not only bridges

Preferred geometry file for FEA: IGES, STEP or SAT?

IGES and STEP are the widely used neutral CAD formats and accepted in nearly all software packages.

IGES or IGS (Initial Graphics Exchange Specification) is the first CAD data exchange standard developed and it is capable of exchanging only geometry and topology information between different CAD systems.

STEP or STP (STandard for the Exchange of Product Data) enables storing, retrieving, sharing and archiving member data (f.ex. profie sections, material properties, etc...) throughout its life-cycle among different databases and STEP has become the industry standard for Data Exchange.

For Finite Element Analysis (FEA), preference for one or the other depends upon several things including:
1) the source CAD that is used to create the neutral format.
2) the target CAD, CAE or FEA package and the use for which the file is intended.
3) the kind and complexity of geometric elements included in the design (e.g., prismatic elements, splines, surfaces, etc.)

In IGES the output is in lines and surfaces while STEP file keeps the assembly hierarchy and output is a mixture of solids,volumes and surfaces. For FEA purposes IGES may be enough as you don't usually need solids/volumes. IGES works ok for importing beams and sometimes works well for plates.
For those that work in the 3D solid modelling world, STEP format may be preferred as it retains the solid geometry and brings the part into the 3D modelling software in solid format. In general STEP is also more robust than IGES for plates and shells imports.

ACIS SAT is very similar to STEP. The kernel of many AutoDesk and Dassault products is partially based on ACIS. For example Abaqus uses ASICS (*.SAT) as it's native kernel so I usually start there and then go to STEP and IGES would be the last choice.

Stupid as it may sound, I generally request more than one format and a screenshot image of what it should look like.

How do I validate FEA results?

First of all, check the numerical consistency of the results:

- Ensure convergence of solution: refine mesh and check if changes in the answer are smaller and smaller.
- As an error estimation look at strain jumps between adjacent elements with results not averaged or smoothed.

Then, compare to other theoretical/numerical results if available. Compare to empirical results if available. For hand-made engineering calculations I recommend MathCAD (or Smath Studio) instead of Excel because the formulas are readeable and with named variables.

Last but not least, compare with systems of increased complexity. For example, if you are analyzing 3D dynamic system with nonlinear material, start with 2D linear static and build confidence in your solutions by adding complexity step by step.

Monster bridge machine

The awesome mega bridge machine SLJ900/32 is being used in China for bridge girders construction.

The manufacturer is Beijing Wowjoint Machinery Co and the main features are:
  Weight: 580 tonnes
  Measures: 91.8 meters long - 7.4 meters wide - 9 meters high

Indeed, China is becoming a great industrial nation.

How to build a gas processing plant

This is an award winning short film for its educational value to the oil and gas industry. It is a pity that the video does not come with music. Included are construction methods, sequence, resources and timelines. Activities covered in the video are earthworks, pipe racks, slugcatcher, steel tanks, heavy lifts, welding & spool fabrication, pavements and fencing.


In the past two years, David H Moloney, an Irish engineer, has made and uploaded short films like this one onto YouTube showing us how civil projects are constructed. These projects range from airports, seaports, motorways and railways to oil and gas pipelines, metros, in-situ piles, bridge construction, post-tensioning, precast yard, quarry, roofing and more. I strongly recommend to visit his channel.

The theories of everything

When you try to be all things to all people, you end up being nothing
(common saying  popularized by Robin Sharma)

I have recently finished watching The Theory of Everything. If you are interested in the borders of science, it is a must see. As Hawking admits, cosmology was thought of as a pseudoscience given that wild speculation was unconstrained due to the lack observations.

More examples of the difficult birth of a new science: the government of the Soviet Union declared that Mendelian genetics or plate tectonics (continental drift) were pseudo-science. Artificial intelligence has been considered as hype and junk science by many. Dijkstra considered software engineering to be a pseudoscience.

On the other hand, there are disciplines that were abandoned. For example, I have been taught elements of NLP - a debunked psycologic theory - many times on prestigious management courses. Why? Because although NLP is pseudoscience it brings under a single unique umbrella a host of useful good interpersonal and conversational techniques. Also, modern city planning has been called pseudoscience by Jane Jacobs and others.

What are the borders between mainstream science, fringe science, junk science, heuristics, useful pseudoscience, science fiction and fraud?