Near the wall and inside the boundary layer the mesh quality is even more important. With a low quality mesh it is impossible to capture the turbulence effect. That is why it is always recommended to have a fine mesh near the walls and inside the boundary layer.

As Mr Massoud mentioned a fine mesh near wall is desired to capture the turbulence effect. For a better understanding see the paper in the link.

http://www.iaeng.org/publication/IMECS2009/IMECS2009_pp2165-2170.pdf

Hi, you can read my paper to know the importance of mesh near the wall.

http://www.sciencedirect.com/science/article/pii/S1875510017300434

Thank you all for replying. As we all know, poor quality cells may cause round-off errors and convergence difficulties which result in non-reliable CFD results. My point is, as Dr. Bensayah mentioned,  if the position of these poor cells is located at the outer wall of a straight duct, where the fluid seems to be completely laminar (see the attached figure). So, can these cells be less problematic at this specific position? Is there a possibility that they may influence the solution in the turbulent flow region after the flow passing the straight duct. That is what I'm thinking about.

if your study is highly dependent on the

low Reynolds flow modeling close to the wall, you should use enhanced wall treatment to model the boundary layer and flow close to the wall accurately. The mesh size close to all can be evaluated using dimensionless wall distance or y plus (A non-dimensional wall distance for a wall-bounded flow) which is a measure to specify

whether the cell is located in viscous sub layer, buffer layer or fully

turbulent region. It can be defined as y plus ¼ u*y plus

w where u is the friction velocity at the nearest wall, y plus is the distance to the nearest cell to wall and w is the local kinematic viscosity of the fluid. For using enhanced wall treatment model, the wall adjacent to the first cell size must be small enough to achieve y plus in order of 1.