Mastering Automation for Structural Engineers: Tools & Strategies

Key Concepts

• Automation in Structural Engineering: The process of using technology to streamline and expedite design, analysis, and documentation tasks.
• AI-Driven Tools: Software and platforms that leverage artificial intelligence to enhance engineering capabilities, moving beyond traditional scripting.
• Parametric Design: A design approach where elements are defined by parameters, allowing for easy modification and exploration of design variations.
• Knowledge-Based Engineering: Integrating engineering knowledge and logic into software to automate complex design decisions and analyses.
• Generative Design: An iterative design process that uses algorithms to explore a wide range of design options based on specified constraints and objectives.
• Web-Based Platform: Software accessible through a web browser, facilitating collaboration and accessibility.
• Custom Web Applications: Tailored interfaces built on a platform to automate specific engineering workflows.
• Finite Element Analysis (FEA): A computational method used to predict how a structure or component will react to real-world forces, heat, vibration, and other physical effects.
• Building Information Modeling (BIM): A digital representation of the physical and functional characteristics of a facility.
• APIs (Application Programming Interfaces): Sets of rules and specifications that allow different software applications to communicate with each other.
• Large Language Models (LLMs): AI models capable of understanding and generating human-like text, used for tasks like code generation.
• Deterministic Tools: Software or processes that produce the same output for the same input, crucial for engineering reliability.
• Agentic Workflows: Systems where AI agents (like LLMs) can interact with and utilize deterministic tools to perform complex tasks.
• Standardization: Establishing consistent processes and logic to facilitate automation.
• Business Case for Automation: Evaluating the return on investment for automating specific tasks based on time savings and efficiency gains.

Victor: Empowering Engineers Through Automation

This discussion with Stein Jansen, Chief Product Officer at Victor, explores the transformative potential of AI-driven automation in structural engineering. The conversation highlights how tools like Victor are shifting the perception of automation from a threat to human judgment to a powerful enabler of efficiency, creativity, and value creation.

The Evolution of Engineering Automation

Stein's background in aerospace engineering, with a focus on parametric and knowledge-based engineering, provided a foundation for understanding how to integrate multiple disciplines to achieve optimal designs. He notes that while aircraft design often involves series production and significant business cases for optimization, the AEC (Architecture, Engineering, and Construction) industry is characterized by unique projects and a more parallel, multi-disciplinary workflow. This inherent complexity and uniqueness have historically contributed to lower digitalization rates in AEC compared to other sectors.

The parallel nature of design in AEC, where disciplines work concurrently and requirements can change, creates significant interdependencies. A change in one discipline, such as a structural beam size, can impact others like HVAC systems. This interconnectedness underscores the need for automation that bridges these disciplinary gaps.

Key Principles for Effective Automation Workflows

Stein outlines several critical steps for creating effective automation workflows that balance speed, accuracy, and usability:

1. Standardization and Explicit Logic: The first step involves extracting implicit knowledge from engineers' heads and formalizing it into explicit logic that can be built into software. This process forces a deeper understanding and standardization of design processes and decision-making. While this might sound like replacing humans, Stein emphasizes that it often leads to a more pleasant and efficient engineering experience by clarifying "what" and "how" tasks are performed.
2. Business Case Analysis: Automation should be approached with a business mindset. Engineers need to assess the investment of time in automation against the expected returns, such as time saved and the frequency of the automated process. This helps prioritize which tasks to automate first, often starting with repetitive tasks like reporting or data manipulation.
3. Step-by-Step Implementation: Rather than aiming for a single "one-push-button" solution immediately, a phased approach based on a clear business case is recommended. This allows for incremental automation and continuous improvement.

