An unconverged column can cause serious problems — especially for the people tasked with troubleshooting it! Continuing from my previous blog post, here are some insights on how to troubleshoot the rate-based calculation in the acid gas columns and specific strategies for the common solvents, such as MDEA and MDEA + PZ.

If you don’t know how to determine where the issue is with your acid gas column, or if you don’t know how to troubleshoot the initialization calculation, please check out the previous post. If you are trying to get your rate-based calculation to converge, you are in the right spot.

The second major step of the calculation in a rate-based column is the rate-based calculation. If the initialization calculation succeeds but the rate-based calculations fail, you can modify some of the rate-based parameters to get to a converged solution. You can access all of the variables by going to Advanced Modeling Convergence Parameter, and by clicking the View button next to the Convergence.

In the example below, the rate-based calculation failed after reaching iteration limit and is not close to convergence:

 
1. Continuation/Homotopy Parameters
The first method is to modify the Continuation/Homotopy parameters. This method could help to switch smoothly from equilibrium to rate-based solution by solving a series of subproblems at the intermediate states. The subproblems are the intermediate steps used to switch from equilibrium model to rate-based model.

a) Iterations
The Continuation/Homotopy Iterations controls the number of subproblems to be solved. The default value is 0, meaning no subproblems to be solved. The recommended value for iterations is 2 – 5, because larger iteration numbers will increase the computation time   .
b) Parameter
The Continuation/Homotopy Parameter dictates how close the subproblem is to the equilibrium solution. The default value is 1, meaning a direct jump to rate-based solution without solving the intermediate states. An increasing value represents the subproblems being closer to the equilibrium solution. When the parameters are in use, HYSYS runs a series of simulations to solve the subproblem by decreasing the Continuation/Homotopy parameters. With large enough parameter, the first subproblem will be very close to equilibrium solution and should converge easily. The solution is then used to initialize the next subproblem, until rate-based solution converges. The recommended value for parameter is 2 – 5 as well.

While the larger parameters will smooth the convergence, they will increase the computation time.

2. Stabilization Method
The second method is to modify the Stabilization Method. The two available methods are Dog-Leg and Line-Search. The Dog-Leg is the default selection. If the convergence failed with the default method, you can try the Line-Search, which is more robust but slower. If you observe the error/tolerance value in the rate-based iterations oscillates with a large variation, AspenTech recommends changing the method back to Dog-Leg.

Now you have learned different strategies and methods to troubleshoot a rate-based column for acid gas systems. The tips vary depending on the solvent used. The table below provides some common examples for different solvents.

In the example of MDEA, common failures can be found in both initialization calculation and rate-based calculation. You can employ the temperature estimate strategy and specify a top stage temperature that is 20 K below that of the top stage feed. If the rate-based calculations fail, try to modify the stabilization method to Line-Search.

Solvent	Problem	Solution
MDEA	Initialization fails Rate-based calculations fail	1. Provide temperature estimates Specify a top stage temperature that is 20 K below that of the top stage feed Specify a bottom stage temperature that is 20 K above that of the bottom stage feed 2. Set the Stabilization Method to Line-Search
DEA	Initialization fails	1. Provide temperature estimates Specify a top stage temperature that is 25 K below that of the top stage feed Specify a bottom stage temperature that is 3 K above that of the bottom stage feed 2. Modify the Liquid Holdup Estimate Field to 1e-7 m3 3. Set the Damping Level to Mild
DGA	Initialization fails	The DGA reaction occurs very quickly.  1. Modify the Liquid Holdup Estimate Field to 1e-9 m3 2. Provide temperature estimates Specify a top stage temperature that is 50 K below that of the top stage feed Specify a bottom stage temperature that is 2 K above that of the bottom stage feed
MEA	Initialization fails	1. Provide temperature estimates Specify a top stage temperature that is 50 K below that of the top stage feed Specify a bottom stage temperature that is 3 K above that of the bottom stage feed
MDEA + MEA	Initialization fails	The reaction rate is very quick because of the existence of MEA.  1. Modify the Liquid Holdup Estimate Field to 1e-5 ft3 2. Provide temperature estimates Specify a top stage temperature that is 20 K below that of the top stage feed Specify a bottom stage temperature that is 5 K above that of the bottom stage feed
MDEA + PZ	Initialization fails Rate-based calculations fail	The PZ+MDEA reactions occur very quickly.  1.  Modify the Liquid Holdup Estimate Field to 1e-4 ft3 2. Modify the Continuation/Homotopy Parameter to 5 3. Modify the Continuation/Homotopy Iterations to 3

 

You should definitely try out some of these strategies if you have an unconverged rate-based column in front of you. Solving the column allows you to take a step closer to a better design and return in capital by optimizing the entire gas plant within Aspen HYSYS^®. You can also view the top 10 questions about our acid gas technology here.

If all of the strategies fail, please contact our amazing customer service support team: they are always happy to help with issues like these. 

Happy column troubleshooting!