Variation in stress values in an element

Question

Hi, 
 
 I am using PLAXIS 3D to model embankments on columns. I made a line cross-section and measured cartesian vertical stresses along the centerline of the columns. But the table shows a wide range of values for an element. I am attaching a screenshot of the results. From the screenshot, it can be seen that for element 33475 the stresses are negative whereas for element 33466, 33445 the stresses are positive. For element 33505 stresses are both positive and negative within the element though these are elements on one column. Also within the element itself, there is a wide range in the stress values varying from -46 to -4000. One node on the column element appears positive and the other negative. I don't know why this is happening because under the embankment crests on the column the load is always compressive. I had used fine mesh initially. Then I found only 6 elements on columns. So I refined the mesh to a coarseness factor of 0.05 for only columns. But still the same issue with the vertical stress values. How do we take the average stress on the column if there is such an issue?

Stefanos Papavasileiou · Answer

Dear Reshma, 
 When you are drawing a line cross-section you are drawing a line that crosses through the elements. It is logical depending on the element size along the cross section to cross an element fewer times than anticipated. You can actually distinguish how many elements you cross if you check the lines in the cross-section: 
 ... 
 Regarding the refinining, it is important to refine in the area you want to draw the cross section otherwise it might not make any drastic impact on the number of results you get. 
 About the results, note that a cross section interpolates or extrapolates results depending on the origin of the raw results, e.g., displacements on nodes, stresses on stress points. That can explain the "jump" in the results as if the element is big enough it can happen that you get an extrapolated result like that. Consider also using the Result smoothing option:

Dennis Waterman · Answer

Hi, 
 Well, first of all, what plot are you looking at? If it's a connectivity plot then it's a visualization feature: in a connectivity plot several measurements (like interface thickness) get a artificial size in order to easier see how things are connected. But that may sometimes look weird. 
 If it's a deformed mesh then the important question is, whether not simply ground level is settling while the piles are not? Hence, it's not said this is a mesh problem, it could in fact be either a visualization feature or a perfectly fine result. However, with the limited data here that is impossible to say. 
 With kind regards.