A comparison of highway driving methods for the Toyota NHW20 Prius

[JimboK]

After much procrastination on my part, I submit for PC members' reading pleasure the following paper describing an attempt at using the scientific method to compare two different highway driving methods. Many thanks to PC members Bob Wilson, Hobbit, and Catgic for their thoughtful review and critique of the draft.

* * * * * *

A comparison of highway driving methods for the Toyota NHW20 Prius:

Electronic cruise control vs. RPM-guided pedal control

Introduction

The question of how best to drive the Toyota Prius at highway speeds for maximum fuel economy (FE) is one of substantial discussion. The consensus appears to be that electronic cruise control (ECC) is best on level terrain. For hills the optimal method is less clear. Some (1,2,3) suggest using ECC in hilly terrain too, whereas others (4,5) have maintained that manual control of the accelerator, holding RPM within a certain “sweet spot” range of efficiency, produces better FE. Obvious disadvantages of the latter method are the need for a tachometer – with which the Prius is not factory-equipped – and the need to continuously monitor RPM and make pedal adjustments. Nonetheless, some FE-conscious drivers might choose this method if a significant FE benefit can be demonstrated.

Though most of my driving is off-highway commuting, I take occasional out of town highway trips. Most are in moderately hilly terrain typical of the central Virginia Piedmont. Experience on those trips suggests that the RPM-guided pedal control (which I will call RGPC here) method may achieve better FE than ECC. But I have not done a controlled test on the same highway at the same average speed in the same or similar weather conditions.

I sought to perform such a test. I hypothesized that a carefully performed driving test using the RGPC method at highway speeds would yield FE superior to that obtained with ECC in a 2005 Prius.


Methods

The test vehicle is a 2005 Toyota NHW20 Prius with the following characteristics, modifications, and accessories:

• Approximately 26,000 miles
• OEM Goodyear Integrity tires inflated to approximately 50 PSI on the front and 48 PSI on the rear
• EV switch, installed as described here
• Engine block heater
• CAN-View
• Window tinting
• Toyota brand front-end vinyl bra
• Toyota brand mud guards, front and rear
• Body side mouldings
• Approximate combined weight of driver and cargo 315 pounds

The vehicle was fully warmed and at operating temperature before testing began.

The test route is part of VA288, a limited access highway in the suburbs of Richmond, VA. The route was chosen because of its gently rolling but long hills, relatively light traffic, and convenience. Speed limit is 65 MPH. The surface is concrete.

The test was between points near the highway’s interchanges with VA10 and US250, approximately 25 miles one way. The start and end points for each segment were a median crossover on the south end and a bridge over a creek on the north end. Both start and end points are at an approximate elevation of 200’. The route peaks at approximately 380’, and its low point is approximately 170’. Details of the test route are shown here.

Weather during the test was pleasant: sunny, relative humidity in the mid 40s, temperatures in the mid to upper 70s, and winds calm or nearly so. A laptop computer with mobile broadband was on board to allow periodic checks of local weather stations at Weather Underground during the test. Significant changes in weather, especially wind and temperature, would require adjustment of test data, and would be done with the Prius MPG Simulator.

Cabin climate control was run on automatic mode as needed for comfort. Initial temperature setting was 78, with adjustments to as low as 76.

CAN-View was used to monitor RPM, trip data, and fuel mileage. CAN-View’s fuel mileage calculation had previously been calibrated after a fuel fill-up, with its reading compared to the calculation from the factory multi-function display (MFD) after a recent highway trip of approximately 200 miles. CAN-View’s calculation was judged to be within 1% of that shown on the MFD.

The RGPC segments were completed first. A speed of 65 MPH was achieved before crossing the start point. Speed was then maintained between 55 MPH and 65 MPH and RPM between approximately 1400 and 2400 as displayed by CAN-View, staying as close as possible to a presumed ideal of approximately 1700 RPM when terrain allowed. On downhills where speed seemed likely to exceed the maximum despite RPM held at 1400+, “warp stealth” (WS) was applied. (The link describes WS in depth, but in brief, it is a term often used to describe a Prius driving technique where the go-pedal is completely released at speeds above 41 MPH, allowing the internal combustion engine, or ICE, to completely cut off, followed by immediate slight pedal pressure to release regenerative drag. The effect is similar to coasting with the car in neutral.) On uphills, progressively more pedal pressure was applied to keep speed at or above 55 MPH. If 2400 RPM was insufficient to maintain 55 MPH, then more accelerator pedal pressure would be applied; 55 MPH was deemed the lowest safe speed for the test.

