Off Road Range for EV? Jeep 4xe Review / Comparison from TFL

[Ssaygmo]

Defender with bigger tires

Given the point of my thread was the electric jeep, but I did comment on the use of stock 20's. He keeps saying "keep in mind this is a factory vehicle, but its a $70k plus factory bronco, so you could buy a defender for 10-20k less and put 35's and a lift on it. Buy the car for the usage you will do 99% of the time, the compromise is the 1% road trips or off-road usage where a cheap beater jeep with locker is probably better.

Sponsored

 

[ajdelange]

The main drift of this thread is that going off road can rip through a battery over the course of only a few miles. The OP hinted at consumption of 6 kWh/mi (!) and I've presented data suggesting that consumption will be around 3.2 kWh/mi in sand for a 2500 kg (5500 lb) truck.The former implies 20 miles in the 120 kW version or 30 miles in a Rivian Max and the latter 37.5 or 56 miles. It's pretty clear that we are not going to be doing extensive "adventuring" in these vehicles unless "adventuring" is mostly on fairly good roads. I think that's the main conclusion for readers to draw from this thread but it raises two very important questions:
1)How do you plan for an trip in order to be sure it is an adventure and not a misadventure?
2)How can Rivians planning and driver information systems give the driver what he needs to know given that he is driving in places where lack of knowledge of the substrate can lead to consumption uncertainties of kilowatthours per mile?

I used to drive down to the Outer Banks from the DC metro area. When down there one of the things we did was take the truck out onto the beach near Oregon inlet and "fish" (rinse mullet and drink beer). The trip down there would be a snap in the Rivian. ABRP says I can get to my buddy's place in Corolla with 20% in the battery with a single charge. If I assume the roads are dry that charge is from 27 to 56% (∆ of 29%) but if there is heavy rain ABRP says I'll be charging form 22% to 60% (∆ of 38%).

Given the paucity of charging opportunities in the Outer Banks my best bet would be to plug into my buddy's clothes dryer outlet and take on maybe 10 hrs worth at 30 amp which would add about 65 kWh to the 36 remaining for an SoC of 101 kWh or 56%. ABRP tells me I can get from his house to Oregon Inlet on 19 kWh. I'd expect to drive no more than 3 miles on the sand and, while I don't have any idea what condition the sand is in today (packed smooth by the tide, torn up by other fishermen, wet from yestrerday's rain, etc) I can allow 9 - 18 kWh for that leaving me 101 - 19 - 18 = 64 kWh which is plenty to get me back to my friends place or, more likely given the way we used to do these trips, 7 miles back up the road to the Outer Banks Brewing Station which has two Tesla destination chargers (can charge the Rivian with a Tesla Tap).

Thus, one can plan reasonably plan a trip that involves a mix of on and off road travel using tools like ABRP with a little manual calculation. But what about the planner in the vehicle? If sitting a mile and a half out on the end of that sand spit how would it react to a request to plan travel from that point back to my friends house? I think it would react in fashion very similar to the way the Tesla system does. I've modified the graph from the previous plot to, I hope, make it more clear what the Tesla system does. In general, the philosophy in prediction is to use the best information you have about the future to propagate further into the future.

