Zhejiang University Team Maps Jet Flow in Exploding EV Batteries
In a groundbreaking study that peels back the curtain on one of the electric vehicle industry’s most feared and least understood phenomena, researchers from Zhejiang University have successfully visualized and quantified the explosive jet flow that occurs when a common lithium-ion battery enters thermal runaway. This isn’t just academic curiosity; it’s a critical step toward engineering safer, more resilient batteries for the millions of electric vehicles hitting the road each year. The team, led by Professor Huang Yuqi and graduate researcher Liu Haodong, didn’t just observe the chaos—they measured it, frame by painstaking frame, under extreme conditions, providing a dataset that could fundamentally alter how battery safety is designed and tested.
The stakes couldn’t be higher. As electric vehicles transition from niche products to mainstream transportation, public confidence hinges on their perceived safety. A single viral video of a battery pack engulfed in flames can do more damage to consumer trust than a dozen positive safety ratings. Thermal runaway, the chain reaction that turns a malfunctioning battery cell into a fireball, is the nightmare scenario. It’s a process that unfolds in milliseconds, hidden inside a sealed metal can, making it incredibly difficult to study. Until now, much of our understanding has been based on computer simulations or post-mortem analysis of charred wreckage. This new research, published in the prestigious journal Chemical Industry and Engineering Progress, offers the first clear, experimental glimpse into the heart of the beast.
The team focused on the “ternary lithium battery,” a chemistry widely used in today’s EVs for its high energy density. While powerful, this chemistry is also more prone to violent thermal events compared to alternatives like LFP (Lithium Iron Phosphate). The researchers’ key insight was that the jet of gas, vapor, and molten debris that erupts from the battery’s safety vent during thermal runaway is not just a symptom of failure—it’s a direct, measurable signal of the internal pressures and temperatures that caused it. By precisely mapping the speed and direction of this jet, you can effectively work backwards to understand the internal conditions of the battery at the moment of failure. This is revolutionary for battery designers, who can now use this data to validate their computer models and optimize the physical structure of the battery cell—its vent size, its internal spacing, its casing strength—to better manage and mitigate these explosive forces.
Capturing this data was no small feat. Imagine trying to film a bullet being fired, but the bullet is made of a chaotic, opaque mixture of superheated gas, boiling liquid, and flying shrapnel, all happening inside a small, violently shaking metal cylinder. Standard high-speed cameras struggle with the blinding glare, the thick smoke, and the sheer speed of the event. The Zhejiang team’s experimental setup was a marvel of precision engineering. They constructed a custom explosion-proof chamber to contain the violent event. A high-powered laser, pulsing at an incredible 3,000 times per second, illuminated the jet stream. A specialized high-speed camera, capable of capturing images with a time interval of just 8 microseconds between frames, recorded the event. This is so fast that in the time it takes a hummingbird to flap its wings once, the camera could capture over 10,000 individual images. The battery itself, a standard 18650 cell (the same type found in many laptops and older EVs), was heated from one side using a copper block fitted with multiple thermocouples to meticulously track its internal temperature rise until the moment of catastrophic failure.
The raw footage, while impressive, was far too noisy and blurry for precise scientific measurement. This is where the team’s second major innovation came into play: a sophisticated, multi-step image processing pipeline. They didn’t just clean up the images; they enhanced them to reveal details invisible to the naked eye. First, they applied a specialized “adaptive Gaussian filter” that intelligently smoothed out random noise without blurring the critical edges of the liquid droplets and solid particles in the jet. Think of it as a smart noise-canceling system for images. Next, they used “histogram equalization” to dramatically boost the contrast, making the faintest droplets stand out starkly against the dark background. Finally, they employed a cutting-edge “cross-correlation algorithm with sub-pixel precision.” This algorithm doesn’t just track large blobs; it can detect the minute movement of individual specks of debris between consecutive frames, calculating their velocity with astonishing accuracy, down to fractions of a pixel. This combination of hardware and software turned a chaotic, blurry explosion into a clear, quantifiable flow field.
What they discovered paints a vivid, dynamic picture of battery failure. The thermal runaway event is not a single, monolithic explosion, but a complex, multi-stage drama. The first act begins the instant the safety valve ruptures. At this critical juncture, internal pressures can exceed 70 atmospheres, launching a high-velocity jet of flammable gas and superheated electrolyte vapor. The team measured initial jet speeds of 70 to 85 meters per second—that’s over 300 kilometers per hour, faster than a Formula 1 car. This initial blast is relatively coherent, with particles streaming out in a unified direction.
