In this work we explore theoretical limits to the self-assembly of terminal structures using molecular dynamics simulations of "patchy particles". We study how the geometry and interaction strength of particles designed to assemble two-dimensional "virus capsids" influences rate and yield of assembled closed capsids made of three identical particles. We extend this work to investigate distinct trimers and advance toward simulation infrastructure that can accommodate "smart" particles that can switch their state, defined by the number and stickiness of their patches. We find that the best assembling particles have “sticky patches” shielded by inert neighbors, which reduces opportunities for defect formation. We also develop automated tools for counting the assembled structures and identify the parameters that have the highest success rate. This work sets the stage for future studies of "smart" particles that can switch their stickiness state to accelerate assembly.