The numbers on the graph are not the same as the numbers I have been discussing for my past "Ichthyology Conferences" (Wives Program: Wives stay home), but the situation is the same if you consider the guy at the end of the sand spit to be at the break point in the heavy red curve i.e. the origin of this problem is 25 miles in to the graph. At this point in the graph the truck has 82.9% SoC. There is no way for the truck to know what value for the rolling resistance coefficient at this point as it's not on a road. Thus it has no speed limit upon which to base adjustments for drag. It does have access to a data base of elevations so it can adjust for that and does so. It does not adjust for wind because it has no wind information. So it uses the rated consumption, adjusted for elevation change, to project to destination estimating that the consumption will follow the dashed line labeled "Forecast at departure". This shows arrival at 65% SoC which is a lousy estimate but that really doesn't matter because as soon as the driver starts to move the vehicle starts to measure consumption and recognizing that recent history is a better predictor of future consumption than rated consumption updates the projections with that. For the particular smoothing algorithm I chose for the picture the algorithm concludes that charge at destination will be 38% and by the time the driver has gone a mile it thinks the consumption is such that the rest of the trip will look like the thin dashed blue line labeled "Forecast at 26 mi." At 2.1 miles in it's forecast is as shown by the line labeled "Forecast at 27.1 mi" which line predicts that the battery will be empty (75 - 25) = 50 miles in to the trip and the destination cannot be reached. What the driver sees is the forecast line slope increasing in the negative direction and this tells him that consumption has zoomed. Eventually the rate of decrease slows and the slope stabilizes so that the forecasts for 30 and 35 miles are pretty close together. What this means is that if conditions persist he is going to be out of gas at around 60 miles (graph miles equal to 60 -25 = 35 miles into his trip). But, in the example trip of the graph, they don't. Our man is assumed back on the bitumen after 10 miles of sand/loose dirt. The algorithm detects the decrease in consumption, blends the new estimates in with the old and makes new predictions as shown by the "Forecast at 38.2 mi." And "Forecast at 40 mi." lines. The red meatball migrates towards the final 48.2% SoC number and is pretty close, in this example, to it for the last 60 miles of the trip.

This algorithm is, as near as I can tell, what Tesla uses. I base this conclusion from study of the graph (see No. 31) presented by the car as I drive. I have concluded that it is very well able to keep the driver aware of this "fuel condition" from early on in the trip. It definitely adjusts for elevation and speed. It does not, adjust for weather but, as this example shows, does correct for both if they arise during a trip. Why doesn't it adjust its predictions for these two factors? First because it has no information from which it can and second because it doesn't need to.

Where this scheme is if he plans to arrive with 10%, is at 12.5% 10 miles from destination (at a quarter percent/mile the last 10 miles would only take 2.5%) and finds sand over the last 10 miles. This 10 miles could eat 17% of the battery (3 kwh/mi x 10 miles)/180) and he's caught out.

 

[ajdelange]

I'm saving this for a separate post as it probably won't interest most though I found it fascinating. It's in regard to the question of the value of a wind report at a location distal in space and time from the place at which an observation was made.

Wind speed is Weibul distributed. I took a bunch of METARs from IAD and found the reported velocities Weibul distributed with shape parameter 1.1 and scale parameter 5.55.

Wind speed decorelates in both space and time. The time decorelation is slow. It takes about a day for decorellatition down to r = 0.2. After an hour and a half r may be around 0.8. Spatial decorellation is faster with stations only a few km apart showing r values of around 0.6 and the sensitivity of r to separation is small. So given that a wind prediction for some spot along a freeway is based on a somewhat aged report from a station some miles away I thought I'd look at prediction error based on r = 0.5.

This picture shows pairs of Weibul variables with the same statistics as the IAD correlated with r = 0.5, For each point the error in using a data point measured at IAD at some time to estimates consumption at a later time and other place, assuming the correlation is 0.5, is the difference between the U2 and U1 coordinates. Thus points that lie on the 45° diagonal represent 0 error but for that to happen is rare. If IAD reports U1 = 10 mph the wind speed (U2) observed an hour later and 15 km away (or whatever combination of time and distance) can be anything from calm up to 50 mph (though that would be very rare).

The magnitude of (U2 - U1) is another Weibul random variable (but with different parameters). The complementary cumulative distribution function for that distribution is shown in the following which shows that the error is greater than 2.56 (the median) 50% percent of the time (the average error is 3.56 mph. Given that the average wind speed for this ensemble is 5.36 mph estimates that assume the wind speed here is the same as it was there some time ago are pretty rotten estimates. That's why Tesla doesn't use them and that's why Rivian won't either.

It is, of course, fair to ask what happens if you have fresh (1 hr or less) data from a station at your location. Assuming this to imply correlation of r = 0.9 we find the median error to be 1 mph and the average 1.6 representing, respectively, 20 and 31% of the average wind speed.

