Fog environment is an essential infrastructure for time-critical and delay-sensitive IoT applications. This paper proposes Load Balancing in Fog Computing Framework (LBFOG) for edge nodes. In LBFOG, the edge nodes in the fog layer are clustered based on their proximity. The IoT nodes will be transmitting two classes of tasks. Class 1 task is delay sensitive, and Class 2 is tolerant of time delay. If the edge node has sufficient energy, it processes Class 1 and Class 2 tasks. However, when the residual energy of the edge node is less than the threshold energy level, it processes only Class 1 tasks since it is delay sensitive and transfers all its Class 2 tasks to the edge nodes with the highest residual energy and greater than the threshold energy. If all the edge nodes in the cluster have less energy than the threshold energy, the cluster's threshold energy is recomputed, and the tasks are further processed. LBFOG was simulated in MATLAB along with static and dynamic task allocation schemes. It is observed that LBFOG conserves 52% and 29% more energy than static and dynamic schemes, respectively.