A series of coronal mass ejections (CMEs) erupted from the same active region between 4‐6 September 2017. Later, on 6‐9 September, two interplanetary (IP) shocks reached L1, creating a complex and geoeffective plasma structure. To understand the processes leading up to the formation of the two shocks, we model the CMEs with the Wang‐Sheeley‐Arge (WSA)‐ENLIL+Cone model. The first two CMEs merged already in the solar corona driving the first IP shock. In interplanetary space, another fast CME presumably interacted with the flank of the preceding CMEs and caused the second shock detected in‐situ. By introducing a customized density enhancement factor (dcld) in the WSA‐ENLIL+Cone model based on coronagraph image observations, the predicted arrival time of the first IP shock was drastically improved. When the dcld factor was tested on a well‐defined single CME event from 12 July 2012 the shock arrival time saw similar improvement. These results suggest that the proposed approach may be an alternative to improve the forecast for fast and simple CMEs. Further, the slowly decelerating kilometric type II radio burst confirm the properties of the background solar wind have been preconditioned by the passage of the first IP shock. This likely caused the last CME to experience insignificant deceleration and led to the early arrival of the second IP shock. This result emphasizes the need to take preconditioning of the interplanetary medium into account when making forecasts of CMEs erupting in quick succession.