Average speed was calculated at the end of each RGPC leg. Those averages were used as the basis for the ECC test speed settings.

The ECC segments were performed immediately after the RGPC test. They began similarly to the RGPC tests, with test speed attained before crossing the starting point. From there, ECC was kept activated with a constant speed for the complete segment.

One round trip test run was completed with each method.

At the conclusion of each segment, the following data were observed from CAN-View:

• Total distance
• Total time
• Average speed
• Average MPG

Data values were spoken into a voice recorder for later transcription into a spreadsheet.


Results

Here are results in distances, times, speeds, and FE:

Method, Segment, Distance (miles), Time, Avg. Speed (MPH), MPG

RGPC, Northbound, 24.8, 0:24:41, 60.3, 55
RGPC, Southbound, 24.8, 0:24:36, 60.5, 61
RGPC, Total/Average, 49.6, 0:24:39, 60.4, 58

ECC, Northbound, 24.8, 0:24:53, 59.8, 57
ECC, Southbound, 24.8, 0:24:48, 60.0, 61
ECC, Total/Average, 49.6, 0:24:51, 59.9, 59

As one can see, there was very little difference in results.

At all times during every segment, the hills’ gentle inclines allowed speed to be sustained between 55 and 65 using the RGPC technique as described above.

There was little fluctuation in weather conditions. Temperature ranged from 76F-78F, and humidity remained in the mid-40s. Wind, whose potential influence was considered to be greatest among weather variables, remained light and variable, with velocity ranging from calm to 3 MPH. Because of the variability of direction, no reliable adjustment in data could be made.


Discussion

The results were somewhat surprising and differed from my expectation. The difference between the two methods can probably be considered insignificant. Given that the RGPC method requires a tachometer and close attention to RPM and pedal position, most drivers likely would choose just to relax and use ECC on a highway like this.

My previous highway driving has been on other routes with presumably steeper hills. On those routes, the RGPC technique employed here often requires either progressive slowing on uphills to below the low speed target, or engine speed considerably higher than 2400 to maintain the target speed. On long downhills, extended periods of WS can be achieved. Downhills even occasionally result in acceleration to potentially unsafe speeds or those well over the speed limit. With practice I have learned to mitigate that tendency somewhat by allowing more speed to decay, with lower RPM toward the end of uphills if safe, or releasing the pedal and beginning warp stealth sooner on downhills. If needed to avoid excess speed, I occasionally release the pedal completely to allow regenerative coasting. Using ECC on these routes, with its aggressiveness in maintaining speed, often causes spikes in engine speed to well over 3000 RPM.

On this test, the RGPC method was able to maintain speed and RPM within their target zones, and opportunities for WS were infrequent. On uphills I never had to choose between an inefficiently high RPM or decelerating below the target speed. On downhills, I usually maintained ICE-on conditions with RPM within its target without using warp stealth.

During uphills on the ECC phase of the test, RPM rarely exceeded 2400. On two occasions it briefly climbed to ~2900. Though above the target, it still was below another RPM efficiency dropoff at 3000 RPM that has been described. On the other hand, it was interesting to note that RPM frequently fell below 1400 on level terrain or slight downhills, occasionally as low as 1200. It has been suggested that the ICE “loafs” inefficiently at these RPM ranges. This loafing, however, seemed not to adversely affect the results in this comparison with a method designed to avoid it.

All of this suggests that this route, even with its hills, allowed relatively efficient operation with ECC, comparable to RGPC.

I have not compared on a topographical map the elevation changes on this route to those on my more typical highway trips. But my findings suggest that the changes are gentler, making it difficult to extrapolate the results either to my typical trips or to those of others on routes with substantial hills. Nonetheless, those whose highway driving is on routes similar to the test route may find these results useful. Even in my own driving, I likely will use ECC more often in the future on segments whose terrain is similar or more level than this test route.

On segments with steeper hills, however, the RGPC method may still be superior. Hobbit has demonstrated that keeping the ICE within its efficient range in hilly terrain produces MPG results well into the 60s in warm weather. RPM may provide a clue as to when to change techniques: RPM values consistently above 2400 with ECC, and especially above 3000, may be a useful signal to turn off ECC and begin manual pedal control. Along with moderation of RPM swings, that would likely increase the opportunity for zero-consumption warp stealth. This is an area for further study.

