Wavelength Division Multiplexing (WDM) can increase the carrying capacity of an optical network without laying additional fibers. However, a disruption in such a high speed and high capacity network can quickly impact the entire network. A fast protection and restoration recovery mechanism is needed to provide uninterrupted data delivery. Implementing IP directly over a WDM optical network, using optical crossconnects and IP routers, is emerging as the preferred method to efficiently utilize the enormous bandwidth offered by WDM networks. However, in such networks, a single link failure in the WDM layer can affect multiple links in the IP layer, which may greatly degrade data delivery. Several solutions have been proposed in the literature to avert such a scenario. These solutions mostly focus on finding paths for the IP connections in the WDM layer in such a way that the failure of a single WDM link does not disconnect the IP topology. Such a mapping is called "survivable". Due to the NP-completeness of the problem, various heuristics based on ILP formulations, tabu search, and shortest path variants have been proposed in the literature. In this paper we study a recent approach by Thiran and Kurant, point out certain attractive features and difficulties with this approach, and present enhancements to their basic approach to achieve better fault coverage and to add robustness to the survivable routing schemes. We provide simulation results evaluating the new heuristics.