My new structures app!

Hi all – I have exciting news…I now have an app about structures in the app store!  It’s for iPhones and iPads.

 

 

 

 

 

 

The app is called “Structures: a visual exploration” and teaches structural behavior all through diagrams, animations, and comparisons.  No text.  No equations.

Many of the concepts in the app came from thoughts developed while writing this blog and I describe them in more detail on my app website here: http://www.understandingstructures.com/development.php

It’s available in all iTunes app stores worldwide and you can go to the app page here.  Please download and enjoy!

Aside on Testing and Understanding

I neither teach nor test my students with a multiple-choice format.  It’s a lazy approach (I did it once 6 years ago and deeply regret it).  My focus is that the architects understand structural behavior rather than learn how to get an answer from an equation. 

I don’t like multiple choice question-and-answer because it’s not a sensible approach to judging the student’s understanding of the subject and their ability to think.  It’s also not realistic.  I don’t know any profession where you’re provided with all required information, and the one and only correct answer is an option presented to you.

The fact is, multiple choice tests are really good at testing if you’re good at taking multiple choice tests.

If architects take multiple choice tests for things like structural engineering then, not only is it a poor test of their understanding, it forms the incorrect opinion that structural engineering is about solving equations and getting the right answer.  No value is placed on judgment, opinion, creativity etc.  Hence, the only way to distinguish between engineers is on fee.  This seems similar to the current situation today.

Maybe worse, engineering students who take multiple choice tests might think solving equations (or performing other repetitive tasks) is what they’re supposed to do.  If that’s the case, good luck with your 11-month career before they hire the same you next year who’s cheaper!

Both my homeworks and exam(s) require the students to have understood what I teach and then think beyond that.  It’s not the most popular approach (every year I receive feedback from a few students, via the MIT evaluation process, that they were not specifically taught what they were questioned on) but it’s what structural engineers do in our professional lives.  And, I would suggest, it’s how we all learn and improve in our daily lives…

Who asks the questions here?

At the same time as teaching this course, I was also a frequent reviewer for an MIT architecture studio (by Meejin Yoon) with many of the architecture students from my class.  At one of my joint architects+engineers class I asked a couple of the architects to bring in their progress models an images and explain what they’re doing.  I told the engineers that they should imagine they are meeting the architect for the first time and the discussion is on structural materials, form, and behavior.

Here’s one of the projects we discussed in class…

Credit: Toshiro Ihara

Honestly, this happens quite a bit in my work life.  I frequently receive emails with images or rhino models asking a mix of general and specific questions.  The value I try to bring to the design is to respond with some sense and general advice.  It’s maybe the hardest thing I do…

So, it was no surprise that the engineering students struggled greatly with this.  They essentially asked many structural and material questions back to the architect who, in turn, was a little stunned as he thought he would get those answers from the engineering students! 

Overall, the engineers felt uncomfortable providing feedback when there wasn’t a clear question.  They wanted to know materials and floor spans before saying whether it would work or not.  Whereas the architect wanted to know what he could or couldn’t do…

I did this on purpose as we’d finished the “primary colors” of structures but we hadn’t started on how to assemble them in to overall forms.  I also did this because, as I’ve mentioned before people who can think and make judgments will always be of value.  Whereas, mechanical/simplified responses to blunt questions with one correct answer (i.e. multiple choice format) are limited and almost useless to the creative design process.

More on this in my next post…

Frame Example 2 of 2: Simmons Hall

This is a brief post to highlight what can be achieved when frame behavior is understood clearly.  Every building has, what we call, a lateral system.  It resists against lateral loads such as wind and seismic.

Typically you don’t see the lateral structural system in a finished building.  Some buildings, however, wear their heart on their sleeve so to speak.  Simmons Hall, an undergraduate dorm building at MIT, is one of those.

The squares you see on the exterior are essentially multiple reinforced concrete moment frames.  Made from precast concrete segments, they were lifted into place and interconnected with couplers to the protruding reinforcement bars. 

An architectural desire (Steven Holl architects) was for the concrete to be the same dimension throughout the building.  Hence, they only variable is the reinforcement inside it.  Reinforcement bars are available in various diameters and we reference this with a number that equates to a multiple of 1/8 inch diameter (for typical bar sizes).  Hence a “number 4” bar is 1/2in diameter.  An engineer in our office, needing an easy reference for the multiple bar sizes throughout the project, created a color chart of the main building elevation.