The Prius is not equipped with a factory tachometer, so drivers using RGPC require an add-on device. In my case it was CAN-View. Others have used a ScanGauge or an inexpensive analog tachometer. For those without the money, desire, or ability to add instrumentation, I have previously proposed a rule of thumb for keeping RPM below its top efficiency limit: Keep the instantaneous MPG (iMPG) as shown on the MFD to a value greater than half the vehicle speed. I first proposed the rule based mainly on my observations below 40 MPH, where it appeared to apply across a broad range of speeds. As a secondary observation on this test I monitored iMPG during several accelerations, both during RGPC and ECC. At no time did I see iMPG drop to less than half the vehicle speed. I could have missed such occurrences; if so, they were rare. Therefore, the rule seems applicable to highway speeds as well as slower ones, and may be of help to those without a tachometer. If during ECC travel the iMPG stays above that threshold, the engine is likely staying below its upper efficiency threshold. On steeper hills, the rule could be applied to prevent inefficiently high engine speeds. In those cases, iMPG is more likely to drop below half the vehicle speed, in which case a change to manual pedal control may produce better fuel economy. This too might be an area for further study and validation.

A substantially different highway driving technique dubbed “Super Highway Mode” (SHM) has recently been proposed. SHM involves a deliberate attempt to keep highway power demands continuously low, effectively keeping RPM below the “loafing” threshold of 1400 or so. Some suggest that maintaining ignition timing (measurable with an aftermarket device like ScanGauge or CAN-View) in a relatively low and specific range is the basis for efficient SHM, although these ignition readings correlate closely with RPM; the latter may be monitored more easily, depending on the vehicle’s instrumentation. My initial anecdotal experience suggests that SHM seems best suited for lightly traveled non-freeway routes on level terrain. Speed will slowly but invariably decay even on level roads, and those who have used it report a maximum speed of 55-60 MPG, with speed decay to as low as 40 MPH – impractical and perhaps even illegal on interstate highways.

Regardless, modification of the RGPC test protocol to use different RPM targets, different speeds, or different methods (such as SHM) might have yielded different results.

Though wind was light in this test, it still could have influenced results. A 3 MPH wind could cause instantaneous fuel economy at 60 MPH to fluctuate by as much as 6 MPG, according to the MPG Simulator. Because of the wind’s variability in direction, however, wind-related adjustments to results could not be made reliably. A day with perfectly calm winds throughout the test would be ideal though difficult to find. Multiple test runs using each method might have minimized the effect of wind shifts on results, and would have added scientific rigor to the test (and possibly yielded different results), irrespective of weather.


Conclusion

Using ECC on a 2005 Toyota Prius achieved fuel mileage results comparable to RGPC in this test on a central Virginia highway with long but modest hills. ECC is easy to use, whereas the RGPC method potentially needs add-on instrumentation, and requires continuous attention to vehicle speed, engine speed, and accelerator pedal position. Therefore, ECC appears to be the superior method on this route and others with similar characteristics, though more controlled testing would allow conclusive judgments to be made. Modifications of the RGPC technique or testing protocol might yield different results.

[usbseawolf2000]

I haven't read through everything in detail but want to share a simple discovery I found while going through all my Prius literature. I am attaching the power/torque curve for 04+ Prius gas engine. It is easy to spot the bump in the torque curve around 2800 RPM.



I've been trying to tap this sweet spot when I need to accelerate onto highway or for passing. Using that sweet spot, the first time in the winter, I was able to get the first 5 mins over 50 MPG without the electric block heater.

I will gather more experience with it and see how it goes.

[patsparks]

Looks like fun guys, I'll stick with my cruise control I think. Thank you, an interesting read.

[efusco]

Yikes, you're getting all fancy pants on us Jim!
Nice work and write up. You can't scrounge up an elevation chart for the route you used for the test run by chance can you? As you suggest the rolling hills seem to be well suited for CC driving. Steep hills as I have around here I feel are often better suited for the RGPC method...but that may just be my sense of things and not based in fact. However I can often avoid the WOT mode by anticipating hills and getting my RPMS up before hitting them and carrying momentum far up the hills rather than hard throttling as the car does when left to it's own devices w/ ECC.

[Neicy]

Excellent write-up JimboK. Thanks for taking the time to do this. It really is helpful.

[JimboK]

Thanks to all for the comments!

I posted a link to this on CleanMPG, where some additional discussion has ensued, FYI.

Seawolf, I'm familiar with that chart and considered including it as a linked reference, but I figured it might be a little technical for some -- like me!

Evan, the page with the route map has two links on the left side labeled Elevation, small or large. Click either of those to see an elevation profile below the map. I agree with you, BTW, that RGPC probably is superior with steeper hills. But I don't have the data to prove it ... yet.

