gsoh/VED

Calculation of Fuel Consumption

Closed this issue · 7 comments

Robotoks commented

Thank you for your contribution. It really helps.
However, few data are available for the Fuel Rate attribute.
Almost all records has no fuel rate value.

gsoh commented

Hi,

Fuel consumptions are obtained through 3 different ways as described in our VED paper (Algorithm 1), using three attributes of the data "Mass-air flow" and/or "Absolute Load" and/or "Fuel Rate".

Note, the data attribute "Fuel Rate" should not be confused with the availability of the fuel consumption information. In this sense, the populated percentage of the attribute "Fuel rate" in the Table 2 does not represent the availability of "fuel consumption rate" or "fuel consumption". Because, for most vehicles, the estimation of fuel consumption & fuel consumption rate is obtained through "Mass-air flow" and/or "Absolute Load".

Please refer to the paper for the details.

Robotoks commented
gsoh commented

Hi,

You could "approximate" fuel consumption using speed based on a set of assumptions about the vehicle types, powertrains, mass in a specific scenario. However, since the approximation involves simplications, you would not obtain as accurate fuel consumption as you would when you use MAF/AbsLoad/RPM with correction factors.

One of the motivations behind making this dataset public is to share accurate fuel-consumption big-data. In this sense, we recommend using Algorithm 1 of our paper to compute fuel consumption as accurately as possible.

Thanks!

Robotoks commented
Robotoks commented
gsoh commented

Hello,

While we appreciate your questions, you will be able to find answers for the questions.

  1. Please refer to the paper for brief explanations on STFT and LTFT. It's up to you which one to use.
  2. You should not take the absolute value of STFT. You should just use STFT.
  3. The "engine configuration & displacement" attribute of VED_Static_Data_ICE&HEV.xlsx and/or VED_Static_Data_PHEV&EV.xlsx files. The paper also mentions it.
  4. You are correct that STFT and LTFT are not always available when MAF is available. One way (but not the only way) to compute the correction term is as follows:

flag = 0
if STFT1 and LTFT1 are available, flag = 1
if STFT2 and LTFT2 are available, flag += 2
if flag == 1:
correction = (1+0.01STFT1+0.01LTFT1)/AFR
elif flag == 2:
correction = (1+0.01STFT2+0.01LTFT2)/AFR
elif flag == 3:
correction = (1+0.005STFT1+0.005LTFT1+0.005STFT2+0.005LTFT2)/AFR
elif flag == 0:
correction = 1

Even though we used the above algorithm to obtain the results presented in the paper, we did not describe the algorithm in the paper simply because we wanted the algorithm to be general, since there are other ways to compute the correction terms.

Thanks.

Robotoks commented