Control of unstable ammonia reactor

 

Thomas Realfsen

Chemical Engineering, Norwegian Univ. of Science and Technology (NTNU)

Trondheim, Norway

November,2000

 

In the paper "Analysis of instability in an industrial ammonia reactor", by John C. Morud and Sigurd Skogestad August 28, 1997, the instability in an industrial ammonia reactor was simulated. A result of this instability was oscillations in the reactor temperatures caused by positive feedback that occurs when parts of the feedflow is preheated with the outlet flow.

 

It was claimed in the paper that it is easy to stabilize the reactor with feedback control and the objective of this note is to confirm this claim and show how it can be done.

 

1. Let us first reproduce the instiability without feedback control. In the figure we show the response in the outlet temperature (result from the paper "Analysis of instability in an industrial ammonia reactor"):

Fig 1. At t = 0 the pressure is reduced from 200 bar to 170 bar; the system remains stable (at least seemingly) and settles at a new steady-state with a somewhat lower outlet temperature.
Then at t = 1200 [s], the pressure is further decreased to 150 bar and the system becomes unstable.
At t = 7200 the pressure is restored from 150 bar to 200 bar.

 

2. One way to stabilize the reactor is to use the quench flow, Q1, to control the inlet temperature to the first bed, see fig 2:

 

Fig 2. Ammonia reactor with feedback control

Simulation of the pressure reduction using this feedback structure with simple proportional control shows that we have a stable system:

 

Fig3. Feedback control of inlet temperature . At t = 0 the pressure is reduced from 200 bar to 170 bar; and

then at t = 1200 [s] the pressure is further decreased to 150 bar . The system remains stable!

At t = 7200 the pressure is restored from 150 bar to 200 bar.

 

3. Now try another simulation (with feedback) where at t = 1200 [s] the pressure is reduced from 200 bar to 150 bar. Again the system remains stable:

Fig4. The outlet temperature stabilized at a new lower steady-state.

How to produce the results in figures 1,3 and 4:

Go to subdirectory control and see matlab files main.m and reguleringsim.m.

-To reproduce the instability in figure 1, the percentage signs in front of section 1 in reguleringsim.m must be deleted (let the percentage signs in front of section 2 and 3 remain).

-To reproduce the results from figure 3, the percentage signs in front of section 1 and 3 must be deleted (the percentage signs in section 2 must remain).

-To reproduce figure 4, the percentage signs before section 2 and 3 must be deleted (the percentage signs in section 1 must remain).

 

To reproduce the results made in this note it's enough to make the changes in the matlab file reguleringsim. The product concentration can also be produced, the changes above must then also be made in the matlab file concprof.m (it's done this way to prevent that the concentration plot follows the iterations made in the ode-solver).