Demand side management and response schemes have the potential to reduce peak demand and improve grid utilization (improving the peak-to-base ratio). Existing methods often rely on accurate predictions or knowledge of the future demand, and most require a bi-directional communication infrastructure. This paper proposes two novel demand management approaches that do not have either of these requirements. The demand management problem is formulated as a constrained optimization problem to meet end-user energy demand subject to the physical limits of the electrical network. The adopted model focuses on the last mile of distribution networks and accounts for transformer load, line load and phase unbalance limitations. The optimization problem is solved iteratively, in a distributed and computationally tractable manner. In the first method, the behaviour of all users is coordinated through a time-varying price signal (a virtual price) that reflects how much power can be distributed given the network constraints. This virtual price signal is broadcast to all end-users by the Distribution Service Operator through a passive uni-directional communication channel. In the second method, an algorithm local to the user approximates the state of congestion in the network using the historical local voltage measurements. This method does not require a communication infrastructure. The performance of both methods is compared and their resulting behaviours are illustrated using a realistic simulation relevant to a typical Australian suburban low voltage distribution network.