The paper presents empirical data from German freeways, showing various congested traffic states near road inhomogeneities such as lane closures, intersections, and uphill gradients. These states can be localized or extended, homogeneous or oscillating, and can coexist, such as moving localized clusters and pinned clusters at road inhomogeneities. The findings are consistent with a theoretical phase diagram for traffic near on-ramps. A novel continuous microscopic single-lane model, the "intelligent driver model" (IDM), is used to simulate these situations, reproducing the observed states by describing inhomogeneities with local variations of a single model parameter. The results are interpreted by formulating the theoretical phase diagram for bottlenecks more generally, showing that a local drop in road capacity has a similar effect to an on-ramp. The IDM model is validated through comparisons with empirical data, demonstrating its ability to simulate real traffic breakdowns at bottlenecks with empirical boundary conditions. The study supports the idea of the suggested phase diagram of congested traffic states and provides insights into the dynamics of traffic breakdowns behind bottlenecks.The paper presents empirical data from German freeways, showing various congested traffic states near road inhomogeneities such as lane closures, intersections, and uphill gradients. These states can be localized or extended, homogeneous or oscillating, and can coexist, such as moving localized clusters and pinned clusters at road inhomogeneities. The findings are consistent with a theoretical phase diagram for traffic near on-ramps. A novel continuous microscopic single-lane model, the "intelligent driver model" (IDM), is used to simulate these situations, reproducing the observed states by describing inhomogeneities with local variations of a single model parameter. The results are interpreted by formulating the theoretical phase diagram for bottlenecks more generally, showing that a local drop in road capacity has a similar effect to an on-ramp. The IDM model is validated through comparisons with empirical data, demonstrating its ability to simulate real traffic breakdowns at bottlenecks with empirical boundary conditions. The study supports the idea of the suggested phase diagram of congested traffic states and provides insights into the dynamics of traffic breakdowns behind bottlenecks.