In the multipath routing (MPR) algorithms for wireless sensor networks (WSNs), disjoint multipath routing (DMPR) approaches perform better in reliability and fault tolerance. However, DMPR poses significant challenges in terms of the optimization of the disjoint paths and the data transmission along the disjoint paths. Considering certain industrial monitoring applications (for example, mine safety monitoring) which have relatively stable network topologies, we propose a centralized-calculating-based 2-disjoint multipath routing algorithm for WSNs (CCDMPR). Based on the global information, the algorithm firstly calculates near to optimize 2-node (link) disjoint paths from a source to the destination using the number of hops and the path quality as metrics, and generates a tiny routing table which is merely composed of the master parent node, secondary parent node couple and the path bit series. Then the tiny routing tables are disseminated to each sensor node along the generated paths. In order to improve the resilience of route maintenance, the algorithm designs a centralized adaptive path maintenance mechanism. In the data routing phase, packets are transmitted according to the path bit series carried in their heads, without any control overhead. The experimental results show that CCDMPR algorithm can shorten the average path length, reduce the total energy consumption and improve the transmission reliability compared with existing approaches.