As @Semiclassical wrote,
$$x+a\log(x)=k \implies x=a W\left(\frac{1}{a}e^{\frac{k}{a}}\right)$$ If you cannot use Lambert function, you will need a numerical method and take into account the fact that $\forall t \geq e$
$$W(t) \geq\log (t)-\log (\log (t))+\frac 12\frac{\log (\log (t))}{ \log (t)}=f(t)$$
$$W(t) \leq \log (t)-\log (\log (t))+\frac e{e-1}\frac{\log (\log (t))}{ \log (t)}=g(t)$$
In the past, I minimized
$$\Phi(a)=\int_e^{1000} \Big[a\, f(t)+(1-a) \,g(t)-W(t) \Big]^2\,dt$$ and obtained $a_{opt}=0.57958$ for which $\Phi(a_{opt})=0.134235$; for comparison, $\Phi(0)=35.6714$, $\Phi(0.5)=0.80423$ and $\Phi(1)=18.8333$.
Making numbers rational, a good approximation is
$$W(t) \sim \log (t)-\log (\log (t))+\frac{11133}{11659}\frac{\log (\log (t))}{ \log (t)}$$ Over the considered range, the maximum error is $\sim 0.007$.
With such an estimate, Newton method would converge to high accuracy within a couple of iterations.
Edit
For illustration, let use use $y=10^{99}$, $b=\frac 12$ and $a=\pi$. This would give $t\sim 4.36783\times 10^{30}$, $W(t)\sim 66.3531$ and then $x\sim 208.454$ while the "exact" solution is $x \sim 208.466$.