acuoci / opensmokepp_surrogateoptimizer

Optimizer for fuel surrogates

Introduction

This Matlab package implements an optimization algorithm to find the best fuel surrogate matching a set of experimental properties, such as molecular weight, H/C ratio, cetane number, TSI (Threshold Soot Index), and distillation curves.

The methodology is inspired by the following papers:

• K. Narayanaswamy, H. Pitsch, P. Pepiot, Combustion and Flame, 165, p. 288-309 (2016)
• K. Narayanaswamy, P. Pepiot, Combustion Theory and Modelling, 22(5), p. 883-897 (2018)

It is written purely in Matlab language. It is self-contained. There is no external dependency.

Note: this package requires Matlab R2016b or latter, since it utilizes a new series of Matlab features.

Installation

1. Download the package to a local folder by running:

git clone https://github.com/acuoci/OpenSMOKEpp_SurrogateOptimizer.git

2. Run Matlab and navigate to the OpenSMOKEpp_SurrogateOptimizer folder, then run the SurrogateOptimizer.m script.

SurrogateOptimizer

FeedBack

If you find any bug or have any suggestion, please do file issues. We are graceful for any feedback and will do my best to improve this package.

Contributing

1. Fork repository
2. Create a new branch, e.g. checkout -b my-stuff
3. Commit and push your changes to that branch
4. Make sure that the test works (!) (see known errors)
5. Create a pull request
6. I accept your pull request

Don't bunch up changes, e.g. if you have bug-fixes, new features and style changes, rather make 3 seperate pull requests. Ensure that you introduce tests/examples for any new functionality

License

Released under GPLv2 license.

Contact

alberto.cuoci at polimi dot it

opensmokepp_surrogateoptimizer's Issues

Adding internal checks about consistency of input data

Internal checks about the consistency of data provided by the user have to be applied.

In particular, in case a genetic algorithm is considered and a list of species is provided, it is important to check if a proper number of species per family is present and consistent with the constraints provided by the user.

Prevent NaN to break fsolve

Native test SurrogateOptimizer.m failing on Matlab R2024a (Mac Sonoma 14.3 M2 Pro) due to NaN returned by objective function @balance in Flash.m

Extending database: viscosity

The viscosity should be added to the database and the objective function in the optimization procedure should be modified in order to accomodate the viscosity.

Fixing database: cetane number and TSI (Threshold Soot Index)

The cetane number and the TSI currently available in the xml database are not correct. They were included just for development purpose.

Input from external ASCII file

Instead of writing the input data directly into the main .m file, it is more appropriate to move everything on external input file (ASCII)
In a preliminary phase, a rigid structure will be adopted.

Reorganizing component families

It would be useful to remove the family field from the database and leave the user to associate an arbitrary family to each individual component in order to increase the flexibility