Part 2 — Modeling Physical Digital Twins: Battery Behavior and Aging with GT‑SUITE and GT‑AutoLion

In Part 1, we explored why solar-charged, battery-powered IoT monitoring devices require reliable prediction of availability, SOH, and RUL, and why telemetry, statistical methods, or machine learning alone are not sufficient. In this second part, we focus on the modeling foundation: how GT‑SUITE and GT‑AutoLion provide the physics‑based structure that allows each digital twin to represent a device’s true operating and aging behavior. Part 3 will show how those models integrate into the customer’s IoT cloud environment for automated daily execution.

Once the necessity of accurate availability forecasting and robust SOH/RUL estimation becomes clear, attention shifts to the underlying models. These models must be detailed enough to reflect real physics, yet efficient enough to operate repeatedly across a large device network for many devices.

Figure 4: GT-SUITE & GT-Autolion model for engineering, software and device validation.

The modeling lifecycle begins during design and validation, when engineers construct a comprehensive system model in GT‑SUITE. This system model includes photovoltaic behavior, MPPT charge control, thermal interactions among enclosure, electronics, and battery, and the device’s consumption profile.

Solar input is modeled using GT‑SUITE’s photovoltaic cell template, which captures how temperature and irradiance reshape the voltage–current curve. Environmental conditions are introduced via a dedicated template capable of querying representative weather data for locations based on their geographical coordinates.

Figure 5: Selected GT-AutoLion aging mechanisms and temperature influence.

At the center of this system model lies the battery, represented through GT‑AutoLion. This model provides a physics‑based description of lithium‑ion operation, including the electrochemical processes that drive both cycle and calendar aging.

Calibration is performed using measured aging data across temperature, SOC, and cycling conditions relevant to the device’s outdoor exposure.

GT-AutoLion’s mechanistic representation captures key degradation processes such as formation, lithium plating, cathode film growth, and active material isolation. This ensures the model remains reliable even when operating conditions move beyond the calibration range.

During the design phase, the full GT‑SUITE + GT-AutoLion model serves as both a predictive tool and a validation instrument. It evaluates expected winter charging behavior, temperature impacts on electronics load, MPPT tracking dynamics, and the combined influence of these factors on long-term degradation. This is where the twin’s physics lineage is established.

When transitioning into operations, the model is distilled into a form optimized for digital twin execution. Elements that are no longer needed, such as detailed solar modeling or enclosure thermal networks are removed, because the device sends the necessary telemetry directly. For an aging digital twin, battery temperature, SOC, current and voltage signals are part of the telemetry data. What remains is the physics core: the electrochemical aging model, the battery’s thermal dependence, and the system interactions necessary to evolve battery state forward in time.

Figure 6: Digital Twin Models: The design model (a) is adapted to used device telemetry data directly for continuous SOH prediction, weather forecast data for SOC prediction (c) or fitted and retrained machine learning models of ambient and device load for RUL prediction (d).

Each day, the twin takes the past 24 hours of telemetry and uses AutoLion to roll the battery’s internal state forward. Aing mechanisms advance according to temperature, SOC, rest periods, and current history. The result is a new estimate of remaining capacity and internal resistance—an updated SOH value that reflects the actual conditions each device has experienced.

While the battery pack in these IoT devices is small, this daily SOH estimation loop creates a continuous, in-use record of performance and durability.

This aligns with the EU Battery Regulation’s Digital Battery Passport concept for industrial and EV batteries ≥2 kWh, where each battery must include lifecycle data such as state of health and usage history, along with static information like identification and chemistry.

Persisting the model version, inputs, and outputs with each run further supports this approach, combining static attributes with continuously updated operational metrics, improving traceability for repurposing/recycling decisions and auditability for authorized stakeholders across the value chain.

Figure 7: Daily SOH estimation for device leverageing GT-Play

From the foundation of a continuously aged battery, forecasting the future becomes straightforward. Statistical models of expected temperature and load patterns can drive the twin forward to estimate Remaining Useful Life. These forecasts provide a distribution of possible outcomes, helping operators plan maintenance and replacement long before degradation becomes operationally limiting.

Part 3 explains how this model—once distilled into individual digital twins—integrates seamlessly into the IoT cloud environment, where it operates as a daily computational service powering device management and decision automation.

To explore how physics-based digital twins can be applied to your systems, connect with our team to see how these models can be integrated into your workflow. Visit our Digital Twin solution pages to learn more about system-level modeling capabilities, and follow us on LinkedIn to stay updated on the latest developments in simulation-driven engineering.