Monday, November 22, 2010

A few lessons from the lab

The past year, I've been conducting an enormous amount of large-scale experiments (my funder is quite generous). Tomorrow I'll have finished large-scale test number 98, and I will have tested 14 slabs (half-scale slab bridges) and 9 wide beams/slab strips. In this past year, I've learned by trial and error how to get a good routine and manage the testing program.

1. Plan
There's much more to plan about doing an experiment than the experiment itself, obviously. You will need a planning to arrange the delivery of your material, the fabrication of your specimens, side testing and then the course of the experimental program itself. Even though your planning will change (mine is changing frequently because of unexpected delays: sickness of technicians, the carpenter being unavailable,... ), you still need a planning to make sure you won't forget anything and to estimate the time every step will take you.

2. Prepare
Don't walk into the lab with empty hands. When I go to the lab, I have my fixed set of documents along with me: a sheet with predicted values to immediately compare my test result to and a table that I fill out during testing in which I note down observations for every load step. I also take the camera along with me to photograph the failure patterns (and sometimes something to snack on).

3. Classify
Don't wait until months after testing to organize and classify your data and notes. Right after a test, I save the raw data and pictures into their respective folders. I add results into tables which I build up test by test. I write a short summary in my lab book. You get the picture: take action immediately to keep it all neatly organized.

4. Automatize
If you carry out a large test series, try to automatize your data processing as much as possible. Even though programming might take you a few days, the benefits will return to you in the long run. I for example wrote a Matlab code that reads my raw data and returns all plots and numerical values I need for the considered test. Programming is by far not my specialty, but I learned something while writing my code, and now I generate all plots in just a few seconds. That sure is a win.

5. Write
Don't wait until months after the experiment to write your report. With my large testing program, I sketched an outline of how I want to discuss every specimen and test and made it a good habit to complete the report after finishing a specimen. Initially, I could remember every detail of every test, but now I notice that every now and then I have to check in my report to verify what happened during a certain experiment. There's a limit to our memory, so you'd better not wait until you start confusing to write down your results and observations into a report.

6. Smile
You probably won't be working all by yourself if you carry out a series of tests. Respect the technicians you work with, don't treat them as "inferior" because they are not pursuing a doctoral degree. Arrive in time when you need to fabricate a specimen and technicians are volunteering to help you. (I'm writing this because I've seen bad examples, unfortunately). It might be hard work (my muscles used to hurt terribly after working in the lab for the first months), but make sure you enjoy. Joke around, have fun, and above all: smile.

No comments:

Post a Comment