Crack width in mortar sample

[[ya...]]

 I am researching crack width in reinforce mortar with carbon fiber samples.  I'm straggling mapping the thickness of the cracks. It microscopic cracks.  I wonder if with the photo resolution I use is enough? I have attached an enlargement of the crack in the software GOM. I would really appreciate if someone can answer me. I attached one the original photo that im import to the analysis and enlargement with "zoom in" for you to see the quality of the crack and if I can measure the width.

[[Iv...]]

Hi Yarden,

the resolution of your surface could be higher to measure more detail. You are losing a lot of pixel because you also measuring the area wide next to your specimen. I would recommend to rotate the camera through 90° and get closer. You are increasing the amount of pixel on your surface and so you can be more precise during the inspection. 

The next point you can improve is the stochastic pattern itself. If you get closer to your surface with your camera you need a finer pattern. You do have a good pattern style on the left side slightly under the middle space (s.attachment). 

I hope this input can help you to reach your measuring goal. 

Greetings, Ivan

 

[[Th...]]

Please sign in to view this username.

 already mentioned some ideas, how to improve your measurement in general.

But as much as I see, you are interested in crack width evaluation, too. So I just want to add a possible workflow for that.

The strain values will only provide an idea where cracks are and do represent the crack width only indirectly by the strain reference length. So if you assume there is no strain left in the mortar material itself after cracks happened, you can get an estimation about crack width by these two workflows:

1. Check the surface component for "Displacement Y".
   1. Apply a "Rigid Body Motion Compensation" (RBMC) to the lower part of the specimen where no crack happened.
      1. Select such an area of your surface component.
      2. Create as Component Region ("OPERATIONS -> Component -> Component Region...")
      3. Apply the RBMC to this Component Region ("OPERATIONS -> Alignement -> Rigid Body Motion Compensation -> Transform By Component..." and select the Component Region).
   2. Create a Section perpendicular to the X-Axis through the Surface Component (Select all points of the Surface Component, "CONSTRUCT -> Section -> Single Section... -> Reference plane: Plane X, Position: ...").
   3. Check this section for "Displacement Y".
   4. Switch on the diagram
   5. Interpret the step height as crack width. / Relate it to the noise and average height of the values in the "undeformed" areas.
2. Check the distance changes of the "undeformed" areas between cracks.
   1. Create two sections perpendicular to the Y-Axis through the Surface Component in "undeformed" areas between cracks - one on each side of a crack. (Select all points of the Surface Component, "CONSTRUCT -> Section -> Single Section... -> Reference plane: Plane Y, Position: ...")
   2. Create a Continuous Curve Distance element ("CONSTRUCT -> Distance -> Continuous Curve Distance..." and select the two sections on both sides of the crack)
   3. Check this element for Distance (or Distance Y).
   4. Switch on the diagram
   5. Interpret the Distance values as crack width. / Relate it to the noise and average height of the values.

I would expect an crack width accuracy of 0.1 Pixel with your images. So as the cracks are visible in the images I suggest they have an overall width in the range of 1-2 Pixel. With the improvements suggested by

Please sign in to view this username.

you can reach an accuracy in the range of 0.02..0.05 Pixel.

[[Pa...]]

Please sign in to view this quote.

Hi Theodor,

 

Thanks you and

Please sign in to view this username.

 for the explanations.

I have a question related to your post.  In the last sentence you mention the expected accuracy of 0.1 Pixels (or 0.02-0.05 Pixels if the improvements mentioned by

Please sign in to view this username.

 are applied. How do you obtain those numbers? How can we get a subpixel accuracy to measure displacements between 19x19 pixel size facets?

 

Thank you!

[[Th...]]

The accuracy values I mentioned are just experienced values. In 2D it is hard to prove them as you have always in depth movement which will scale your coordinates, but with calibrated 3D systems from two cameras you can make a zero-deformation test by just moving a stiff object and check for the deformation noise you get. So on a 1m long field of view you usually get noise-deformations of ~20µm with a 1MPixel camera. So 1 Pixel represents more or less 1mm and 20µm are 0.02 Pixel therefore.

Such an accuracy is reached by DIC algorithms using grey value gradients and fitting them through the pattern. So if the pattern only moves slightly the gradients will change even if the pattern doesn't move a whole pixel aside.

[[ya...]]

Please sign in to view this quote.

Hi Thepdor, 

I am almost unfamiliar with the software at all so it is very difficult for me to perform analyzes even though I have tried to watch tutorial videos. Can I send you individual pictures of one sample and be able to videotape the sequence of actions? I know this is a strange request but I am helpless

[[Th...]]

Sorry for late reply.
If this is still an issue for you, you can upload some data to https://filetransfer.gom.com/105/807/public.html
Is there a way to get in direct contact to you, to provide some evaluated data or additional information how to use this data in free GOM Correlate?