Historical data is the foundation. Physics AI adds the layer engineers can explain. See how Braskem is bringing SPM to its operations with UPTEC and Akselos.

Braskem is bringing Structural Performance Management (SPM) to its operations because its engineers need AI they can trust, and models trained only on historical data aren’t enough. SPM is built on Physics AI. It starts from engineering physics, runs on the sensors a plant already has and follows API and ASME standards.
That combination is what Braskem’s engineering team was looking for. As Aloísio Azevedo, Global Optimization Manager at Braskem, puts it: “Anchoring our AI applications in robust physics-based models gives us the confidence to run our assets more safely, profitably and reliably.”
The work builds on the partnership with UPTEC Solutions announced in September, which brings SPM to petrochemical assets in Latin America.
Historical data is the foundation
Historical data covers conditions a plant has already run in. And capturing this data is the baseline to build a layer that enables optimization decisions to push equipment somewhere new.
A higher rate, a different feed, a longer run to the next turnaround. Each process change moves the steel into a different condition that results in a direct impact to its reliability and remaining useful life. A statistical model can extend the trend, but it lacks the fundamental understanding of the physical reality of the asset, making its predictions insufficient for a safety-driven operation .
Braskem reached the same conclusion. “Generic models that only use historical data don’t seem enough to us,” says Aloísio Azevedo, Global Optimization Manager at Braskem.
The data matters more than ever. It describes how assets have behaved under different process conditions. Physics enables a different layer of decision making using this data.
What makes physics-based results trustworthy to engineers?
Engineers trust a result they can trace back to methods they already use. Physics AI gives them a view into the mechanics of how the model builds insights.
Azevedo names the reason Braskem selected Akselos: “it combines physics-based models with existing sensors and rigorous engineering standards.”
In SPM, each part of that statement does specific work. A structural digital twin of the asset calculates how the steel responds to real operating loads, pressures and temperatures amongst other process and operational conditions. Existing plant sensors and data historians feed it, so there is no need for a new instrumentation campaign to get results. The calculations are based on API and ASME engineering standards, the same ones integrity engineers apply in their own assessments, for example Fitness-for-service projects. RB-FEA, Akselos’ patented solver, runs this analysis at full-asset scale, 1000× faster than conventional FEA with the same accuracy.
This method, in Azevedo’s words, “… helps our engineering team trust the results.”
An AI result only truly helps engineers run a unit if the engineer who is responsible for the equipment understands the insights and can confidently explain them to their peers and leaders. Calculations that follow API and ASME give Braskem’s engineers that confidence.

Diagram showing operating data from existing sensors feeding a physics-based structural digital twin, calculated to API and ASME standards, producing structural limits for operating decisions
How is Braskem applying Structural Performance Management?
Braskem is bringing SPM to its operations, with UPTEC Solutions supporting the work in the region and Akselos providing the system.
For Braskem, the value of the collaboration is operational. Azevedo sums it up in three words: “more safely, profitably and reliably.” It also builds on what the plant already has, combining “physics-based models with existing sensors and rigorous engineering standards,” so the engineering team can trust what the results tell them.
UPTEC Solutions brings that capability to the region. Hildebrando Maia, CEO of UPTEC Solutions, describes the goal as “bringing unique, innovative, and tested solutions to the Brazilian and Latin American market.”
Azevedo frames the collaboration as a long-term direction: “This is how we are redesigning the future and applying SPM to our assets.”
Explore more: Akselos and UPTEC Solutions partnership announcement
What does this mean for petrochemicals in Latin America?
It puts physics AI in the hands of the region’s own engineering teams, supported locally.
“Industrial AI is top of mind for energy leaders in Latin America,” says Ricardo Pinto, VP Europe and LATAM at Akselos. The partnership with UPTEC Solutions is about “bringing SPM for critical infrastructure in Latin America right here.”
For UPTEC Solutions, the partnership goes beyond following a trend. “It is much more than applying physics AI models to follow market trends. It is about resetting Latin America’s benchmark to transform industrial performance,” says Hildebrando Maia, CEO of UPTEC Solutions.
For Braskem, the partnership is part of a wider ambition. “This is how we are shaping the future of petrochemicals in Latin America, in partnership with UPTEC and Akselos,” says Azevedo.
Key takeaways
- Models built only on historical data lack the foundational understanding of the physics that govern that asset. This is what Physics AI is able to do.
- Physics-based models, existing sensors, data historian and API and ASME standards together give Braskem’s engineers insights they can explain.
- An AI insight only helps make decisions on how to run an equipment if the engineer can explain the insights to peers and leaders.
- Braskem is bringing SPM to its operations with UPTEC Solutions and Akselos, to run them more safely, profitably and reliably.
Ricardo works with energy operators across Latin America, and Europe to connect structural integrity priorities with broader business outcomes. He helps organizations identify where structural clarity can unlock longer asset life, reduced uncertainty, and stronger operational performance.
