The use of mixture model theory in CFD for the chemical reaction between CO2 and soda lime in closed circuit rebreather scrubbers
© Cunningham et al.; licensee Springer. 2013
Received: 19 April 2013
Accepted: 18 September 2013
Published: 30 October 2013
A mixture model simulation is presented by modeling the axial scrubber in a Closed Circuit Rebreather (CCR). The mixture model is a good substitute for the full Eulerian multiphase model because the interphase laws are unknown in this case. Analysis of mesh size, mesh type and inflation are made to independently characterize their accuracy by means of convergence before further comparisons with experimental data. The importance of mesh refinement is demonstrated near the wall with satisfactory results seen on the near grid wall of the boundary where a finer mesh is utilized. The contribution of inflation and grid independence to the accuracy of the model is presented in the results section.
KeywordsSCUBA Closed Circuit Rebreathers Modelling CO2 Mixture model theory
Prior to Clarke (Clarke 2001), the kinetics of CO2 absorption in scrubbers was poorly understood. A stochastic method simulating a bed containing a minimum of 200,000 volume elements or cells was employed and within each cell, the temperature and quantity of CO2 stored for each time increment was defined. The model was constrained by physics, the chemical absorption within each cell with its resulting temperature are probabilistic as opposed to the chemical properties of the absorbent or CO2 which means mass and heat transfer are also determined stochastically. The outputs of the analysis were comprised of a model simulating CO2 absorption and thermal fluctuations however the model is only applied to axial design based scrubbers.
Dongsik, Fumin and West (Dongsik et al. 2011) analyzed imperfect CO2 removal mechanisms of CO2 scrubbers. Their work introduced a stochastic model for three CO2 related rebreather faults: (i) CO2 bypass, (ii) scrubber exhaustion and (iii) scrubber breakthrough. The authors proposed a stochastic process driven by a Poisson counter to characterize the concept of CO2 channeling and the three stated rebreather faults. In probability theory, a Poisson process is a stochastic process which counts the number of events and the time that these events occur in a given time interval. The aforementioned work also advances the understanding of breathing dynamics associated with CCRs and how to maximize performance in terms of breathing and peak to peak pressure. This model is constrained as it does not model the chemical reactions occurring within the scrubber.
where ∆V is the change in volume during a time increment ∂ and ∆ρ is the change in density during time increment ∂.
The average of (T1, T5, T9), (T2, T6, T10), (T3, T7, T11) and (T4, T8, T12) are taken to obtain a temperature for that layer of soda lime granules for any given time.
Proposed modeling approach
Taking the density of CO2 gas as 1.87 kg/m3, the Re number can be shown to be 0.189 for a particle diameter of 0.0011 m, velocity of 0.0126 m/s and a dynamic viscosity of 0.0001372 kg/ms. The laminar conditions apply up to Re=10 (Rhodes 1989). Within the laminar model, mixture model theory and Henry’s law were applied in the simulation.
Mixture model theory
The mixture model is a type of multiphase system fined as a mixture of the phases of solid, liquid and gas. Multi-phase flow phenomena are typically dominated by one phase and another non-dominating phase e.g. dust in air (Manninen et al. 1996. However in the case presented in this paper the secondary phase or non-dominating phase cannot be neglected due to the influence on the fluid dynamic behavior of the mixture. The model contains an air|liquid pairing where the air is the inlet gas and the liquid is a reacting component. The decision of modeling threaction as a liquid-particle mixture is based on literature which states absorption as “the removal of one or more selected components from a mixture of gases by absorption into a suitable liquid is the second major operation of chemical engineering that is based on interface mass transfer controlled largely by rates of diffusion”(Sinnott 1996). Gas absorption occurs when a mixture of gas comes into contact with a liquid for the purpose of dissolving one or more components of the gas mixture in the liquid. Thus the absorption of CO2 occurs with the NaOH component of soda lime in the liquid phase (Physical and Engineering Data 1978).
where α K the volume fraction of the phase K, ρ K is the average material density, u K is the local instant velocity of phase K and ρ m is the local density of the mixture (Ishii 1975).
where p is the partial pressure of the solute in the gas above the solution, c is the concentration of the solute and kH is a constant.
The three step reaction
Water is required to initiate the CO2 absorption (Eq. 8). However water is a by-product of the chemical reaction that takes place within the canister (Eq. 9). If the incoming gas stream is saturated with water vapour, an excess of water vapour will remain in the canister. This excess water coats the soda lime granules and cause blockages in the pores. The CO2 does not absorb as efficiently and this may also cause caustic vapour in the loop which could burn the diver’s throat. Conversely if the incoming gas stream is too dry, the commencement of the reaction may be limited or the absorbent bed may be too dried out, thereby preventing absorption. Moisture levels of the incoming gas stream should be maintained above 70% RH when using soda lime.
