bryanwweber/cansen

SENKIN-like wrapper for Cantera

cansen's Issues

Strange behavior when too many products are specified for eq. Ratio and multi

When --multi is specified and extra products are in the CPROD lines (e.g. CO2 for H2 O2 combustion) the processes are spawned but nothing seems to happen. Cases without this work fine

Use an exception class instead of sys.exit()

Rather than printing the error to the console, properly raise an exception. This should probably be a new exception class relevant to the particular error.

Volume profiles aren't applied properly

It seems that the volume profile specification does not result in the proper pressure and temperature being maintained through a process. This may be due to low (time) resolution of the volume trace, errors in the numerical derivative, or both. Possible fixes include fitting a continuous function and taking the derivative to get the velocity, or using a higher-order numerical derivative.

Add ability to process multiple input cases in serial/parallel

It would be nice to be able to specify several cases to run in a single file and have CanSen read them. These cases would then be run automatically, in parallel if the user desires, since it should be embarrassingly parallel.

Verify that equivalence_ratio functions properly

There may be a bug in the calculation of O atoms. Also need to update how to check that all required elements are in the complete products.

Update documentation for the location of executables

Hi,
I recently had an issue searching for the cansen executable under windows. It is not mentioned where to find it and how to use it. It maybe would be a good idea to inform the ignorent user that it is located in the scripts directory.
Best,
Abhi

Add unit tests

There should be a framework in place to ensure that changes don't break existing functionality. The nature of the testing needs to be decided and added.

Add try-except clauses to the import statements

Check if input file and chemistry files exist in `cli_parser`

Also remove file check in convert_mech

Refactor the problem choice into a new function

Return the reactor, n_vars, wall, and tempFunc

Add setup.py to install to site-packages

Add distutils setup to install as a package. Add entry points so that it can be used as a script

Implement user defined time interval for writes to save file

Decide how to save binary output

In SHA: ee97b3c code was added to save output as a PyTables table instead of a PyTables array. The code in that commit was commented out because of an incompatibility with how the interpolation for printing is done. The following problems were noted in that commit:

1. Interpolating is more complicated (have to read data off disk to interpolate, or not flush the table on every time step)
2. Save file might be larger (save file with the table was 2kB larger than with array)

However, several benefits were noted:

1. Set up of table is simpler than setup of array - doesn't require assigning to temp variable on every timestep
2. Doesn't require numpy stacking and reshaping operations, which may be slow. No evidence has been found of this being true, but it seems logical

With this issue creation, the following additional benefit is noted:
3) If a table is used, the sensitivities can be stored in their own array of their natural shape and would not require reshaping on every time step, plus an additional reshape on reading the data back in.

Furthermore, it may be useful to save additional solution variables, in particular netw.n_vars, because of the possibility of reshaping the output. These variables don't have a natural place in the array output as it stands. However, a separate table could be created in the output file to store variables such as this. This would not affect the array.

Prevent multiple problem types from overwriting each other

Specifying multiple problem types currently sets the last one to occur. Instead, we should throw an error and inform the user.

Write the command line help

Add interpolation for sensitivity on last time step

Interpolation of the sensitivity matrix needs to be done when the last time step does not match the end time. Probably the previous sensitivity will need to read from disk.

Update all documentation for 1.1 release

Decide if its possible to break the simulation when ignition happens with derivative

Associated with #12 for automatic ignition delay calculations, it should be possible to break the simulation when the ignition delay is found. There are several potential problems with a generic peak finding algorithm:

1. There will be noise in the derivative
2. How to distinguish and not break for first-stage ignition
3. How to tune the "sensitivity" of the peak finding algorithm

Need to decide how to handle these problems.

Port code to be compatible with Python 2

Conversion from inp to cti is broken

Add documentation for supported keywords

Implement Other SENKIN reactors

• CONV
• CONP
• VTIM
• VPRO
• ICEN
• COTV
• CONT
• TTIM
• TPRO

Add internal combustion engine model model

Need keywords

• ICEN - Internal combustion engine keyword
• CMPR - compression ratio, default 15
• DEG0 - Starting crank angle, default 180 degrees
• VOLD - Displacement volume
• VOLC - Clearance volume
• LOLR - Ratio of the length of the engine connecting rod to the crank radius, default 3.33
• RPM - Revolutions per minute, default 1500

Move documentation out of main code branch

The documentation folder should not be on the main code branches. Look up a tutorial for Sphinx to figure out how to do this.

Add optional ICEN keywords

Optional ICEN keywords

• DTDEG - Maximum time step in terms of crank angle degrees, default STPT
• NCANG - Run the simulation for x crank angle degrees. Use one of TIME, NCANG, NREV to specify the end time
• NREV - Number of revolutions to run the simulation for. Use one of TIME, NCANG, NREV to specify the end time

Switch to h5py for HDF5 output?

Investigate h5py instead of pytables

Add post-processor examples

Need to add examples of how to read the binary save file.

Package for Conda?

Could this be packaged for conda (please)?

Properly use exceptions

Instead of using sys.exit(1) when an error has occured

Move mechanism conversion out of the SimulationCase

So that multiple SimulationCases can be run in embarrassingly parallel mode.

Temperature profiles are not applied to the reactor

Setting the Solution state does not affect the reactor after the reactor has been initialized

>>> gas = ct.Solution('h2o2.xml')
>>> reac = ct.Reactor(gas)
>>> reac.T
        300.0
>>> gas.TP = 1000, None
>>> gas()
       Report with T = 1000
>>> reac.T
        300.0
>>> gas()
       Report with T = 300

Add enhanced sensitivity analysis

Updated based on comments from #3 :

• Parse additional CanSen specific keywords, tentatively placed after END
• Modify run_case so the proper things are changed based on the keywords

Add ability to specify pressure profile through volume profile

Since specifying a pressure profile directly is not possible in the current Cantera 2.1.1, we can alternatively specify the volume profile that would achieve a given pressure profile. This might be useful for simulating shock tube experiments where the pressure change is a constant function of time (e.g. 1%/ms). This feature should be in addition to specifying the pressure profile directly because once Cantera is fixed, that will be a useful case as well.

Implement new automatic ignition delay calculation methods

Keywords DTIGN and TLIM are implemented. Need to implement

• TIFP - Temperature slope = maximum
• UIGN - User specified function
• KLIM - Species maximum

Add documentation of and within functions

Documentation needs to be added for each function and for the operations within each function.

Rename variables to be consistent with PEP8

should_be_with_underscores not camelCase

Add checking for valid but not implemented keywords

Print warning to user that the keyword will be ignored.

Store CTI file with the HDF output

It would be very useful to store the CTI file in the binary output file, so that post-processing would not rely on having two files present.

Add examples of more complicated uses (TTIM, VTIM) to the online docs

Implement ignition delay reporting

• Decide what to do when ignition happens
• Give users option to break, just report delay and continue
• How to handle printing in the case of ignition?

Inconsistent definitions for velocity profiles

In profiles.py, VolumeProfiles outputs dV/dt as velocity, whereas ICEngineProfile calculates the actual piston velocity dx/dt. I stumbled across this issue looking at PyKED's RCM example. Presumably both classes are thought as plugins for cantera's Wall objects, which uses the wall velocity as input (in m/s). where for the RCM the area has to be set to 1: otherwise, the velocity will not be calculated correctly.

I am not certain to what extent CanSen is under development, but the PyKED example draws from here, and certainly uses incorrect units. Curiously, neither bore diameter nor piston area are stored in the current PyKED format, so it is not an easy fix.

Implement sensitivity

• Parse sensitivity keyword
• Add sensitivity reactions to reactor
• Save sensitivity output