Mazda to Debut Next-Generation Direct Injection Gasoline and Diesel Engines, Transmission and Kiyora Concept at Tokyo Motor Show 2009
Tropical Storm Ketsana Kills More Than 240, Displaces More Than 609,000 in Philippines

National Labs Developing Methodology for Estimating Real World Fuel and Electricity Consumption of Plug-in Hybrids Based on Dynamometer Data

NREL’s proposed adjustment method assumes that the increase in gasoline use during CD mode is the same as the increase calculated for CS mode. This works well for blended PHEVs that have lower electric power capabilities. Source: Gonder et al. Click to enlarge.

Researchers from the US Department of Energy’s (DOE) National Renewable Energy Laboratory (NREL), Idaho National Laboratory (INL) and Argonne National Laboratory (ANL) are cooperating to develop and test a method for predicting the real-world fuel and electricity consumption of plug-in hybrid electric vehicles (PHEVs) by adjusting dynamometer test results. After examining data on the only PHEV currently available in large numbers, the new adjustment method shows promise for reasonably predicting PHEV average fuel and electricity use, despite differences in design.

Current rules for conventional vehicles do not work for plug-in hybrids because the vehicles run on both electricity and gasoline; industry debate centers on the rules for estimating miles per gallon. This was highlighted by the reaction to GM’s announcement that the Chevy Volt would attain 230 mpg in the city cycle, given a single charge per day, along with combined cycle electricity consumption of 25 kWh/100 miles, based on a draft EPA methodology. (Earlier post.)

PHEV testing is further complicated by the fact that these vehicles operate in two different modes based on the distance they are driven (initially depleting energy from the large vehicle battery, and eventually sustaining the battery charge for longer distance driving). Consensus is building on techniques to handle these first two complications, but one question that remains is how to adjust raw certification cycle test results to best predict a PHEV’s average real-world energy use.

Official fuel economy testing for all vehicles is conducted on chassis dynamometers, which are basically treadmills for cars and trucks. One subtlety of chassis dynamometer testing is that vehicle fuel economy measurements using decades-old standard speed profiles may be overly optimistic compared to today’s average on-road fuel use. Official methods exist to adjust the test cycle fuel economy of conventional vehicles to better estimate expected real-world fuel use, but a similar adjustment method has yet to be finalized for PHEVs.

—NREL Research Engineer Jeff Gonder

In an effort to objectively predict on-road energy use of a PHEV, NREL developed and is proposing a method to adjust the standard test cycle results from each mode—charge depleting (CD) and charge sustaining (CS)—of PHEV operation. The adjusted values are then combined into a single fuel economy prediction.

The Blended Method...assumes that the increase in gasoline use during CD mode is the same as the increase calculated for CS mode. This works well for blended PHEVs that have lower electric power capabilities for CD mode, and would thus require additional engine power (blended with the electrical power output) for more aggressive driving. The downside of this method is that PHEVs with high electric power capabilities may not need help from the engine and therefore would not use more gasoline in CD mode but would simply deplete their battery energy over a shorter distance. It is also possible that a blended PHEV would actually increase its depletion distance in the event that the vehicle controller commanded the added engine output in CD mode to be high power (to achieve high engine efficiency) and thus prolonged its battery depletion.

Even so, such tradeoffs between CD fuel consumption and depletion distance should somewhat balance out through UF [utility factor] application. For instance, though this Blended Method for applying adjustments may penalize the high electric power PHEV with some excess CD fuel use, the method assumes a longer CD distance than the vehicle actually achieves. This gives it an inflated UF weighting for CD fuel displacement (relative to its CS fuel use). These two factors may roughly balance each other out when calculating the total combined consumption. A similar balance (in the reverse direction) could work out for the longer depletion distance blended PHEV.

—Gonder et al.

NREL has applied this technique to its PHEV analysis for several years, but until recently has not been able to validate it against data on a large number of PHEVs operating on the road. Partnering with the two other DOE laboratories provided that opportunity.

INL monitors fleet fuel use of advanced technology cars as part of DOE’s Advanced Vehicle Testing Activity, and has accumulated more than a year’s worth of data on roughly 100 PHEVs of the same design. (Because of limited purpose-built PHEV availability, these vehicles are production hybrid vehicles modified with aftermarket PHEV conversion kits). ANL had collected dynamometer test data on the same type of vehicle to evaluate PHEV test procedures over the standard certification cycle speed profiles.

This analysis provided a great example of collaboration among three DOE labs. NREL applied the adjustment technique we developed to PHEV data from ANL dynamometer testing and compared the fuel economy predictions to on-road data from INL’s large fleet evaluation effort. After accounting for how frequently the PHEV’s in the INL-monitored fleet actually plug in, we found excellent agreement between the adjusted test cycle predictions and the actual fleet fuel and electricity use.

