How long will the cyclone system in a Cat Cracker reactor or regenerator last?
Cyclones systems in Fluidized Catalytic Cracker units (FCC) are typically designed to a creep allowable stress, where the stress field at various locations of the system has been determined by linear elastic analysis. The basic allowable stress for such internal structures (not on the pressure boundary) is commonly taken from B31.3 or API RP-530, where the allowable stress is dependent on the design life, often 100,000 hours, approximately 11 years. Such analyses may take into account stress classification to give an allowance for elevated stresses of secondary nature (self-equalizing stresses), but this practice often under-estimates the actual stress redistribution that takes place as the structure permanently deforms from creep. Once the time in operation has surpassed the design life, it would seem that the creep life of the structure should be consumed and that the cyclone system would be due for a replacement or at least extensive repair in the load bearing components, hanger straps, outlet tubes, plenum floors etc. However, this is often not the case and many cyclone systems have had useful lives in excess of 20-25 years. If the operating temperature is less than the design temperature with only occasional excursions, then the creep damage on the cyclone system will be significantly less compared to the design expectation and the cyclone system will last much longer. However, if the operating temperature is always near design temperature and there are frequent excursions, the creep life will be consumed at a higher rate, but the useful life will still mostly be well in excess of the nominal design life.
One methodology to determine a more accurate creep life prediction is to analyze the cyclone system with “real” operating condition with an explicit creep model using Finite Element Analysis (FEA). Becht Engineering has used the explicit creep model with finite element method to solve several complex high temperature creep problems.
Let’s look as some of the advantages an accurate simulation of the progression of creep damage through FEA can provide and discuss the challenges in putting it into practice.
- Enables estimates of remaining creep life that are far more accurate than those obtained with linear-elastic analyses. Such estimates are of great importance in equipment strategy developments.
- Obtaining good understanding for where in the structure creep damage will first manifest itself, which can help prioritize inspection efforts.
- Ability to simulate actual historic operating condition (if data is logged!) and anticipated future conditions in any possible combination of normal operating, emergency loading (flooded cyclones), temperature upset conditions etc.
The finite element analyst will encounter a number of challenges with this approach, but they can be overcome.
Such challenges include:
– A detailed FEA model is required
- The shell element approach typically employed in linear-elastic analysis of these structures does not reliably provide results of acceptable accuracy when used in creep simulations. 3D solid elements are needed in creep regions that do not exhibit near uniaxial loading, which requires very detailed meshes.
– Selecting creep properties of the actual materials
- Testing of actual material specimens is ideal for establishing the creep properties, but such data is rarely available. There is a large spread in the published creep data and one must usually resort to selecting the lower bound of the scatter band to ensure conservatism.
– The creep protocol endorsed by Fitness-For-Service standard API-579-1/ASME-FFS-1 is the MPC Omega model. The uniaxial formulation of this model has shown good correlation to test data, but is not incorporated into commercial FEA codes.
- Extensive user coding (and testing!) is required to successfully implement this model. If starting from scratch, one should realistically expect such development to take several weeks, if not months.
– API-579-1/ASME-FFS-1 lists the acceptable creep damage as 80%, but does not offer guidance on allowances for localized effects.
- An approach put into practice in the Industry is to permit creep damage to be at, or even exceed, 80% as long as at least half of the cross section of a load carrying component shows less than 80% creep damage. This approach is particularly useful when coupled with a material softening model that reduces the stiffness of creep damaged material, which ensures stress redistribution away from the damaged regions is accounted for. This approach makes the analysis similar to limit-load analysis in that, if the FEA solver finds a converged solution, it is demonstrated that the parts of the structure that have less than 80% creep damage can adequately carry the load.
Becht Engineering has experience in the detailed creep analyses described here and employs an implementation of the Omega protocol in the ABAQUS FEA code that has proven both robust and reliable. Recent work includes a cyclone system of a FCC regenerator for a major European refinery and converter vessel cyclones for an African petrochemical plant. Both of these analyses demonstrated that the remaining creep lives far exceeded those predicted by the original design and the useful life will likely be governed instead by refractory degradation and/or cyclone body erosion. This enables the Owner to make favorable adjustments to equipment strategy, inspection planning and future operating envelopes.
Left Figure: Model mesh for studying creep of cyclone outlet tube
Right Figure: Cyclone outlet tube creep damage after simulating historic and anticipated future conditions. Peak damage is at 48%, far less than predicted by original design.