Particle size is connected to many of the other parameters of seed particles in a PIV system. With size affecting visibility, flow conformity, and being integral in relation to pixel size. A rough number for ideal particle size is 1-100um though sizes in the nm and mm?s have been used for certain PIV applications. With smaller sizes being necessary for micro-PIV methods and larger sizes being a requirement for large scale flow visualization. The importance of size is related to how truly the tracer will follow the flow with particle diameter having the largest effect on stokes number, which is a representation of flow tracer fidelity. Though when particle size becomes too small it can be difficult to confirm that the tracer is not being affected by minor currents or other factors within the fluid. Also as size decreases visualizing the spheres can become quite challenging. However, the stokes number can provide a decent representation of how well particles follow the flow. Though, the stokes number is an approximation based on assumptions and therefore can only provide a useful representation rather than a confirmation of tracer fidelity.
Particle size distribution:
A parameter that should be considered in conjunction with particle size is distribution. As particles in the sizes used for PIV are so small that no meaningful quantity of tracers can be produced in a specific size and rather size ranges need to be considered. With tighter size distributions, there will be less error attributable to differences in visibility of particles and a better approximation of how well each particle being used will conform to the flow. For example fluorescent red polyethylene has multiple size ranges available (10-22um, 10-45um, 10-90um, and 10-150um). With tighter size distributions being more difficult to obtain and as such being more expensive. Raising the question of what the trade off between price and size distribution is. Wide distributions can be used within PIV, however they may necessitate further image processing and may reduce accuracy of measurements. Therefore, there is no perfect size distribution choice. Though, with the understanding of what is available the choice of a correct size and size distribution can be determined.