/sim_battery_system

Primary LanguageMATLABGNU Lesser General Public License v3.0LGPL-3.0

sim_battery_system

A simulation framework for lithium-ion battery systems.

Developed at the Institute of Automotive Technology, Technical University of Munich.

Contact: Christoph Reiter

Introduction

This is a model for the simulation of lithium-ion battery systems of any number of serial and parallel cells. Everything is set up using parameters, so no changes of the model itself are necessary to adapt to different system architectures.

Further information about the modelling approach and the validation of the model can be found in the following publication:

Reiter, Christoph; Wildfeuer, Leo; Wassiliadis, Nikolaos; Krahl, Thilo; Dirnecker, Johannes; Lienkamp, Markus (2019): A Holistic Approach for Simulation and Evaluation of Electrical and Thermal Loads in Lithium-Ion Battery Systems. In: 2019 Fourteenth International Conference on Ecological Vehicles and Renewable Energies (EVER). Monte-Carlo, Monaco, 08.05.2019 - 10.05.2019. IEEE: IEEE, S. 1–17.

DOI: 10.1109/EVER.2019.8813640

Features

These are the currently planned features of the simulation framework:

  • Electric cell simulation using an ECM from 0 to 4 RC elements
  • State dependent ECM parameters influenced by SOC, temperature and current rate of the cell
  • Battery system simulation of any desired number of cells connected in parallel or serial
  • Individual respresentation of cells inside the battery system --> Monitor individual cell behavior inside a battery pack
  • Simulate passive balancing of cells connected in serial.
  • Automatic assignment of statistical parameter deviations to all cells inside the battery system
  • Simulation of irreversible and reversible cell heat dissipation
  • Simulation of thermal cell state using a lumped mass approach
  • Heat dissipation simulation from cell to its environment
  • Simulation of heat fluxes between cells inside the battery system
  • Simulating the behavior of temperature sensors placed in the battery system
  • Automated logging, plotting and saving of simulation results
  • In-depth analysis of cell load using load spectrum analysis

All the things listed here that are not included as of now will be made available as soon as they are thoroughly tested and streamlined.

Use and Expansion of the Framework

We are very happy is you choose this framework for your projects and research! We are looking forward to hear your feedback and kindly ask you to share bugfixes, improvements and updates in this repository.

Please cite the paper above if you use the framework for your publications!

The model is released under the GNU LESSER GENERAL PUBLIC LICENSE Version 3 (29 June 2007). Please refer to the license file for any further questions about incorporating this framework into your projects.

Requirements

The model was created with MATLAB 2018b. If you want to commit an updated version using another toolbox please give us a short heads-up.

Required Toolboxes:

  • Simulink 9.2
  • DSP Systems Toolbox 9.7
  • Simulink Coder 9.0

The insitute usually follows the 'b' releases of Mathworks, so the framework may get updated to 2019b in the future.

How To Use

The framework in this repository is provided with sample data and should run 'out of the box'. Please note, that all provided parameters must be regarded as arbitrary sample data for testing purposes only and should not be considered as having anything to do with reality. Long story short: If you want to use the framework for your projects you must provide your own data!

Basic steps

Use the following steps to start the simulation:

  • Set simulation_framework\ as MATLAB's current folder
  • Open sim_battery_system.prj --> This adds all relevant paths and checks if all files are where they have to be.
  • Run main_sim_battery_system.m --> The simulation starts.

How to adapt the framework to other battery systems

Generally it is a good idea to go to main_sim_battery_system.m and work your way down since everything the model needs gets called from within this script. Some further details for the case you want to use the "simple" features:

