Analysis of Heat Dissipation in Li-Ion Cells & Modules for Modeling of Thermal Runaway
The 3rd International Symposium on Large Lithium Ion Battery Technology and Application May 15,
2007 Long Beach, California
Gi-Heon Kim (firstname.lastname@example.org) Ahmad Pesaran (email@example.com) National Renewable Energy Laboratory Golden, Colorado
Supported by Energy Storage Program Office of FreedomCAR and Vehicle Technologies Office of Energy Efficiency and Renewable Energy U.S. Department of Energy
Motivations for This Study
Li-Ion batteries need to be safe and abuse tolerant for high penetration into hybrid vehicle market. Understanding abuse behavior of Li-Ion batteries and addressing it are part of the DOE/FreedomCAR and USABC Programs. Thermal abuse behavior of Li-Ion batteries are expected to be affected by the local distributions of heat and reacting materials. The impact of cell-internal-structures and spatial variations of temperature and materials is expected to be critical, especially in large format cells. Modeling the multi-dimensional thermal/chemical phenomena in Li-Ion batteries could provide a better understanding of thermal runaway.
Objectives of This Study
To develop 3D Li-Ion battery thermal abuse “reaction” models for cell and module analysis. To understand the mechanisms and interactions between heat transfer and chemical reactions during thermal runaway for Li-Ion cells and modules. To develop a tool and methodology to support the design of abuse-tolerant Li-Ion battery systems for PHEVs/HEVs. To help battery developers accelerate delivery of abusetolerant Li-Ion battery systems in support of the FreedomCAR’s Energy Storage Program.
Thermal Runaway: Which processes could be modeled?
External Abuse Conditions
External Heating Over-Charging Over-Discharging
Causing or Energizing Internal Events or Exothermic Reactions
Thermal Runaway Thermal Runaway
Lithium Plating If Heating-Rate exceeds Dissipation-Rate
High Current Charging Nail penetration Crush External Short Decompositions Internal Short Circuit Electrochemical Reaction
Gas Venting Flames Rapid Disassembly
Approach and Model Capabilities
Reviewed literature for chemical reactions in Li-Ion batteries and consulted with experts (Robert Spotnitz of Battery Design). Incorporated exothermic component reactions commonly accepted. Formulated and implemented exothermic chemistries into the 3D model. Collected physical/chemical cell properties and parameters for the model. Constructed thermal-chemistry coupled models in cells and modules. Performed simulations of oven and localized heating
Capturing thermal paths inside a cell by addressing actual geometries/properties of cell components. Simulating behavior under various thermal conditions by considering local cooling/heating effects. Predicting the thermal runaway propagation through a module.
Not an electrical or electrochemical model.
We reproduced the approach to modeling thermal abuse of lithium-ion cells provided by Hatchard et al. (J. Electrochem. Soc., 148, 2001) Extended it for our three-dimensional cell and module analysis.
Our model compares well with literature model for Oven Heating (155oC) LiCoO2/graphite cell
SEI decomposition Negative-Solvent reaction Positive-Solvent reaction Electrolyte decomposition
NOTE: Lithium metal involved reactions (that are important in overcharge conditions) and combustion reactions were not considered in this version of the model. We plan to include them in the next version.
Hatchard et al
3D Battery Model
Addressing: Effect of non-uniform distributions Effect of thermal/electrical path design inside cells/batteries Effect of localized heating/cooling Effect of geometries; shape and dimensions of cell component
Cell Level Thermal Abuse Analysis
3D Oven Test Simulation
Battery is initially at a normal operating temperature (35oC). Battery is placed in an oven which is preheated at the desired test temperature. Oven temperature is kept constant during test.
? Model with Symmetry Plane
Heat Sources SEI decomposition Negative-solvent reaction Positive-solvent reaction Electrolyte decomposition
Exterior Surface Boundary Condition Natural convection Black-body irradiation Gray surface Conduction
Core Material Cylindrically orthotropic properties (direction dependent) The model was developed based on Finite Volume Method.
The can/case is electrically and thermally connected to one of the terminals.
Oven Test - Lumped
Oven Temperature & Size Impact Onset of Events
18650 Cells 18650 Cells
Thermal Runaway : Occurred
Oval Cell Oval Cell
Vjr: Jelly Roll Volume A/Vjr: Heat Exchange Area per Volume Small cell did not go into thermal runway at 150?C
3D Oven Test
Temperature, Heat P/S N/S
Although oven test is not a highly multidimensional phenomena, it still demonstrates the noticeable spatial distribution especially in large cells.
Temperature, Heat P/S N/S
*D18H65: Diameter of 18 mm, Height of 65 mm
P/S: positive/cathode -solvent N/S: negative/anode-solvent
Simulating Localized Heating
? Model with Symmetry Plane
Hot-Spots Localized energy would be released in a short period of time at an arbitrary small region inside a cell core.
(It may represent internal short circuit!)
Heat Sources SEI decomposition Negative-solvent reaction Positive-solvent reaction Electrolyte decomposition No resistive/Joules heating Core Material Cylindrically Orthotropic Properties
Note The can is electrically/thermally connected to the core at the bottom
Localized Boundary Conditions Each boundary section can have various boundary conditions independently Natural/forced convection Gray-body radiation
Abuse reaction propagation depends on cell internal structures and materials.
D50H90 size cell A certain amount of energy (equivalent to 15% of stored energy in the presented case) is released at small portion of core volume (0.5% of total jelly volume) for 1 sec.
Delay between measuring external temperature and internal event, external sensing may be too late.
Reaction Front Propagation
SEI decomposition reaction completion surface Propagates Initially in azimuthal direction Forms hollow cylinder shape reaction zone Center axis zone starts to react Finally reaction goes further in outer radius cylinder zone
Module Level Analysis of Cell-to-Cell Thermal Runaway Propagation If one cell goes into thermal runaway, will it propagate to other cells and how?
