**MPG-Based Approach**

Tom,

Thanks for checking your model against the EPA MPG test profiles. In my earlier post I assumed GM would just use the US06 profile to calculate the range. Upon further investigation, I agree this is incorrect as US06 is too aggressive. Although I do find it encouraging that your model predicts the Volt would still achieve almost 30 miles all electric using that profile.

Anyway, inspired by your results, I dove head first in to the EPA Final Technical Support Document EPA420-R-06-017, which details the revisions to improve the calculation of fuel economy estimates that became effective with model year 2008.

**www.epa.gov/fueleconomy/420r06017.pdf**

If I were GM, I would base my range estimate for the Volt using a “combined range estimate” extrapolated from the technique the EPA uses to calculate the combined MPG. The EPA calculates the combined fuel economy by weighting the city at 55 percent and the highway at 45 percent using the following equation:

FE combined = (55% x City FE) + (45% x HWY FE)

Using this equation and substituting range for MPG, results in the following “combined range” equation:

Range Combined = (55% x city range) + (45% x hwy range)

Plugging in the raw range numbers your model predicted for City and Hwy range yields a combined range of

**.**

__39.9 miles__Range Combine = (55% x 40.2) + (45% x 39.6) = 39.93.

For pre-2008 model year vehicles, the city and highway tests (I understand these are the 2 profiles you ran in your model, in addition to US06) were adjusted downward to account for typical road conditions that can affect fuel economy, such as higher speeds, cold temperature, and use of air conditioning. The city results were adjusted downward by 10 percent and the highway results were reduced by 22 percent. I know this is deep into the specifics, but didn’t you state your model includes allowances for headwinds, grades, A/C load, etc? I believe the answer is yes, and if correct, your model’s range predictions would appear to agree with GM’s claims!

HOWEVER, for 2008 and later model year vehicles, the fuel economy label estimates are determined by either performing the entire set of five tests on the test vehicle (see

**http://www.epa.gov/fueleconomy/420f07066.htm#calc**), and using that data to calculate city and highway estimates, or by applying mathematical derived linear regression equations to the basic data from the city and highway tests to obtain estimates that simulate the effects of all five test cycles. The second approach is referred to as the mpg-based approach, since the resultant label adjustment will vary depending on the measured fuel economy (i.e., mpg) of a vehicle over the FTP and HFET tests, and will not require any additional tests. (See pages 2 and 5, EPA420-R-06-017)

The mpg-based approach is the following:

MPG based city FE = 1 / (.003259 + 1.18053/FTP FE),

where FTP FE = City test/profile

MPG based highway FE = 1 / (.001376 + 1.3466/HFET FE),

where HFET FE = Highway test/profile

For example, substituting ranges for fuel efficiencies, and using the range values predicted by your model yields the following ranges:

MPG based city range = 1 / (.003259 + 1.18053/CITY RANGE)

MPG based city range = 1 / (.003259 + 1.18053/40.2)

MPG based city range = 30.7 miles

MPG based highway range = 1 / (.001376 + 1.3466/HWY RANGE)

MPG based highway range = 1 / (.001376 + 1.3466/39.6)

MPG based highway range = 28.3 miles

And using these values to calculate a MPG based combined range yields

MPG based combined range = (55% x 30.7) + (45% x 28.3)

MPG based combined range = 29.6 miles

As a check, it would be interesting to remove the "typical road conditions" (headwinds, grades, A/C load, etc.) parameters in your model , and recalculate the HWY and City Ranges. Then plug these values for HWY and City ranges into the MPG based approached as a check? I’m guessing this is what GM is doing, and predict it should also equal approximately 40 miles. If correct, this result would be additional confirmation that your model and assumptions are pretty darn good, thus it would be reasonable to assume the ranges it is predicting at different sustained speeds is also accurate.