Claus Process Reactor Simulation

Joel Plawsky, Arun Khuttan, Max Bloomfield Rensselaer Polytechnic Institute Troy, NY

2015 CFES Annual Conference 1 Claus Process

8H S + 5O ® SO + 7S + 8H O 2 2 2 2 • The Claus process is the largest volume gas desulfurizing process and is used to recover elemental sulfur from hydrogen sulfide produced during the hydrodesufurization process.

• H2S is burned and then reduced to form elemental sulfur. Often ammonia is 2 present in the feed and needs to be converted to N2. Claus Process

• The project considered only the first part of the process, the furnace. • The furnace environment is very harsh and temperatures are high.

• Acid gases from H2S and often NH3, from hydro-denitrogenation, are produced.

3 Claus Process Reactor

Checkerwall Vectorwall • Claus furnaces contain a checkerwall to protect a downstream heat exchanger from direct contact with the furnace “flame” and to help mixing. • Checkerwalls are primarily an obstruction and their lifetime is limited. • Project was designed to determine the effects of introducing a static mixing element, a Vectorwall™, into the reactor. • Preliminary data suggests the Vectorwall™ provides > 40% improvement in throughput and yield. 4 Claus Process Reactions

Reactions and Rate Laws

3 k 3 k 3 3 1 H S + O ¾¾1 ® SO + H O NH + SO ¾¾5 ® S + H O + N 2 2 2 2 2 3 4 2 8 2 2 2 2 2 1.5 0.25 0.5 r1 = k1PH S PO r5 = k5CNH CSO 2 2 3 2 3 k 3 1 k NH + O ¾¾2 ® H O + N CH + 2O ¾¾5 ®CO + 2H O 3 2 2 2 4 2 2 2 4 2 2 0.2 1.3 0.75 r = k C C 6 6 CH O r2 = k2 PNH PO 4 2 3 2 k 1 k 1 7 f H + O ¾¾3 ® H O H + S ¾¾¾® H S 2 2 2 2 2 ¬¾k ¾¾ 2 2 2 7 r 0.5

r3 = k3CH CO r7 = k7 f PH PS - k7r PH S PS 2 2 2 2 2 2

1 k CO + O ¾¾4 ®CO 2 2 2 0.25 0.5 r4 = k4CO CCOCH O 2 2

• The basic model was chosen to consist of 7 non-elementary reactions with 11 separate species. Reactions are very fast and highly exothermic. Ammonia reaction is slower and new systems are designed around ammonia destruction. • The model was designed to solve the fluid mechanics, heat transfer, reaction kinetics, and mass transfer processes governing the behavior of the reactor. • Thermodynamic properties – NASA polynomial format. • Transport properties – kinetic theory approximations. 5 Claus Process Reactor Transport Modeling

• Fluid Mechanics – Temperatures are high so the system can be assumed to be in the gas phase and behave as an ideal gas mixture. – Flow rates of reactants are high and any obstruction, like a checkerwall leads to highly turbulent flow. Use k –  turbulence model to describe the flow field.

• Heat Transfer – Assume all heat is generated via the chemical reactions. – Reactants enter at relatively low temperatures. – Reactor outer wall is insulated so all heat leaves via convective flow. – Neglect internal radiative exchange at present.

• Mass Transfer – Gaseous components form an ideal mixture and diffuse throughout the reactor. – Process accelerates due to the large temperature increase. 6 Claus Process Reactor Geometry

• Proceeded in stages using simplified 1-D, 2-D, and currently full, 3-D geometries.

• A class project solved the kinetics in ideal continuous stirred tank and plug flow reactors.

• Solved the full, coupled system in dispersed plug flow and 2-D representations. • Reactions are extremely fast and

H2S burns as a flame, so finding solutions is difficult and mesh adaptation is critical. • System on right is > 100,000 elements. 7 Claus Process Reactor – 1-D Simulation

• Modeling of real-world kinetics in a flame-like environment. – Fractional rate orders are not handled well by most software. – The large differences in reaction rates require external control of concentrations to insure bounded values and a “soft” landing. – Solutions need to be approached parametrically. In this case that meant incrementally increasing the heat generation rate.

Gas Velocity Heat Generation Rate

8 Claus Process Reactor

H2S Profiles • Implemented a 2-D formation to provide the first approximation to Vectorwall™ formulations and to compare Vector. • Great differences in rates and distribution of species depending on the insert geometry. All geometries have same open area for flow. • Sulfur conversion is greater in the Vectorwall™ reactor.

Checkerwall Vectorwall™

Hydrogen Sulfide Concentration Profiles 9 Claus Process Reactor Flame Fronts

• Reactions actually take place in a flame. Comsol simulation was able to show the flame front. • Look at different Vectorwall™ configurations. Specifically, whether it is better to have a central opening or central obstruction.

