Hello Elmira. Beer-Lambert (or strictly Bouguer's Law - formulated in 1729) basically deals with pretty dilute systems (solutions of colored material or dilute suspensions) - typically in the ppm region up to, maybe,1 volume % - and simply indicates that the fall off in light within such systems is exponential. The implicit assumption is that light is only lost once. With multiple scattering (particles in suspension) - probably you're thinking of light scattering here - then light is scattered off another particle before it reaches the detector. This is to be avoided in light scattering as it has the effect of diminishing the apparent size of the particulate system. The effect of this in a Beer-Lambert plot is that there is no longer a linear dependence (when the correct parameters are plotted - say transmission versus concentration) and the graphical plot flattens as if it were approaching some saturation. Thus in this region, the Beer-Lambert law is not valid. For some more you can do see:
June 22nd 2010 Determination of the optical properties of CeO2 and CaCO3 by the volume concentration experiment - the 3 S's, Beer-Lambert and Mie
Nothing really important to add to the former answer, but you may find interesting the approach to the B-L law found in works dealing with transmittance and reflectance of layered media, in which both absorption and scattering are taking place. Simulations (ie via Monte-Carlo) produce a clear departure from B-L transmittance behavior when the scattering probability is increased.
Some of these simulations are easy to reproduce. A description of the construction of such codes -made for Kubelka-Munk reflectance calculation- can be found in: JOSA-A, 33, N10, 2016 ; http://dx.doi.org/10.1364/JOSAA.33.002053
Section 4.3 reproduces the B-L behavior found in transmission for low-scattering conditions.