Training Day

Training Day

January 26, 2018

We started a full day of testing with a 7 A.M. coffee run, followed by loading up the many suitcases of equipment and driving to test our first structure. Although it was very early and I was not fully caffeinated, I was very excited to learn from and work alongside so many experts.

The plan was to focus on buildings with simpler configurations, and today we started with schools that sustained a small amount of non-structural damage in the September 2017 earthquake. The research team is comprised of faculty and students from five different universities, each with their own approaches to sensor set-up and data recording. In light of this, the full team worked together on the first school building to establish a baseline for data collection, exchange tips, as well as develop a instrumentation and testing protocol that would make our subsequent 3-4 hour test per building run as smoothly as possible.

Sensor set-up for the first building took longer than expected; too many people were working on the same tasks and communication via phone, walkie talkie, and even yelling became a struggle. Information crawled out from our honorary Commander-in-Chief of Data Acquisition (DAQ) through the network of people involved. It was certainly a learning process.

Figure 1: First School Building Tested by the Full Research Team

Let’s pause for a moment and discuss what our research objectives are. In my first blog post, I briefly mentioned that we are measuring the dynamic response of buildings subject to ambient vibrations produced by sources such as wind, cars passing by, or perhaps people walking on the floors if the building is occupied. The goal is not just to collect dynamic response and damage data for the small subset of buildings we are able to test during this research trip, but to use our data to develop prediction models to determine basic dynamic properties of other untested buildings in the building stock here. To measure and record ambient vibrations, we strategically position a number of accelerometers (comparable to the size of a D battery with connections for data output and mounting) throughout the building for an extended period of time, typically 30-90 minutes. In some cases the ambient vibrations the buildings are insufficient to generate large enough vibrations, and our team of engineering experts resorted to producing their own random forced vibrations by jumping like frogs in the middle of floor slabs and running suicides with sudden stops on roofs.

The sensor set-up process for ambient vibration testing seems straight forward— position accelerometers, plug them into their respective channel, run a sensor-data acquisition systems check, and sit back while the data records; yet, in reality it is a lot of work. First, the DAQ Commander sets up the main laptop while someone else determines the best locations for the accelerometers, usually with a few locked rooms between us and these ideal locations. Next, we run up and down to various floors of the building like little spiders with hundreds of feet of BNC cables in tow to connect the main laptop station to the selected accelerometer locations, while keeping track of where each of the cables lead to plug them into the correct data channel. Then we decide which accelerometers are the most reliable and place those in locations deemed critical to collect the dynamic response of the building. After that, we take our aluminum angle mounts and screw the accelerometers in a perpendicular orientation. Once the accelerometers are very carefully placed into their mounts, all of the wires are connected to accelerometers at one end and DAQ at the other. The whole set-up is triple checked to insure the sensor – cable – DAQ channel connection is correct and the sensors and DAQ equipment are functioning properly. We can now begin acquiring data, finally!

Figure 2: Deployment of Cable and Accelerometers

In real-time, we were able to monitor the recorded structural behavior by filtering and conducting a fast Fourier transform, FFT, on a portion of data coming in from each of the accelerometer channels. Each channel of data is plotted in a frequency graph that appears in the matrix on the computer screen pictured below. In most cases, the peak in the graphs indicates a higher response of the building at that particular frequency due to the random vibrations, likely because it corresponds to the resonant (or natural) frequency of the building. The ability to rapidly assess data in the field allows us to ensure data fidelity without conducting a full, more time-consuming, analysis. Often our initial Fourier transform helped us identify whether it was necessary to retest accelerometers or investigate the source of unexpected results. After troubleshooting these issues, we can finally relax for the 30-90 minutes of data collection until it is time to do the whole set-up process in reverse when we pack up. 

Figure 3: Real-time Monitoring of Building Response

After testing the first building with the full team, we split up and were planning for each sub-team to test two other buildings; however, daylight was running short and each only measured one more building. The collective four buildings completed on the first test day was a success, and the professors were ready to move on to the exciting part: examining the data to see what they could learn about building response and the quality of our sensor set-ups.

We did see a very interesting frequency appearing in nearly everyone’s data, which was particularly strong for ground measurements. Based on the warnings from local engineering faculty and practitioners yesterday of soil-structure interaction, we postulated this may be the natural frequency of the soil on which the building is built. Though we are still debating the exact details of what is occurring in the data, we are all in agreement that the soil is intervening in the measured results. We are all pretty excited to see something unusual in the data and have some indication of what is generating these results. There is certainly more we need to do to understand Mexico City’s unique soil condition and how it affects structural response. After returning from our research trip here in Mexico City, we will need to carry out more in-depth analyses to obtain more accurate building natural frequencies, mode shapes, and soil-structure interaction.

I will end this with some pictures of the second school building we saw today. We are looking to repeat the process for another two schools tomorrow!

Figure 4: Second School Building Tested on January 27, 2018