AIT Austrian Institute of Technology your ingenious partner The Electric Energy Storages Library e-monday am 22. September 2014 H. Kapeller, D. Dvorak
Overview Short Introduction of AIT Austrian Institute of Technology Presentation of the Electric Energy Storages Library 2
AIT Austrian Institute of Technology Owners 50.46% Republic of Austria BMVIT Federal Ministry for Transport, Innovation and Technology 49.54% Federation of Austrian Industries Employees: 900 plus 200-250 on contract basis, thereof 95 PhD students Financial Goal: 30% Cooperative Research, 30% Contract Research, 40% Basic Funding 3
4
Electric Drive Technologies Research Focus and Systemic Approach Validation Prototyping Component design Virtual design Operational strategies Component specification EM Electric Machines PE Power Electronics VT Vehicle Technologies EES Electrical Energy Storages
AIT Austrian Institute of Technology
Outline The Electric Energy Storages Library: Multi-Physical Battery Simulation Introduction The Electric Energy Storages (EES) Library Battery Models Basic Concept & Main Modules Modeling Electric Models / Thermal Models / Parameter Adaption Block Validation Based on Battery Measurements Conclusion and Outlook 7
Introduction Motivation Ongoing activities in the electrification of the conventional vehicle technology offer a great chance by helping to reduce and keep CO 2 emissions low Due to the need for a convenient range for electric vehicles (EV), lithium-ion technologies (Li-ion) are very attractive for such systems: Li-ion provide proper operating characteristics such as high energy density and cycle stability Thus a growing demand in special simulation software for lithium-ion battery cells has been recognized in recent years 8
Introduction The Electric Energy Storages (EES) Library For development, modeling and validation of multiphysical battery models Modelica was used Modelica is a modern, object-oriented programming language which is specialized on modeling complex multiphysical systems All developed battery models are provided in the Electric Energy Storages (EES) library which Allows simulating the operating behavior of lithium-ion batteries on cell and stack level Allows considering mutual influences of electric, thermal and aging effects 9
Introduction The Electric Energy Storages (EES) Library Library for multiphysical simulation of li-ion batteries ElectricEnergyStorages library: Extensive applications Easy configuration Simple usage Main use cases Electric simulation Thermal investigation Function development Aging prediction 10
Introduction The Electric Energy Storages (EES) Library The EES library is aimed for Cell & battery stack design Simulating a wide range of different battery technologies Various practical user applications Considering key aspects like electric, thermal and aging effects which can be evaluated with respect to the operating mode of the battery Structural overview of the EES library (left) & advanced electric cell model in Modelica (right) 11
Battery Models Basic Concept & Main Modules Each battery model consists of the following three main modules Electric model (V) Thermal model (T) Parameter adaption block (P) Basic modular concept of the battery models The electric model (V) is used to map the dynamic material characteristics to an electric equivalent circuit The thermal model (T) is used to consider thermal aspects during operation The parameter adaption block (P) calculates the electric, thermal and aging dependencies of the internal battery parameters Integrated parameter sets: Lithium-Iron-Phosphate (LFP) Nickel-Manganese-Cobalt (NMC) 12
Modeling Electric Model: Simplified & Advanced Models The EES library offers two categories of battery models on cell and stack level Simplified ( static ) and Advanced ( dynamic ) models Static models consist of an electric equivalent circuit with Reduced complexity Advanced models can be used to perform very realistic simulations of the operating behavior of a lithium-ion battery considering Highly complex processes like open circuit voltage hysteresis Double-layer and diffusion effects Self-discharge and aging mechanisms 13
Modeling Electric Model: Simplified Models The StaticResistance Models Simplified electric equivalent circuit consisting of A serial resistance and an open-circuit voltage source Corresponding stack model StaticResistanceScaled available A stack consists of ns serially and np parallel connected cells Instanced values of U Cell, I Cell, R Cell scaled up to the equivalent serial and parallel stack size U Stack, I Stack, R Stack shown in (1), (2) and (3) U SSSSS = U CCCC nn (1) I SSSSS = I CCCC nn (2) R SSSSS = R CCCC nn (3) nn Creating and calculating only one scaled cell instance results in very short simulation time even for stacks with a high number of cells 14
Modeling Electric Model: Advanced Models - The DynamicImpedance Models Advanced electric cell model in Modelica (parameter adaption block: configures electric circuit, cell diagnostics block: calculates unmeasurable values e.g. Soc) Advanced electric cell model Two RC-parallel-circuits in order to consider double-layer and diffusion effects Considering aging mechanisms, self-discharge and an open-circuit voltage hysteresis for charging and discharging Advanced stack model DynamicImpedanceInstanced Each cell is instanced separately Different operating behavior of each individual cell can be analyzed Realistic behavior of a whole battery stack in any serial and parallel connection can be simulated 15
Modeling Thermal Model Electric losses generate heat flow within cells The heat flow can be considered by using the thermal battery models of the EES library The EES library offers scaled prismatic and cylindrical cell models Each cell model can be scaled in its size Independent of the shape each thermal cell model can be used to build up an entire battery module and an entire stack Thermal cell interconnections can be modelled in a flexible way as required from the individual application Thermal models can be developed independently of the electric models 16
