Biomolecular light-harvesting antennas operate as nanoscale devices in a regime where the coherent interactions of individual light, matter, and vibrational quanta are nonperturbatively strong. The complex behavior arising from this could, if fully understood, be exploited for myriad energy applications. However, nonperturbative dynamics are computationally challenging to simulate, and experiments on biomaterials explore very limited regions of the nonperturbative parameter space. So-called quantum simulators of light-harvesting models could provide a solution to this problem, and here we employ the hierarchical equations-of-motion technique to investigate the recent superconducting experiments of Potočnik et al. [A. Potočnik et al., Nat. Commun. 9, 904 (2018)] used to explore excitonic energy capture. By explicitly including the role of optical driving fields, nonperturbative dephasing noise, and the full multiexcitation Hilbert space of a three-qubit quantum circuit, we predict the measurable impact of these factors on transfer efficiency. By analysis of the eigenspectrum of the network, we uncover a structure of energy levels that allows the network to exploit optical "dark" states and excited-state absorption for energy transfer. We also confirm that time-resolvable coherent oscillations could be experimentally observed, even under the strong, nonadditive action of the driving and optical fields.