With increased variable renewable generation and retirements of dispatchable thermal plants, net load variability and forecast error will increase. Many ISOs and RTOs have implemented or are considering implementing ramp products in their real-time markets to maintain both reliability and efficiency with the changing resource base. The design of any real-time market enhancement, however, must account for the fact that ISOs and RTOs use separate real-time engines for commitment, dispatch, and pricing. Depending on which engine is modified, there will be different impacts on reliability, system dynamics, system cost and prices. We present a rolling horizon simulation framework for the PJM Real-Time Market, consisting of a day-ahead model and real-time commitment, dispatch, and pricing engines that account for generator outages, net load variability and rolling short-term forecast error for a 2050 future capacity scenario. We compare the impacts across ramp product designs on system costs, reliability, prices, generation mix and generator revenues. We analyze the sensitivity of these results to the quantity of ramp product procured in the real-time commitment vs. dispatch and show that a balanced requirement added to both the commitment and dispatch engines provides the best performance across multiple metrics.