Chemical Looping Combustion


Chemical Looping Process


Chemical looping is a process to indirectly oxidize fuels with air, converting the chemical energy in fuels to thermal energy.   In contrast to direct oxidation with air,  carbon dioxide and nitrogen are in different exhaust streams.  This facilitates the sequestion of CO2 without a separate system for gas separation before (e.g. oxyfuel combustion) or after the combustor (e.g. link<post-combustion CO2 capture>).  The system only requires a condenser to remove the water.

On the left, is a diagram of CL system based on two fluidized bed reactors.  Solid particles are circulated around the loop which oxidizes in the air reactor and reduces (provides oxygen to oxidize the fuel) in the fuel reactor.   Typical carrier materials include metals such as copper, iron or nickel.   The reaction in the air reactor is typically exothermic, while the reaction in the fuel reactor can be slightly endothermic or exothermic depending on the fuel gas composition and the carrier used.  For endothermic systems, many proposed designs use the thermal energy from the circulating the solids to provide the necessary heat to maintain the temperature in the fuel reactor.

Experimental Systems

From the perspective of and modeling simulation, NETL has three systems onsite for validation of chemical looping models.  There are several smaller systems  (TGA and fixed bed) which are used for calibration of reaction models.




The leftmost figure shows the “single fluid bed” reactor.  The reactor has 2.5 inch inner diameter and is insulated and heated to maintain a specified operation temperature.   This apparatus is used to investigate interactions between flow dynamics and reactivity in a smaller more controlled scale than the CLR.

The center figure shows the chemical looping reactor (CLR).  The system is several meters high and has design capacity of 50kW thermal.  The air reactor is 6 inches in diameter and the fuel reactor is 8 inches in diameter.

The rightmost figure shows the cold flow (CLR).  It has approximately the same geometry of the CLR.  This system is used to provide guidance to the operation of full system as well as for general exploration of the dynamics of circulating chemical looping systems.  The system can run in both batch a circulation modes.


“Cold Flow” Full Loop Simulation




Above are three animations from an Euler-Euler simulation of the cold flow experiment. The two animations on the left show the solid volume fraction, the one on right shows the gas pressure.  The solid circulation is counter clockwise.

As seen in the animation, the fuel reactor is operating in the bubbling regime, while the lower part of the air reactor is in the fast fluidized regime.  Additional injection ports at the side the air reactor, provide the additional gas flow to elutriate the solids from the air reactor transport the carrier through the riser and cross-over to the cyclone.  Most of the solids drain into the loop seal which is also operating in the bubbling regime and then into the fuel reactor.  The solids leave the fuel reactor through the L-valve.  The flow rate in the L-valve using the additional regulated by changing the flow rate to the three inlet shown in pink.

These types of simulations are computationally demanding, requiring several months of computational time on 100’s of CPU-cores.  The main driver is the relatively long physical time which must be simulated for the solid distribution and pressure to reach an equilibrium is relation to the fast local physical process which must be resolved for accuracy and stability.

REF: ICMI Annual Report Rev. 0, November 25, 2014, URS-RES-1-1098


SFB Flow Dynamics – Distributor Geometry


Euler-Euler simulations were performed of the single fluid bed reactor (SFB) using four different gas injector geometries.  The objective was to  investigate the changes in the dynamics of the reactor.

The SFB injector (upper figure) consists of four bubble caps (6.35 mm diameter, 3.5 cm long) each with 3 1.1 mm holes placed slightly below the head nut.   The bubble caps are inserted into the distributor plate so that bottom of the head nut is 9.7mm from the bottom of the reactor

Three different configurations of the distributor plate are investigated (middle figure): uniform gas injection, four bubble caps and six bubble caps.  To greatly reduce the mesh size (and computational expense),  two different approximations are used for the small gas injection holes.  For the “ring” injection gas is injected uniformly  along the cells highlighted in red on the left figure.  For the “discrete” injection gas injected uniformly through the cells at three locations  highlighted in red on the right figure.


SFB Flow Dynamics – Visualizations










The animation above shows the predicted “bubble pattern”.  The iso-surfaces are generated from the solid volume fraction ( α) and highlight the boundary between the dilute (α <0.2 ) and dense (α >0.2 ) parts of the system.  The dilute portion is on the “inside”.  The gas velocity vectors are also shown colored by the magnitude.

The bubble pattern for the uniform gas injection (porous plate) shows large bubbles rising periodically following the centerline.  Multiple small bubbles in bed lower region merge (at mid-height) and form single “large” bubble (centerline rise).  An interesting item for further investigation is why the bubbles  seems to preferentially form at only one or two locations.

In contrast the bubble patterns for the bubble cap models in characterized by the successive merger of the bubbles create by bubble caps.  The formation location of the single large bubble is a higher height than the  uniform injector.  No significant differences were seen between the simulations between the “ring” and “discrete” injection.



SFB Flow Dynamics – Bubble Analysis



The results above summarize further analysis of the bubble contours for each of the four simulations. This analysis highlights further differences in the dynamics between due to the geometry changes.

The differences is bubble size between uniform and distributed injection is quite evident in the figures, particularly for the bubble diameter.   The upper right figure shows that the predicted bubble diameter profile fairly is consistent with the standard experimental correlation (Darton).

The differences in the bubble rise velocity profiles are less dramatic.  This is may be due to wall effects which alter the bubble growth and rise velocity.



Konan et. al. 2015,  Numerical study of gas-solid fluidized bed dynamics with distributor design,   2015 Pittsburgh Coal Conference.

Weber et. al. 2013, Fluid bed characterization using Electrical Capacitance Volume Tomography (ECVT), compared to CPFD Software’s Barracuda, Powder Technology, vol. 250, pp. 138-146.