Aggregation within particle suspensions is often sought to enhance settling. Aggregate settling velocity and density predictions are directly related to the drag coefficient. Despite extensive studies on aggregate drag forces, there remains a lack of explicit correlations suitable across a range of properties and flow conditions; most employ implicit approaches assuming non-spherical, porous aggregates are rigid spheres. Particle-resolved Direct Numerical Simulation (DNS) was used to directly study drag on aggregates with various properties (void fraction, shape, orientation, fractal dimension) for Reynolds numbers between 0.13 and 2256. Testing previous implicit models against DNS results, the smallest error achieved was ~21%. From simple modifications accounting for aggregate shape and orientation, three new implicit models are proposed that substantially improve drag prediction, reducing error to ~11%. Using DNS, a new explicit correlation is proposed to calculate drag coefficients for porous aggregates of arbitrary shape spanning a range of properties with only ~6% error.