For long-term lithium-ion battery simulation, the Thevenin model is generally the strongest choice. The Rint model is computationally simple but cannot represent transient polarization. The PNGV model captures short-term dynamic behavior well, yet its cumulative capacitance can produce substantial voltage drift during extended cycling. The Thevenin model provides a better balance of dynamic accuracy, numerical stability, and computational cost.
The Rint model is best for simple steady-state analysis, the PNGV model is useful for short dynamic events, and the Thevenin model is usually the most suitable for long-term dynamic simulation and validation.
Why Equivalent Circuit Model Selection Matters
The simulation problem
Lithium-ion battery voltage is not determined only by open-circuit voltage and ohmic resistance. During a current pulse, the terminal voltage also reflects polarization, charge redistribution, and relaxation.
A useful model must therefore reproduce both the immediate voltage drop and the slower transient response without becoming unstable during long simulations.
Why equivalent circuit models are widely used
Equivalent circuit models use voltage sources, resistors, and capacitors to approximate battery behavior. Compared with detailed electrochemical models, they offer lower computational complexity and are easier to implement in battery-management and real-time testing systems.
Their accuracy depends heavily on parameter identification across state of charge, temperature, current level, and battery aging condition.
How the Three Models Represent Battery Dynamics
Rint model: the simplest representation
The Rint model consists primarily of an open-circuit voltage source and a series internal resistance. Its terminal-voltage response is dominated by the instantaneous ohmic voltage drop:
[ V_{\text{terminal}} \approx U_{\text{OCV}} - I R_{\text{int}} ]
This structure is easy to execute and can produce a relatively low average error under straightforward or steady-state conditions.
The limitation of the Rint model
The Rint model has no dynamic polarization branch. Consequently, it cannot accurately reproduce voltage relaxation, transient overvoltage, or the gradual recovery that follows a current pulse.
It is therefore unsuitable when the goal is precision dynamic pulse modeling, such as HPPC testing, power-limit estimation, or high-fidelity transient validation.
Thevenin model: adding transient polarization
The Thevenin model extends the Rint structure with a parallel resistor-capacitor branch. This RC network represents the battery’s dynamic polarization and voltage relaxation behavior.
A typical structure includes:
- An open-circuit voltage source
- A series ohmic resistance
- One or more parallel RC polarization branches
The RC time constant determines how quickly the modeled polarization voltage responds and relaxes:
[ \tau = R_p C_p ]
This allows the model to represent both the immediate ohmic response and slower dynamic voltage behavior.
PNGV model: adding cumulative voltage behavior
The Partnership for a New Generation of Vehicles, or PNGV, model adds a bulk or accumulative capacitance to the Thevenin-like structure. This element is intended to represent changes in open-circuit voltage associated with accumulated load current.
The model can capture short-term polarization effectively because it includes both ohmic and transient components. However, the accumulative capacitance introduces a state that can build voltage over time if its behavior is not accurately bounded or calibrated.
Direct Comparison of Dynamic Performance
Rint: strong simplicity, weak transient fidelity
The Rint model has the lowest computational and parameter-identification burden. It is appropriate for basic energy calculations, preliminary studies, or systems where transient voltage accuracy is not critical.
Its main weakness is structural: no parameter-fitting process can make a single-resistance model fully reproduce dynamic polarization that the circuit does not physically represent.
PNGV: accurate short-term response, long-term drift risk
The PNGV model can provide strong voltage tracking during short pulse sequences. Its additional capacitance helps represent cumulative voltage effects that are absent from the Rint and basic Thevenin structures.
However, the reference reports that voltage error can accumulate during extended dynamic cycling because of voltage buildup on the Cpb parameter. In the cited testing, the error reached as much as 3.87 V, or 6.7% of rated voltage, after six DST cycles.
This makes PNGV performance sensitive to parameter calibration, initial conditions, current integration, and the duration of the simulation.
Thevenin: accurate dynamics without the same cumulative drift
The Thevenin model captures dynamic polarization through its RC network while avoiding the same long-term accumulative-capacitance mechanism associated with PNGV voltage buildup.
Under peak pulse discharge conditions of up to 4 C, the reference reports a maximum voltage error below 1.5 V. More importantly for long simulations, it maintains dynamic accuracy without the observed continuous-cycle error accumulation of the PNGV model.
Why the Thevenin Model Fits Long-Term Simulation
Numerical stability over repeated cycles
Long-term simulations repeatedly integrate current profiles and update internal model states. Any small bias in an accumulative state can become a significant voltage error after many cycles.
The Thevenin model is generally more stable for this use because its polarization states describe transient relaxation rather than continuously accumulating voltage in the same way as the PNGV bulk-capacitance state.
