They have masses of J-7s, J-8s, Q-5s, and even early JH-7s and J-11s to replace in coming years. Something like 500 J-7s, 200 J-8s, 120 Q-5s, a currently small number of JH-7s that will grow as the fleet becomes older, and about 100 legacy J-11s.
That is something like 300 long range fighters, 500 medium weight shorter range fighters, and 120-150 strike aircraft.
-J-20 is obviously a necessary plane for the future, but even then its numbers will be limited. 200-300 at most for a full production run, if even that.
-J-10B/C still obviously has a role in PLA as a lighter, shorter range fighter, so we can't cut J-10s out of production anytime soon given the number of short range J-7s that still need to be replaced. I'm not saying J-10s will replace all J-7s given how many J-7s there are -- I expect an FC-31 derived fighter to eventually supercede J-10 production in the medium weight fighter role.
-Then there is the question of what will replace long range air to air (J-8s and early J-11s) as well as strike (held by Q-5s and early JH-7s). J-20s can make a dent into the long range air to air component of the fleet which need to be replaced, but it'll probably be post 2030 for its full 200-300 production run to end, and by then other flankers like J-11Bs may start to be replaced by J-20 too not counting the earlier J-8s and J-11s.
So there will be a deficit in long range air to air that will be felt earlier rather than later, and that is what J-11D is meant to alleviate imo. I can foresee a production run of 200 J-11Ds, with accompanying upgrades for older J-11Bs which will all occur during the first four or five years of J-20 production, all to be set for both PLAAF and land based PLANAF. J-16 on the other hand will look to replace the JH-7s and Q-5s, with a final production run of maybe 150 aircraft. Of course, it might be possible for them to buy only J-16s or only J-11Ds to replace the J-8s, J-11s, JH-7s and Q-5s with a single aircraft, But that would mean over 300 of only one type of either J-16 or J-11D, and if the two aircraft are inevitably more suited to strike and air to air, then PLA might end up with overcapacity in one role and undercapacity in another -- it is where multirole may not be cost effective, especially say if J-16s are more expensive with its additional cockpit, more strengthened structure, more A2G oriented avionics, compared to J-11D, with that same cost distributed for all 300+ spaces to be filled. But separating the roles, while using a common but less structurally reinforced airframe for J-11D, with J-16 derived avionics and radar oriented for A2A and removing some of the fancier A2G functions, and removing the second cockpit, may allow better specialization of roles while reducing the extra cost of having to pay for the added bits and pieces on all 300+ spaces, but also retaining as many common components to both aircraft as possible making the lifetime cost of both aircraft lesser than if they were fielding two completely different unrelated airframes with unrelated avionics.