 

[Riventures]

This so depressing, seriously. And other sources also confirm your calculations within 5-6% variance, not to mention the new 4Xe video of TFL

 

[ajdelange]

I don't think people should be depressed. I think they should be realistic. I think if someone wants to do the kind of stuff depicted in the video in No. 45 he should buy one of those vehicles, do the after market mods necessary to handle that kind of terrain and have at it. I don't think the R1T is the truck for that sort of adventure. Even so, note that the trail those guys drove on was only 3..1 miles long. Note that in my post where I spoke of driving on sand the distance involved was only about 3 miles. When I drove out to my hunting "lodge" (tar paper shack) in WVa it was mostly paved and only a mile or so on dirt and only a mile on rough track. Isn't this going to be the case for most of the adventures contemplated? So I'd say think about what you'd like to do and consider how much of the driving will be at 3 - 6 kWh/mi. Will it be so much that you will not be able to return to the trail head with sufficient SoC to get to a charger? I know that for some the answer will be that it is but I really don't think these vehicles are for applications like Rommel's Afrika Corps.

A very important aspect of driving a BEV is that you quickly learn what the displays shown in No. 31 and discussed in No. 47 can tell you. The piece of beach you may be thinking of driving on may not have C as bad as 0.3 and you favorite logging trail may not have C as big as 0.1 after all. But we do have to allow that they could be that bad or worse. The point is that you'll find out as soon as you drive them. I assume in all this that the Rivian will have a display like this one. It almost has to IMO. It will certainly have some way of monitoring battery consumption rate.

 

[Riventures]

@ajdelange

I understand the points you are making, but even in my most simplistic use case scenario, things may not be looking very good. I certainly won't be doing any of those "off roading" shown on TFL on a regular basis.

Let's take the most simplistic use case for me:

Day trip to Assateague Beach. It is 170 mi. from our home to the point of entrance of the beach (38.189400, 75.159024). Starting on the MD side driving through some of the most beautiful beaches towards VA side is about 27 miles on sand. We would stop have some dinner, enjoy the sound of the ocean and breeze, take a dip, and get out of the beach from VA side. The nearest charge point is 52 mi. away. Thankfully, it is a 60kWh station with two chargers. From there to home again is another 170 mi.

So, with max pack, let's assume we'll have 400 mi. range, let's exclude payload changes for now.

• 400 - 170 = 230 mi left by the time we get to the beach.
• Let's assume 27 mi. beach driving will consume 70% of that charge, which will leave us with 69 mi of range.
• If we get to the supercharger, we still should have some juice left worth about 15 mi.
• To have enough power left to get home, we would have to spend 155/60 = 3 hours at the charger, assuming it is available and someone else is not using it.

I really am hoping you will point out all my mistakes and bad assumptions and show how my numbers are way off, I sincerely do. If I am right with the very rudementary and rough calculations, this little "beach" excursion would turn into a misadventure for the family. Driving 3.5 hours to get to the destination, about 3-4 hours to enjoy the beach, 3 hours at the charge station waiting, and another 3.5 hours to get home (assume easy bridge traffic), won't fly for any of the family members

 

[ajdelange]

You do have a challenging mission for sure. You want to do a run on sand that will probably gobble half your battery and you want to do it in that charging desert known as the Delmarva.

My approach would be to get on the island with 90% SoC. I assume that this will be just south of OC and there are chargers there - just not very impressive ones. If it's 27 miles to Chincoteague i would get off there with 40% remaining and then you proceed to Salisbury on another 16% arriving there with 24%. Then recharge there just enough to get to one of the big gun chargers at Annapolis.

Thus it seems the big downer is charge time on the Delmarva with its 50 kW chargers. I expect this to change with time. The first bit of good news might come if Tesla goes through with its plan to open SC to Rivian and other CCS vehicles. This might even happen pretty soon. The Delmarva isn't exactly covered with SC either but there is one at OC.

 

[Riventures]

You are right on point. If Tesla does really open their chargers, that would be awesome as there at least three in that area.

thank you for commenting.

 

[Inthedmv]

You are right on point. If Tesla does really open their chargers, that would be awesome as there at least three in that area.

thank you for commenting.

