Amine Regenerator Control

Part 2: Strategies for Controlling Lean Loading

Pedro L. Ott, Nathan A. Hatcher and Ralph H. Weiland

Optimized Gas Treating, Inc.
Cimarron Park Loop
Buda, TX 78610

There are several control strategies for ensuring good amine regenerator performance. As will be
seen, some work better than others, and some do not work at all. But all function by monitoring and
controlling the temperature somewhere in the regeneration system. In the first part of this article, it was
shown that anywhere from the overhead vapour line to just below the feed tray are satisfactory measuring
points. In this Part 2 we address various approaches to control. The article uses specific operating cases
to show the benefits and shortcomings of several strategies. Discussion of the various approaches is made
concrete by using MDEA as a selective solvent in a typical natural gas processing plant; however, most of
the discussion applies equally well to most other amines and gas treating applications.
Reboilers in the amine regeneration section of a treating plant come in a number of varieties,
including kettle, once-through thermosiphon, and recirculating thermosiphon, and they can be steam
heated, hot oil heated, process-gas heated, and direct fired. Regardless of type though, reboilers have the
main purposes of providing:
 Sensible heat to raise the rich amine feed to the boiling temperature inside the regenerator,
 Heat to reverse the reactions between acid gases and the amine, and
 Energy to generate the dilution steam necessary for providing the acid gas partial pressure that
directs dissolved gases from the solvent into the vapour.
Energy requirements to reverse the acid gas-amine reactions are very amine specific. Energy needed to
raise solvent temperature and generate steam are less so because the solvent heat capacity is dominated
by water as is the heat of vapourisation (mostly of water) from the solvent.
Regenerator performance is usually measured in terms of the solvent lean loading(s) produced.
Strategies for controlling lean loading include:
 Manipulate reboiler steam or hot oil flow to meet a target regenerator bottoms temperature or
reboiler temperature,
 Manual reboiler steam or hot oil flow control,
 Maintain a fixed flow ratio of steam or oil to the solvent flow, and
 Allow the ratio of steam to solvent flow rates to be biased or reset to control overhead temperature.
Each of these strategies is examined in turn using ProTreat® mass transfer rate based simulation as a
guide to the quantitative response and behaviour of an example amine unit.

Manual Flow Control

Provided the absorber is operating in a region where its performance is not controlled by solvent
capacity limitations, and it is not running close to an operational cliff, absorbers and amine systems are
usually very well behaved and they can lend themselves to manual control of the heating medium flow rate
to the reboiler, as well as solvent temperature and circulation rate. However, operators need to have a
good understanding and grasp of how the system responds to change and how parameters interact.
Without such understanding it is very easy to move a set-point in a direction that intuitively seems correct,
only to find that it is the opposite of what is actually needed. If tight specifications on gas purity do not have
to be met and treating demands do not change rapidly, manual control may well be adequate; otherwise,
automatic control is almost essential.

