The global energy release in a solar flare is comprised of a large number of "elementary bursts", represented by an arcade of flare loops formed and heated by reconnection during the impulsive phase. Decades of observations have shown that the total emission of these flare loops may persist for several hours after the impulsive phase. How much energy is used to heat these loops? Why do they stay bright for hours after reconnection? To answer these questions, we explore observation-constrained modeling approaches. Flare loops are anchored to chromospheric ribbons, which respond prominently and almost instantaneously to reconnection energy release. Therefore, we can infer heating rates in hundreds of flare loops from spatially resolved light curves of flare ribbons. With these rates, we model the evolution of individual loops using the EBTEL code (Klimchuk et al. 2008) and compute synthetic total flare emissions in the corona. This method has been applied to model flare observations on the AIA scale and quantify heating energies in AIA loops. Most recently, we modeled a flare observed by the Goode Solar Telescope (GST), and derived heating rates in 77,000 GST-scale threads. Our studies suggest that the heating rate in a flare loop (of size 0.6") or thread (0.1") consists of an initial impulsive pulse lasting  about a minute followed by a prolonged tail heating for approximately 10 minutes. The prolonged tail heating maintains loop temperature and density, producing the persistent flare SXR and EUV emissions. Comparing the derived heating rates and loop properties across the AIA and GST scales provides interesting insight into the impact of the filling factor on flare modeling. Furthermore, we have developed an approach to reconstruct post-reconnection magnetic field, by fitting thousands of observed EUV flare loops to extrapolated magnetic field lines anchored in the flare ribbons. Our first experiment on a two-ribbon flare demonstrates that, as the flare progresses, newly formed loops anchored to newly brightened ribbons are described by field lines with decreasing twist parameter $\alpha$. This indicates that the post-reconnection magnetic field is not linear force-free. This study demonstrates, for the first time, the physical nature of the long observed strong-to-weak shear evolution in two-ribbon flares, and implies incomplete relaxation by fast reconnection. It also verifies a long time delay (20 - 30 min) between reconnection forming a flare loop and its appearance in EUV passbands, suggesting that the prolonged heating of post-reconnection flare loops is likely driven by continuous dissipation of current-carrying fields after fast reconnection.