Tag Archives: Genetic Algorithms

The Lie of the Perfect Cylinder (Part 3): Embracing Chaos (Stochastic Optimization)

Posted on by
Reply

Why do bridges fall down? Why do roofs collapse? Usually, it’s not because the engineer got the average math wrong. It’s because of an “outlier.” A single joint that was weaker than expected, or a load event that exceeded the “average” prediction [1].

In traditional design, we fear these outliers. We try to hide from them behind huge Safety Factors (as discussed in Part 1).

But in Stochastic Optimization, we don’t hide. We invite the outliers into the model. We design specifically for the chaos. This approach, widely used in aerospace and financial engineering, is the frontier of structural design for natural materials [2].

In a standard Grasshopper script, a number is a scalar value: `Diameter = 10`.
In a Stochastic script, a number is a Probability Density Function (PDF) [3].

Instead of telling the computer “Bamboo is 10cm thick,” we tell it:
“Bamboo is a bell curve. It is usually 10cm. Sometimes (68% of the time) it is between 9cm and 11cm. Rarely (1% of the time) it is 7cm.”

This is a much more honest way to describe nature. Natural materials like bamboo do not have a single “strength” value; they exhibit statistical variability that follows specific distribution patterns (often Weibull or Normal distributions) [4].

So, how do we optimize for a “curve”? We use a brute-force method called the Monte Carlo Simulation [5].

Imagine we have a design for a bamboo truss. To test if it is robust, the computer plays a game of dice.

The Iteration Loop:
The computer builds a virtual model of our truss. But for every single strut, it randomly assigns properties based on our probability curve [5].

  • Strut A gets assigned “Weak.”
  • Strut B gets assigned “Average.”
  • Strut C gets assigned “Strong.”

The Stress Test:
It applies the load. Does the truss break?

Repeat x 1000:
It resets and tries again with new random values. It does this 1,000 or 5,000 times.

The Result:
We don’t get a simple “Pass/Fail” result. We get a Probability of Failure (Pf).
“This design failed in 4 out of 1000 simulations. It has a 99.6% Reliability Index.”

Now, we hook this into our Genetic Algorithm (Galapagos or Wallacei).

Usually, GA looks for the lightest structure. But a Stochastic GA looks for the most Robust structure [6].

What is robustness?
A “Strong” structure might hold a heavy load, but if one member is slightly weak, it collapses. A “Robust” structure is resilient. It has redundancy. If one bamboo pole is weaker than expected, the forces redistribute to its neighbors. The structure survives. This concept is critical for bamboo, where local defects are common [7].

This brings us to the end of our three-part exploration on “The Lie of the Perfect Cylinder.”

  • Part 1 showed us that Safety Factors are safe but wasteful. They treat bamboo like bad steel.
  • Part 2 showed us that Scan-to-BIM is precise but logistically difficult.
  • Part 3 showed us that Stochastic Design is the mathematical middle ground. It allows us to design safe, efficient structures by embracing the statistical reality of nature.

Evolution of Computational Strategy. A comparison of the three dominant approaches to material uncertainty. While ‘Safety Factors’ remain the industry standard for compliance, ‘Stochastic Optimization’ offers the highest research value for maximizing structural efficiency without compromising robustness.

As we move forward in 2026, my research will be heading in this direction. I want to move away from drawing “ideal” shapes and start coding “robust” systems. Because in the end, architecture shouldn’t be about fighting nature’s chaos. It should be about finding the order within it.

Reference

[1] R. E. Melchers and T. Beck, *Structural Reliability Analysis and Prediction*, 3rd ed. Chichester, UK: John Wiley & Sons, 2018.

[2] M. Papadrakakis, V. Papadopoulos, and N. D. Lagaros, “Structural reliability analysis of elastic-plastic structures using neural networks and Monte Carlo simulation,” *Computer Methods in Applied Mechanics and Engineering*, vol. 136, no. 1-2, pp. 145-163, 1996.

[3] S. S. Rao, “Engineering Optimization: Theory and Practice,” 4th ed. Hoboken, NJ: John Wiley & Sons, 2009.