—Jeff Gonder

For the Hymotion Prius PHEVs, NREL’s method estimated fuel consumption of 4.2 L/100 km (55 mpg) and electricity consumption of 5.5 kWh/100 km (89 Wh/mi). This represents an average decrease in gasoline fuel consumption of roughly 20% relative to the comparable HEV Prius window sticker value.

The validation has so far only been possible on a single aftermarket conversion PHEV design—a Prius converted with a Hymotion pack. It will be important to repeat the analysis once dynamometer testing and substantial on-road fleet data becomes available for different PHEV designs, particularly those with greater electric driving capability, NREL notes.

While this initial comparison is promising, further evaluation of the proposed adjustment method will be required on other vehicles and PHEV platforms. In particular, it will be important to evaluate how well the method predicts in-use performance for non-blended PHEVs that possess high electric power capability. As discussed, the proposed method utilizes a simplifying assumption that the PHEV will deplete its stored battery energy over the same distance in real-world cycles as in the dynamometer tests, which may not be the case in reality. This paper described how the UF application may help offset such differences, but further evaluation will need to confirm how well this works out.

—Gonder et al.

In the meantime, NREL plans to extend the analysis by simulating “virtual” fleets with a variety of PHEV powertrains operating on GPS driving profiles obtained from conventional vehicle travel surveys.

While this process seeks to predict the average on-road fuel and electricity use from a large number of PHEVs, fuel economy will vary greatly based on how the vehicle is driven and it will be important to educate PHEV drivers on how to obtain the best results, NREL notes.

The results of this research were presented at the recent 5th IEEE Vehicle Power and Propulsion Conference (VPPC’09) in Dearborn, Mich. Gonder published an earlier paper discussing broader PHEV fuel economy reporting issues and serves on the SAE subcommittee (chaired by ANL) that will shortly be issuing an official recommended practice for measuring PHEV fuel economy.




Any single method could be mostly non-representative because users will use very different daily e-distances.

A common method would have to account for various situations, such as mpg or (Km/L) based on 10% e-distance, 20% e-distance, 30% e-distance, etc etc.

This way, the mpg for a Toyota Prius PHEV could go from 60 to infinity, depending on daily e-distance used.

Would it be fair to average (weighted or non-weighted) all possible situations to arrive at one single figure? Why not, as long as every country and manufacturer agree to do it the same way. Internationally, a liquid fuel consumption given in Km per Liter, (Km/L) would be more appropriate. That method is already used in many Asian countries.


Will someone involved in the industry make a comment about the value of these tests and calculations?

It sounds like interesting work but not innovative. And I'm not sure what problem it will solve.

Henry Gibson

There needs to be a definate concensus statement from all national laboratories that it is impossible to use only the BTU content of a liquid fuel and compare it to electricity with a BTU content of 3412 BTU per kWh. Electricity is used two to ten or more times more efficiently in automobiles than gasoline mostly because its use is not subject to the second law of thermodynamics. ..HG..

Henry Gibson

On the other side of the issue, heat pumps can easily take 3412 BTU of electricity and convert it into 3*3412=10236 BTU of hot water with EcoCute heat pumps. This is three times as much heating as can be done with an electric heating element, but the cost of the unit is still very high. ..HG..


Henry, newer heat pumps are hitting COP's of 5-6.

This means replacing natural gas fired heating with a heat pump powered by electricity from a combined cycle gas turbine (60% efficient) will use about 1/2 to 1/3 as much gas as the original system and heat pumps are good match for wind power as it is generally more windy in the winter and the electricity can be stored as heat inside the buildings (much cheaper than storing electricity)

Electric vehicles and heat pumps can together make use of off peak electricity which will hopefully mean nuclear plants can replace coal as baseload. Make up the rest with wind and solar and balance the system with combined cycle gas turbines running on a mix of natural gas, bio-methane and synthetic natural gas from other hydrocarbons.

You can learn more about Idaho National Laboratory's energy projects at

Nat Pearre

It is useless to try to come up with a single fuel consumption number for a vehicle that has more than one more-or-less independent energy sources.

There needs to be a range on electricity (cty/hwy on the existing drive cycle), and a mileage on liquid fuel (as we have now). Attempting to average the numbers out for all possible use cases in the end gives no information at all, since I have no idea how my driving relates to that average.

I think Americans are smart enough to figure out a 4 number system. We've done just fine with a two number system.

Some PHEVs will have a blended-operation mode when the battery is full, but even then there is still some amount of available energy in the battery, and some value for energy consumption per mile, so you can still quantify the range you get out of your wall socket.

The comments to this entry are closed.