W. Russell, Blasch Precision Ceramics, Albany, New York
A computational fluid dynamics (CFD) study was performed on a theoretical reforming furnace to determine the comparative results of flow uniformity in flue gas tunnels with and without a standing tunnel in the outer position. The furnace modeled was 42.5 ft in height, 52.5 ft in width and 52.5 ft in length. Along the furnace floor ran eight flue gas tunnels, 52.5 ft long and 72 in wide with a 26-in internal tunnel width. Ninety-six flue gas burners were modeled to provide heat to 336 catalyst tubes (FIG. 1).
Two furnace models were compared, one running a full set of eight tunnels and another without the outer tunnel but with the other seven tunnels (FIG. 2). This is intended to replicate a situation where a tunnel has collapsed during a campaign. During a maintenance outage, the debris was removed, but a new tunnel was not constructed, likely with the plan to rebuild the tunnel during a future turnaround with more time available for construction.
Mesh and CFD inputs. The furnace’s size in the model was large enough to dictate a course mesh size, and the detail required for desired outputs was small enough to dictate a fine mesh size, so a variable mesh approach was used. A symmetric plane condition was used in the middle of the furnace, between tunnels four and five, to increase the refinement level for this study. Because the main focus of the study was the area around tunnel one (the outermost tunnel), a finer mesh was applied to the area.
Flue gas entered the model through the burner openings on the furnace ceiling and traveled down toward the floor, entering the tunnels if present and exiting the model out the openings at the end of the tunnels (FIG. 3).
For this analysis, the reaction occurring inside the catalyst tubes was not modeled, but the tubes were assumed to act as a heat sink with convection governed by standard heat transfer equations. It is important to note that this assumption will limit the result of this model’s accuracy regarding absolute temperatures, particularly in areas directly adjacent to the catalyst tubes. Results from this model pertaining to absolute temperatures should be considered comparative only.
Results: Reforming furnace with all tunnels. Flow trajectories from the CFD results showed that a reforming furnace operating with all tunnels will have a flow field covering the entire furnace, as designed, providing an accurate control case (FIG. 4).
This theoretical reforming furnace showed some recirculation areas and did not have a perfect flow field, suggesting that it could benefit from burner and tunnel port balancing to optimize the overall furnace profile.
The residence time of the flue gas coming through the burners over the top of the outer tunnel can be measured in this model by running a secondary particle study. This study simulates a single particle inserted into the furnace through a burner inlet and measures the time it takes for that particle to travel through the furnace to the exit (FIG. 5). This particle insertion is replicated hundreds of times across the 12 burners over the top of the outer tunnel. Then, it is charted in a histogram to show the overall distribution. Uniform residence time is an essential aspect of furnace efficiency since the total heat transfer depends on time (FIG. 6).
A temperature profile can be used to view the uniformity within the reforming furnace. A vertical profile in the middle of the outer tunnel shows the flow field of the burners. The temperature in each tunnel can be represented graphically by taking data points along the length of the reformer, 18 in. above the furnace floor. The tunnels can each be graphed with an individual line plot, as shown in FIG. 7. As previously mentioned, this analysis was run assuming the catalyst tubes will act as a simple heat sink instead of a complex reaction location, so the absolute temperatures shown are much more stable than field results typically indicate.
Results: Reforming furnace without an outer tunnel. The second furnace can be modeled without the outer tunnel and analyzed for the same furnace conditions. Flow trajectories show that a reforming furnace without an outer tunnel will have non-uniform flow conditions.
These non-uniform conditions result in a wider range on the residence time graph (FIG. 8). The blue bars represent the controlled furnace with all tunnels, and the red bars represent the furnace without an outer tunnel. The absence of the outer tunnel enables a portion of the flue gas to exit the furnace rapidly and creates recirculation of another portion of the flue gas, leading to a wide range of residence times (FIG. 9). It can also be noted that the flue gas exiting the outer furnace opening is not limited to the burners in the outer row, but also from adjacent burners within the furnace.
A vertical temperature profile in the middle of the outer tunnel shows the burner’s flow field. The furnace without an outer tunnel shows a clear flow bias from the first two burner banks toward the exit, recirculating the remaining flow from the third burner bank (FIG. 10).
Temperature data points taken along the length of the middle of each tunnel are plotted here (FIG. 11). Each line represents one tunnel, with the outlier red line coming from the section of the furnace with a missing outer tunnel. The average temperature inside the furnace with all tunnels intact was 1,753°F (FIG. 12). The average temperature inside the furnace without an outer tunnel was 1,681°F, and the average temperature inside the area where an outer tunnel would be was 1,537°F.
The gas composition output of a steam methane reformer (SMR) is directly related to the temperature. The expected composition can be calculated for a tunnel with all tunnels (1,753°F) and a tunnel without an outer tunnel (1,681°F and 1,537°F). Several assumptions must be made for this calculation. First, assume that the flue gas temperature at the points measured is the exact temperature of the reforming gas composition inside the tubes. This is, of course, not the case, but since both furnace outputs are being treated with the same assumption, it should provide comparative information. Second, the temperature data points taken for this CFD study were near the furnace floor where the tunnels are present, and it was assumed that the temperature near the roof of the furnace was identical in both models, so these areas are averaged together when looking at the gas composition output. This is an area where more analysis could be utilized to get a more accurate result.
Considering the above-mentioned assumptions, this CFD model suggests that the percentage of hydrogen (H2) in the furnace output’s gas composition decreases if all the tunnels are not in place, resulting in a decreased furnace temperature (FIG. 13). The percentage of H2 could decrease by 9.7% in the furnace area adjacent to the missing outer tunnel, and the percentage of H2 could decrease by 0.95% across the entire furnace. A chemical plant producing 2,000 metric tpd of ammonia could theoretically decrease production by 19 metric t if the primary reforming output is a limiting factor. If the reforming furnace is run for a year before the tunnel is replaced, the resulting production loss would be 6,916 metric t at a bulk sale price of $300/metric t. The theoretical annual impact of running the furnace without an outer tunnel is $2,074,800.
Takeaway. Flue gas tunnels play an important role in the overall distribution of flow in a reforming furnace. Running a furnace without an outer tunnel results in a wider distribution of residence time because flue gas from the front burner banks can exit the furnace very quickly, and other flue gas from the back burner banks creates recirculation zones. This reduces the uniformity of the temperature distribution across the furnace, directly influencing the gas composition output. The theoretical annual economic impact of running an SMR without an outer tunnel is more than $2 MM. HP