But the drama quickly intensifies. Within mere hundreds of microseconds, the second act unfolds. As the jet interacts with the cooler ambient air, the superheated electrolyte droplets begin to violently boil and fragment. Surface tension, air resistance, and internal turbulence cause larger droplets to shatter into a fine mist. The once-unified jet begins to spread out, its velocity dropping as its energy dissipates over a wider area. The researchers observed distinct regions of high-speed flow migrating outward from the vent, while swirling eddies and turbulent zones began to form at the jet’s periphery. In some areas, evaporating droplets created localized density differences, triggering small-scale convective currents.
By the time the event reaches its third act, around 600 to 800 microseconds after venting, the violence begins to subside. The battery’s internal chemical reactions, having consumed much of their volatile fuel, start to wane. The jet velocity drops significantly, and the flow becomes more stable, though still complex. The team even observed the formation of strange, gelatinous residues—partially reacted electrolyte components that had cooled and coagulated mid-air, a haunting signature of the extreme chemistry that had just taken place. The entire explosive sequence, from the first puff of vapor to the final settling of debris, lasted approximately 12,000 microseconds—just 12 milliseconds. Yet within that infinitesimally small window, the researchers captured a wealth of data that tells a story of immense pressure, searing heat, and violent fluid dynamics.
The implications of this research are profound and far-reaching. For battery cell designers, this is a goldmine. They can now see exactly how their vent designs perform under real-world failure conditions. Does the vent open cleanly, or does it fragment and create dangerous shrapnel? Is the jet directed in a way that minimizes damage to neighboring cells in a pack? Does the internal cell structure promote or hinder the buildup of catastrophic pressure? These are no longer theoretical questions; they are empirical ones that can be answered with data.
For engineers developing Battery Management Systems (BMS), this study provides critical early-warning signatures. By understanding the precise sequence of events and the associated pressure curves that lead to jetting, BMS algorithms can be refined to detect the precursors of thermal runaway with greater accuracy and speed, potentially triggering cooling systems or isolation protocols before the point of no return.
For fire safety engineers, the detailed velocity field maps are invaluable. Knowing the exact speed and direction of the jetting flame and debris allows for the design of more effective firewalls and suppression systems within a battery pack. It informs where to place fire retardant nozzles and how to design cell spacing to prevent a single cell’s failure from cascading into a pack-wide inferno.
Even for policymakers and safety regulators, this research provides a concrete, scientific basis for establishing more rigorous safety standards. Instead of relying on pass/fail tests, future regulations could mandate specific performance criteria for jet velocity and direction, pushing the industry toward inherently safer designs.
This work also stands as a significant methodological advancement. Previous studies often relied heavily on computer simulations, which, while powerful, need real-world data for validation. Other experimental attempts struggled with the sheer complexity of the three-phase (gas, liquid, solid) jet, often yielding low-quality, unusable imagery. The Zhejiang team’s image processing pipeline—combining adaptive filtering, contrast enhancement, and sub-pixel tracking—provides a robust, repeatable method that can be adopted by other research labs around the world. It transforms an almost impossible measurement into a routine procedure.
Of course, the researchers are the first to acknowledge the limitations and the path forward. Their study focused on a single type of battery cell under controlled laboratory conditions. Real-world EV batteries come in many shapes, sizes, and chemistries, and they fail under a variety of triggers—mechanical crush, electrical short, or external fire. The next step is to apply this same sophisticated technique to a wider array of battery types and failure modes. The team also suggests using additional sensors, like hot-wire anemometers, to provide point-validation of their optical measurements, further bolstering the dataset’s credibility.
Furthermore, while this study mapped the two-dimensional flow field, the real event is three-dimensional. Future work will likely involve tomographic techniques to build a full 3D model of the jet, providing an even more complete picture. The ultimate goal is to create a comprehensive “failure atlas” for lithium-ion batteries—a detailed catalog of how different cells fail under different conditions, providing an indispensable reference for the entire EV ecosystem.
In the high-stakes race to electrify transportation, safety is not a feature; it’s the foundation. Every innovation that makes batteries more powerful or cheaper is meaningless if it comes at the cost of reliability. The work by Liu Haodong, Zhang Pengfei, and Professor Huang Yuqi at Zhejiang University represents a giant leap in our ability to understand, predict, and ultimately prevent the most dangerous failure mode in an EV battery. By illuminating the darkness inside a failing cell, they are helping to build a future where electric vehicles aren’t just cleaner and more efficient, but fundamentally, unquestionably, safer.
By Liu Haodong, Zhang Pengfei, Huang Yuqi, College of Energy Engineering, Zhejiang University. Published in Chemical Industry and Engineering Progress, 2024, 43(2): 703-712. DOI: 10.16085/j.issn.1000-6613.2023-1363.