When the architect saw this in a meeting, it formed the inspiration for the cladding color scheme…

Frame Example 1 of 2: Why understanding frame behavior is important

This post gives a small detailed example of frame behavior and why it’s important to understand it. The next post will be about frame behavior on a larger (building-size) scale.

We were asked to investigate a sudden breakage of a glass infill panel in a shopping mall. Interestingly enough, it happened between Christmas and New Year but quite a few years ago.

The breakage occurred at nighttime while the mall was shut and was discovered by the night security guard (who took photos 1 and 2 shown above). The mall placed plywood in that area before opening and was, of course, concerned to know why this had happened, and whether it was a one-off or a repeatable occurrence. As we were preparing to go to the mall we reviewed the photos and noted the glass seems to have fallen directly below, rather than in or out of, its vertical plane. We also noted (photo 2) the supporting angle clips were rotated at their connection to the verticals. We asked various questions: had this occurred before? (no), can the floor cleaning equipment impact the glass given the glass hangs beyond a floor-level upstand? (no), did they suspect vandalism? (hopefully not) etc.

At the mall we observed many of the existing angle clips were rotated (see photo 3). Rotation of the angle clips is not surprising as there is only one bolt to the vertical.

Photo 3

We can easily draw the behavior of the system under its own self-weight. See diagram below. The pinned supports can only provide vertical support, so the bending (which exists because the glass is offset laterally from the supports) occurs at the clip-to-glass connection and in the glass itself. This is an odd method of support. We normally have fixed rotation at the main support (the shoulder in my previous analogy) not at the thing we’re supporting.

To manage this fixed rotation point, the as-built clip-to-glass connection creates a push-pull between its top contact edge and its central bolt (see photo 3 again). Unfortunately, the existing plastic washer between the clip and glass was not the full height of the clip so the top edge of the metal clip was in direct contact with the glass.

Direct metal to glass contact is not a good thing. It’s even worse when that area of glass is under bending stress.

So, a poor choice of the overall structural system, combined with incorrect local detailing, produced overstress and damage to the glass, causing it to shatter.

We reviewed various remedies (adding a laminate, switching the glass out for metal mesh etc.) and, on balance, decided the best solution was to make the base connection (metal clip to vertical) fixed for rotation. We achieved this by welding along the edges of the angle clips. Of itself, this wouldn’t magically remove in-built stresses in the glass. So, each glass panel was removed and the angle clips were correctly aligned, before any welding took place. The glass panels were then reinstalled.

It’s been eight years since that breakage event and no other panels have broken or exhibited signs of deterioration.

Class on Frames to architects and engineers

So, as the second part of two on bending moments, I covered frames and their behavior under loads.

Frames work in bending; which is a structurally inefficient way to carry load (see shoulder analogy from my last post).  But, the majority of structures are not the most structurally efficient (and nor should they be) as we have greater issues to deal with such as needing flat floors, vertical walls, and open passage between the walls.  Hence, the dominance of frames in our lives.

Overall, this is the most technical of classes so bear with me as we go through some of this…

My approach, as before, is to think about the direct loadpath of any applied load to the supports.  Once we have that, any deviation produces bending.  The greater the distance from structure to loadpath, the greater the bending.  This is why I started this whole course with cables (and a bit on arches) as it’s the clearest way to think about structural behavior.

Here are some examples from simple to a bit trickier.  I provide some explanation and follow with diagrams.  The diagrams have blue for the frame and load, red for the direct loadpath, and green for the bending moment.

1)      So let’s take the simplest frame.  Pin joints at the base and a pin joint in the middle of the beam with a single load.  The direct loadpath consists of straight compression lines to the supports (i.e. what a cable would do in reverse).  So there is no bending moment at the pins (by definition) and linearly increasing bending moment as we move to the frame corners.  These are the critical bending locations.  Now, we draw bending moments perpendicular to the steel members but the frame corner is the frame corner…i.e. it has one bending moment.  So I like to draw it curved around the corner to remind everyone.