Boundary conditions and assumptions
The simulation was performed using the CFD software program Ansys 13.0 CFX and the following assumptions are made for the model; (i) the CO2 is absorbed fully without loss until breakthrough, where breakthrough is defined as the time until the canister effluent/CO2 passes through the soda lime granules unscrubbed, (ii) the CO2 gas is uniformly distributed throughout and (iii) a constant CO2 injection rate is employed. The axial scrubber is analyzed using CFX and is comprised of a packed bed of soda lime granules modeled as a porous media in which exhaled breath (5% CO2, 16% O2, 78% N2) and traces of water vapour (William et al. 2009) is passed through the scrubber at an inlet velocity of 0.0126m/s. When the composition of exhaled breath passes through this porous media, a series of chemical reactions take place absorbing the CO2 and producing water vapour and heat. The inlet gas has a static temperature of 25°C initially at atmospheric conditions.
The continuous governing equations are converted into algebraic equations by using the finite volume method. These subdomains or boxes allow the analysis of flow in each box individually and then these fluid portions can be collated to yield a complete picture of fluid flow in the entire domain of the scrubber.
Modelling the geometry
Table of initial mesh details
Duration of computation
10 h 52 mins
17 h 54 mins
4 h 18 mins
In order to initiate this comparison the different types of mesh, the number of elements and the duration of computation are given in Table 1. The results varied in computational duration with the tet mesh the most computationally expensive.
Table of modified mesh details
Duration of computation
19 h 2 3mins
17 h 54 mins
16 h 48 mins
The computational run time using this hex mesh was recorded at 19 h 23 mins. This mesh is the least accurate and the most time consuming. The tet and swept mesh both exhibit convergence and accurate similar values in comparison with the experimental data. The swept mesh is computationally less time expensive and structured so it was deemed the best system.
There is also a strong interaction between modeling errors and the time and space resolution of the grid. The quality of the mesh can be determined by many different factors (i) mesh type, (ii) convergence criteria, (iii) inflation, (iv) aspect ratio and (v) skewness of the mesh.
Details of modified inflation layers with a constant quad grid
Figure 14 illustrates the need for inflation in laminar flow against the walls of the axial scrubber segment as the temperature values particularly in 'No Inflation 2’ are not only too high but there is a greater time lag observed. Due to the slow nature of heat transfer, there is no difference between 3, 5, 8 or 10 inflation layers as the results collapse onto each other in both graphs. It is shown there is a need for inflation against the walls when analyzing heat transfer. Due to the difference in convergence, 10 inflation layers are used for the grid independence test even though it is 0.978 more than a good aspect ratio from Table 3.
Grid independence test
Details of modified edge-sizing grid with constant inflation layers
Grid convergence study using Richardson extrapolation
Details of the grids used for GCI
Order of accuracy and grid convergence index
Temperature (°C) MAE
The paper presented a CFD model of a CCR axial scrubber using the mixture model theory to analyze the chemical reaction between soda lime and CO2 as an alternative technique to current methods. The mesh density influences the accuracy of the results and thus a benchmark with experimental data of the final mesh was conducted. The first parameter influencing the mesh is the type of mesh chosen for the model. A structured quad mesh was identified as being the optimum as it showed satisfactory convergence and comparison with the experimental data. It was also less computationally expensive to run. The analysis of inflation was conducted on the boundary walls of the model. The level of inflation was varied for the different models and compared against experimental data where it is shown that inflation on the walls will contribute to the accuracy of the model. A grid independence test was carried out to analyze how the fineness or coarseness of the mesh grid influenced the results. The finer the mesh, the more accurate the solution becomes however at significant computational cost. The point at which similar results are seen between two meshes acts as validation in the choice of selecting a 600 grid with 10 inflation layers as the optimum grid. This mesh density is a close match in temperature and produces better results over the time on the x-axis. The aspect ratio and skewness of the cell are also used as a means of validating the mesh independent of experimental results. For the final mesh both the aspect ratio and skewness are acceptable values at 9.24 and 0.6. The relationship between the actual mesh and the experimental data show that the predicted results lag behind the actual experimental profile. This may be attributed to a lag needed in the simulation where the model needs to reach a steady state phase before the temperatures correlate to the experimental data as is seen with the second set of monitor points. The overall trend of the model prediction agrees well with the experimental data. The mesh density; including type, grid size and inflation coupled with the aspect ratio and skewness provide a method of characterizing the mesh. The GCI value of 5% is also an acceptable result. The method presented allows an independent validation of the mesh quality which is further validated with the experimental results.
