Concrete failure analysis is a complex process. That’s because engineering is a conservative profession. After all, who wants us to take unnecessary risks when public safety and millions of dollars are at stake? Engineers use factors of safety to take into account everything from variability in material strengths to magnitudes of applied loads. So when things go badly wrong with a concrete structure, there’s usually more than one cause. In large construction projects that almost always means more than one entity contributed to the failure. How do we sort out what caused the problem and who is responsible?
A thorough, systematic scientific methodology helps us make sure we don’t miss anything in our concrete failure analysis. It’s helpful to start with the big picture, then move progressively to smaller and smaller details. Once we think we have all the details, we put them back together to see whether they make a complete, coherent, non-contradictory whole. If we see any gaps or contradictions, we investigate further until they are resolved.
Get the big picture
A key question is why we need a concrete failure analysis in the first place. Observations from the client such as what they noticed when can help guide the investigation.
Before going to the site, the investigator needs to get as much background information as possible. Most of this information will be in the form of documents: project plans and specifications, meeting minutes, correspondence, change orders, concrete batch tickets, field notes, test data.
Additional background information may include weather data from a website such as Weather Underground. Weather conditions before, during, and immediately after concrete placement, along with any soil data, help define the environment. Sanborn maps can provide information about potential environmental contamination from previous uses of the site and the surrounding area.
To make the most of the site visit, it is helpful to compile a booklet with maps of the area, at least one copy of the site plan, data sheets, and a photo log. That way, everything is in one place.
Sometimes legal agreements or distance from the office limit access to the site. In that case, you will need to take samples on the first visit. It is important to determine how to extract and handle them. Certain sampling methods may destroy or obscure the very information we are interested in. Beton consults with the laboratory scientists who will be examining them to alert us to any precautions we should take.
Visit the site
On site, the first task is to get an overview by performing a though visual examination. This will set the context for the rest of the investigation. We may sketch a map of the cracks, sound the surfaces to detect delamination, perform nondestructive tests, or extract cores. We note all of these observations on the site plan, along with the locations of any photographs.
In some cases, the cause of the failure is entirely structural – that is, the imposed load exceeded the load-bearing capacity. The crack pattern will indicate whether the failure was due to flexure, shear, or compression. Nondestructive testing will show the location of the reinforcing bars. Some cores may be necessary to provide information about the strength of the concrete, or to confirm that the bar sizes were as specified.
Where the observed distress cannot be fully explained by structural causes, or the performance of the materials is in question, materials science comes into play.
Selection of the locations for coring is essential to a good concrete failure analysis. It is almost never possible to remove and test enough cores for a proper statistical representation of the site. Instead of a random selection, nondestructive testing or visual observations can provide guidance for the core locations.
If we want to know the cause of delamination, for example, we need cores of areas about to delaminate, not where the surface has already detached. Sounding will indicate where these areas are. Sketches of the delaminated areas and core locations on the site plan provide the context for the testing.
Some engineers submit “blind” samples to the testing laboratory, thinking that this practice will elicit “unbiased” opinions. Beton prefers to give the technicians and scientists as much information as possible about the site and the nature of the investigation. This information helps us determine which tests to conduct and how to prepare the samples.
For example, if we suspect corrosion, microscopy specimens must be prepared without water (which will dissolve the salts) and without epoxies that contain chlorides (which will be impossible to distinguish from chlorides that were present already). If we suspect sulfate attack, specimens for the electron microscope should be coated with carbon rather than gold, which interferes with the detection of sulfur.
In planning the laboratory investigation, it is helpful to keep in mind the sample dimensions for each type of laboratory test so that the larger scale can set the context for the smaller. Site observations can be on the scale of tens of meters down to millimeters. Tests of cores for engineering properties such as compressive strength are on the scale of hundreds to tens of millimeters.
The stereomicroscope provides information on the order of hundreds to tenths of millimeters. Optical- and electron microscopes provide complementary information on roughly the same scale, from tens of millimeters down to tens of micrometers.
Although the specimens for the various analytical techniques are minute, they can represent a much larger scale. It may make sense to crush a large sample and take a small, representative portion of it for testing to give an average concentration. At other times it’s better to get the concentration near the surface separately from the concentration farther from the surface. For example, the sampling method from ASTM C1556 can help us predict the remaining service life of a bridge deck.
It is helpful to have some overlap between test methods as a “reality check”. Even the best test methods can have biases or blind spots. Testing the same sample in more than one way, or examining companion specimens with overlapping techniques, can provide needed correction, supply additional details, or lend confidence to the data. Once all the results are available, the whole investigative team works together to make sure we have a complete, coherent explanation of all of the data.