Can existing software handle industrial optimization problems? optimization of the kemira's (B) ammonia synthesis loop

Abstract

We test an existing software on the real case of the Kemira's (B) ammonia synthesis loop, especially when this works at low capacity. We use optimization to examine cases where we just seek to maximize ammonia production and also to maximize ammonia production taking into account steam cost and the loop's bleed valorization. The software tested is the commercially available BELSIM-SIMU, where a routine using a Successive Quadratic Programming (SQP) algorithm is used for optimization. A “two-level” approach is used: at the lower level, the unit modules are calculated in a sequential modular fashion while tears, supplementary equality and inequality constraints and optimization are handled simultaneously on the upper level. Results show that, when the ammonia synthesis loop is considered alone, production can still increase by almost 3% while power supply diminishes. For the optimization software we can conclude that it is both efficient and reliable. Great flexibility (through calculation modes and possibility to have either nested loops or one outer loop) is available in SIMU but numerical calculation of the derivatives is rising calculation time. Mention will be made to an optimization software, called OPTIM ( Kontopoulos, 1993), newly developed in the LASSC, which also uses the “two-level” approach but also analytical derivative information, chainruling and automatic unit calculation sequence detection. Significant computer time savings are possible in this case (going up to 60% in the ammonia synthesis loop examined here, and up to 95% in other cases examined).