- ANSYS CFX: ANSYS CFX. Canonsburg, PA: Academic Research Release 11.0. ANSYS, Inc; 2006.Google Scholar
- Baker A: Lecture 7-Meshing applied computational fluid dynamics (Fluent). 2002.Google Scholar
- Bowen RM: Theory of Mixtures, Part I. In Continuum physics vol. 3. Edited by: Eringen AC. New York: Academic Press; 1976.Google Scholar
- Clarke JR: Computer modelling of the kinetics of CO2 absorption in rebreather scrubber canisters. Proc OCEANS 2001, 3: 1738-1744.Google Scholar
- Dongsik C, Fumin Z, West M: Diagnosis and prognosis of scrubber faults for underwater rebreathers based on stochastic event models. Prognostics and Health Management (PHM), 2011 IEEE Conf 2011, 1-8. doi: 10.1109/ICPHM.2011.6024353Google Scholar
- Farajzadeh R, et al.: Mass transfer of CO2 into water and surfactant solutions. Pet Sci Technol 2007, 25(12):1493-1511. 10.1080/10916460701429498View ArticleGoogle Scholar
- Farajzadeh R, Zitha PL, Bruining J: Enhanced mass transfer of CO2 into water: experiment and modeling. Ind Eng Chem Res 2009, 48(13):6423-6431. 10.1021/ie801521uView ArticleGoogle Scholar
- Ishii M: Thermo-fluid dynamic theory of two-phase flow. NASA STI/Recon Technical Report A 1975, 752: 29657.Google Scholar
- Johnson G, Massoudi M, Rajagopal K: Flow of a fluid—solid mixture between flat plates. Chem Eng Sci 1991, 46(7):1713-1723. 10.1016/0009-2509(91)87018-8View ArticleGoogle Scholar
- Joseph DD, et al.: Ensemble averaged and mixture theory equations for incompressible fluid—particle suspensions. Int J Multiphas Flow 1990, 16(1):35-42. 10.1016/0301-9322(90)90035-HView ArticleGoogle Scholar
- Katch FI, Katch VL, McArdle WD: Exercise Physiology: Energy, Nutrition and Human Performance. Philadelphia: Lea and Febiger; 1996.Google Scholar
- Klos R: Principles of work of different types of underwater breathing apparatus. Pol Mar Res 2008, 15(4):72-84.Google Scholar
- Manninen M, Taivassalo V, Kallio S: On the Mixture Model for Multiphase Flow. Finland: VTT Publications. Technical Research Center of Finland; 1996:288.67.Google Scholar
- Nuckols ML, Purer A, Deason GA: Design guidelines for carbon-dioxide scrubbers. Revision A. Technical manual. Panama City, FL (USA): No. AD-A-160181/4/XAB. Naval Coastal Systems Center; 1985.Google Scholar
- Oberkampf WL, Blottner FG: Issues in computational fluid dynamics code verification and validation. AIAA J 1998, 36(5):687-695. 10.2514/2.456View ArticleGoogle Scholar
- Olutoye M, Eterigho E: Modeling of a gas absorption column for CO2-NaOH system under unsteady-state regime. Leonardo El J Pract Technol 2005, 4(7):49-54.Google Scholar
- Physical and Engineering Data: Shell Internationale Petroleum Maatschappij. The Hague; 1978.Google Scholar
- Poling BE, Prausnitz JM, John Paul O'C, Reid RC: The properties of gases and liquids. Volume 5. New York: McGraw-Hill; 2001.Google Scholar
- Pordal HS: Practicing the science of computational fluid dynamics. Stress Engineering Services, Inc; 2006:16. http://www.stress.com Google Scholar
- Pordal HS: Practicing the science of computational fluid dynamics. Stress Engineering Services, Inc; 2006:10. http://www.stress.com Google Scholar
- Rhodes M (Ed): Introduction to particle technology. The Atrium, Southern Gate, Chichester, West Sussex PO19 8SQ, England: John Wiley & Sons Ltd; 2008.Google Scholar
- Roache P: Perspective: a method for uniform reporting of grid refinement studies. Trans ASME, J Fluids Eng 1994, 116: 405-405.View ArticleGoogle Scholar
- Sexton PG, Nuckols ML: Computer simulation of breathing systems for divers. J Eng Ind – T ASME 1983, 105: 54-59. 10.1115/1.3185864View ArticleGoogle Scholar
- Sinnott RK: Chemical Engineering, Volume 6: An Introduction to Chemical Engineering BT-M.TECH-E&T-SRM-2013-14 Design. 4th edition. Edited by: Coulson JM, Richardson JF. UK: Butterworth-Heinemann Ltd; 1996:530-550.Google Scholar
- Stern F, et al.: Comprehensive approach to verification and validation of CFD simulations: 1: methodology and procedures. J Fluid Eng 2001, 123(4):793-802. 10.1115/1.1412235View ArticleGoogle Scholar
- W.R. Grace and Co.: The SODASORB manual of carbon dioxide absorption. Fifth printing. Lexington, MA: W.R. Grace & Co., Dewey and Almy Chemical Division. Section P-1, Chemical and Physical Processes in Carbon Dioxide Absorption; 1986.Google Scholar
- Wang TC: Temperature effects on baralyme, sodasorb, and lithium hydroxide. Ind Eng Chem Process Des Dev 1975, 14(2):191-193. 10.1021/i260054a017View ArticleGoogle Scholar
- Wilcox DC: Turbulence modeling for CFD. Volume 2. California: La Canada: DCW industries; 1998:103-217. 91011Google Scholar
This article is published under license to BioMed Central Ltd. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.