This work describes experiences porting the UTCHEM chemical flood simulator from a serial environment to the nodal environment of distributed memory massively parallel computers. Two conversion strategies have been explored. The first approach required the least amount of effort. The serial version of the code was modified to execute independently on each processor. This version permits many different problems to be evaluated simultaneously but does not confer any computational advantage. In the second version, the program was altered, using domain decomposition, to distribute the data, and message passing communication, to couple the node computations. This allowed for execution of a single problem across all of the computing nodes. This investigation shows that the message passing version of the code speeds up well, as more computational nodes are employed, for problem sizes with a small surface to volume ratio. A more efficient linear solver is implemented in this work, which promises both good parallel efficiency and robustness for large-scale simulation problems.