From a basic kinetic theory perspective the A-B collision rate $\theta_{AB}$ is given as a function of the mean speed and the number density.$$\theta_{AB} \propto \bar{C}_A n_B$$
If we are considering particles moving in 3D space, with velocity components ($C_1,C_2,C_3$) described by a Maxwellian distribution. Then what is the mean collision rate of the particles as a function of their $C_1$ velocity?
As a starting point, I derived an equation for the mean speed of particles with one given velocity component $C_1 = a$ $$ \bar{C}(a) = \frac{1}{\int_a^\infty \chi(C)dC} \int_a^\infty C\chi(C)dC$$ Where $\chi(C)$ is the speed distribution. This gives the following relation: $$ \bar{C}(a) = a+\frac{\sqrt{\pi}}{2} \big(\frac{2kT}{m}\big)^{1/2} e^{(ma^2/2kT)}\Big[1 - erf\sqrt{\frac{ma^2}{2kT}}\Big]$$
Using the basic model of the collision rate developed from kinetic theory suggests that the collision rate would be proportional to this mean speed value. However, from my data this seems to significantly overestimate the number of collisions. Results seem to agree better with a $\sqrt{\bar{C}_A^2+ \bar{C}_B^2}$ scaling. Does anyone have a 3D hard sphere particle simulation that could be used to shed any light on this?