We study optimal time-allocation strategies for a cellular architecture with relays that can store and forward information from the base station to the terminals on the downlink and from the terminals to the base station on the uplink and determine the consequent capacity gains. We focus on the case when the cellular structure is one-dimensional and evaluate the sensitivity of the capacity gains to key system parameters such as relay power, self-noise, relay location and terminal locations. We show that the optimal time allocation problem on both uplink and downlink with the objective of maximizing the maximum common throughput of all the users in the cell, can be formulated as a linear program. We derive several analytical conditions that any optimal allocation must satisfy yielding simpler linear programs. We also present a sub-optimal allocation algorithm that does not require the solution of a linear program.