HRSG Ammonia Injection Optimization

HRSG Ammonia Injection Optimization

Gas turbine engines used for power generation often have a Heat Recovery Steam Generator (HRSG) to increase efficiency. The hot turbine exhaust is used to create steam in heat exchanger tube banks, which is then used to create additional electricity. The emissions control system to clean the gas turbine exhaust consists of both carbon monoxide (CO) and nitrous oxide (NOx) reduction through catalyst elements. The CO is reduced by a dedicated CO catalyst. The NOx reduction, via Selective Catalytic Reduction (SCR), requires that ammonia be injected upstream of the catalyst and thoroughly mixed with the turbine exhaust gases. The catalytic reaction of ammonia and NOx converts the NOx to nitrogen (N2) and water vapor (H2O) which then exhausts out the plant stack.

To optimize the performance of these catalytic reactions, and thus minimize plant emissions, there are several key design criteria related to the flow characteristics of the HRSG. In particular, after the exhaust gas exits the turbine, it must be carefully controlled such that:

  1. Velocity distribution is uniform through the catalysts within 15%
  2. Temperature profile is within +/-20° through the catalysts
  3. Correct stoichiometric ratio of ammonia to NOx is provided through the SCR
  4. Pressure drop is minimized.

All these factors play a role in the resulting emissions, with the goal of minimizing CO and NOx emitted from the plant. This must be achieved while injecting the minimal amount of ammonia to avoid discharge out the stack (also called “ammonia slip”).

HRSG Performance Improvement Project

HRSG Plant

Figure 1. HRSG plant in western U.S.

A western U.S. plant (Figure 1) with two HRSG units that handle the exhaust from 501F turbines was having issues with ammonia distribution and NOx catalyst performance. A redesign and installation of a new ammonia injection grid (AIG) was required to improve the situation. Concord Environmental of Voorhees, New Jersey performed the engineering, procurement, and construction of the AIG, as well as installation of a new catalyst. Airflow Sciences Corporation of Livonia, Michigan performed the flow system design optimization. This involved conducting a computational fluid dynamics (CFD) flow study to analyze the current system and recommend design changes to the HRSG and AIG. The goal of the flow modeling was to determine inefficiencies with the current system and develop a new design to meet the goals of optimizing the ammonia, velocity, and temperature distribution through the SCR catalyst while minimizing pressure loss.


HRSG Geometry

Figure 2. Geometry of HRSG from turbine outlet to stack showing internal tube banks, catalysts, and AIG.

The geometry of the HRSG is shown in Figure 2. Flow exits that gas turbine and passes through 2 tube banks (red), the duct burner (gray), and 4 more tube banks before encountering the CO catalyst (yellow). The ammonia is injected through the AIG and combined flow passes through the SCR catalyst (blue) and additional tube banks before exiting the stack.


The flow modeling was performed using the Azore® CFD program for Azore Software LLC. This is a 3D polyhedral CFD tool that includes flow and heat transfer simulation. The heat addition from the duct burners and the heat removal from the tube banks was included in the simulation.

Baseline CFD Results

The baseline CFD results at full unit load confirm poor ammonia distribution at the SCR catalyst face. Figures 3 and 4 show the overall flow patterns through the full HRSG. Figure 3 provides flow streamlines colored by the gas velocity. This indicates that flow exits the turbine at very high velocity, on the order of 300 ft/s (90 m/s), and then decelerates as the cross sectional area increases. The flow velocity through the tube banks and catalysts is on the order of 15 ft/s (4.5 m/s) and is well behaved.

Figure 4 shows a side view of the HRSG with velocity contours indicating magnitude and velocity arrows/vectors indicating directionality. Figure 4 has a different color scale than Figure 3 and thus shows the velocities through the tubes and catalysts more clearly.


Baseline CFD Streamlines

Figure 3. Baseline CFD results represented as flow streamlines.

Baseline CFD Gas Velocities

Figure 4. Baseline CFD results, side view, indicate how the gas velocities change through the HRSG.


Figure 5 is a close up view of the region starting at the AIG, and depicts the ammonia concentration in several planes downstream of the AIG. Figure 6 is a plan view of the HRSG from the CO catalyst to the stack inlet. Note the high levels of ammonia concentration near the walls of the unit. A fair amount of ammonia does not get mixed into the flow stream, but continues to be trapped near the walls, as the gases flow toward the stack inlet. The large red areas would correspond to areas of high ammonia slip. The CFD results indicate that non-uniform flow near the AIG is causing the ammonia to be concentrated in recirculation zones, due mainly to an expansion in the cross-sectional area of the HRSG ductwork in the region of the AIG.


Baseline Ammonia Concentration

Figure 5. Baseline ammonia concentration, with areas in red being undesirable.