Flame Fronts

(heat generation rate)

10 Claus Process Reactor 3-D Geometry

• 3-D geometry consists of hexagonally arranged openings with hemispherical heads representing the VectorwallTM Elements. • Vector elements in the system below are designed to move material in a clockwise direction. • Each vector element can be individually oriented so intense turbulence can be generated.

11 Claus Process Reactors 3-D Results - Velocities • Compare three configurations; Straight reactor with choke ring, hex wall reactor, and VectorwallTM reactor. Choke ring system is most common. • Ideally looking for uniform, plug flow velocity profile.

Choke Ring Hex Wall Vectorwall™ 12 Claus Process Reactors 3-D Results - Velocities • Streamlines provide a better representation of how the various fuel and oxidant feed streams travel throughout the reactor. • Vectorwalls introduce swirl downstream, but also introduce mixing upstream.

Green – fuel

Blue – air

Choke Ring Hex Wall Vectorwall™ 13 Claus Process Reactors 3-D Results - Velocities • Vectorwall™ efficiency may be related to how close it is located to the reactant inlet. • Locating the wall further upstream provides for more dead space upstream of the wall. • Commercial reactors have a side stream inlet. Mostly inefficient since it does not interact with the main flow. Solution: Feed secondary reactant into Vectorwall™ .

Side Stream Vectorwall™ High Vectorwall™ Low 14 Claus Process Reactor 3-D Results – Mixing & Reaction

• Attempting to assess how to characterize the effect of the Vectorwall™. • One way to do that is to assess the standard deviation as a function of axial position.

2 Deviation z = éc x, y - c ù dx dy ( ) ë ( ) avg û òò

Top of diverter heads

15 Vectorwall™ Coal Combustors

• Coal combustors also often suffer from inadequate mixing

leading to enhanced NOx and CO production. • Incomplete combustion often results from “roping” where the coal particles accumulate as they follow the flow streamlines. • Vectorwall™ design can be used to reduce particle accumulation and enhance combustion via enhanced turbulence.

16 Vectorwall™ Coal Combustors

Reaction Rate Law (mole/s)

3 k 1.5 1 H S + O ¾¾® SO + H O r1 = k1PH S PO 2 2 2 2 2 2 2 k 1 2 HCN + O ¾¾® NO + CO + H r2 = k2CHCN CO 2 2 2 2 1 k 3 H + O ¾¾® H O r3 = k3CH CO 2 2 2 2 2 2 1 k 0.25 0.5 4 CO + O ¾¾®CO r4 = k4CO CCOCH O 2 2 2 2 2 1 k 2 5 r = k 4pr N P C + O2 ¾¾®CO 5 5 ( p p ) O 2 2 k 0.2 1.3 5 CH + 2O ¾¾®CO + 2H O r6 = k6CCH CO 4 2 2 2 4 2 k 2 7 C + CO ¾¾® 2CO r7 = k7 4prp N p PCO 2 ( ) 2 k8 2 C + H O ¾¾®CO + H r8 = k8 4prp N p PH O 2 2 ( ) 2

• Reaction set for coal combustion is very similar to Claus. Conditions are not as severe. Tracking particle size is improtant since combustion is a heterogeneous reaction. 17 Coal Process Reactor Plug Flow Reactor

• We use a similar process for solving these reactions, beginning with an idealized reactor to insure the set of reactions run and that we can integrate heat generation. • Coal process leads to a more sedate temperature rise relative to the Claus process. • Reactions are slower since we are burning a solid particle rather than a gas. • Inlet temperature into the combustor was set at 1350 K.

18 Coal Process Reactor Plug Flow Reactor

• Heterogeneous burning of the coal particles is the most important process. Modeled as a shrinking core though we don’t actually chart ash creation. • Set inlet particle size to 75 microns which is broadly representative of commercial systems. • We can understand how fast the particles burn and how large a combustor is

required.

Particle Size (m) Particle Size

19 Conclusions and Future Work

• We successfully simulated the Claus process in ideal chemical reactors, in a dispersed, plug flow reactor, in a two-dimensional flow reactor with pseudo-Vector elements, and are finalizing three-dimensional reactor simulations with checkerwall and VectorWallTM inserts.

• The next steps will be: – Complete the 3-D simulations on simple systems to incorporate the full heat generation rate associated with the reactions. • May be able to understand what we need via parametric sweeps in isothermal reactors. – Define the optimal arrangement of Vectorwall™ elements and the optimum location of the Vectorwall™. – Port the results over for industry use.

• Funding for this project was provided by: New York State Pollution Prevention Institute

20