you might research the early NACA reports for the introduction of CSK rivets. Or maybe "Fluid Dynamic Drag" by Hoerner.
I remember an old story where some engineers wanted to replace protruding head rivets with CSK for the drag reduction on some post-WW2 bomber. The wise chief designer asked "what is the performance gain from CSK rivets ?" answer "4 mph"; 2nd question "what is the penalty of the upper turret ?" answer "about 40mph"; conclusion "I'll keep the protruding head rivets (and save all that production time) and re-design the turret (now go away before I get cranky)".
The answer to your question is both simple and very hard !!?? On one level the impact is small (like fuselage rivets, with the heads deep in the boundary layer), on another (like wing Leading Edges) the impact is very large as they'd trip the Boundary Layer.
Start with defining the boundary layer 'transition-line' for every component... at specified flight conditions.
For attached/smooth boundary layer areas [forward of transition line], protruding/rough-anything... including rivets and rough paint can create drag increments. Exception: drag from [for instance] rivet-heads in-a row that all align within a single disturbed airflow field are have drag that is not linearly additive.
For detached/turbulent boundary layer areas [aft of transition], protruding/rough-anything... including rivets and rough paint etc... within the separated boundary layer, essentially contribute no drag increment to the Acft.
WWII fighters were built with this philosophy: In the flow critical areas forward use flush riveting only; however, in the boundary transition, aft, OK to use protruding head rivets. This philosophy led to low-cost, speed-of-assembly [large numbers, lightest construction]... and no significant effects on high speed performance. NOT '100% flush-riveted pretty'... but functional.
Now simply define the numbers of protruding head rivets [X] in the flow critical areas, define the average 'drag cost' per rivet head [Y]; and then do the math... X*Y=Z-total rivet-drag [that must be] overcome by thrust. NOTE: properly installed flush fasteners in the critical boundary layer, have essentially NO drag increment.
REALLY IMPORTANT... Read Hoerner's Fluid Dynamic Drag for a much better understanding of all this complexity.
PS... The drag-effect of a bad paint job can sometimes be worse than protruding head rivets, for all Acft.