Sufficient dynamic detail
The Thevenin model is more expressive than Rint without requiring the full complexity of an electrochemical model. Its RC branch can reproduce the key behavior needed for many practical simulations:
- Pulse-induced voltage drop
- Polarization buildup
- Voltage relaxation
- Dynamic power response
- Repeated charge-discharge behavior
For applications requiring greater fidelity, the model can be extended with multiple RC branches or other validated enhancements.
Practical computational cost
Thevenin models remain lightweight enough for simulation, parameter sweeps, controller development, and many real-time applications. This makes them a practical middle ground between an overly simple static model and a computationally intensive electrochemical model.
Parameter Identification Determines the Result
Parameters must be measured across operating conditions
No equivalent circuit model is universally accurate with one fixed parameter set. Resistance, capacitance, and open-circuit voltage vary with SOC, temperature, current, and aging.
Parameters should therefore be identified from controlled dynamic tests rather than assumed from nominal datasheet values.
Dynamic testing is essential
Pulse profiles such as HPPC or DST provide the current-voltage responses needed to estimate ohmic resistance and polarization parameters. Least-squares fitting and time-constant optimization can then be used to determine values that reproduce measured voltage behavior.
For the Thevenin model, fitting the RC time constant and polarization resistance is particularly important because these parameters govern the transient response.
Cell manufacturing consistency also matters
Parameter quality depends on the repeatability of the tested cells. Consistent electrode pressing, cell assembly, and test conditions improve the reliability of extracted resistance and polarization values.
A well-structured model cannot compensate for inconsistent experimental data or poorly controlled test conditions.
Understanding the Trade-offs
The Rint model can be too simple
The Rint model is attractive when speed and simplicity dominate. Its limitation is not merely lower precision; it cannot represent the physical pattern of transient polarization and relaxation.
Using it for high-current pulse validation can lead to misleading conclusions about voltage sag and available power.
The PNGV model can accumulate error
The PNGV model’s cumulative capacitance is useful for representing certain voltage changes, but it can also create long-term drift. This is especially problematic when the model is used over many dynamic cycles without appropriate correction, bounding, or recalibration.
Short-term accuracy should therefore not be mistaken for long-term simulation reliability.
Thevenin models still require calibration
The Thevenin model is not automatically accurate simply because it is structurally better. A poorly identified RC time constant, inadequate SOC dependence, or omitted temperature effects can still produce substantial error.
Its advantage is that it offers a stable and effective foundation for calibration and enhancement.
More complexity is not always better
Adding more RC branches may improve fit quality over a specific test profile, but it also increases parameter count and identification effort. Excessive complexity can reduce robustness when the model is applied outside the conditions used for calibration.
The preferred model should be the simplest one that meets the required accuracy over the intended operating range.
Making the Right Choice for Your Goal
Choose the model according to the duration, dynamic intensity, and accuracy requirements of the simulation.
- If your primary focus is computational simplicity or basic steady-state analysis: Use the Rint model, provided that transient polarization and pulse-voltage accuracy are not important.
- If your primary focus is short-duration pulse-response analysis: Consider the PNGV model, but verify its cumulative-voltage behavior and sensitivity to long simulation periods.
- If your primary focus is long-term dynamic simulation and validation: Use a calibrated Thevenin model, preferably with validated enhancements such as additional RC branches when the application requires them.
- If your primary focus is real-time SOC, SOH, or SOP estimation: Start with a parameterized Thevenin model because it offers a practical balance between dynamic fidelity and computational efficiency.
For robust long-term lithium-ion battery simulation, a properly calibrated Thevenin model provides the most dependable balance of accuracy, stability, and implementation practicality.
Summary Table:
| Model | Dynamic Performance | Long-Term Stability | Computational Cost | Best Use Case |
|---|---|---|---|---|
| Rint | Poor transient response; no polarization | Good; no drift but lacks fidelity | Low | Steady-state analysis, simple energy calcs |
| PNGV | Good short-term response; captures polarization | Poor; voltage drift up to 6.7% after cycles | Medium | Short pulse testing (HPPC) with calibration |
| Thevenin | Excellent; balances ohmic and polarization dynamics | Good; minimal drift over time | Medium | Long-term dynamic simulation, BMS/EST |
Enhance your battery simulation with KINTEK precision equipment. Our cell fabrication tools ensure consistent electrode pressing and assembly, yielding reliable data for accurate model parameterization. Whether you're in battery R&D or advanced materials research, our portfolio supports your long-term simulation goals. Contact us today to discuss your needs and elevate your research!