  1. Provide the cell data in parameters\
    • Examples are given in cell_parameters_NCR18650PF_2RC.m and cell_parameters_NCR18650PF_4RC.m
    • Provide a lookup-table for your cell's open circuit voltage
    • Decide how many RC-elements you want to have (currently: 0-4), set unneeded ones to false and provide dummy data for those.
    • Provide the dynamic ECM parameters depending on temperature, SOC and current rate. You need at least two breakpoints for each value. If you don't want a dependency modelled use two arbitrary breakpoints, e.g. 25 °C and 25.01°C. The model doesn't extrapolate to the value will be treated as static.
    • If needed: Provide the thermal cell parameters.
    • If needed: Provide the standard deviations for all cell parameters. No deviation for a parameter? Set the value to 0.
  2. Set up the system parameters in parameters\
    • Sample Data is provided in system_parameters.m
    • Set up the number of cells connected in serial and parallel.
    • Set up the parameters for passive balancing (or disable cell balancing).
    • Set up the system's initial state in system_initialization.m. At least the SOC is needed and - if in consideration - the cells temperatures. This must be set for every individual cell!
    • If you want to use any kind of thermal simulation provide the ambient temperatures around each cell and the heat transfer coefficient in ambient_temperature_parameters.m. The sample implementation uses static ambient temperatures, the model also accepts dynamic inputs. The latter must be provided in the Simulink model.
  3. Provide the load cycle in load_cycles\
    • The variable must be named load_cycle
    • The structure is composed of column vectors [time [s], voltage [V], current [A]]
    • The voltage is just for comparison purposes, only the current is used as simulation input
    • The example implementation uses load data from a single cell and scales that depending on the number of serial and parallel cells. Of course you can directly provide data for your system. In this case edit the scaling in main_sim_battery_system.m
  4. Select the things that are logged in the Simulink model sim_battery_system.slx in the subsystem Monitoring and Logging
    • Go to the block Monitoring and Logging and comment in all signals you need
    • Go to the block Load Spectrum Analysis/Classification and do the same for load spectrum analysis. Further settings for this feature can be found in parameters\LSA_parameters.m
  5. Set up the simulation in main_sim_battery_system.m
    • Choose the cell and system parameters and the load cycle
    • Choose the simulation step-size and the max. simulation time
    • Choose if you want plot or save the results or globally disable logging and load spectrum analysis
    • Select the types of thermal simulation you want to use. If you want to simulate the heat transfer between cells and/or temperature sensors you need to provide further information. This is explained in the script.
  6. Start the simulation
    • If you chose to save your plots and/or simulation results you can find those file in the folder simulation_results\ after the simulation.

About the implementation of the state-of-charge

The model does not limit battery SOC, so it is possible to get SOCs >1 and <0 in the simulation. This is done deliberately to make errors in charge/discharge control detectable that would result in deep-discharging or overcharging the battery system.

Some information about the thermal simulation

The model supports two types of thermal simulation. The first type disregards any thermal interaction with the environment and just simulates heat transfer for each cell to its environment, defined by the temperature gradient between cell and its surroundings and a heat transfer coefficient. The second option works in addition to the first and also simulates the heat fluxes between the cells. Heat fluxes in all spacial direction are supported, so for this case it is not sufficient to just provide the number of cells in serial and parallel. The model also needs information about how the cells are oriented. This is done by defining the number of layers the cells are placed in. Refer to geometric_system_parameters.m for further information.

Currently the framework assumes constant ambient temperatures and heat transfer coefficients for each cell. Those can also be dynamic and therefore be used for simulation of a battery thermal management system (BTMS). This is implemented in sim_BTMS building upon this simulation framework. Refer to the linked repo for further information.

Roadmap

This model was created for my dissertation. Here is a (incomplete) list of ideas for features that I messed around with but that in the end were out of scope of my dissertation. Therefore they aren't properly tested, the code is a mess and therefore is not included in this release. If you are interested in those features send me an E-Mail and I will happily share everything I have regarding these features. --- Christoph

Features:

  • Assign individual parameter sets to cells: This was used for the validation of the parallel connection model. The implementation was done based on 4D-lookup-tables of the parameter sets which leads to massive memory demands and some adaptions to the simulation models.
  • Optimization method for determining the thermal system parameters: It is not straightforward to adapt the thermal simulation to an existing system, because of the complex thermal interaction between the cells and with the environment. The idea was to induce thermal gradients to a battery system in a real measurement setup and then tune the simulation parameters using an optimization algorithm. First results have been promising, but we only used this for a very simple system.
  • Improved handling of temperature sensor placement: As of now every temperature sensor has to be assigned by hand. We build a working prototype allowing many different possibilities of automatic sensor placement. This includes the uniform distribution of sensors in the battery system, automatic setting of a measuring range for each sensor and the automatic placement of sensors in hot-spots in the battery system.

Authors and Maintainers

  • Christoph Reiter
    • Idea, structure, underlying concepts and algorithms except where noted otherwise.
    • Research of the theoretical background.
    • Supervision of the contributing student's thesises.
    • Final revision and benchmarking.
  • Leo Wildfeuer
    • Cell measurements and parameter fitting of the 4 RC ECM.

Contributions

The authors want to thank the following persons, who, whith their brilliant contributions made the framework possible.

  • Johannes Dirnecker
    • Initial implementation in Simulink and MATLAB and development of the matrix-based parallel circuit simulation as part of his term project.
  • André Thomaser
    • Initial implementation of the submodel simulating heat fluxes between cells as part of his bachelor's thesis.
  • Thilo Krahl
    • Initial implementation and development of the submodel for load spectrum analysis as part of his master's thesis.
  • Sebastian Huber
    • Initial implementation of the submodel for the simulation of temperature sensors as part of his term project.
  • Thilo Wurster
    • Initial implementation of the method for assigning individual parameter sets to the cells as part of his master's thesis.
  • Michael Baumann
    • Providing advice for cell modelling and cell measurements and parameter fitting of the 2 RC ECM.
    • Being an awesome colleague and office mate at the insitute.