[4] F. Faris, “Reliability analysis of WBM MSE wall based on tensile strength variation,” *ASEAN Engineering Journal*, vol. 12, no. 4, pp. 15-22, 2022. Available: https://journals.utm.my/aej/article/download/17320/7866

[5] G. I. Schuëller, “On the treatment of uncertainties in structural mechanics and analysis,” *Computers & Structures*, vol. 85, no. 5-6, pp. 235-243, 2007.

[6] H.-G. Beyer and B. Sendhoff, “Robust optimization – A comprehensive survey,” *Computer Methods in Applied Mechanics and Engineering*, vol. 196, no. 33-34, pp. 3190-3218, 2007. Available: https://doi.org/10.1016/j.cma.2007.03.003

[7] P. Faber, “Robust design optimization of structures under uncertainties,” in *Proceedings of the 12th International Conference on Applications of Statistics and Probability in Civil Engineering (ICASP12)*, Vancouver, Canada, 2015.

My First Year Mapping the Intersection of Code and Climate

Posted on by
Reply

building structure transitioning from a digital parametric wireframe into a real-world bamboo pavilion

Consistency is often more difficult than intensity. It is easy to sprint; it is hard to walk every day for a year.

Today marks a small but meaningful milestone for me: I have successfully published a blog post every single month for the past 12 months. One year of consistent writing.

To some, this might seem trivial. It’s just a blog, right? But for me, this represents a discipline I’ve been trying to cultivate. In a world of instant updates and fleeting social media stories, the act of sitting down to write a thoughtful, long-form piece once a month feels like an act of resistance. It’s a commitment to deep thinking over quick scrolling.

When I started this commitment a year ago, I had a few hopes.

For Myself: Writing forces clarity. You think you understand a concept—like computational design or sustainable bamboo construction—until you try to explain it to someone else. Writing these posts has been my best method of study. It forces me to research deeper, structure my thoughts, and articulate my arguments.

For My Students: I wanted to create a resource that extends beyond the classroom. A lecture lasts 100 minutes, but a blog post lasts forever. Students can revisit these ideas about parametric design, environmental responsibility, or professional ethics whenever they need them.

For the Institution: I hope this blog contributes, in a small way, to the scientific culture of Universitas Medan Area. Academic discourse shouldn’t just happen in closed journals; it should be accessible, public, and engaging.

For the Public: Architecture can feel elitist or inaccessible. I try to write in a way that bridges the gap – making complex ideas about resilient cities or design technology understandable to anyone who cares about the built environment.

Looking back at the archive, I see a map of my own intellectual journey this year.

We explored computational design – demystifying Grasshopper not just as a tool for making weird shapes, but as a way to think algorithmically.

We dived into bamboo architecture, discussing how traditional materials can be optimized with modern technology.

We tackled climate resilience, especially after the floods of November. The post “Designing for Cyclones” wasn’t just an article; it was a response to a real crisis we all faced.

We reflected on education, asking hard questions about why hydrology isn’t foundational in design schools.

Each post was a snapshot of what I was learning, questioning, or fighting for at that moment.

I don’t know who reads every post. Analytics give numbers, but they don’t tell stories.

But then, something surprising happened in October.

Someone approached me on campus – someone I didn’t know – specifically to discuss bamboo. They weren’t a student in my class, but they had read my blog post about bamboo construction joints. They came with specific questions, ready to discuss preservation techniques and structural details.

I was genuinely surprised.

To be honest, sometimes writing a blog feels like shouting into the void. You press “publish” and wonder if anyone actually cares. But that conversation in October proved that words travel. It proved that there are people out there – students, practitioners, enthusiasts – who are hungry for this kind of specific, technical knowledge.

That moment was a turning point for me. It shifted my perspective from “I have to write this for my schedule” to “I get to write this for a community.”

It is the best kind of reward. Not the traffic numbers, but the real, human connection that starts with a shared idea.

I hope this blog serves as a small spark.
A spark for students to read more than just captions.
A spark for colleagues to share their own expertise publicly.
A spark for anyone to start writing their own thoughts.

Because knowledge that isn’t shared is knowledge that stagnates. Writing keeps it moving.

So, here is to consistency.

To showing up at the keyboard even when I’m tired.
To researching topics that challenge me.
To pressing “Publish” even when I’m not sure if it’s perfect.

Thank you to everyone who has read, shared, or discussed these posts over the last year. You are the reason I keep writing.

Let’s see what the next 12 months will teach us.

Keep reading. Keep writing. Keep learning.