This paper proposes a reference model for defining security benchmarks for the safety assessment of Cooperative Driving Automation (CDA) applications. Our reference model provides a systematic approach to benchmark the resilience of CDA applications against malicious attacks through extensive system simulations. It enables the test repeatability and comparison of results across different implementations of CDA applications. In our approach, a benchmark is defined as a series of tests that expose the target system to specific attacks while recording its response. Using this model, we define a benchmark for evaluating the resilience of Cooperative Adaptive Cruise Control (CACC) algorithms against barrage jamming attacks targeting the physical layer of the IEEE 802.11p communication standard. We apply this benchmark to assess and compare the performance of four CACC algorithms: P1, Flatbed, Ploeg, and Consensus. The benchmark measures reveal that the Consensus algorithm demonstrates the highest resilience against jamming attacks, primarily due to its heavy reliance on onboard sensors and the use of sensor data from all other vehicles for decision-making. In contrast, the P1 algorithm, which depends mainly on vehicle-to-vehicle (V2V) communication, proves to be the most vulnerable. Furthermore, the results indicate that vehicles are most susceptible to jamming attacks during acceleration phases, making these periods critical for security evaluation. These findings validate the effectiveness of our benchmarking framework in identifying strengths and vulnerabilities of CACC algorithms under cyberattacks.