Baseline Ammonia Concentration

Figure 6. Plan view of baseline ammonia concentration, showing excess ammonia along the walls and entering the stack, i.e. ammonia slip.


The typical AIG consists of an arrangement of pipes that feed vaporized ammonia. A single feed pipe from the vaporizer splits to a number of headers, each of which feeds a number of AIG lances. The lances are arranged over the cross section of the HRSG to provide uniform coverage of injected ammonia. Each lance has a large number of nozzles through which the ammonia is injected into the flow stream. In this case, there were 9 headers feeding 36 lances, on either side of the duct. Each lance had 16 nozzles, for a total of 1,024 ammonia injection points over the HRSG cross section. When designing an AIG, common practice is to carefully consider the diameter sizes for the header and lance piping, along with the nozzle size, as well as the layout and number of lances and nozzles. These design factors are critical to achieve balanced flow from the thousands of nozzles in order to obtain sufficient mixing and a uniform ammonia distribution at the SCR catalyst.

Detailed AIG Modeling

In modeling an HRSG, it is not correct to simply assume that equal flow exits each nozzle of the AIG. Airflow Sciences' technique is to create a second detailed CFD model of the AIG itself to simulate the internal flows and quantify the flow split between nozzles. Figure 7 shows an example of AIG geometry and the velocity of vaporized ammonia within the header and lances. Figure 8 is a closeup of the nozzle locations on the lances. One of the benefits of this detailed model is an ability to analyze the amount of heat transfer from the gas flowing past the lances to the ammonia within the lances. This abrupt increase in temperature results in a density change for the ammonia mixture, affecting the pressure and velocity within the lances. All of these factors were considered for the redesign of the AIG, including a change to the spacing of the lances and nozzles. The general goal of ammonia injection is for the amount from each nozzle to be within 2% of other nozzles to ensure uniform distribution.


AIG Design

Figure 7. Ammonia velocity through a typical header and lances, with the data used as input for the full HRSG CFD model.

AIG Design Closeup

Figure 8. Pressure profile in header and lances, with sufficient pressure being required to eject the ammonia from the furthest nozzle.


CFD Design Optimization

The detailed AIG model data was used for more accurate representation of ammonia injection in the full HRSG ductwork CFD model. The non-uniform flow concerns near the AIG that were discovered in the baseline analysis was corrected with the addition of flow control baffles to redirect flow inward, resulting in an elimination of recirculation zones and a reduction in AIG bypass. In addition, the AIG was redesigned including a local static mixer to improve ammonia distribution at the SCR. The mixer had very low pressure loss (~0.1 inch H2O / 25 Pa). The CFD model was run of the final design; Figures 9 and 10 show the ammonia concentration results for the same views as the baseline Figures 5 and 6. Most notable is the absence of high ammonia concentrations near the walls. A comparison of the plan views shows more uniform distribution of the ammonia across the width of the HRSG. The amount of ammonia slip is significantly lower in this redesign.


Redesigned Ammonia Concentration

Figure 9. Redesigned ammonia concentration, showing elimination of AIG bypass.

Redesigned Ammonia Concentration

Figure 10. Plan view of redesigned ammonia concentration, showing more uniform distribution.


An isometric view of the ammonia concentration for the new design geometry is depicted in Figures 11 and 12 for comparison. In addition to the redesign of the AIG and the baffles at the edges to redirect flow, the position of the AIG was moved upstream. This new placement allowed for more residence time for mixing prior to the SCR catalyst and prevented ammonia from flowing upstream. The baseline ammonia RMS of 69% at the SCR catalyst face was reduced to 7% RMS with the redesign, indicating a significant improvement in uniformity.


Baseline Ammonia Concentration

Figure 11. Baseline ammonia concentration, showing AIG bypass at the injection locations as well as along the edges.

Redesigned Ammonia Concentration

Figure 12. Redesigned ammonia concentration, indicating improved ammonia flow into the gas stream and uniform profile at the SCR catalyst.


Plant Results

The CFD modeling conducted by Airflow Sciences successfully solved the poor ammonia distribution issue in the HRSG. The recommended AIG and flow control devices were installed by Concord Environmental in late 2017. The unit has operated well since coming back online. Plant operating experience has confirmed that since the installation of the redesigned AIG, NOx control has improved and that ammonia usage has decreased 15-25% depending on load and other conditions. In addition, ammonia salt formation had dropped off significantly on the tube banks downstream of the SCR catalyst. This was evident during a recent outage when less debris needed to be cleaned from the tubes compared to previous operating experience. The plant representative stated that the AIG redesign “appears to be a great success”.

Azore CFD Software


Airflow Sciences Equipment