(371a) A CFD Model-Based Optimization of a Process Burner Geometry

Burner design and placement in process heaters, crackers, and reformers continues to be an active area of study as the temperature and heat flux distribution from the resulting flames can vary significantly between designs. It has become even more important in recent times to design necessary adjustments at existing cracker plants to adapt to the new feed types emerging from the booming shale gas industry.  The design of a burner plays an important role in product mix and overall performance of the crackers.  It also directly impacts the process efficiency, emissions control, and the mechanical durability of the equipment and its components.

 Traditionally, optimization of such systems is performed by looking at a limited set of design variations (3 or 4), and picking the best one out of those. However, true rigorous optimization done with the help of an adaptive optimization algorithm lets one to not only explore the entire design space, but also find the most optimum configuration. Such a methodology helps in two ways: one is the pure benefit of finding the best design out of all possible designs. Secondly, the arrival at the optimum design is much quicker than experimental trials or even sequential simulation trials.

In this paper, we will first describe why a robust, hybrid and adaptive algorithm (SHERPA) is necessary for such a difficult problem.  SHERPA uses a blend of search strategies to adapt to the design space within user-defined constraints and objectives. Then we will show with the help of an example, how an improved design out of thousands can be discovered without having to simulate or experiment with the thousands of possibilities.

Following design variables were used:

1. Burner tile angle
2. Primary injector angle
3. Primary injector spacing
4. Primary port spacing
5. Secondary injector spacing
6. Secondary injector angle

The flow, heat, and chemistry taking place on the fire box side of a process heater is modeled using STAR-CCM+, a computational fluid dynamics software code. The heat transfer to the process tubes and walls takes into account both convective and radiative modes, with the radiation being modeled using the discrete ordinates method (DOM).  The design objectives for this study were to minimize the CO and NOx emissions.