Re: [sumo-dev] Miscalculation of Rotational Kinetic Energy, Curvature Energy Loss, and Auxiliary Energy Demand

Hi,
I just fixed the main issues and I can confirm that the changes in our
tests are below 1%. The only thing with major deviations is the
aggregated emission output (meandata) when used with a different step
length than 1s (a case which had no tests at all until now).

Best regards,
Michael

Am 2020-10-16 09:40, schrieb Jakub Ševčík:

Dear all,
thank you for your comments on the energy model in SUMO. I am just

internally answering my colleague Jan to his questions but think that you
could be interested in some of my observations.

Despite bugs mentioned by Burford, their correction (that is certainly

important in SUMO) will only slightly impacts our energy results for these
reasons:
1. Change of Rotational Energy << Change of Kinetic Energy

since the internal moment of inertial << mass of the vehicle (for example
for our simulated trolleybus J_int = 0.03*mass)


2. The calculation of the Curvature Energy Loss is applicated only 3 times per trolleybus No. 13 line circulation in simulation without surrounding traffic. The reason for such behavior is that the diff of route angle is
not zero only at three time-steps of the simulation from total ca 5k.

3. Fortunately we use the default SUMO length 1 s of the time-step and

therefore, the confusion of power [W] vs energy [Ws] vanishes. (It counts both for the bug in the constat power of aux system calculation and for the order logic of energy/power calculations in SUMO's emission/energy model).

Finally, the paper [ I. Sagaama, A. Kchiche, W. Trojet and F. Kamoun,

"Proposal of More Accurate Energy Model of Electric Vehicle For SUMO] you
mentioned proposes three changes for SUMO model
b. calculation of efficiency as \eta_bat*\eta_inv*\eta_mot

I don't understand the difference between such calculation in comparison with using only one constant eta, that can be precalculated of course using
the above-mentioned formula.

Maybe in our case would be interesting to consider different values for the
efficiency of propulsion using energy from battery

(battery->inverter->motor) and from overhead wire (owhd wire -> inverter -> motor) but again fortunately these values are almost the same in practice.


a. using the proposed modified equation for calculation of the energy of auxiliary drives corresponds to the common approach in SUMO if the ambient
temperature is 20°C
c. the modification of efficiency of regenerative breaking is already
integrated into the SUMO

https://github.com/eclipse/sumo/blob/4358260686c4a331c75bce1b5d7702e3eda5cdde/src/utils/emissions/HelpersEnergy.cpp#L132
.

Best,
Jakub


On Thu, Oct 15, 2020 at 11:55 PM Jan Přikryl <prikryl@xxxxxxxxxx> wrote:


Dear Joerg,

I am just curious: Do you have (or will you have) any comparison with
measurements of real vehicles in real traffic? We are namely using the

original HelpersEnergy subsystem to compute energy flows in the ElecHybrid
device and, to be sure, we have tried to validate the outputs of SUMO

against telemetry that we have from trolleybuses -- and, given the natural
variance of the data that you get from such a vehicle in real traffic

(different saloon occupation and hence different mass, different traffic conditions, different daily temperatures resulting in heavy variations in
auxiliary power consumption and so on) and given that for the modelled
vehicle you get at least three different sets of drag and resistance

parameters (one set provided by the manufacturer of the vehicle, a second set from the provider of the chassis, and the third one from the vehicle
operator), we were quite happy with the results even with the current

implementation (the only change is that we try to limit the acceleration of
a vehicle based on its power limits).

On the other hand, there is a paper from 2018 (
https://ieeexplore.ieee.org/document/8450371) that reports a 23-29%
difference

in energy consumption compared to a test with a real electric vehicle. I
did not read it yet, I just came across it recently.


That leaves me wondering whether our comparison is correct and whether we
should not crosscheck it one more time...

Jan

On Thu, 15 Oct 2020, at 10:23 PM, joerg.schweizer@xxxxxxxx wrote:
> Thanks Michael for putting your hands on this, and sorry it took time to
> find the file.
>
> It has been quite crucial for the project as the determination of the
> energy consumption has been the main goal.
>
> Hope it gets fixed soon.
>
> Thanks again,
>
> Joerg
>
>
> On 15/10/20 21:53, Michael Behrisch wrote:
> > Hi,
> > thanks for the report and sorry for the delay. Unfortunately we have
> > several places in the code which calculate energies. I suppose you are
> > referring to HelpersEnergy::compute
> >
https://github.com/eclipse/sumo/blob/master/src/utils/emissions/HelpersEnergy.cpp#L57
> > ?
> > This function seems to have some problems indeed not only the ones you
> > mention but in total it should return power and not energy because the
> > time scaling is done outside the function when used as an emission
> > model. For the battery device however energy is expected. Sorry for the
> > hassle, it is now tracked here:
https://github.com/eclipse/sumo/issues/7695
> >
> > Thanks again for finding out!
> >
> > Best regards,
> > Michael
> >
> > Am 11.10.20 um 06:40 schrieb Burford Furman:
> >> Hello,
> >>
> >> We are running SUMO ver 1.6.0 under Windows 10, and we just discovered
> >> that the SUMO source code is calculating the Rotational Kinetic
Energy,
> >> the Curvature Energy Loss, and the Constant Auxiliary Energy Demand
> >> incorrectly. The SUMO documentation lists the correct equations, but
for
> >> some reason the source code is not correct in ver 1.6.0. Our findings
> >> are summarized below:
> >>
> >> image.png
> >>
> >> image.png
> >>
> >> Thank you.
> >>
> >> Burford Furman
> >>
> >> --
> >>
************************************************************************************
> >> Burford J. Furman
> >> Professor
> >> Department of Mechanical Engineering
> >> San Jose State University
> >> One Washington Square
> >> San Jose, CA 95192-0087
> >> Phone: (408) 924-3817 Email: Burford.Furman@xxxxxxxx
> >> <mailto:Burford.Furman@xxxxxxxx>
> >> Web page: http://www.sjsu.edu/people/burford.furman/
> >> <http://www.engr.sjsu.edu/bjfurman/>
> >>
************************************************************************************
> >> "We must remember that intelligence is not enough. Intelligence plus
> >> character--that is the goal of true education." - Dr. Martin Luther
King
> >> Jr., 1947
> >>
> >> "/Instrumental or mechanical science
> >> is the noblest and above all others, the most useful.../" - Leonardo
da
> >> Vinci
> >>
> >> _______________________________________________
> >> sumo-dev mailing list
> >> sumo-dev@xxxxxxxxxxx
> >> To unsubscribe from this list, visit
https://www.eclipse.org/mailman/listinfo/sumo-dev
> >>
> >
> _______________________________________________
> sumo-dev mailing list
> sumo-dev@xxxxxxxxxxx
> To unsubscribe from this list, visit
> https://www.eclipse.org/mailman/listinfo/sumo-dev
>

--
Jan Přikryl
prikryl@xxxxxxxxxx





• References: