You can get very accurate estimate by using any root finding algorithm, e.g. bisection.
At given dimensions:

total area of 3 circles is 94247.7796. We can get very close to the same area for dissolved polygon by simply playing with radius:
import arcpy, math
from arcpy import env
resultFC="c:/scratch/scratch.gdb/dissolved"
target=94247.779608
result=arcpy.GetCount_management("POINTS")
nNodes=int(float(result.getOutput(0)))
##find lower limit for radius and set upper one
low=math.pow(target/nNodes/math.pi,0.5)
high=5*low
g=arcpy.Geometry()
iterCount=0
while True:
if (high-low)<0.01:break
mid=(high+low)/2
pgon=arcpy.Buffer_analysis("POINTS",g, mid, "FULL", "ROUND", "ALL")[0]
curArea=pgon.area
if curArea<target:low=mid
else:high=mid
iterCount+=1
arcpy.AddMessage("\nTarget = {:6.2f}\nArea = {:6.2f}\nRadius = {:6.2f}\nIterations = {}\n".
format(target,curArea,mid,iterCount))
arcpy.CopyFeatures_management([pgon],resultFC)

Script messages:
- Target = 94247.78
- Area = 94245.82
- Radius = 110.97
- Iterations = 16
As one can see the difference between target and result is a couple of square meters or 0.002%.
pi * r * r
, so r issqrt( area / pi )
, certainly, but if your points' buffers overlap you've got an NP Hard problem on your hands. It doesn't make sense that this would have needed to be solved. Perhaps you should take a step back and explain why you need this.