Note how the compressive thrust is countered by a reaction in the opposite direction. i.e. there is a vertical and horizontal component to the reaction (just as there is for a hanging cable).

2)      Now, if we keep everything the same but have the beam without a central pin, we still have straight compressive lines but they start higher up and cross through the beam.

What does this mean?  Well, a number of things:

1. First it means there are locations of zero bending moment in the beam.  We saw this with fixed end beams in the last post.
2. More importantly, we can immediately see the horizontal reaction at the base supports has been reduced.  The thrustline is steeper – it’s like having a cable with greater sag.
3. This all results in lower bending moments at the corners which were the critical locations.  And this is as it should be for an overall structural frame that is now stiffer than before.

3)      Last step for a basic frame with central singular load, if we make the base supports fixed, then our thrustlines cross both the beam and the columns.  This tells us a great deal about the deflected shape of the frame.  Remember how I explained about “smile” and frown” bending moments and their relationship to curvature in the last post.  Well, this lets us immediately draw the deflected shape as shown below.

Notice how I haven’t done any equations or used numbers.  This has all been qualitative work based on fundamentally understanding how structures carry loads.  This can be extended to other locations of the load or multiple loads.

As a last example, here’s the same type of diagram for a lateral load.  Notice how our thrustlines have both tension and compression.

Class on bending moments to architects

Over two classes (this and the next one) I covered bending moment and frame behavior. These are the two hardest concepts to understand as, unlike what we’ve seen so far, the behavior is not immediately visible. Appropriately, it also brings us to the end of the primary colors of structure. These complete the “types” of structural elements that exist in all structures.

Why do we care about bending? Simply because that’s how most things break.

It’s easy to get large stresses by bending. If you want to break a pencil, you don’t pull it, crush it, or shear it…you snap it. You hold it between your thumbs and index fingers (what we call: 3-point bending). Similarly when you want to break something larger you snap it over your knee. Lever action is a powerful thing.

I didn’t get to cover much in this class beyond bending moment diagrams for beams. My emphasis was in understanding the behavior so we know where critical locations are for the beams (which typically drives the structural depth required). I always like to refer back to the cable analogy. Cables can’t carry any bending, just axial load. Given bending is caused by an axial load at a distance from the support (think about holding a weight at arm’s length and how much harder your shoulder has to work than if your arm is by your side) the shape of the bending moment on a beam is the same shape a cable would take for the same load.

Hence, for a simply-supported beam (i.e. a beam just propped on end supports) we get the shape of cables we’re used to…
(Beams and loads are in blue. Bending moments are in red. Deflected shape in black.)

Now, if we grab hold of the ends of the beam and don’t allow them to rotate (what we call “fixed” ends) then the shape of the bending moment diagram stays the same, but its position is raised so we have less bending moment in the middle as the fixed ends can now help.

Notice this means there are locations where there is zero bending moment. That’s good for us to know when designing or evaluating an existing structure (assuming that’s the only load). But it also tells us something about the shape of the beam. This is because bending moment is directly linked to curvature.

You will appreciate that curvature means it’s curved (of course) but a beam can have positive and negative curvature. Put visually, it can be curved like a smile or like a frown. A simply-supported beam is all smile. But a fixed-end beam has a smile in the middle which blends into frowns at either end. So, at that crossover, there’s a moment when the beam is not curved. It’s inclined but not curved.

We talked a lot in class about gradient, curvature, and deflected shape and I found it interesting that many of the architecture students who are typically comfortable doing 3d double-curved geometry in rhino did not know about positive and negative curvature. However, more importantly, I also covered the following points:
• What are physical examples of a pinned support (i.e. “just propped”)
• Why do structural engineers draw conceptual beams with one fixed end and one roller end?
• What are physical examples of a fixed ends for beams?
• How can overhanging cantilevers act to offset bending moment at the midspan?
• How does bending moment relate to placement of reinforcement in concrete beams?

One question I found particularly fascinating: Why do we draw bending moment diagrams always going back to “zero” at the end? The fixed end of a beam has a bending moment but we (structural engineers) tend to draw a closed vertical end. I think this is just for neatness reasons but many of the architecture students were very confused by this. I’ve always taught bending moment diagrams in frames to avoid this (see next post) but hadn’t thought about it for just beams. Funny how easy it is to misrepresent something…