Controlling to Regenerator Bottoms or Reboiler Temperature

The material in a reboiler is at its boiling point which by definition is the temperature at
which the total vapour pressure of the (loaded) solvent is equal to the total pressure in the reboiler. In any
well-stripped solvent, the partial pressure of the acid gases in the solvent is almost always very low;
otherwise, high gas purity would be impossible to achieve. (The exception is amines used for carbon
capture where 10 to 15% of the CO2 is to be slipped through the absorber, so lean loadings are purposely
kept quite high and CO2 exerts a significant partial pressure.) So in almost all gas treating applications, the
acid gas partial pressures are very low and the boiling point is barely affected at all by lean solvent loading.
For a well-stripped solvent, the boiling temperature is almost solely a function of the solvent strength and
the column pressure, i.e., T = f (P, %wt amine).
Figure 1 is a schematic of the regenerator with relevant process data. If the regenerator is part of a
tail gas treating unit (TGTU) the target lean loading on H2S will likely have quite a small value; however the
high rich solvent loadings exemplified here would not pertain. A high reflux ratio (high reboiler energy flow)
will be needed to produce a low lean loading. On the other hand, if H2S is being selectively removed from
a high pressure natural gas, solvent lean loading will not have to be as low and can be achievable with
more modest reflux ratios and reboiler energy demands. But regardless of the application, over the H2S
loading range from 0.0003 to 0.004 produced using reboiler duties from 1.4 to 7.3 MW (5.0 to 25 MMBtu/h),
the reboiler temperature in this example does not vary by more than 0.05°C (0.1°F). This is far outside the
capability of any commercial temperature measuring device in process service. Even the tower sump
temperature varies by only slightly more than 0.5°C (1 °F). As discussed in Part 1 of this article, this
realisation makes absolutely ludicrous the whole notion of controlling regeneration by using reboiler or
tower sump temperature as a measure of lean loading. Such a strategy has no possibility of working
reliably. In order to see a reasonable response to the controlled variable, the lean loading has to be grossly
off specification. Letting the unit go out of environmental compliance for the control strategy to work is not a
viable strategy at all.
 Figure 1 Regenerator Specifications

Controlling Overhead Temperature

Solvent lean loading is highly responsive to reboiler duty; however, lack of instrumentation for
measuring this quantity makes it almost impossible to use it directly in a dynamic or process control setting.
And neither reboiler nor tower sump temperature is a suitable substitute. However, regenerator overhead
temperature shows significant, useful variation with reboiler duty, i.e., with lean loading. For the case
already discussed, Figure 2 shows how H2S and CO2 lean loadings can be controlled by using the
regenerator overhead temperature as a proxy. Of course, the loadings of the two acid gases cannot be
individually controlled. Once the regenerator is operating to produce a specified lean loading with respect
to one acid gas, the other is then completely determine by the now-established operating conditions, by the
vapour-liquid equilibrium, and by the mass transfer characteristics of the tower internals as they pertain to
the second acid gas.
Properly calibrated (not just zeroed and spanned) instrumentation can usually be relied upon to
measure and control temperature to within 1°C (2°F) so, in the present case, measuring and controlling
to the overhead temperature will likely provide adequate control of the critical lean loading and, therefore, of
treating itself in most instances.
The main detractor against this control strategy is that the control is entirely feedback based. Rapid
changes in amine circulation may limit the effectiveness of the controller response to the inherent time
delay in the system.
 0.004 0.00025

0.00020
0.003
H2S Lean Loading

CO2 Lean Loading

0.00015
0.002
0.00010

0.001
0.00005

0.000 0.00000
80 90 100 110 120
O/H Temperature, °C

Figure 2 Regenerator Overhead Temperature as a Proxy for Solvent Lean Loading

Setting the Heating Medium to Solvent Flow Ratio

A rough guideline for setting reboiler duty in amine regenerators is to use a value from 120 and 180
kg of 3.5 barg steam per standard m3 of solvent (1-1.5 lb of 50 psig steam per standard USgal). Figure 3
shows the H2S and CO2 loadings produced over a tenfold range in solvent rates when the reboiler energy
input is maintained at the somewhat high value of 200 kg of 4 barg steam per standard m3 of solvent
(needed to reach a low enough H2S lean loading in this gas plant). The trays were kept hydraulically well
balanced (equal jet and choke flood values) by assigning 10% downcomers; flood ranged from 9.7 to 97%
in the 1200-mm diameter column. The simulated overhead vapour temperature over this range of
circulation rates varied by less than 0.5°C, probably because the regenerator is being boiled quite hard. A
somewhat wider temperature range might be expected with more modest boilup. Regardless, the variation
in overhead vapour temperature can be expected to be small using such a control strategy.
 Figure 3 How a Fixed Amount of Reboiler Energy per Unit of Solvent Keeps H2S
Loading in a Narrow Range over a Wide Range of Circulation Rates