I live in Montgomery County, MD and drive my Tesla M3 to the eastern shore often. One thing I tell people about driving an EV is that you have to get out of the mindset of driving until empty and then filling the tank. You need to have situational awareness of what your charging options are, and you will get a feel for it once you have driven the vehicle for awhile. You also need to consider not only how much charge you need to get somewhere, but how you are going to get back where you came from (or to your next location).

I understand the concern about driving on sand and the power consumption that will require, and that there may not yet be Tesla Supercharger-level recharging options on the RT 50 corridor to OC/Assateague. If you decide to go with the 300+ range RT1, my recommendation would be to drive to the eastern shore, hit one of the level 2 chargers between Queenstown and Salisbury for a bit to get a bit more range, then do the same thing on the flip side. That will minimize your charging time. Plugshare is a good site for identifying charging options.

I also thought I read somewhere that Rivian will be putting one of their RAN stations on Kent Island. There is a Supercharger at the RoFo, but there is also a WaWa that would be a good charger location. Fingers crossed and good luck with whichever option you go with...

 

[ajdelange]

You need to have situational awareness of what your charging options are, and you will get a feel for it once you have driven the vehicle for awhile.

You do indeed and that is the thing that frightens me the most when I think about the future of BEV adoption in the USA. You know and I know that there are lots and lots of people out there that will never get that feel. Until and unless charging opportunities are as plentiful as gassing up opportunities I fear we won't get there.

 

[Inthedmv]

You do indeed and that is the thing that frightens me the most when I think about the future of BEV adoption in the USA. You know and I know that there are lots and lots of people out there that will never get that feel. Until and unless charging opportunities are as plentiful as gassing up opportunities I fear we won't get there.

Last week I just drove to NYC and back (with a detour in the Philly area on the way back) and a quick glance at the supercharger map told me I didn't really have to plan my charging because of the number of superchargers between DC and NYC. That is the exception and not the rule nationwide, but we are making progress. Two years ago I drove to Florida and you had to plan, six months ago I drove to Florida and back and there were more options. We will get there...

 

[DuckTruck]

I'm saving this for a separate post as it probably won't interest most though I found it fascinating. It's in regard to the question of the value of a wind report at a location distal in space and time from the place at which an observation was made.

Wind speed is Weibul distributed. I took a bunch of METARs from IAD and found the reported velocities Weibul distributed with shape parameter 1.1 and scale parameter 5.55.

Wind speed decorelates in both space and time. The time decorelation is slow. It takes about a day for decorellatition down to r = 0.2. After an hour and a half r may be around 0.8. Spatial decorellation is faster with stations only a few km apart showing r values of around 0.6 and the sensitivity of r to separation is small. So given that a wind prediction for some spot along a freeway is based on a somewhat aged report from a station some miles away I thought I'd look at prediction error based on r = 0.5.

This picture shows pairs of Weibul variables with the same statistics as the IAD correlated with r = 0.5, For each point the error in using a data point measured at IAD at some time to estimates consumption at a later time and other place, assuming the correlation is 0.5, is the difference between the U2 and U1 coordinates. Thus points that lie on the 45° diagonal represent 0 error but for that to happen is rare. If IAD reports U1 = 10 mph the wind speed (U2) observed an hour later and 15 km away (or whatever combination of time and distance) can be anything from calm up to 50 mph (though that would be very rare).

The magnitude of (U2 - U1) is another Weibul random variable (but with different parameters). The complementary cumulative distribution function for that distribution is shown in the following which shows that the error is greater than 2.56 (the median) 50% percent of the time (the average error is 3.56 mph. Given that the average wind speed for this ensemble is 5.36 mph estimates that assume the wind speed here is the same as it was there some time ago are pretty rotten estimates. That's why Tesla doesn't use them and that's why Rivian won't either.

It is, of course, fair to ask what happens if you have fresh (1 hr or less) data from a station at your location. Assuming this to imply correlation of r = 0.9 we find the median error to be 1 mph and the average 1.6 representing, respectively, 20 and 31% of the average wind speed.

Clearly, once Rivians start rolling down America's highways, byways, and trails, Las Vegas will sit up and take notice.

I believe the major casinos' actuarials and gambling experts will see the profit potential in developing platforms for us to place bets related to our Rivians, our behaviors, and our results.