Cell-Cell Propagation in a Module….
….is a result of INTERACTION between the distributed chemical resources and the thermal transport network through a module. Heat Exchange Modeling in a Module
Radiation Heat Transfer: Long range heat transfer through electro-magnetic
wave emission/absorption at cell surfaces.
Conduction Heat Transfer: Thermal diffusion through cell to cell electric
connector (and/or conductive structures).
Convection Heat Transfer: Heat transfer due to bulk motion of heat transfer
Approach to Thermal Runaway Propagation
A Multi-Node Lumped System Model for Module Propagation Analysis
Each cell in a module and ambient (or a container box) were considered as thermally lumped systems; [N+1] nodes were solved. Thermal-chemistry coupled equations are solved at each node.
Venting and convection heat transfer due to the venting were not considered in this study. Model does not consider flames due to venting of the flammable electrolyte. Structural integrity of cells and module are assumed to be intact even at high temperatures.
Eventually all cells went to thermal runaway
Module consists of 20 D50H90 Cells 5 x 4 aligned array spaced by 3 mm Radiation & terminal conduction Assume one cell is gone in to thermal runaway
Note: Lumped approach tends to underestimate the thermal runaway propagation due to the lack of capability addressing local heating. 17 .
with immediate neighboring cells?
Base Case – Explaining the Results: Heat Sharing
Heat flows from the trigger cell to its neighbors through radiation and connector conduction ( neighboring cells went into thermal runaway).
Radiation Heat Conduction Heat
+ : Heat Rejection from trigger cell - : Heat Absorption at trigger cell
To North Cell
To West Cell
From Trigger Cell
To East Cell
To South Cell
Heat Generation Heat Out to Ambient
Base Case – Explaining the Results: Heat
generation and transfer at neighboring cells
Enough heat reached the surrounding cells to trigger their thermal runaway.
Radiation Heat Conduction Heat Abuse Reaction Heat Gen
Impact of Smaller Cell-to-Cell Connector Size
Base Case Base Case Smaller Connector Heat conduction or concentrated heat to neighboring cell is much higher in this case
In this case, less number of cells went into thermal runaway.
Less conduction, higher radiation distributed heat through the entire module
Impact of How Cells are Connected Electrically Connector configurations can be electrically identical, (Configuration) but thermally different.
Base Case (Thermally serial)
Node # Time (min)
Accumulated Heat Reaction Heat Q/Mcp [oC]
It appears that only a few neighboring cells went into thermal runaway in this case .
Impact of Cell Size on Thermal Propagation
Base Case (D50H90)
Node # Accumulated Heat Reaction Heat Time (min)
Heat is more quickly transported in a small cell system, so that the released heat is more evenly shared in a module.
It appears that in a module with small cells, thermal runaway did not propagate.
Small Cell Case (D18H65)
Impact of a Thermally-Conductive Matrix as Medium
Base Case - Air
It appears that the very conductive medium reduced the chance for propagation.
No other cell went into thermal runway.
PCM/Graphite Imbedded Matrix*
* A highly porous graphite structure that is impregnated with phase change material (PCM) S. Al-Halaj, et. al
How Thermally-Conductive Matrix Help?
Heat-flow time integration for first 10 min
Heat Inflow Heat Outflow Radiation Heat Connector Conduction Heat Matrix Conduction Heat Node # [oC] Accumulation = Generation + (In-Out) Q/Mcp PCM/Graphite Matrix Imbedded* Node # Base Case
Accumulated Heat Reaction Heat Node # Node #
High conductivity matrix provides “High-Speed” thermal network among the cells in a module. Matrix conduction (distributing network) dominates heat transfer between the thermal nodes over the connector conduction (concentrating network).
What is the message?
Concentrated Delivery Neighbors
Is Fast or Slow Heat Transfer from “Hot” Cell Good or Bad?
Distributed Delivery Neighborhood
Fast: Bad Slow: Good
Fast: Good Slow: Bad
Example: Thin series connector and Thick parallel connector are good for propagation-resistive design
Which one is preferred for thermal design?
Electrically 6 in Parallel, 4 in Series
Is the answer always true?
Concluding Remarks on Propagation in Module
Thermal runaway propagation is determined by the competition between heat dissipation through thermal network and localized heat generations. Even with a “High Speed” thermal network, if a large amount of heat is released from a localized source that exceed the capacity of “the network”, then the thermal propagation occurs in a greater speed through this “High Speed” network.
Propagation in module is …
A result of INTERACTION between the thermal transport network and the distributed chemical resources through a module.
Li-Ion Reaction chemistry was implemented into a finite volume 3D cell model addressing various design elements.
Simulated “oven test” indicated that cell size (to say precisely, heat transfer area per unit volume) greatly affects thermal behavior of a cell. Simulated a localized heat release, similar to an internal shortcircuit.
Propagation of abuse reaction through a module was simulated.
A complicated balance between heat transfer network and distributed chemical resources. This balance is affected by cell size, configuration and size of cellcell connectors, and cell-cell heat transfer medium.
A feature designed for improved normal operation may or may not be advantageous to prevent cell-to-cell thermal propagation and vise versa.
? ? Improve model by comparing with experimental data from other Labs. Address limitation of the model
– The impact of convective heat transport on module thermal runaway propagation by properly quantifying its contribution – include venting.
Expand the model capability to address various chemistries and materials such as iron phosphate. Investigate internal/external short by incorporating thermally coupled electrochemistry model into the three dimensional cell model. Work with developers on specific cell and module designs.
DOE and FreedomCAR Program Support
? Tien Duong ? Dave Howell
? Robert Spotnitz (Battery Design, LLC)