Consider a universe with a nonzero curvature and matter.
One can write the Friedmann Equation in this universe as such:
$$\frac{H(t)^2}{H_0^2} = \frac{\Omega_0}{a^3}+\frac{1-\Omega_0}{a^2}$$
Where $H(t)$ is the Hubble parameter, $\Omega_0$ is the density parameter for matter (since it is the only component in the universe aside curvature, I will ignore the subscript) and $a$ is scale factor.
Now consider a universe where $\Omega_0$ is greater than 1 (positive curvature), leading in a negative term of $\frac{1-\Omega_0}{a^2}$. It is obvious that the universe will expand first, then come to a halt when $H(t) = 0$.
However, when $H(t)$ turns negative, LHS of the equation is again positive, which is no difference from the expanding scenario. Thus I conclude this universe will expand forever, however this universe will meet its end in the Big Crunch. Which means the universe will first expand until it reach a maximum of scale factor, then contract until everything crash together.
Where did I made a mistake?