Revolutionizing Bridge Engineering

with Generative AI: The OpenBrIM

Journey

OpenBrIM Platform

Cagin Yakar Chief Executive Officer

A Note Before We Begin

With due respect, deep familiarity with a subject can sometimes make it easy to overlook the challenges of being new to it. In this presentation, I'll aim to convey the information clearly, using basic examples to help bridge that gap.

OPENBRIM GENERATIVE OBJECTS

OpenBrIM Generative Objects

HOW OPENBRIM MAKES THE IMPOSSIBLE POSSIBLE?

How Did We Get Here?

Under Development Since 2015 FHWA Bridge Publications

  • Transforming bridge practice from paper delivery to digital delivery
  • Developing digital BrIM standards
  • Developed Parametric Library Objects

WHAT IS A HARD PROBLEM FOR AI?

Chess: A Historical Milestone

  • Early Benchmark: Once seen as a test of intelligence.
  • AI Breakthrough: Deep Blue's victory over Kasparov in 1997.
  • Current Status: Chess is now a manageable problem for AI, mastered through advanced algorithms and computational power.

Is Chess Still Hard?

  • Finite Problem: Despite complexity, it's a finite, rule-based game.
  • AI Mastery: Chess is no longer a challenge, with AI dominating the game.

Real-World Decision Making: A True Challenge

  • Tasks like autonomous driving require real-time decisions in unpredictable settings, far more complex than chess.

SIMPLIFYING BRIDGE ENGINEERING FOR AI: SINCE 2015

"Any intelligent fool can make things bigger, more complex, and more violent. It takes a touch of genius—and a lot of courage—to move in the opposite direction." — Albert Einstein

On Mathematical Maturity Thomas Garrity

OPENBRIM LIBRARY: BRIDGE ENGINEER'S FUNCTION CONTAINER

  • Transparent parametric dependencies
  • Customizable dependencies according to different project and company requirements
  • Reusable and modular way to define parametric relationships

THE SYNERGY OF PARAMETRIC ENGINEERING AND HOLISTIC APPROACH

Defining a Bridge with Parameters

  • Automatic Model Generation: Key parameters drive the creation of the bridge’s 3D and mathematical model (FEM).
  • Seamless Analysis Integration: The analysis engine simulates behavior.
  • Automated Specification Checks: Each bridge model undergoes specification checks during the process.

THE SYNERGY OF PARAMETRIC ENGINEERING AND HOLISTIC APPROACH

With OpenBrIM, we're empowering bridge engineers in ways never before possible.
Any sufficiently advanced technology is indistinguishable from magic.

The Exponential Growth Of Combinations & Computational Challenges

  • Total Combinations: 1,166,200
  • Time Required per Evaluation: 0.5 seconds
  • Total Computation Time per Scenario: ~97 hours
  • Geometric Growth: Number of alternatives and required resources increase geometrically.
  • Human Limitations: Given the vast number of combinations, evaluating every scenario would far exceed human lifetimes.

How Can We Achieve Optimal Solutions Within Feasible Time and Resource Constraints?

  • Define Your Goal:
  • Objective Function: Designed to minimize a specific parameter (e.g., area of the section).
  • Maximum number of solutions: Limits the scenarios the optimization process will evaluate in detail, focusing resources on the most promising options.
Parameter Name Description Min Value Max Value Increment
dw_Input Web depth 72 105 1
tw_Input Web thickness 0.5 1.25 0.125
tft_Input Top flange thickness 0.75 1.5 0.125
tfb_Input Bottom flange thickness 0.75 1.5 0.125
bft_Input Top flange width 12 30 2
bfb_Input Bottom flange width 12 30 2

Cutting Down Time and Resource Demands with Neural Networks / Genetic Algorithm / Parametric Bridges

The Art of Simplification: Experience and Education in Simplifying Solutions

  • Human: Just as humans learn to filter out unnecessary solutions based on education, experience, intuition, and critical thinking.
  • AI – Neural Networks: AI systems are trained to recognize and discard less relevant options, focusing only on the most promising paths to efficiently solve complex problems.

OPTIMIZE DESIGN PARAMETERS

Neural Network Space Reduction

  • Purpose: Eliminate unnecessary solutions without evaluating them.
  • Process: Focuses on relevant solutions, discarding irrelevant data early.
  • End Result: A more efficient model with only the most relevant solutions.

AI IN THE DRIVER'S SEAT: HOW MUCH CAN WE TRUST ITS DECISIONS?

Oversized Stop Signs Confuse Tesla’s Autopilot’s Weakness: A Stop Sign T-Shirt

MERGING AI WITH TRADITIONAL METHODS

  • 3D MODEL- We, as OpenBrIM Engineers, are the Architects. 2D DWG
  • We created OpenBrIM.
  • And we’ve been waiting for you.

Revolutionizing Bridge Engineering

with Generative AI: The OpenBrIM

Journey

OpenBrIM Platform

Cagin Yakar Chief Executive Officer