13 July 2017: "T-bone" release (RETRACTED!)
About the T-bone release:
This release is the latest stable CO2MPAS model release and the first to take place after the recent publication of the correlation legislation . This release will become official by the 1st of September 2017. No additional updates of the physical model and the optimizer of CO2MPAS are foreseen from now until 1/9/2017 unless it is deemed necessary due to critical bugs that might be reported. This release can be used for Dicing purposes as of 27 July 2017 as explained in the following document released by DG Clima .
Certain important updates have taken place. We kindly invite you to read carefully the report, changelog and known limitations regarding this release.
Please report any bugs or issues you encounter while using this release preferably in the GIThub platform or the official CO2MPAS mailbox ([email protected]).
Model-changes
- Speedup the model avoiding certain unnecessary identifications during the prediction phase.
- Resolved stability issue (very small changes in bag CO2 values leading to disproportional changes in NEDC predictions under certain conditions).
Thermal model
- Add a filter to remove invalid temperature derivatives (i.e., abs(DT) >= 0.7) during the cold phase.
Clutch model
- Some extra RPM (peaks) have been verified before the engine’s stops. This problem has been resolved filtering out clutch_delta > 0 when acc < 0 and adding a features selection in the calibration of the model.
Engine model
- The auxiliaries_torque_losses are function of engine_capacity.
CO2 model
-
Implement a fuzzy rescaling function to improve the stability of the model when rounding the WLTP bag values.
-
bugfix: in tau_function when a hot cycle is given.
-
Set co2_params limits to avoid unfeasible results.
-
Rewrite of calibrate_co2_params function.
-
Use the identified_co2_params as initial guess of the calibrate_co2_params. Update CO2 optimizer enabling all steps in the identification and disabling the first two steps in the calibration. Optimize the parameters that define the gearbox, torque, and power losses.
IO & Data:
-
bugfix: excel validation formulas on input-template & demos did not accept vehicle-family-id with single-digit TA-ids.
-
Add cycle scores to output template.
-
Add model scores to summary file.
-
Report co2_rescaling_scores to the output file and the summary file.
-
Disable input-file caching by default (renamed option --override-cache --> use-cache
-
Known Limitations: the CO2MPAS modelgraph command cannot plot flow-diagrams if Internet Explorer (IE) is the default browser. Copy the URL and paste it in another browser to visualize the diagrams.
GUI
-
feature: co2gui command does not block, and stores logs in temporary-file. It launches this file in a text-editor in case of failures.
-
feature: remember position and size of the window between launches (stored in persistent JSON file).
All-in-One
-
Detect 32bit Windows early, and notify user with an error-popup.
-
Possible to extract archive into path with SPACES (not recommended though).
-
Switched from Cygwin–>MSYS2 for the POSIX layer, for better support in Windows paths, and package manager update manager. Size increased from ~350MB –> ~530MB.
-
Use [AIO] to refer to the ALLINONE base-folder in the documentation.
The Dice
Please read carefully and pay particular attention to the guidelines provided on how to set up and use The Dice. function.
-
Dice commands won’t start without the configuration files; remember to transfer your configurations from your old AIO or request them to the JRC if it is your first use.
-
enhance handling of email encodings on send and receive
-
add --subject and --on options for search criteria on the tstamp recv and project trecv subcommands.
-
bugfix: BCC-addresses were treated as CCs; --raw output was corrupted.
-
new feature: possible to modify selectively logging output with ~/logconf.yaml file; Having more detailed logs generally improve error handling.
-
Known Limitations
Microsoft Outlook Servers are known to corrupt the dice-emails; depending on the version and the configurations, most of the times they can be fixed. If not, as a last resort, another email-account may be used, either to receive the tstamps (added in the tstamp_recipient param), or to send the dice-email in the first place.
A permanent solution to the problem will be provided when the the Exchange Web Services (EWS) protocol is implemented in CO2MPAS.
On Yahoo servers, the TstampReceiver.subject_prefix param must not contain any brackets ([]).
Using GMail accounts to sent Dice may not(!) receive the reply-back “Proof of Posting” reply (or it may delay up to 4 hours). It is recommended to have a 2nd email address in the tstamp_recipients to receive the dice-tstamp email within minutes.
Additionally, Google’s security provisions for some countries may be too strict to allow SMTP/IMAP. In all cases, you need to enable allow less secure apps to access your account.
Datasync
-
Datasync was producing 0 values in the first and/or in the last cells. This has been fixed extending the given signal with the first and last values.
-
bugfix: remove buggy interpolation methods.
Kindest Regards,
CO2MPAS-Team