Victor's Approach to Automation Tools

Victor is presented as a web-based platform designed to make it as easy as possible for engineers to automate workflows. Its core functionality lies in enabling the creation of custom web applications. Key features include:

• User Interface Building: Victor allows for the creation of intuitive interfaces that go beyond simple scripts. These interfaces can present visualizations of results, maps, 3D models, and allow for input of parametric data, requirements, loads, and material properties.
• Data Input Flexibility: Information can be entered directly into the Victor interface or imported from external sources like Excel spreadsheets.
• Integrations: The platform offers integrations with various engineering software, including finite element analysis packages and BIM models, enabling the automation of end-to-end processes.
• Publishing and Collaboration: Once an application is built, it can be published within a company's environment, granting colleagues access to use these automated tools.

Victor's slogan, "Automate the Boring, Engineer the Awesome," encapsulates its mission to free engineers from repetitive tasks, allowing them to focus on more complex and creative problem-solving. This increased speed of design iteration not only helps meet deadlines but also facilitates design optimization, leading to material savings, reduced CO2 impact, and cost reductions. The ability to present multiple design options to clients in the same timeframe significantly enhances value and collaborative decision-making across disciplines.

AI and Human Judgment: A Synergistic Relationship

The discussion addresses the concern of AI replacing human judgment. Stein clarifies that while AI, particularly LLMs, can be probabilistic and non-deterministic (meaning the same input might yield different outputs), this is not acceptable for critical engineering design.

Victor's approach to AI integration involves:

• Deterministic Processes: The platform is built on Python, a deterministic programming language. AI is used to generate Python code based on prompts, which then forms a deterministic tool. This ensures that once the code is reviewed and validated, it will consistently produce reliable results.
• Quality Assurance: Similar to human review processes, a robust review mechanism is essential for AI-generated code. This ensures that the automated tool aligns with all norms and calculations.
• Empowerment, Not Replacement: AI is positioned as a tool to accelerate the creation of these deterministic automation tools, saving engineers time on coding and allowing them to focus on the engineering judgment required for validation and higher-level problem-solving.

Future Trends and Agentic Workflows

Stein expresses excitement about the future of AI in engineering, particularly the convergence of LLMs with deterministic tools, leading to "agentic workflows." This involves:

• LLMs as Tool Callers: LLMs can be trained to call trusted, deterministic tools (e.g., beam calculations, FEA). This allows the LLM to leverage validated engineering processes to perform tasks.
• Structural Agents: The vision is to create AI "agents" that act as colleagues, assisting engineers with tasks like exploring design options or performing calculations. These agents can be seen as a form of "intern" with specific knowledge, with the ongoing question being how "senior" these agents can become.
• Addressing the Engineer Shortage: This increased productivity through AI assistance is crucial for addressing the global shortage of engineers and enabling them to make a greater impact.

The Power of Python and Accessible Automation

The conversation also touches upon the limitations of traditional tools like Excel for complex engineering tasks. Python is highlighted as a powerful alternative due to its ability to integrate with FEA packages via APIs, enabling automation of processes that are cumbersome or impossible in Excel.

Recognizing the learning curve associated with Python, Victor's app builder aims to democratize coding. By allowing engineers to generate Python code from prompts, it lowers the barrier to entry, making automation more accessible. The platform's browser-based nature further simplifies adoption, eliminating the need for local installations.

Embracing Innovation: A Call to Action

Stein's final advice to engineers and AEC leaders is to take calculated risks and embrace innovation. He acknowledges the inherent risks in the AEC industry but argues that resisting automation due to perceived risk is counterproductive. He draws parallels to past innovations like the shift from paper drawings to CAD and from 2D to 3D modeling, where initial hesitation was overcome.

The recommendation is to:

• Start Small: Begin with pilot projects, perhaps exploring AI for a specific task or automating a small, well-defined workflow.
• Invest Time: Dedicate a small amount of time to experimentation and learning.
• Focus on a Clear Case: Identify a specific problem that automation can solve.
• Dare to Dream Big: Maintain a forward-looking perspective and envision the potential impact of automation.

The overarching message is that by embracing automation and AI, engineers can significantly enhance their productivity, creativity, and overall impact, ultimately shaping a more efficient and future-ready AEC industry.