Using a reboiler duty that is directly proportional to the circulation rate allows the unit to respond
well to solvent flowrate changes. The solvent H2S lean loading varies only a little over a tenfold range of
flow rates. In a relative sense, the CO2 lean loading varies by a factor of ten; however, in terms of absolute
loading values the CO2 loading response is quite similar to H2S: both experience a change in loading of
0.0002 to 0.0003 loading units. Reboiler conditions play a large role in determining regenerator
performance because a large fraction of solvent stripping actually takes place there. From the lowest to the
highest solvent rates, reboiler pressure varied from 1.149 to 1.279 barg (16.67 to 18.55 psig) and reboiler
temperature ranged from 127.9 to 129.9°C (262.2 to 265.8°F).
The mass transfer performance of trays improves with both vapour and liquid flow rates. Both are
increasing as the circulation rate through the fixed size regenerator increases. On that basis, one might
expect both the H2S and CO2 lean loadings to improve with solvent rate. The H2S loading does; the CO2
loading does not—instead it grows worse, by about the same number of loading units (0.0002) as H2S
loading improves (0.0003). Although both H2S and CO2 mass transfer rates benefit from the higher mass
transfer coefficients that accompany increased traffic through the regenerator, the effect is stronger on H2S
desorption than on CO2 because of a higher reaction enhancement effect. Somewhat more of the
additional energy, therefore, is used to strip just a little more H2S, leaving slightly less available for
removing CO2. This is the reason for the counteracting effect, and for the relative total stripping remaining
roughly unchanged.
Adjusting Steam-to-Solvent Flow Ratio to Control Overhead Temperature
Maintaining a constant steam-to-solvent ratio allows simple, accurate control of stripping under
varying conditions of solvent flow rate. As also already shown, adjusting the steam flow allows control of
solvent lean loading under varying rich amine loading conditions. Therefore, it only stands to reason that a
combination of the two will allow control of lean loading when both solvent rate and rich loading are
changing. This is done via cascading the control system and, of course, it is used to control the
regenerator’s overhead (or feed tray) temperature as a proxy for lean loading. The two individual controls
have already been discussed and this approach is just a marriage of the two.
As an example, a sudden increase in rich loading will cause overhead temperature (more or less
equivalent to the boiling point) to drop. The physical response should be to increase reboiler steam flow to
bring the temperature back up to the set point value. However, the rich solvent flow has not changed so
the flow-ratio controller wants to keep the same steam rate; thus, a control system with only a simple flow-
ratio controller cannot respond except by violating the prescribed flow ratio. The basic strategy is to
institute a second level of control to override the set flow-ratio controller and allow the steam rate to be
increased, all the while resetting the ratio to the new value needed to maintain the set point temperature.
The reboiler steam flow changes in response to the measured solvent flow, while the steam-to-solvent flow
ratio adjusts in response to rich loading as indicated by the temperature measured at an appropriate
location in the regenerator.

Summary
Most attention is usually focused on the absorber because this is the equipment that actually
processes the gas. However, absorber performance is controlled to a very large extent by the regenerator
and how it performs in providing a satisfactory lean amine solvent. Thus, one could easily take the position
that solvent regeneration is really the key operation, and its accurate and reliable simulation is at least as
important as the absorber’s.
The ProTreat® simulator uses a mass transfer rate-based regenerator model similar in every
respect to that for the absorber, allowing it to predict the performance of the entire amine unit with uncanny
accuracy. This includes the effect of heat stable amine salts (HSSs) on both retarding absorption and
enhancing regeneration, leading to either better or worse overall amine unit performance, depending on the
circumstances.
Satisfactory absorber operation often depends critically on achieving the right solvent lean loading.
Because it is hard to measure on-line, temperature is the usual proxy for lean loading. Part 1 of this article
discussed where the best places are to measure temperature in a regenerator; this article has discussed
several ways to control lean loading by controlling temperature. But probably the most beneficial tool one
can use for analysing and monitoring an amine unit is a truly mass transfer rate-based simulator.