A novel model based on the Eulerian-Lagrangian approach was introduced in order to capture the physics of particle-fluid, particle-wall, and colloidal particle-particle interactions through four-way coupling in low-flux nanofluid-based direct absorption solar collectors (DASCs). This was accomplished by means of a computationally efficient particle-particle-interaction detection algorithm implemented within the frame of an in-house fluid-particle coupling algorithm that allowed for the simultaneous evolution of the carrier and particulate phases. The optical, thermal, and dispersion effects of particle-wall and colloidal particle-particle interactions were investigated over a range of particle volume fractions (phi(p)) and flow Reynolds numbers (Re). It was found that the outcome of a particle-wall collision is either particle deposition or rebound, depending on the balance between kinetic energy of the incoming particle and particle-wall van der Waals potential energy. Brownian motion was established to be the main deposition mechanism for nanoparticles in the boundary layer. As Re decreased, Brownian motion of nanoparticles dominated over convection, which led to higher deposition rates. Deposited particles were found to cause greater attenuation of solar radiation near the collector upper surface, especially in the visible region of the solar spectrum. Whereas in the infrared region, radiation attenuation became less sensitive to particle deposition. The drop in collector efficiency when particles are allowed to deposit relative to the case when particles are not allowed to deposit was highest for low Re and phi(p). Yet, regardless of Re, an increase in particle surface potential led to a better dispersion of nanoparticles. Furthermore, colloidal interparticle interactions were found to have a negligible effect on collector performance. The results highlight the important effects of particle deposition and non-uniform particle distribution on the performance of DASC systems, which cannot be captured using conventional single-phase models.