In the BAUT Forum there's is topic about binary planets , and about  the question if binary planets are possible .
http://www.bautforum.com/space-astronomy-questions-answers/90064-binary-planets. html
Of course this subject demands for a gravity simulator simulation  
The file in attachment starts with 100   0.01 Jupiter sized planets at 5AU around a central    
solar mass. ( in red ) . To this system were added 100 Earth sized planets at 4.9 AU ( in blue ) .  
All planets are in the ecliptic plane and have initial zero eccentricity , with random mean anomaly .  
So we have two rings of objects .    
One can think of it of being a proto planetary disc .

Lets zoom in to body 59 and center it ( red spot in the middle) .  
 
This body merged before the start of the animation with another 0.01 MJup, but this is not really important .  
Watch however the two Earth sized planets approach from the left .  
As they are approaching Body59 they clearly interact at the end .  
One body gets absorbed , the other one changes its course and ...indeed is captured !!
 
This is the first time I was lucky enough to see a capture in the sims I've made so far  .  
I think in the above sim there's  another capture with body 78 , but I haven't animated it .  
 
It's interesting to see that the capture is the result of the interaction of 3 bodies ( in fact 4 if the sun is also counted) .  Captures where fewer bodies are involved are almost impossible or even impossible I think .  
 
The above sim was run for some more hundreds of years . The number of bodies decreases due to further merging or expulsion. Fewer active bodies remain , decreasing the chances of having multiple encounters .  
I saw no captures anymore after the initial few years.