Wireless sensor networks that operate on batteries have limited network lifetime. There have been extensive recent research efforts on how to design protocols and algorithms to prolong network lifetime. However, due to energy constraint, even under the most efficient protocols and algorithms, the network lifetime may still be unable to meet the mission's requirements. In this paper, we consider the energy provisioning problem for a two-tier wireless sensor network. In addition to provisioning additional energy on the existing nodes, we also consider deploying relay nodes (RNs) into the network to mitigate network geometric deficiency and prolong network lifetime. We formulate the joint problem of energy provisioning and relay node placement (EP-RNP) into a mixed-integer nonlinear programming (MINLP) problem. Since an MINLP problem is NP-hard in general, and even the state-of-the-art software and techniques are unable of offer satisfactory solutions, we develop a heuristic algorithm, called SPINDS, to address this problem. We show a number of novel algorithmic design techniques in the design of SPINDS that effectively transforms a complex MINLP problem into linear programming (LP) problems without losing critical points in its search space. Through numerical results, we show that SPINDS offers very attractive solution and some important insights to the EP-RNP problem.