After reading this thread and many, many others, I'm confident in betting the farm on the following wagering lines:

RUNNING OUT OF CHARGE in first 100,000 miles:

AJDelange: Over/Under: 0.5 times

DuckTruck: Over/Under: 12.5 times

I'm comfortable betting my pension, my 401k, and my Roth IRA on the Under line for AJ.

While Vegas likely will not let me bet on my own results (see the "Pete Rose still doesn't get it!" story from a few decades back), if someone wants to place the Over bet on me, we can discuss the appropriate split beforehand. I know that number is attainable! I'll just need to mortgage the house and cars first, as well as sell my Taylor Swift doll collection. It includes neutered versions of several of her famous ex-boyfriends, as well as all three of her adorable cats, Meredith Grey, Olivia Benson and Benjamin Button! Some lucky person can have it all (well, except for the neutered parts). The price (like me) is high, but negotiable.

Any takers for the Over bet on me, or for the TSwift collection, please I.M. me at either @shakeitoff or #youneedtocalmdown

This could be more profitable than the upcoming sure-to-be-a-big-hit IPO!

Don't miss the boat by missing the bet!

 

[SeaGeo]

I'm saving this for a separate post as it probably won't interest most though I found it fascinating. It's in regard to the question of the value of a wind report at a location distal in space and time from the place at which an observation was made.

Wind speed is Weibul distributed. I took a bunch of METARs from IAD and found the reported velocities Weibul distributed with shape parameter 1.1 and scale parameter 5.55.

Wind speed decorelates in both space and time. The time decorelation is slow. It takes about a day for decorellatition down to r = 0.2. After an hour and a half r may be around 0.8. Spatial decorellation is faster with stations only a few km apart showing r values of around 0.6 and the sensitivity of r to separation is small. So given that a wind prediction for some spot along a freeway is based on a somewhat aged report from a station some miles away I thought I'd look at prediction error based on r = 0.5.

This picture shows pairs of Weibul variables with the same statistics as the IAD correlated with r = 0.5, For each point the error in using a data point measured at IAD at some time to estimates consumption at a later time and other place, assuming the correlation is 0.5, is the difference between the U2 and U1 coordinates. Thus points that lie on the 45° diagonal represent 0 error but for that to happen is rare. If IAD reports U1 = 10 mph the wind speed (U2) observed an hour later and 15 km away (or whatever combination of time and distance) can be anything from calm up to 50 mph (though that would be very rare).

The magnitude of (U2 - U1) is another Weibul random variable (but with different parameters). The complementary cumulative distribution function for that distribution is shown in the following which shows that the error is greater than 2.56 (the median) 50% percent of the time (the average error is 3.56 mph. Given that the average wind speed for this ensemble is 5.36 mph estimates that assume the wind speed here is the same as it was there some time ago are pretty rotten estimates. That's why Tesla doesn't use them and that's why Rivian won't either.

It is, of course, fair to ask what happens if you have fresh (1 hr or less) data from a station at your location. Assuming this to imply correlation of r = 0.9 we find the median error to be 1 mph and the average 1.6 representing, respectively, 20 and 31% of the average wind speed.

Did you happen to look at this with directionality included?

 

[ajdelange]

No. As you can probably guess the correlation data I found for wind comes from the wind generation industry. A wind turbine is always headed into the wind so what is important to them is the correlation of the magnitudes. A car going down a highway must head the way the road goes. This means we'd have to analyze the correlations between wind components in a particular direction. That adds another degree of freedom (to the wind) resulting in even weaker correlation. Correlation is small enough in the more optimistic case to convince me that you can't practically project even wind magnitude from one place to another let alone a particular component of wind speed.

 

[emoore]

Last week I just drove to NYC and back (with a detour in the Philly area on the way back) and a quick glance at the supercharger map told me I didn't really have to plan my charging because of the number of superchargers between DC and NYC. That is the exception and not the rule nationwide, but we are making progress. Two years ago I drove to Florida and you had to plan, six months ago I drove to Florida and back and there were more options. We will get there...

I predict in 10 years that the charging and gas station anxiety will flip. It will start to be more and more difficult to drive across the county with an ICE vehicle.

Sponsored