This paper investigates low-complexity resource management design in multi-carrier rate-splitting multiple access (MC-RSMA) systems with imperfect channel state information (CSI) for ultra-reliable and low-latency communications (URLLC) applications. To explore the trade-off between the decoding error probability and achievable rate, effective throughput (ET) is adopted as the utility function in this study. Then, a mixed-integer non-convex problem is formulated, where power allocation, rate adaption, and user grouping are jointly taken into consideration. To solve this problem, we first prove that ET is a monotone increasing function of rate under the strict reliability constraint of URLLC. Based on this proposition, an iteration-based concave-convex programming (CCCP) method and an iteration-free lower-bound approximation (LBA) method are developed to optimize power allocation within a single subcarrier. Next, a dynamic programming (DP)-based method is proposed to determine near-optimal user grouping schemes. Besides, a CSI-based method is further proposed to reduce the complexity and obtain important insights into user grouping for MC-RSMA systems. The simulation results verify the effectiveness of the CCCP and LBA methods in power allocation and the DP-based and CSI-based methods in user grouping. Besides, the superiority of RSMA for URLLC services is demonstrated when compared to spatial division multiple access.