s FC is 15 sen (cents) per km. Since I drive an average distance of 104.94 km per day, this means I spend an average of RM15.48 per day on petrol.
Fig. 1 shows my car’s average FC for seven groups of daily average distances. Up to 145 km per day of travel, my car’s average FC tends to remain unchanged because the confidence intervals (CIs) for the first four groups of distances overlap one another.
But for greater distances (>185 km per day), my car’s FC would improve (i.e., become smaller). Note that for these great distances, I only have one FC measurement for each distance group, so I could not calculate each of these group’s CI (at least two measurements are needed).
Nonetheless, their lower FC averages (6.7-7.0 L per 100 km) compared to that for the shorter distances (7.5-8.1 L per 100 km) suggests that my car’s FC would fall if I were to drive my car for more than 185 km in a day.
That my car’s FC would improve if I traveled greater than 185 km per day can be explained. The further I traveled in a day meant that I was using the highway roads more. Driving on highway roads tends to decrease FC because of better road conditions, lesser traffic, lesser braking-and-accelerating movements, and our cars can be driven at the optimal speeds for longer periods (note: for Nissan Grand Livina, the optimal speed for lowest FC rate is between 70 to 75 km per hour).
Fig. 2 shows my car’s FC based on individual petrol brands. For BHP, my car’s FC would start to fall if I traveled greater than 105 km per day. However, for the other three petrol brands, there was no clear trend of any change in my car’s FC for traveling distances less than 145 km per day.
For Petronas-X, only one FC measurement was done for 305-325 km per day, whereas for Shell, one FC measurement was done 185-205 km per day and another for 525-545 km per day. It is unfortunate that I have very few FC readings for these greater traveling distances. Thus, for these great distances, I could not calculate their CIs due to insufficient measurements.
Nonetheless, the lower FC averages for these greater distances (>185 km per day) suggests that my car’s FC would indeed fall. This is especially true for Shell where my car’s average FC was measured at 6.7 L per 100 km. More measurements are needed to confirm this trend.
What about FC differences between petrol brands? Take the first group of 65-85 km per day. The average FC for BHP, Esso, Petronas-X, and Shell for this group was 8.4, 8.2, 7.8, and 7.7 L per 100 km, respectively. Their averages may differ, but their CIs overlap one another – and not just for 65-85 km per day group but also for other groups as well (85-105, 105-125, and 125-145 km per day).
My results show that there is insufficient evidence that one petrol brand is better or worse than another brand. Instead, it is likely that there is no difference between petrol brands in terms of
Old vs. new Petronas formulation
My first six months of study reported that the old Petronas formulation was best for longer traveling distances. So how did the new Petronas formulation (Petronas-X) stack up against the old formulation?
Fig. 3 shows the FC for the old Petronas formulation (using same data as in my first report, just expressed differently). Compare Fig. 3 with the Petronas-X chart in Fig. 2. Initial examination may seem that the old formulation gave better FC than the new formulation for long traveling distances. For instance, the old formulation’s average FC for the 105-125 km per day group was 6.6 L per 100 km, which is lower than the new formulation’s average FC of 7.4 L per 100 km for the same group. But look at their CIs. The CI for the old formulation was large, and it overlapped with the CI for the new formulation.
For the 85-105 km group, both formulations have similar average FCs and their respective CIs also overlap each other. For the 65-85 km group, the old formulation had an average FC of 9.5 L per 100 km, but this value was based only on a single measurement (thus, its CI could not be calculated). Perhaps with more measurements, I might have obtained an average FC more similar to that by the new formulation (7.8 L per 100 km) and a CI that overlapped the CI for the new formulation.
My results suggest that there is little difference, if any, between the old and new Petronas formulations.
Remarks on the results
Unless we can control the environment in which we test the various petrol brands, it is crucial we do many, many measurements over prolonged periods. People often make too few measurements before deciding on the best petrol brand. We cannot also decide on the lowest FC petrol brand simply by comparing FC averages. We also need to compare the petrol brands’ CIs which show the margins of error for the FC measurements.
This update after one and a half years of continuous measurements is more pessimistic than my first (initial) report. With more than 100 FC measurements, my study reveals that there are likely to be little differences between petrol brands in terms of FC. Although my car’s FC would likely fall if I drove more on highway roads, no one petrol brand show any advantage over others.
That said, however, BHP warrants more FC measurements especially for longer daily travelling distances. Fig. 2 shows that BHP tended to give lower FC starting at a shorter 105 km of distance traveled per day. For the other petrol brands, they are only likely to start giving lower FC at further distances of 185 km per day onwards.
Perhaps petrol brands are different from one another, but my study suggests that these differences could be too small to be noticeable in everyday driving conditions. In a controlled environment, these petrol brands may give clear and measureable differences in terms of FC, but in the uncontrolled and varied conditions of the outside streets, these differences may become less pronounced or significant.
My study cannot prove or disprove anything due to the lack of scientific rigor in my study setup. As mentioned previously, my study can only offer speculative or tentative findings. I end this update by saying that more measurements are required to verify my results and to determine the stability of my latest round of results.
Until my next update, ciao.