search

rl-institut / simbev

MIT License
10 stars 4 forks source link

Fix results ending too early #67

Closed mosc5 closed 2 years ago

mosc5 commented 2 years ago

A more thorough look at outputs is happening in #41

For now. this PR fixes #66 by changing the last event end (park_end or drive_end) to the last possible timestep.

nesnoj commented 2 years ago

Thanks for the quick proposal! Some stats:

==================================================================
=== Corrupted files (end with driving): 4 of 32688
Range of missing timesteps (end with driving): 
drive_start_timesteps: min=35028, max=35040
drive_end_timesteps: min=35039, max=35053
=== Corrupted files (end with parking): 6930 of 32688
Range of missing timesteps (end with parking): 
park_start_timesteps: min=34907, max=35061
park_end_timesteps: min=34982, max=35040
==================================================================

Problem 1 As drive_start_timesteps never exceeds 35040, your fix in #67 should work for trip files that end with driving. However, the consumption is flawed then.

Problem 2 There are numerous EVs where the last event's park_start_timesteps exceeds 35040, example: bev_medium_02203_90kWh_SR_Mitte_events.csv In this file, the previous driving event already ends 35058 which exceeds 35040, so the last event should be removed and the previous one adjusted..

nesnoj commented 2 years ago

Problem 2 has been solved.

Also, about 30% of EVs show park_start >0 in the first event. I dirty-fixed this too..