Simulating Destruction: Building Virtual Earthquakes

EarthquakeSim’s YouTube channel has gone viral by exploring how visual storytelling could help disaster preparedness.
 

By Lauren A. Koenig, Ph.D., Science Writer (@Lauren_A_Koenig)
 

Citation: Koenig, L.A., 2025, Simulating Destruction: Building Virtual Earthquakes, Temblor, http://doi.org/10.32858/temblor.357
 

Growing up in Romania, a country beset by active earthquake zones, Michael Ronnet trained as a classical pianist. But when he experienced his first earthquake at 8 years old, that event instilled a curiosity about seismic forces that he just couldn’t shake.

“I was fascinated by the idea that an earthquake could generate enough energy to move entire buildings, cities, or even countries,” said Ronnet. “Since I was little, I would always have these questions, like, if I lived in a brick building, how would my house shake? Or, what if I was in a classroom? How would the benches move?”

As a child, he even stacked toys on a table and conducted rudimentary experiments, shaking that furniture to see what would happen. As an adult, he realized that his younger self had actually built makeshift shake tables.
 

 

Years after those at-home experiments, Ronnet’s curiosity persisted, but existing resources left him unsatisfied. Scientific papers lacked dynamic visuals, while the video simulations he encountered were too simplistic. By chance, he stumbled upon open-source software he could use to answer his childhood questions. While still pursuing a music career, Ronnet taught himself how to create detailed earthquake simulations and finally found a way to visualize the scenarios that had intrigued him for so long.

Ronnet now has nearly 400 videos posted to his YouTube channel, EarthquakeSim, which he created only two years ago. He claims to have the world’s largest 3D earthquake simulation channel on YouTube.

Using 3D modeling software, he creates visualizations of how buildings and structures respond to seismic shaking. Many simulations depict iconic landmarks and cityscapes trembling, cracking, or collapsing under earthquakes of varying magnitudes. Others take viewers inside everyday spaces – a classroom, a living room – exposing their vulnerabilities to different levels of ground motion.
 

 

Ronnet’s videos have drawn a global audience, amassing over 25 million views, 90,000 subscribers, and 1.1 million hours of watch time as of this writing. Featured in documentaries and international news segments, the simulations have extended their impact beyond YouTube.

For Ronnet, the videos are more than visual experiments – they’re tools for education and preparedness. EarthquakeSim emerges from a meticulous creative process, seeking to answer the question of whether visual simulations can effectively capture both scientific accuracy and the public’s interest.
 

How earthquake data becomes 3D visuals

Ronnet’s earthquake modeling started as a side project. While building a successful music-focused YouTube channel, “Pardon my Piano,” he discovered Kai Kostack’s channel about destruction physics, as well as a physics simulation plugin tool for a 3D visualization software program that Ronnet now uses. Kostack and Martin Felke initially designed the plugin to simulate how buildings collapse during disasters, with the goal of identifying structural weak points and helping first responders locate potential air pockets in the debris.

Those online tools piqued Ronnet’s interest and evoked fond memories of the basic shake tables that he built as a child. While experimenting with the plugin, Ronnet realized its potential for building even more detailed models. “I could create endless variations of structures with customized breaking points to simulate different construction materials,” he said.

Once he realized how he could use this plugin for earthquakes, Ronnet began to create his now famous simulations. He starts each simulation by browsing existing architectural models for inspiration and pulling ground motion data from publicly available databases. He converts this data into acceleration measurements, which are then fed into the plugin.

The process is highly manual. Ronnet builds each virtual environment piece by piece, programming rigid body objects with material properties and breaking points. Larger models, like those simulating downtown Los Angeles, can contain over 25,000 objects. Although occasional glitches require troubleshooting, most simulations are completed within a month.

To enhance accuracy, Ronnet cross-references his simulation results with historical earthquake damage patterns, such as the soft-story collapses observed in the 1994 Northridge earthquake. He then compares the simulation outcomes to the Modified Mercalli Intensity (MMI) scale as an additional benchmark for realism. The MMI scale measures how intense shaking feels to people on the surface, and scores are assigned based on observable damage to buildings, infrastructure, and the landscape.
 

 

Today, Ronnet collaborates with Victor Vescu, a Caltech Ph.D. student who serves as EarthquakeSim’s assistant manager, to create more sophisticated models. In a recent project, they simulated ground liquefaction during the 2011 Tohoku earthquake.

By pushing the limits of a plugin that’s under continued development, Ronnet and Vescu are working to simulate earthquakes in a range of locations and indoor settings. With hundreds of scenarios, viewers are likely to find one that resonates with them, perhaps sparking a deeper interest in the subject, Vescu said.

YouTube had an impact on Vescu’s own path to science. “I am convinced,” he says, “that before long, the first aspiring geophysicists who grew up watching Michael’s channel will [eventually] reach university classrooms.”
 

Seeing seismic stories

Many of EarthquakeSim’s viewers aren’t focused on the technical mechanics behind the simulations – they’re captivated by the visuals and the unfolding mini-dramas in each video.

Ronnet carefully designs each simulation, treating them like puzzles in which every detail serves a purpose. Often, he builds simulations backward, placing structures that remain standing in the background to create a clearer view of the destruction as it unfolds.