Modeling Thermal Model: Prismatic Cell Model Thermal prismatic cell model-concept (top) and implementation in Modelica (bottom) Simplified thermal model with one cell discretization Measurements on prismatic cells show that ohmic losses are concentrated in the center of the cell Hence a thermal capacitance in the middle of the model is used Thermal conductance from the center to each surface by thermal conductivities in three dimensions Anisotropic thermal cell behavior can be modeled Thermal connector on each surface Enables the integration of the cell model into a thermal stack model 17
Modeling Thermal Model: Cylindrical Cell Model Similar to the Prismatic Cell model Thermal conductance between the center of the cell and the surface or the two electrical pins by thermal conductivities in radial and axial direction respectively As in the Prismatic Cell model thermal connectors enable the integration into a thermal stack model Thermal cylindrical cell model-concept (top) and implementation in Modelica (bottom) 18
Modeling Parameter Adaption Block Calculates the dependencies between each parameter value of the electric components and the operational parameters (e.g. Soc, cell temperature) Independent of the chosen model (static/dynamic; cell/stack) Configuration of operational dependencies either via Polynomial functions Lookup tables 19
Modeling Parameter Adaption Block Polynomial Functions A polynomial function as shown in (4) can be parameterized via a vector p, consisting of the polynomial coefficients p = c n, c n 1,, c 2, c 1, c 0 n y = c i x i i=0 (4) Hence constant, linear or any other correlation that can be described via a polynomial function is possible 20
Modeling Parameter Adaption Block Lookup Tables 2 nd option for the user to input operational dependencies of parameters One dependent variable value (e.g. serial resistance, open circuit voltage etc.) for each value of independent variables (battery Soc, temperature) Original measurement data can be used The parameter adaption block includes two additional subblocks Cell diagnostics Aging 21
Modeling Parameter Adaption Block Cell Diagnostics Block Calculation of cell-relevant values that cannot be measured directly, e.g. Soc SSS is a value that defines the remaining electric charge Q BBBBBBB in relation to the maximum cell capacity CC, calculated according to (5) SSS = Q BBBBBBB CC = Q IIII i dd CC (5) Cyyyyy = i 2 CC dd (6) Q IIII initially available charge i current withdrawn from battery 22
Modeling Parameter Adaption Block Aging Block Add-on for parameter adaption Independent variables: Soc, T, Cycles Considers aging effects Cell capacity decrease Resistance increase Calendaric aging algorithm according to (7), based on Arrhenius equation T AAA = A 0 e SSS SSS 0 B e T T 0 C t (7) Cyclic aging modeled via polynomial function 23
Validation Based on Battery Measurements An FTP-72 load cycle is applied to the configured battery model and the real battery in order to compare the simulated and the measured operating behavior Test setup: climate chamber For calibrating and validating the battery model, the battery cell Headway 38120P is cycled with two different load profiles while the cell voltage is measured The first load profile is used to extract the parameters of the electric equivalent circuit which represents the dynamic behavior of the battery The cycle must be applied to the battery over the entire operational range in order to characterize the cell dynamics over the entire Soc Extracted parameters Open circuit voltage Ocv, serial resistance Rs, resistance and capacity of the first RC-parallel-circuit (Rp1, Cd1), resistance and capacity of the second RCparallel-circuit (Rp2, Cd2) 24
Validation Based on Battery Measurements The extracted parameter values are noisy, especially at the upper and lower end of each graph (solid line) Hence a polynomial function of 7 th degree was fit to the extracted data (dotted line) Estimated parameter set for the "Headway 38120P" battery cell 25
Validation Based on Battery Measurements A second load profile, derived from the FTP-72 cycle is scaled to a current of 16 A and applied to the cell A simulation configured with the extracted parameters has been set up applying the same current profile Comparison of the measured (solid) and simulated (dotted) cell voltage Max. voltage difference (about 31 mv) at 400-500 s Simulation Measurement Measured and simulated battery voltage during FTP-72 cycle 26
Validation Based on Battery Measurements The absolute value of the relative voltage difference U rrr between the simulated (U S ) and the measured (U M ) difference is calculated according to (8) U rrr = U S U M U M (8) The accordance of the simulation with the real measurements strongly depends on the quality of the parameter set used to configure the battery model 27
Validation Based on Battery Measurements Maximum deviation about 0.95 % between simulated and measured battery behavior (peak values of U rrr remain below 0.75 %) Realistic operating behavior of the battery model was verified Absolute value of the relative difference between the measured and simulated battery voltage 28
Conclusion and Outlook Multiphysical li-ion battery model Electric, thermal and aging simulation Modular structure, flexibility The EES in projects: Comparison of material characteristics Vehicle 2 grid simulation Aging simulation Simulation of battery stack including cooling circuit Battery management system design Hardware in the loop application Generic parameter set for thermal model Continuous improvement of the EES library 29
Thank you for your attention! AIT Austrian Institute of Technology Mobility Department Electric Drive Technologies hansjoerg.kapeller@ait.ac.at http://www.ait.ac.at