With the high penetration of renewable energy, the conventional solution to balancing supply and demand requires substantial reserve generations and, thus, curtails the environmental and financial benefits. To mitigate the overuse of reserves, an alternative approach, widely referred to as demand response, has been attracting increasing attention. The essence is to exploit the flexibility in demand to compensate for the variability in supply. Among the various forms of demand response, of particular interest to us is the so-called duration-differentiated energy service, in which a load requires a constant power level for a specified duration of time. In this paper, we further explore this by taking the charging/discharging interactions among the loads into account. The introduction of peer-to-peer charging facilitates power allocation and enlarges the set of adequate supply profiles. We propose an algorithm for power allocation and show that a given supply profile is adequate if and only if the algorithm produces a feasible allocation. We also relate this algorithm to dynamic programming. In the case of an inadequate supply, the adequacy gap can be obtained via a slightly modified algorithm.