Camera angles and object placements, like a door blocked by a collapsed chimney, are meant to make viewers think about hazards and alternative escape routes in their own homes.

Unlike media coverage of earthquakes, which often focuses on human interest stories, Ronnet’s simulations offer a more detached perspective that’s focused on structural responses. By leaving people out of the picture, Ronnet notes that this provides a clearer view of seismic forces at play (and perhaps makes this type of content less distressing for sensitive viewers). Without the shaky hand-held camera, he says the viewer can also get a more accurate sense of the relative intensity between different earthquake magnitudes.

Rather than voiceovers, Ronnet’s videos rely on text notes and background music to deliver key information. Although some viewers might prefer narration, Ronnet believes the visuals speak for themselves. As the simulations have evolved, he’s worked with Vescu to add more on-screen explanations and safety tips for their audience.
 

 

Ronnet also engages directly with his audience, using polls and requests to shape his content and responding to questions in the comments.

“These simulations really pave the way for open discussions,” said Ronnet. “If we want to teach people, we have to find an engaging way to make it happen, because if you just listen to it in a passive way, I don’t think that’s really learning,” he said.
 

Understanding trade-offs in seismic modeling

Ronnet’s work lands him in the middle of a scientific debate over the strengths and limitations of simulations. All models require trade-offs—balancing detail, accuracy, and computational constraints. Thus, the choices made by modelers can shape interpretations of seismic risk.

Earthquake simulations, both computational and physical, help fill gaps in observational data. Despite Earth’s long seismic history, the first seismogram was recorded only in the late 1800s. Modern simulations allow scientists to map ground motion data from one location to another, test how different regions might respond to various types of shaking, evaluate new earthquake-resistant designs, and produce scenarios that can’t be replicated using physical experiments.

While shake tables are the gold standard for simulation equipment, they are costly, in high demand, and limited to smaller, less destructive scenarios. In contrast, computational models allow researchers to push structures to failure without grappling with those constraints. These models are cheaper and can be quickly modified to test different scenarios, making them valuable for comparative studies.

Still, both approaches involve oversimplifications that may lead to misunderstandings, said Saeed Fathali, vice president of structural risk and resilience at Nova Group, an environmental and engineering advising firm. On platforms like YouTube, he worries that valuable context may be missing for viewers.

“The [model] input is a bunch of assumptions,” Fathali said. He noted that estimates about material imperfections, randomness in ground motion, and structural variability can introduce uncertainty into the model results.

Ronnet underscores that simulations shouldn’t be taken as definitive predictions. “It’s always important to mention that the damage is entirely hypothetical and is not guaranteed to represent what might occur in a different earthquake scenario,” he said.
 

 

Fathali says that Ronnet’s channel is a valuable tool for illustrating the relative effects of different seismic forces. He noted that using simulations for comparing potential, but not absolute, effects is common in earthquake engineering, especially for simulations that aren’t powered by high-end, industrial-grade computational systems.

Vescu hopes that in the future, these simulations can have additional utility. “Michael’s simulations could be used in [calculations] of simulated ground motion [versus] those observed after natural quakes,” he said.
 

Encouraging empowerment

Although EarthquakeSim is only two years old, Ronnet has made ambitious plans to expand his collaborations with researchers and support earthquake-affected communities, while continuing to play the piano — his other passion.

Ronnet is also building partnerships with organizations that benefit from his simulations. For instance, Matei Sumbasacu, founder of Re:Rise – an organization focused on reducing seismic risk in Romania – has used Ronnet’s Budapest simulation to address normalcy bias, the tendency to underestimate earthquake risk due to a lack of recent events. “Michael’s videos are a very good tool that removes this level of abstraction,” Sumbasacu says. “They send the message that earthquakes are real, they can happen, and we need to prepare…it really does have the potential to change people’s behavior much more than wordy presentations or photos.”

Sumbasacu believes that when it comes to reducing seismic risk, effective visual communication is “a must.” Ronnet concurs. “I hope to inspire people not only to learn, but also to take proactive steps to reduce seismic risk and better prepare for earthquakes,” he said.

Ronnet and Vescu are also collaborating with Corridor Digital, a production studio known for viral content, to simulate how the Los Angeles studio itself might hold up during a major earthquake.

By strengthening connections with viewers, Ronnet hopes to inspire people to take proactive steps toward earthquake preparedness, such as securing objects in their homes and choosing safer evacuation routes. He emphasizes that instead of feeling helpless, people can be empowered to take charge of their own safety, particularly in areas with limited government support. “You can’t prevent an earthquake,” he noted. “But you can plan for a safer future.”
 

 

Further Reading

KaiKostack. (n.d.). GitHub – KaiKostack/bullet-constraints-builder: Add-on for Blender to connect rigid bodies via constraints in a physical plausible way. GitHub. https://github.com/KaiKostack/bullet-constraints-builder
 

Copyright

Text © 2025 Temblor. CC BY-NC-ND 4.0

We publish our work — articles and maps made by Temblor — under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International (CC BY-NC-ND 4.0) license.

For more information, please see our Republishing Guidelines or reach out to news@temblor.net with any questions.