[Ovmsdev] CHARGEBY / CHARGEAT timing

[Tom Saxton]

Utilities are discovering that eventually there's going to be a big
problem with EVs starting their charging en masse right when the billing
rate drops. The graphs are astounding; in San Diego a huge chunk of the
vehicles participating in the EV Project start charging at exactly
midnight causing a giant spike in the EV utility load. The only reason it
isn't a problem now is because there are so few EVs.

Now the preferred model is to use CHARGEBY instead of CHARGEAT. The
advantage is that the start times get spread out by how much each person
drove that day and even the end times will spread out some because the
charge time prediction is always going to be at least a little bit off.

Eventually, we may get a more sophisticated demand response model where we
get an additional incentive to let the utility handle the CHARGEBY timing
and also be able to start the charge early when there's excess power on
the grid, as from wind turbines putting a bunch of power into the grid.

   Tom

On 10/17/13 11:21 PM, "Wiro Zijlmans" <wirozijlmans at gmail.com> wrote:

>Mark,
>I use the chargeat option as offered by the car itself mostly for
>starting the charge at a time that the electricity rate is lower which is
>typically around 9 pm. To support this a threshold of several hours would
>be useful. The build in software also uses a threshold which is at least
>4 hours. Have never noticed to have gone beyond this threshold so do not
>know what it is. Can try it.
>
>Groet / Regards
>Wiro Zijlmans
>
>> Op 18 okt. 2013 om 03:00 heeft Mark Webb-Johnson
>><mark at webb-johnson.net> het volgende geschreven:
>> 
>> 
>> I'm seeking people's opinions for the ACC CHARGEBY / CHARGEAT timing
>>behaviour.
>> 
>> The problem I have is to decide what algorithm to use for the case that
>>people plugin after the scheduled charge time (or in the case of
>>CHARGEBY, plugging in with a charge predicted to take to long to
>>complete by the scheduled charge time).
>> 
>> For example, say I have CHARGEAT set to 19:00:00. I get home, and
>>plugin, at 19:10:00. What should we do? Schedule the charge for 19:00:00
>>the next day, or charge immediately? The logical choice seems to be
>>charge immediately, but how far do we allow that 'nice' behavior to go?
>>19:10:00? 20:00:00? 23:00:00? etc?
>> 
>> I guess we need some threshold, but what do people think is a sensible
>>value for that threshold?
>> 
>> For CHARGEBY, it is slightly more complicated. Say I plugin at
>>18:00:00, with a CHARGEBY of 21:00:00, but the prediction says 4 hours
>>is required? What to do? What if, for the same case, I plugin at
>>22:00:00?
>> 
>> The current algorithm in ACC does not allow any safety threshold. If
>>CHARGEAT is 19:00:00 and you plugin at 19:01:00, tough luck - the charge
>>will start at 19:00:00 tomorrow. For CHARGEBY, we do check if there is
>>enough time to complete the charge before the scheduled time in the same
>>day - if not, we send a charge alert message and start charging
>>immediately.
>> 
>> What do people think? Remember that this is a small embedded processor
>>without much ram/flash, so we need to keep things simple.
>> 
>> Regards, Mark.
>> 
>> _______________________________________________
>> OvmsDev mailing list
>> OvmsDev at lists.teslaclub.hk
>> http://lists.teslaclub.hk/mailman/listinfo/ovmsdev
>_______________________________________________
>OvmsDev mailing list
>OvmsDev at lists.teslaclub.hk
>http://lists.teslaclub.hk/mailman/listinfo/ovmsdev
>