[douglas001001]

Nice write-up, interesting to see your opinion.

I agree that on unknown terrain that most would get about the same as cc.

But, in your daily commute, or on routes you know like the back of your hand is where the techniques can see improvement depending on route and speed. The slower the avg speed the more opportunity to hold onto the low/no fuel states. The hillier the terrain, a wider range of speed is needed to minimize burns on ups and take advantage of downs.

I bet if you repeated several times, noting the good and bad things that cruise control does, holding onto the good and eliminating the bad that you would see positive results.

I would guess that one of the things fighting against manual driving is the computer trying to keep battery charge at 60%. To see improvement over the entire route, the sum of all of your decisions has to be better, sometimes picking the highest at the moment might not be the best thing later on in the drive (for instance, if you WS glide too much you might burn more later with less battery when you could have used shm to save/add battery for that later part of your route).

To me, the most important thing to understand is how to control the car, knowing what each pedal input tells the car to do based on the many variables (speed, battery charge, IGN/RPM, etc). If you know that and your route, you can get to the most efficient way to drive a route if you have the desire to figure it out.

[bwilson4web]

Excellent report and methodology!

Originally Posted by JimboK

. . .
The test route is part of VA288, a limited access highway in the suburbs of Richmond, VA. The route was chosen because of its gently rolling but long hills, relatively light traffic, and convenience. Speed limit is 65 MPH. The surface is concrete.

The test was between points near the highway’s interchanges with VA10 and US250, approximately 25 miles one way. The start and end points for each segment were a median crossover on the south end and a bridge over a creek on the north end. Both start and end points are at an approximate elevation of 200’. The route peaks at approximately 380’, and its low point is approximately 170’. Details of the test route are shown here.
. . .

The test route is consistent with typical beltway or expressway routes with a few modest hills. Folks in hilly areas (aka., Little Rock AK) might have a different result. However, I think your route is consistent with what I would expect to find in the DC area and certainly Huntsville AL. It really looks like the type of profiles found on Interstate highways.

GOOD JOB!
Bob Wilson

[bestmapman]

From this expeiment it would seem that, for highway driving, aerodynamic improvements would yield more results than driving tehniques.

It seems from my prospecive, a lot of people have made significant mechanical mods. Has anyone made any aerodynamic mods, such as tail cone or wheel skirts.

[bwilson4web]

One of the best things Jim did was to hold the end-to-end time constant, the actual trip speed:

Originally Posted by JimboK

. . .
Results

Here are results in distances, times, speeds, and FE:

Method, Segment, Distance (miles), Time, Avg. Speed (MPH), MPG

RGPC, Northbound, 24.8, 0:24:41, 60.3, 55
RGPC, Southbound, 24.8, 0:24:36, 60.5, 61
RGPC, Total/Average, 49.6, 0:24:39, 60.4, 58

ECC, Northbound, 24.8, 0:24:53, 59.8, 57
ECC, Southbound, 24.8, 0:24:48, 60.0, 61
ECC, Total/Average, 49.6, 0:24:51, 59.9, 59

. . .

I have no problem with any proposed, driving strategy as long as the block-to-block time over a known route is compared to a reference sample, in this case ECC. If someone wants to define a well quantified, alternative driving strategy and test it against the equivalent block-to-block time standard, then those results will have merit, actual value, especially if replicated by others.

Jim correctly pointed out that there are speed ranges not yet investigated throughly, rpm of 3,000 and above. There are other, more radical driving strategies but any evaluation needs to compare equivalent, block-to-block times. So let's look at how one might design an experiment to measure the effect of 2,800 rpm acceleration.

First find a test route with a non-trivial distance and traffic conditions conducive to controlled tests. Jim did that with a 25 mile, divided highway. It also needs to be a route where 2,800 rpm acceleration can be tested against ECC managed acceleration. This means either some hills or if flat, some given speed range, minimum and maximum values.

First take a series of test runs using 2,800 rpm for acceleration and record the block-to-block times. Use these runs to 'standardize' the protocol so consistent block-to-block times are returned. One of Jim's better techniques is to enter the test track at a fixed speed on each runs versus starting from a dead stop. Be sure to record the MPG data. Using the block-to-block times and distance, calculate the equivalent ECC speed and repeat the test using ECC.

This will let us see to what extent acceleration at 2,800 rpm has an advantage, equal, or disadvantage compared to ECC managed acceleration.

Bob Wilson

ps. In the spirit of the holidays, I'll hold off on commenting about other driving protocols that have yet to be rigorously tested.